[From the U.S. Government Printing Office, www.gpo.gov]
RESOURCE DAMAGE ASSESSMENT OF THE T/V PUERTO RICAN OIL SPILL INCIDENT APPENDICES Property of CSC Libraz? U.S. DEPARTMENT OF COMMERCE NOAA COASTAL SERVICES CENTER 2234 SOUTH HOBSON AVENUE CHARLESTON , SC 29405-2413 April 1986 United States Department of Commerce National Oceanic and Atmospheric Administration National Ocean Service Office of Ocean and Coastal Resource Management Sanctuary Programs Division Washington, D.C. 20235 TABLE OF CONTENTS Page APPENDICES 1. List of Species Found in the Gulf of the Farallones Region ................................. 1-1 2. Annotated Species List of Marine Birds Affected by the T/V Puerto Rican .......... ... 2-1 3. Species Selection of T/V Puerto Rican Oil Spill Risk Analysis Canputer Modeling..................... 3-1 4. Oil Spill Model Description ................................ 4-1 5. Bird Mortality Resulting from Oil Spilled by the T/V Puerto Rican .............. .......... . ... 5-1 6. Damage to the Plankton of the Gulf of the Farallones Caused by T/V Puerto Rican Oil Spills..................... 6-1 7. Background Documents - Biological Hindcast of the Effects of the T/V Puerto Rican Oil on Three Seabird Species ...................... 7-1 - Special Oil Spill Risk Analysis of Tank Vessel Puerto Rican Incident ............................ 7-29 - Trajectory Analysis and Modeling Support for the Puerto Rican Oil Spill...... 7-41 The Impacts of the T/V Puerto Rican Oil Spill on Marine Bird and Mammal Populations in the Gulf of the Farallones, 6-19 November 1984 ........... 7-83 - Aerial Surveys of Seabirds and Marine Mammals in the Gulf of the Farallones Region, 11-20 November 1984 ...... 7-167 - Aerial Surveys of Seabirds and Marine Mammals in the Gulf of the Farallones Region, 30 November 1984 ......... 7-188 - Aerial Surveys of Seabirds and Marine Mammals in the Gulf of the Farallones Region, 12 December 1984 ......... 7-191 Page - The Effects of Oil from the T/V Puerto Rican on Recruitment and Recovery of Barnacle Populations ........... 7-199 - An Analysis of the Inmediate Effect of the Puerto Rican Oil Spill on the Intertidal Marine Life of Marin County, California ................... ................... 7-235 - Toxic Effects of Four Refined Oils Spilled into the Gulf of the Farallones on Natural Communities of Plankton Under Simulated In Situ Conditions ......................... 7-257 - Necropsy Protocol for Marine Animal Mortality Along the Point Reyes National Seashore. November 3, 1984 (Puerto Rican Oil Tanker Sinking) to March 7, 1985 ......... 7-349 - Documentation of Administrative Research Costs............. 7-359 8. Curriculum Vitae of Participants., ............................ 8-1 I I I__ _ _ APPENDIX 1 I 1 ~~~~~~~~~SPECIES LIST I I I I I I I I I I I I I I Appendix 1 Marine bird species found in the Point Reyes-Farallon Islands National Marine Sanctuary (source: Ainley, 1976; Winzler and Kelly, 1977; Uduardy, 1977.) COMMON NAME SCIENTIFIC NAME Fork-tailed storm petrel Oceanodroma furcata Leach's storm-petrel Oceanodroma leucorhoa Ashy storm-petrel Oceanodroma homochroa Double-crested cormorant Phalacrocorax auritus Brandt's cormorant Phalacrocorax penicillatus Pelagic cormorant Phalacrocbrax pelagicus Western Gull Tarus occidentalis Ccmmon murre Uria aalge Pigeon Guillemot Cepphus columba Marbled murrelet Brachyramphus marmoratus Cassin's auklet Ptycnoramphus aleuticus Tufted puffin Fratercula cirrhata Brown pelican Pelicanus occidentalis Arctic loon Gavia arctica Red-throated loon Gavia stellata Western grebe Aechmophorus occidentalis Red-necked grebe Podiceps grisegena Eared grebe Podiceps nigricollis Horned grebe Podiceps auritus Northern fulmar Fulmaras glacialis Black scoter Melanitta nigra Surf scoter Melanitta perspicillata White-winged scoter Melanitta fusca California gull Larus californicus Herring gull Larus argentatus Ancient murrelet Synthliboranthus antiguus Red phalarope Phalaropus fulicarius Skua Catharcta maccormicki Bonaparte's gull Larus philadephia Sabine's gull Xema sabini Common tern Sterna hirundo Buller's shearwater Puffinus bulleri Black-footed albatross Diomedea nigripes Black legged kittiwake Rissa tridactyla Pink-footed shearwater Puffinus creatopus Sooty shearwater Puffinus griseus Elegant tern Sterna elegans Common loon Gavia immer Rhinoceros auklet Cerorhinca monocerata Spotted sandpiper Actitis macularia Least sandpiper Erolia minutilla Western sandpiper Caladris mauri Willet Catoptrophorus semipalmatus Long-billed dowitcher Timnodromus scolopaceus Great blue heron Ardea herodias 1-2 (Cont'd) COMMON NAME SCIENTIFIC NAME Snowy Plover Charadrius alexandrinus Dunlin Erolia alpina Killdeer Charadrius vociferus American Coot Fulica americana Black brant Branta bernicla nigricans Diving ducks. Anas Black-bellied plover Squatarola Squstorola Semipalmated plover Charadrius semipalmatus Sanderling Crocethia alba Whimbrel Numenius Phaeopus Heermann' s gull Larus heermanni Forster's tern Sterna forsteri American white pelican Pelecanus erythrorynchos Pomarine jaeger Stercorarius pomarinus Glaucous-winged gull Larus glaucescens Northern pintail Anas acuta Ruddy duck Oxyura jamaicensis Greater scaup Aythya marila Red-tailed hawk Buteo jamaicensis 1-3 Appendix 1. Marine mammal species found in the Point Reyes/Farallon Islands National Marine Sanctuary (source: California Department of Fish and Game, 1979. Living Marine resources of the proposed Point Reyes-Farallon Islands Marine Sanctuary). IOMMON NAME SCIENTIFIC NAME PINNIPEDS: California sea lion Zalophus californianus Steller sea lion Eumetopias jubatus Harbor seals Phoca vitulina Northern elephant seal Mirounga angustirostris Northern fur seal Callorphinus ursinus Fissiped: California sea otter Enhydra lutris nereis Cetaceans: Blue whale Balaenoptera musculus Sei whale Balaenoptera borealis California gray whale Eschrichtius robustus Finback whale Falaenoptera physalus Humpback whale Megaptera novaeangliae Pacific pilot whale Globicephala machorhynchus Killer whale Orcinus orca False killer whale Pseudorca crassidens Sperm whale Physeter catodon Baird's beaked whale Berardius bairdi Curier's beaked whale Ziphius cavirostris Ccunon dolphin Dellhinus' delphis Risso's dolphin Grampus griseus Pacific white sided dolphin Lagenorhynchus Obliquidens Northern right whale dolphin Lissodeiphis borealis Harbor porpoise Phocoena phocoena Dall porpoise Phocoenoides dalli 1-43 Appendix 1 * Invertebrate species found in the Point Reyes- Farallon Islands National Marine Sanctuary (source: Jones and Stokes, 1981.)3 COMMON NAME SCIENTIFIC NAME Echinodermos Giant red sea urchin Strongylocent-rotus franc iscanus Bat star Patiria miniata Ochre star Pisaster ochraceus Large- spined sea star P . Tgigateus Short-spined sea star P. brevispinus Marine Molluscs Market squid Lolig opalescens Abalone spp. Haliotis spp. Limpets Acmaea spp. Periwinkles Littorina spp. Bay Mussel Mytilus edulis California mussel M.clfrianu Falcate date mussel Adula falcata Pacific razor clam Siliqua patula Conrad's piddock and flat-tipped piddock Pinitella conradi and P. penita California brackish water snail Tryonia imitator Soft-shell clanmy arenaria3 Gaper clam Tresus nuttallii Pismo clan Tivela stultorum Carnwin Washington clam Saxidcanus nuttallii Purple-hinged rock scallop Hinnites giganteus Shipworm Teredoc navalis Marine CrustaceansI Copepod Acartia tonsa Copepods Calanus spp. Acorn barnacle Balanus spp. opossum shrimp Neomnysis mercedis Gr ibble Limnoria spp.3 Long-horned beach hopper Orchestoidea californiana Euphaus iids Thysanoessa spinifera and Euphausia pacifica Bay shrimp Crangon franc iscorum Peneid shrimp Sergestes similus Spot Prawn and pink shrimp Pandalus platvceros and P. jordani Spot Prawn P. platyceros Lined shore crab Pach-ygrapsus crass ipes Dungeness crab Cancer mnagister Rock crab C. antennarius3 MoDle crab Emer4i`ta a~naioga Appendix 1. (Cont'd) Invertebrate species found in the Point Reyes- Farallon Islands National Marine Sanctuary (source: Jones and Stokes, 1981.) COMMON NAME SCIENTIFIC NAME Marine Worms Fan Worm Eudistylia spp. Ornate tube worm Diopatro ornata Mussel worm Nereis vexillosa Bloodworm Euzonus spp. Sand tube worm Pharagmatopoma californica Coelenterates Giant green anemone Anthopleura xanthogrammica Strawberry anemone Corynactis californica Orange cup coral Balanophyllia elegans Brown cup coral Paracyathus stearnsii Appendix 1. Species of Fish Found in the Point Reyes-Farallon Islands National Marine Sanctuary (source: Jones and Stokes, Associates, 1981). COMMON NAME SCIENTIFIC NAME Pacific lamprey Entosphenus tridentatusI Leopard shark Triakis semifasciata Green sturgeon Acipenser medirostris White sturgeon A. transmontanus American shad Ailosa sapidissima Pacific herring Clupea harengus pallasii Northern anchovy Engraulis mordax Chinook salmon Oncorhynchus tshawytscha Coho salmon 0. kisutch Steelhead rainbow trout S~almo gairdneri gairdneri Longfin smelt Sprnchus, thaleichthys Eulachon T~haleichthys pacificus Pacific hake Merluccius productus Thpsmel t Athieri1n-ops affinis Threespine stickleback. Gasterosteus aculeatus Striped bass Morone saxatil-is Barred surfperch Amphistichus. argenteus Shiner surf perch Cymatogaster aggregata Pile surfperch Rhacochilus vacca Monkeyf ace eel Cebidichthys violaceus Bay Goby Lepidogobius leid~us Albacore Thunnus alalunga Bluefin tuna T. thynnus Swordfish Xiphias gladius Rockfishes Sebastes M Bacacc io S. Paucispinis Chil ipepper S9. goodei Yellowtail rockfish 'E. -flavidus Canary rockfish S. pinniger Vermilion rockfish S. Mi-ni-atus Widow rockfish S. entamelas Dark-blotched rockfish S. crameri Splitnose rockfish 'S. diploproa Stripetail rockfish ~.saxicoia Shorthelly rockfish S.jordani Sablefish S. Aoplopoma fimnbria Kelp Greenling Hexagranmns decagraimmus5 Lingcod Ophiodon elongatus Pacific staghorn sculpin Leptocottus armnataus Cabezon Scorpaenichthys ma rmo ratus Sanddabs Citharichthys nmarmnoatus California halibut Paral ichthys cal iforniCUS Pacific Halibut Hippoglossus stenolepis Starry Flounder Platichgtys stellatus 1-7 Fish Species List (Cont'd) Speckled sanddab Bat ray Brown smoothhound White surfperch American shad California tonguefish Northern midshipman Pacific pcmpano Pacific tomcod Red Irish lord Saddleback gunnel Whitebait smelt Angel shark Big skate Pacific electric ray Sevengill shark White shark Blue shark I I I _ _ _ APPENDIX 2 I 3 ANNOTATED MARINE BIRD SPECIES LIST I I I I I I I I I I I I I I Erratum: In Appendix 7; Background Documents, Add: Qualitative Assessment of Ecological Damage to the Point Reyes-Farallones National Marine Sanctuary Resulting from the T/V Puerto Rican Oil Spill ........................... Page 7-65 2-1 Appendix 2: Annotated species list of selected marine birds affected by the T/V Puerto Rican oil spill (source: Briggs et al, 1983). ORDER GRAVIIFORMES (Common Loon (Gavia inmer) This loon is an unccmmon migrant and winter visitor which resides primarily on protected coastal waters, bays, and estuaries. A few individuals usually oversummer. Arctic Loon (Gavia arctica) Nesting on freshwater lakes along the Arctic Circle and wintering along the Pacific coast from southeast Alaska to southern Baja California, this is the most abundant loon found off California and the most pelagic. Though they are present off California in all months, Arctic Loons are least numerous in summer, when only a few hundred non-breeders remain. Despite wide occurrence in migration, preferred coastal wintering areas were relatively few. Red-throated Loon (Gavia stellata) This loon is an uncommon to canmmon migrant and winter resident along the coast; a few nonbreeders remain through summer. They prefer waters very near the coast and are the least frequently seen loon >5 km offshore. ORDER PODICIPEDIFORMES Western Grebe (Aechmophorus occidentalis) With major nesting colonies on lakes and ponds from northern California through the Rocky Mountains into Canada, Western Grebes are a prominent component of the seabird fauna wintering off central and northern California. Migration to salt water occurs from late September through autumn, and return migration occurs mainly in April and May. The portion of the wintering population occurring off California varies but little each year, though substantial shifts occur along the coasts of Washington and British Columbia. Red-necked Grebe (Podiceps grisegena) Red-necked Grebes are a rare but regular visitor along the coast. Most sightings occur during winter and migration from Monterey Bay north. They prefer estuaries, harbors, and bays, but can occur along the cpen coast, even offshore. Horned Grebe (Podiceps auritus) Horned Grebes are uncanmon along the coast during migration and winter and are rare during surmmer. They are least numerous south of 2-2 Point Sur. Preferring protected coastal waters, lagoons, etc., Horned Grebes are rarely found seaward of the littoral zone. They are about as abundant coastally as the Eared Grebe and much less so than the Western. They exhibit stronger affinity for salt water than do Eared Grebes. Eared Grebe (Podiceps nigricollis) These grebes are uncommon to common along the coast in most seasons, but are rare in sunnuer. Eared Grebes are more gregarious than Horned Grebes and often are distributed farther south. This species prefers protected bays, estuaries, and harbors, but during winter can be quite numerous at offshore islands including the Farallones. Northern Fulmar (Fulmarus glacialis) Nesting abundantly throughout the Arctic, fulmar populations in Alaska are estimated to total about 12 million, including 38% non-breeders. After nesting they disperse to California waters, where they constitute an important part of the offshore fauna each winter. Numbers here are reported to vary considerably from year to year. Sooty Shearwater (Puttinus griseus) Many recent authors regard Sooty Shearwaters to be among the most important of avian predators in the North Pacific during spring and sumner. Nesting both off South America and in the southwestern Pacific, Sooty Shearwaters arrive off California to early spring and research peak populations in may through July. Flocks numbering in the tens or hundreds of thousands have been reported. Return migration occurs after August, with most birds departing by mid-September. Sooties, presumably young birds, may be found off California in any month. Brandt' s Cormorant (Phalacrocorax penicillatus) Brandt's Cormorants are, to a large extent, endemic to California Current waters. Their 59,000-bird nesting population is third largest among central and northern California species; the Farallon Islands harbor the world's largest colonies. Formerly thought to be sedentary throughout their range, it now appears that after August, Brandt's Cormorants disperse both northward and southward from colonies in northern California. This is the most abundant cormorant offshore, but usually is found over waters less than 100 m deep. Pelagic Cormorant (Phalacrocorax pelagious) With a total nesting population of almost 16,000, Pelagic Cormorants rank as California's sixth most numerous breeding species. Unlike Brandt's Cormorant, the Pelagic nests in low density on precipitous cliffs, occupying many hundreds of sites. This species is perhaps the least pelagic of the cormorants and appears to undertake no major post-nesting population movements. 2-3 Black Scoter (Melanitta nigra) This is a rare to uncommon migrant and winter resident. Though they are quite scarce south of central California. Black scoters may be locally numerous north of Point Reyes in some years. Occasionally, individuals oversummer as far south as San Diego. The North American population is thought to be about 0.5 million birds. Surf Scoter (Melanitta perspicillata) The most abundant of marine waterfowl in coastal waters and migration, Surf and White-winged Scoters together maintain North American nesting populations of over one million birds. All Surf Scoters in the world nest in Alaska and Canada. Scoters migrate over neritic waters and feed in shallow waters near the shoreline White-winged Scoter (Melanitta tusca) See Surf Scoter above. ORDER CHARADRIIFORMES Red Phalarope (Phalaropus tulicarius) Red Phalaropes and Red-necked Phalaropes nest in large but unknown numbers throughout Arctic North America and migrate through and winter off California. Much circumstantial evidence suggests that Red-necked Phalaropes migrate earlier and closer to shore than do Reds. These two species usually are impossible to distinguish during aerial surveys. Western Gull (Larus occidentalis) Ranking as the state's fourth most numerous nesting species, the Western Gull breeding population tops 50,000 birds, with more than 40 colonies of greater than 50 birds. Western Gulls are ubiquitous along the shoreline and in neritic waters, but appear not to venture inland or much beyond the continental slope. The world population of this California Current endemic is centered on the Farallon Islands. Common Murra (Uria aalge) The common Murre is California's most abundant nesting seabird, occupying about 20 colony sites from Hurricane Point (Monterey County) to Castle Rock (Del Norte County). The major California colonies are located from Trinidad Head to Castle Rock (Humboldt and Del Norte Counties and at the Farallon Islands. Common Murres are usually encountered over self waters, where they are abundant; fewer are observed offshore. Numbers are highest during winter, when immigrants from northern colonies enter California waters. The world population has been estimated at 15 million birds, of which, about 10 million reside in the North Pacific. The Alaskan population numbers about 5 million birds. 2-4 Pigeon Guillemot (Cepphus columba) A common nesting resident at hundreds of sites all along the coast, the state nesting total is about 14,700 birds. The nesting season is April-August. A few birds persist through autumn and winter, but most birds are absent from October through February. The winter range is virtually unknown, though banding evidence suggests north- ward movement out of the state. The population probably is stable. Usually Guillemots are found near land, but they may be seen any- where over the continental shelf. The Alaskan nesting population numbers around 200,000 birds. Marbled Murrelet (Brachyramphus marmoratus) This is an uncommon nesting resident of central and northern California, where it apparently nests in coastal conifers and forages to about 2 km offshore. They are found throughout the year north of Monterey but only rarely south of there. The state's nesting popula- tion is about 2,000 birds. Ancient Murrelet (Synthliboramphus antiquus) A rare to uncommon winter visitor off California, numbers of Ancient Murrelets appear to vary widely between years. They usually arrive in autumn and depart in January through April (depending on winter oceanoographic conditions), but records exist for all seasons. This species usually is most numerous from Monterey Bay northward. The Alaskan nesting population numbers around 400,000. Cassin's Auklet (Ptychoramphus aleuticus) This abundant alcid occupies an extensive breeding range in the north Pacific. The Alaskan population numbers about 600,000, and there are about 1.06 million nesting in British Columbia. Of the 131; (-170 Cassin's Auklets nesting in California, 80% (105,000) occupy Southeast Farallon Island, 17% (23,000) use the northern Channel Islands, and the remainder nest at Castle RDck and Green Rock in Del Norte and Humboldt Counties. Cassin's Auklet comnonly occurs offshore during winter (year-round near the Farrallones). Rhinoceros Auklet (Cerorhinca monocerata) Rhinoceros Auklets are seasonally abundant in California waters when large numbers of wintering birds enter the region from the north. This species is distributed discontinuously around the rim of the North Pacific. In Alaska, where predation by introduced Arctic foxes probably eliminated many former colonies, the population numbers about 200,000 birds. The California nesting population totals only about 180 pairs, including 100 pairs at Castle Rock (Del Norte County) and 50 pairs at Southeast Farallon Island. I I I__ _ _ I ~~~~~~~~~APPENDIX 3 1 ~~~~~~ SPECIES SELECTION FOR T/V PUERTO RICAN OIL SPILL RISK ANALYSIS COM4PJJER MO)DELING I I I* I I I I I I I I I I I 3-1 Appendix 3: Selection of seabird and marine mamnal species for oil spill risk analysis Five seabird species and one marine manmal have been selected for inclusion in the oil spill impact analyses. The species have been selected on the basis of several criteria relating to the importance of the different species within the marine bird and mammal fauna of the Point Reyes-Farallon Islands National Marine Sanctuary, their susceptibility to adverse impact due to contact with floating oil, and their seasonal status on the Farallon Islands. The criteria for seabird species are as follows: 1) The species having the largest breeding populations within the Sanctuary because of their overall importance in consumption of oceanic prey, their heavy utilization of the islands for nesting, and their visibility to the public. 2) Seabird species which dive to obtain food because of the increased exposure they may have to floating oil. 3) Species having Federal or State designations as Rare or Endangered because of the implications of the Endangered Species Act. 4) Species for which data exists and which are sufficiently abundant at sea within the Sanctuary to facilitate computer modeling of populations. Under these criteria, the species selected include the largest nesting populations on the Farallon Islands (and within the State of California), three of which are diving birds, and two additional seasonal visitors to the Sanctuary, one of which is listed as Endangered by both the State and the Federal Government. Brief descriptions of the status of each key species appear below. Species Notes Common Murre, Uria aalge Nesting, diving birds Cassin's Auklet, Ptychoramphus Nesting, diving birds aleuticus Western Gull, Larus occidentalis Nesting birds Brown Pelican, Pelecanus occidentalis Seasonal visitor; endangered species Arctic Loon, Gavia arctica Seasonal visitor; diving birds 3-2 COMMON MURRE (Uria aalge) Numbering somewhere around 100,000 breeding birds on the Farallones in summer 1982, murres also nest at about a half dozen other sites around the rim of the Gulf of the Farallones. Following the nesting season, adults accompanied by their dependent young move to the waters within a few kilameters of the mainland shoreline and disperse north and south to make use of nearshore prey such as the northern anchovy. Numbers decline in autumn near the Farallones, then build once again about December, both as a result of early visitation to the colony sites and due to an influx of visitors from waters as far north as British Columbia. The latter may be primarily juvenile birds and may be more common near the mainland than near the shelbreak (the vicinity of the tanker tragedy). In general, murres occur in densities exceeding 50 birds per sq. km. through the waters of the inner and middle continental shelf and in much lower densities seward of the 100-meter isobath. This is a species that obtains prey underwater by "flying" in the manner of penguins. Murres commonly are found to be heavily impacted by floating oil slicks and constituted almost almost a third of all birds recovered dead on Gulf of Farallones and adjacent beaches in the three weeks after the sinking of the T/V Puerto Rican CASSIN'S AUKLET (Ptychoramphus aleuticis) These auklets are the second most abundant of the seabirds nesting in California and the most abundant breeder on the Farallones. The total numbers of birds associated with the islands now are estimated to exceed 180,000. This is a diving species like its relative the murre, but takes primarily euphausiids and other small crustaceans for food. Most birds are encountered in the waters of the outer continental shelf and upper continental slope, where densities vary from about 10 to greater than 50 birds sq. km. Areas having the heaviest concentrations of auklets include the outer shelf from North Farallon Island to Cordell Bank and from South Farallon to about Asension Canyon, 60 km to the southeast. Cassin's Auklets appear to move slightly farther offshore during fall and winter than in the nesting season (roughly Apil through July) and the numbers of locally nesting birds are augmented in fall and winter by visitors from Pacific Northwest colonies. Not often encountered in beached bird surveys (whether or not oil has been spilled), it is likely that most mortalities of this species go undetected at sea, where carcasses sink or are eaten by fish or other birds. WESTERN GULL (Larus ocidentalis) The only gull nesting in large numbers along the coast of California, the Western gull is a conspicuous member of the nearshore bird community. About a half of the State's 50,000+ total nests on the Farallones. Adult birds depart the islands for a few months after conclusion of the nesting season in July and August, but return in the thousands by midwinter. Both the necessity of establishing territorial claims and the lure of carrion associated with winter pupping activities of the northern elephant seal may be involved in the return to the islands. Like other large coastal gulls, the Western gull eats a wide variety of naturally occurring 3-3 have been recorded during beach censuses following oil spills, perhaps because as aerial predators they see and avoid oiled waters. A recent study campleted under MMS sponsorship in Santa Barbara Channel found, hcmTever, that juvenile gulls ware considerably nare apt to come into contact with floating oil than were (experienced adults). Additionally, several studies have indicated that transfer of small quantities of petroleum to gull eggs during incubation can lead to significant declines in hatchability and survival of the young. Gulls of this species occur throughout the shelf and slope waters of the Sanctuary and adjacent regions. BROWN PELICAN (Pelecanus occidentalis) Now rebounding from a severe and prolonged decline in colony productivity, the California Brown Pelican is currently under annual consideration from change in its status as an Endangered Species. Pelicans are abundant and visible visitors to the shores and waters of the Sanctuary from about mid-May through December. Populations found at shoreline roosts vary seasonally fram none in January to about 5,000 to 10,000 in September (for the area from Point Reyes to Ano Nuevo Island), 100 km south of the Farallones). This species obtains its prey by plunge-diving from heights of up to 15 to 20 meters; detection is limited by turbidity of the water. Like the gull studies mentioned above, juvenile pelicans appear to be more apt to contact floating oil than are adults. Foraging takes place mostly over waters within 15 to 20 km from the nearest shoreline roost, but we have encountered feeding birds in densities up to about 2 birds/sq. km. in the area where the stern of the T/V Puerto Rican now rests. ARTIC LOON (Gavia arctica). 1 The North American population of Arctic Loon (G. a pacifica) breeds on freshwater lakes along the Artic Circle in Alaska and Canada, and winters on salt water along the Pacific coast from southeast Alaska to southern Baja California. Arctic Loons are found on California waters throughout the year, but large numbers occur only during migration; the population is smallest during summer when only non-breeders are present. The size of the North American population is unknown; data from California during migration suggest that the Alaskan population may range into the hundreds of thousands. They are most conspicuous during migration when large flocks travel near the coast during daylight hours. In winter the birds tend to be more sedentary, when scattered individuals and small groups inhabit protected inshore areas. Arctic Loons are the most abundant gaviiforms in California waters, and the most pelagic in distribution; they may be outnumbered by Red-throated Loons (Gavia stellata) within 0.5 km of shore. Loons obtain all of their small fish prey by diving and underwater pursuit; their diet on marine waters is very poorly known. 1/ From: Briggs, Tyler, Lewis, and Dettman. 1983. Seabirds of Central and Northern California, 1980-1983 - status, abundance, and distribution. August, 1983. Pacific (CS Region, Minerals Management Service, U.S. Department of the Interior. 3-4 Despite their pelagic reputation, we found Arctic Loons to be most numerous on waters of the continental shelf in all seasons. They were most widely distributed during migration, but even then average density was 3 to 40 times higher on shelf waters than over the slope (average for November surveys was 4.69 birds/km2 on shelf versus 0.32/km2 on slope). Small numbers of Arctic Loons were also observed traveling over waters seaward of the slope. During the winter, Arctic Loons were even more restricted to protected coastal waters; none were observed beyond the slope. Areas preferred by wintering loons included the following: Point Montara to Bodega (including Tcmales Bay) accounted for 35.9% of all wintering loons on average; Monterey Bay averaged 17.3%; and Point Buchon to Point Piedras Blancas averaged 10.5%. In contrast, the 300+ km stretch of coast from Point Arena to Oregon accounted for only 15% of the total. NORTHERN FUR SEAL (Callorhinus ursinus) The criteria for the one marine mammal selected, the northern fur seal, Callorhinus ursinus, are as follows: 1) Northern fur seals are extraordinarily vulnerable to death from oiling. This fact results from their need for clean, air-filled fur as protection against cold ocean waters. Laboratory studies have convincingly demonstrated that oiled fur seals experience elevated metabolic rate and hypothermia during immersion in sea water at normal environmental temperatures; .depending upon the proportion of the body oiled, hypothermia would lead to death.2 2) Northern fur seals have a large population at sea within the Point Reyes-Farallon Islands National Marine Sanctuary, and the larger area affected by the T/V Puerto Rican incident. Aerial surveys conducted over a three-year period, 1980-1983, indicate that the northern fur seal populations within the area defined for risk analysis is at a seasonal peak of over 11,000 animals in February and March, and averages nearly 8,000 animals from January through May (standard error of 1,200)3 3) The northern fur seal is an economically important species. A commercial harvest of young male fur seals occurs annually under the regulation of the North Pacific Fur Seal Commission; nearly 26,000 animals were taken by the U.S. in 1983.4 At present, sale of skins produces about $1 million in net revenue.5 2 Kooyman, G.L., R.L. Gentry, and W.R. McAlister. 1976. Physiological Impact of Oil on Pinnipeds. final Report Res. Unit 71 to OCSEAP, Boulder, Colorado. 3 Bonnell et al., 1983. Pinnipeds of central and northern California. Final report to the MMS 84-044. 4 National Marine Fisheries Service. 1984. Marine Mammal Protection Act of 1972. Annual Report 1983/84. 5 Gentry, R.L. 1981. Northern Fur Seal in Ridgeway and Harrison, eds. Handbook of Marine Mammals, Vol. I. Academic Press, NY. 3-5 4) The northern fur seal is a candidate species for listing as threatened under the Endangered Species Act of 1973 (for purposes of management and conservation, candidate species are afforded the same protection as listed species). The population of fur seals in the eastern North Pacific has been declining at a rate of 4-8% per year; studies have implicated entanglement in debris such as trawl nets as a factor in this decline.6 NMFS 1984; Fowler, C.W. 1982. Interactions of northern fur seals and commercial fisheries. In K. Sabol, ed. Transactions of the 47th North American Wildlife and Niat--ural Resources Conference. PortlandF Oregon. Wildlife Management Institute, Washington,, D.C. I I I__ _ _ APPENDIX 4 I OIL SPILL MODEL DESCRIPT'ION I I I I I I I I I I I I I I I 4-1 Appendix 4: Oil spill model description Model Description The model for estimating bird mortality is divided into * ~~three sections: 1) Oil Slick Movement Model: A detailed description of the history of the slick including the location, size, I ~~~~shape, and percent coverage at different points in time. Descriptions of the slick are derived either from observational data, or from hindcasts made using 3 ~~~~HAZMAT's On Scene Spill Model (OSSM). 2) Bird Contact Model: A simulation of bird distribution and movement in the area of and around the slick. This section estimates the number of birds which actually encountered the oil. * ~~~3) Bird Fate Model: A model which tracks the fate of birds which encountered the slick. These birds may or may not become seriously ailed. of those which were, I ~~~~some sank or were scavenged, some were carried out to sea, some were washed ashore on beaches which were searched, and some were washed ashore on beaches which * ~~~~were not searched. The output of the fate section of the model provides esti- I ~mates of the number of birds which died but were not counted. 3 ~The overall model structure is diagrammed in Figure 1. The following are more detailed accounts of the three sections of the * ~model. Oil Slick Movement Model I ~~A detailed description of the location, shape and movement * ~of an oil slick is required in order to estimate the number of birds encountering spilled oil. This description was obtained using maps of the observed slick at various points in time and interpolating between successive observations to describe the I ~behavior of the slick during intervening'time periods. ~~I / Ecological Consulting, 1985. 2/ See discussion of actual versus observed mortalities in I ~~previous section. 4-2 Figure 1I Flow diagram showing the relationship among the three sections of the model used in this analysis. ovals indicate programs, rectangles indicate input or output data. j.Oil Slick Observations MOVEMENT MODEL Digital Maps of Slick Movement * Behavioral Data IRD CONTACT a Distributional Data MODE i~~~~~~~~~~~~~ i Locations and Numbers of Birds IEncountering Slick �Bird Trajectory Model *At Sea Loss Rates *Beached Bird Data Estimates of Total Mortality 4-3 Description of Oil Slick Data observations of the position of the slick from the T/V I ~Puerto Rican were summarized from observations by the Coast Guard, PR60 and Pt.Reyes N.S. The observations consisted of the position and shape of the slick at each point in time. Twenty-one such observations were used, spanning the period from the vessel breakup on November 6, to the last sighting of a slick north of I ~Bodega Head on November 12. These data were input to the model as * ~sets of ordered pairs of latitude s and longitudes describing the shape and position of the slick at a given time. in addition to I ~these data, we also used estimates of the percentage of the slick which was actually covered by oil -- since typically so-called slicks contain large areas of open water, and these open areas increase in proportion with the age of the slick. Estimates of coverage were provided by the MAS group of NOAA, Seattle. I ~~An alternative approach was also used based on a HAZMAT hindcast of the spill using the On Scene Spill Model, OSSM. The hindcast was based on real time wind data collected during the period of November 6-11 at weather stations in and around the Gulf of the Farallones. For this analysis, OSSM was conditioned I ~to include the response of the slick to the shelf-edge jet which occurred on November 5 and 6, causing the slick to reverse direction and move rapidly northward toward the Farallon islands. The results of the HAZMAT hindcast were recorded at 6 or 24 hour intervals depending on the rate of movement of the slick. These I ~results consisted of estimates of the shape and position of the slick at each point in time, and were input to the bird contact model in the same way as the real observations. 4-4I The results of the hindcast are generally very similar to the observations. There are two ways in which the observed and hindcast slicks differ. (1) The hindcast shows the main body ofI the slick remaining east of the 200 meter depth line (which defines the edge of the continental shelf) prior to the reversal of November 5 and 6. The observations show the main body moving west of the shelf break prior to this time. (2) The sizes of the slicks predicted from the HAZMAT results are about 66% higherI than those observed. It is unclear whether the hindcast or the observations are more accurate in this case, since some portions of the actual slick probably went unreported. Interpolation of Slick Movement in order to completely describe the path of the slick, it was necessary to "fill in" the parts of the path between observa- tions. This entailed pairing each observed piece of the slick with a piece observed at a later time and interpolating betweenI the two irregular shapes. Paired observations of the shape of the slick were input in the form of two sets of ordered latitudes and longitudes defining two closed polygons. Lines forming all possible connections between the two sets of vertices were constructed, and the two lines containing no intersections withI other lines were chosen to represent the boundaries of the path of the slick-during the period between the observations. if an oil slick was observed at times T-1 and T-2, the area swept out3 during the period T-1 to T-2 was defined as the area between the 4-5 two non-intersected lines and the edges of the two polygons formed by the slick at T-1 and T-2 plus the area of the slick at time T-2. The slick area at time T-I was not included in the * ~estimate because its area would have been included in calculating the area swept out between times T-O and T-1. The method used to I ~estimate the area swept out by the slick is illustrated in Figure 2. Bird Contact Model I ~~we assume that birds resting or feeding on the water within the region where the passage of the slick was observed or where the passage of the slick was interpolated were at risk of oil contact. A "risked" bird, however, does not necessarily become oiled for several reasons. Areas reported as being part of an I ~oil slick are generally only partially covered, birds may deliberately avoid oil contact, or the oiling of a bird may be so light as to have little immediate consequence. Estimates of * ~mortality are based on the assumption that some percentage of the birds on the water in the region of the slick will become oiled I ~sufficiently that they will become debilitated and die. The * ~method used to estimate this percentage is described in Model Calibration. 4-6 Figure 2 The interpolation of the passage of an oil slick between times T-1 and T-2. The new area affected during the period T-1 to T-2 would be the stipled region plus the cross- hatched region. SLICK AT ~-TIME T-2 I 4-7 Distributional and Behavioral Data For the hindcast, biological input parameters for each species were derived frcm survey data for November 1984. Aerial surveys were carried out by PRBO biologists under contract to NOAA using the methodology developed by UCSC biologists. Surveys were conducted on November 14 and on November 17, 1984. Transects are shown in Figure 3. Three biological zones in the survey area were identified as follows: North Shelf: That area between 380 N and 390 N, west of the California coast and east of the 200 meter depth line. This corresponds generally to the area of the continental shelf north of Pt. Reyes and south of Point Arena. Central Shelf: That area between 370 15'N and 380 N, west of the California coast and east of the 200 meter depth line. This corrsponds generally to the area of the continental shelf north of Bean Hollow State Beach and south of Point Reyes. Pelagic: That area within the north-south boundaries of both above shelf areas combined, eastward from the 200 meter depth line to the limits of our transect data at the 2,000 meter depth line. Distributional Data: The distributional maps for each species (Figures 4 through 11). The spatial distribution of Common Murres, Cassin's Auklets, and Arctic Loons are summarized in Figures 4 through 6. Input parameters derived from the UCSC data base and used in the model analysis are described below. 4-8 Figure 3 Aerial transect lines and biological zones used in compiling data on seabird distributions. Dotted lines denote boundaries of biological zones, dashed lines denote positions of transects. 'Fort Bragg 39- Pt. Arena ~Tra~ecli~:1.......%p :ranset 49 I 1,Bodega Head . i : ~~NORTH SHELF :Tr'~ ~Transect 5462. i P~t. aReyes ZONE ............. CENTRAL SHELF T ransect 4 1 Head FaraZlon Is. v ZONE Sa Francisco ; Transect 4 1 , Ree Transect 39 ........ iTrans et . .......... ................... , , , ,. , , , - Santa Cruz 3 " 124' 123' 4-9 3 Mean Density of Groups: Mean density of groups of birds was calculated of birds was calculated separately for each species in each biological zone. Densities are expressed as groups per km2. Distribution of Group Sizes: Mean group size and the distribution of group sizes was determined separately for each species. These distributions did not vary appreciably by biological zone; therefore zones were combined for this parameter. Daily Activity Pattern: There are two components to the daily activity pattern, hours of daylight and percentage of birds on the water (as compared with flying) at any given time. In early November, the Farallones region has approximately 11 hours of daylight. All speicies considered here are assumed to be active during this time period. Percentage of birds on the water during the active period was derived from NOAA survey data for November 1984. Birds are assumed to be resting on the water during the hours of darkness. Movement Data: Estimates of movement rates of Cacuon Murres and Cassin's Auklets are based on data for radio-tagged Cassin's Auklets collected by UCSC biologists. The mean distance moved per three hour time interval was 4.3 km. Arctic Loons are engaged in a southward migration at this time of year and periodically stop to rest. On average 60% of the Loons observed in the November 14 and 17, 1984 transect surveys were on the water, so it is assumed that at this time Loons spent 40% of the daylight hours in flight and 60% at rest. Values used as input to the Bird Contact Model are given separately for each species considered (Tables 1 through 3). 4-10 Table 1. Input Values for Common Murre, T/V PUERTO RICAN Hindmast Data for November 1984 Mean Density of Groups: Biological Zone Groups/km2 North Shelf 2.28 Central Shelf 5.929 Pelagic 0.11 Distributiion of Group Sizes: Group Size % Frequency 1 39.5 2 16.1 3 7.3 4 2.2 5 8.0 6 4.4 7 1.5 8 0.7 9 1.5 10 5.1 11-15 5.8 16-20 2.2 21-30 2.1 31-40 0.7 41-50 0.7 51-60 0.7 61-70 1.5 Mean Group Size: 6.066 Daily Activity Pattern: Percent Percent on the water in the air 11 hours active 97.4 2.6 13 hours quiescent 100.0 0 Movement Data: Mean distance moved per 3 hour time interval: 4.3 km 4-11 Table 2. Input Values for Cassin's Auklet, T/V PUERTO RICAN Hindcast Data for November 1985 Mean Density of Groups Biological Zone Groups/km2 North Shelf 0.96 Central Shelf 0.53 Pelagic 1.19 Distribution of Group Sizes: Group Size % Frequency 1 64.0 2 18.0 3 10.0 4 6.0 5 2.0 Mean Group Size: 1.64 Daily Activity Pattern: Percent Percent on the water in the air 11 hours active 64.2 35.8 13 hours quiesent 100.0 0.0 Movement Data: Mean distance moved per 3 hour time interval: 4.3 km 4-12 Table 3. Input Values for Arctic Loon, T/V PUERTO RICAN Hindcast Data for November 1984 Mean Density of Groups: Biological Zone Groups/km2 North Shelf 1.456 Central Shelf 8.966 Pelagic 0.324 Distribution of Group Sizes: Group Size % Frequency 1 61.0 2 22.2 3 2.8 4 2.8 5 2.8 6 0.0 7 2.8 8 2.8 9 0.0 10 2.8 Mean Group Size: 2.083 Daily Activity Pattern: Percent Percent on the water in the air 11 hours active 60.8 39.2 13 hours quiescent 100.0 0.0 Movement Data: 40% in flight in southward migration at any given time during the active period. 4-13 Figure 4 Density Distribution of Common Murres During November 1984 382a * ~~~~~~383V. I ~~~~~37 5& - 373 -s 37 23. -- 3~ ~ ~~ ~~~l 37 Ili13 21 1 23 5 122 58 1 23e122 I Birds/sq km )0-~ Birds/sq km -5 Bird/sq km 5-1Birds/sq 1w 59-188 Birds/sq km 1 9Srd/qk Undefined A Farallion Is. I Sunken Stern 4-14 ~~Figure51 Density Distribution of Cassin's Aukiet During November 1984 38 25 ---7 -------+- + . 4 1~~~~~~~~~~~~ 38 3- -- -- -------------~ ~ ~ ~ ~ ~ --------- + ++ -- a------ 23 5 13--------2----12-3 - 12 2 8 Brd/s k 8--Brd/sqkm5-5 Brd/s k 2-58Bidssqkm5818 Brd/s k )18 Brd/s k 3 I50 Uneie+ aalnI.I Sne tr ----------------- ~~~~ 4-15 3 ~~~~~~~~~~~~~~~~~~~~~Figure 6 Density Distribution of Arctic Loon During November 1984 __ 38a5*.. - --- - - 3 ~~~~~~38 3- 37587 -..--.- 3 - 3736 .4. +~~ -: - 3 ~~~37 23- --. '-- - ---- 1 37 ~~~~~~123 33 123 29 123 5 12 o ~ 36 1~22 2 J2 Birds/so kw )0-5 Birds/sq km 5-25 Birds/so kim 3 ~~~~~~~~~~25-58 Birds/sq km 50-IfiS Birds/sq km )188 Birds/sq km 3 ~ ~ ~ ~ ~ ~ ~~ ~~Undef ined A Farallon Is. I Sunken Stern 4-16 ~~Figure 71 Density Distribution of Common Murres During 3 Entire Year 3817 37 IT � + +4* - - - - - -- 373& --------- 37 23" 37I3 112 --:-- 23 35 n 123 ~~~5 122 58 12236121 I 7 9 Birds/so kx WS- Birds/sq kg 5-25 Birds/sq km3 25-51 Birds/sq km 50-188 Birds/sq 1111 )100 Birds/sq km Undefined A Farallon is. z Sunken Stern 3 4-17 Figure 8 Density Distribution of Cassin's Auklet During Entire Year 3825 -a 3~~_ __=- -- - - - - 37 __ _ U. ---- - -- _=37: ' _ _------------- rr tf3 n - - - - - - - - - - -233 23 2. ' L--- - .....1 21 3 8 Birds/so ke )8-5 Birds/sqo 5-25 Birdsksq km 25-50 Birds/sq km 58-1s88 Birds/sq km irdss m .I~ W Undefined ' Farallon Is. I Sunken Stern I~~3 3..._:-' 4-18 Figure 95 Density Distribution of Brown Pelican During Entire Year 325 38 17- + + +. 38 3.1-- + - 37 So. 37 36 + ++ 37 23- -- 137 35 123 20 123 5 122 56 122 36 12221 5 1 d B irds/sq km }8-5 Birds/sq km j ~5-25 Birds/sq km 25-50 Birds/sq km 58-108 Birds/sq km )110 Birds/sq km Undefined A FarallIon Is. I Sunken Stern3 4-19 Figure 10 Density Distribution of Western Gull During Entire Year 31�-- - - 3 ~~~-.-------------- -------- --- ------ ------- ------- d----. W .-:-+ --------------- ~ ~ ~ ~--- 383---�-- - - - - - -- - - - - - - - - -- - - - -- I~~~~~~~~ - Pt, -------- -~~~~~~~~~~ - -------- - ----------------------- ------------------------ -� --------------------- ~~~- - - - ---------- ~~~~~~~~~~- - ------------- ~~~~~~~~- - - ----------------- ~~~~~~~~~~- - ------------------ ~~~~~~~~- - - ----------------- - - - - - - - - - - -- -/-q-k - - -5-B-r-s-s- - I~~~~~~~~ - -idss k- - j33j )-5 -Bi ---- ----d I~~~~~~~~ 5-5 B-dss k- - jj 5-18 Bidss -m -18 B-dss -km --------- 3~~~~~~~~ -~Un -f -e - -aalo -s - -u~e - St---eri---- --- I~~~~~~75- 4-2 0 Figure 11 Density Distribution of All Birds During Entire Year 38 3. 37 23~~~~~~~~- 123 35 1~~~3 N 123 5 s o 0 ~ 36 I Birdsls'q km )0-5 Birds/sq km 3 -25 Birds/so kmu B55 irds/sq km 51-198 BirdS/sq km fjjj )100 Birds/sq km Undefined A Farallon Is. I Sanken Stern3 4-21 The likelihood that a bird will contact oil depends on whether or not it is on the water when in the vicinity of the slick. Members of the species considered here were assumed to be resting on the water during the hours of darkness, and to be in flight for part of the daylight hours -- about 11 hours in November at the latitude of the Farallones. Percentages of birds on the water during the daylight hours were calculated from unpublished data provided by Dr. Kenneth Briggs. These data were collected during November 1984 in the same geographic area using the aerial strip census protocol described previously. Common Murres spent 97.4% of their time on the water, Arctic Loons 60.8%, and Cassin's Auklets 64.2%. Estimates of movement rates for Common Murres and Cassin's Auklets are based on data for radiotagged Xantus' MurreletsI/. The mean distance moved per three-hour time interval was 4.5 km. Arctic Loons are engaged in southward migration at this time of year, but periodically stop to rest on the water. No data are available describing short term (i.e. on the order of hours) movement patterns of migratory loons or any migrating seabird species. However, since migratory distances tend to be very large relative to the size of the slick, we assumed that Loons either did not move at all -- i.e., were resting -- or moved a I/ Hunt, et al., 1976. 4-22 Estimating Numbers of Birds oiled The total area swept out by an-oil slick multiplied by-the density of birds in the region through which it passed provides a minimal estimate of the number of birds at risk of oil contact.3 The area swept out by the slick during a given time interval was estimated using the methodology described in Interpolation of Slick Movement. Bird density estimates were based on the observ- ed densities in the three zones described above. If the slick crosspd between zones during a time interval, the average density3 was computed based on the relative area which lay in one zone or the other.3 Although the density of birds on the water provides a partial estimate of the number of birds at risk, the actual number at risk may in fact be much greater. Seabirds are highly mobile, and although the population of birds present in a parti- cular area of ocean might remain relatively constant, the indi-B -viduals comprising that population could change completely during a given time interval. The turnover rate within the region of an oil slick was estimated based on the distribution of distances3 moved per three-hour time step, assuming that the movement of an .individual or group could be modelled as a random-walk process.I One thousand simulated bird groups were distributed uniformly across a region the shape of the slick. At three-hour intervals 1 ~~~~~~~~~~~~4-23 3 ~each group was moved a random distance in a randomly selected direction. The distance moved was selected by sampling from a distribution of distances moved per three hours. Each group was 3 ~moved until it had passed out of the region and the number of time steps required to do so was noted. This process is illus- 3 ~trated in Figure 12 The average number of time steps required to leave the region from a random starting position represents an estimate of one-half the average time spent in an area of ocean * ~the size and shape of the slick (since groups where already within the region when they were placed in motion, this time is 3 ~multiplied by two). The turnover rate is therefore the time step divided by the mean length of time spent in the region. For example, if a groups spends on average 5 hours in the region 3 ~before exiting, and the time step is 3 hours, the estimated turnover rate would be 0.6 -- i.e., there would be 60% replace- I ~ment of old animals with new animals during one time step. The number of birds encountering the slick would then be 1.6 times the number of birds on the water. Note that this is an estimate 5 ~of the turnover in an unpolluted region of ocean. We assume that the encounter rate would remain the same if the region were I ~covered by the slick, although in some cases flocks might turn aside after initially encountering it. Bird Fate Model * ~~Birds which were killed or incapacitated by oil are assumed to have had one of four fates: either (1) they were carried out I ~to sea by winds and current,s (2) they sank or were scavenged before reaching shore, (3) they came ashore along stretches of 4-24 Figure 12 Illustration of the method used to estimate the length of time a flock of birds would spend in a region the size and shape of an oil slick. Circled numbers indicate the position of a hypothetical flock at three- hour intervals. The distance moved is based on the relative frequency of distances moved; the turning angle is random. In this case, the flock required about 6.2 intervals or 18.6 hours to leave the region. ,'ii!!!i!!:................iiiiiii l: ii:.:: .......... �-::�:::::::ii:~~ 3 ~~~~~~~~~~~~4-25 3 ~coastline which were not searched, or (4) they washed up or swam ashore on beaches which were regularly enumerated. The purpose of the bird fate model is to partition birds among these various * ~fates. Trajectories of Oiled Birds 3 ~~The HAZMAT oil spill trajectory model was used to compute families of possible trajectories assuming that dead or incapa- citated birds move at 2.2% of the wind velocity. This value is 5 ~based on published experiments with dead, oiled auks in the Irish Sea. 1' The value is lower than the 3.0% typically used for 3 ~surface films of oil, probably because of the relatively large subsurface resistance of the carcasses. Eight typical regions were selected from which to launch circular clusters of model bird bodies.-' Groups of 100 birds were launched from each region and followed until they either came ashore along one of 3 ~six coastal segments, or were washed out to sea (See Figure 13 for a map of launch regions and coastal segments). A matrix was constructed containing the probability that a bird which was 3 ~oiled in a given region would come ashore along a given stretch of coast and the length of time that would be required to do so 3 ~~(see Table 4 The number of birds estimated by the bird contact model to have been oiled within a given launch region were partitioned 3 ~according to the probability that they would come ashore in one I 1~/ Hope Jones et al., 1970. 2/ J. Galt, personal comimunication. 4-26 5 Figure 13 Simulated Bird Carcass Launch and Recovery Areas Used in Bird Fate Model OSSM LAUNCH POINTS FOR SIMULATED BIRD CARCASSES ) 37023.3 122057.4 3 Nov 1200 () 37018.O 122049.5 5 Nov 1200 @37043.4 122053.6 7 Nov 1200 ()3756.7 122055.1 9 Nov 1200 ( 38� 0.7 122055.1 9 Nov 1200 ) (37059.6 1230 4.9 9 Nov 1200 .( 38014.5 1230 7.0 10 Nov 1200 )37016.7 1230 5.2 5 Nov 1200 390 RECOVERY AREAS FOR Pt. Arena SIMULATED BIRD CARCASSES () South of Bolinas ....... * (4)Bolinas to Pt. Reyes (Pt. Reyes to Tomales Pt. !+~~ (~~)~Bodega Area (Tomales Pt. to .Jenner area) Jenner to Pt. Arena North of Pt. Arena ........ :....... . (~) Farallones ()Out to Sea (Never Beached) Bodega Head �: O �3S ' O iPt. Reyes 3s:.............. B) - Farallon Is. an Francisco 374.' 24D t3 Sana Cr. 3T' '4 ' ' ,,Santa Cruz 14123' 4-27 Table 4 Probability (expressed as percent chance) that a bird carcass launched in a given area will come ashore on a particular stretch of beach (for launch areas and recovery areas shown in Exhibit 33. Percent chance of bird carcass recovery if launched from a given launch area Launch Area Recovery Area 1 2 3 4 5 6 7 8 A 0 0 0 0 0 0 0 0 B 2 0 0 0 100 0 0 0 C 44 3 21 23 0 0 0 0 D 53 4 4 17 0 0 0 0 E 1 93 75 60 0 100 100 29 F 0 0 0 0 0 0 0 0 G 0 0 0 0 0 0 0 26 H 0 0 0 0 0 0 0 45 4-28 of the six coastal segments. Of these coastal segments, all were3 carefully searched during the period of the spill except the coastline north of Jenner and South of Point Arena. This area is composed primarily of steep cliffs with little or no access to3 the intertidal, and except for one location was never searched for beached birds.1' Loss of Oiled Birds at Sea Although an oiled bird might drift in the direction of a particular beach and ultimately come ashore there, it might also3 be lost to natural processes before reaching shore. of 106 murre carcasses released by PRBO biologists in 1980 and 1981 within3 1,500 m of shore, only 4 were ever recovered (PRBO unpublished data). However, a large proportion of the weighted drift bottles and gull carcasses released at the same time were eventually3 recovered, indicating that the disappearance of the Murres did not result from being carried out to sea, but rather that the carcasses sank or were scavenged along the way. The experiment carried out by Hope Jones et al. /. resulted in 20% recovery even though bird carcasses were at sea ten or more days and is the3 source of the loss rate estimate used here. We assume that the loss of bird carcasses is a function of how long they are at sea, and that the fraction lost each day is constant. For example, if one-half of the carcasses were lost each day, there would be on-half remaining after one day, one-3 I/ G. Page, personal communication. 2/ Hope Jones et al., 1970. 4-29 quarter remaining after 2 days, one-eighth remaining after 3 days, etc. If 20% were still floating after 10 days as Hope Jones et al. found,-1 the estimated loss rate would be 15% per day. We use these data to generate a matrix of probabilities that a bird oiled in a given region would be recovered along a given segment of coastline. Let L.. be the probability that a carcass launched from region i will come ashore in segment j, S be the loss rate (sinking or scavenging) per day, and D be the number of days required to float from i to j based on the results of the HAZMAT trajectory model. In addition, let C. be a vari- J able which assumes the value 0.0 if a beach was not searched, and 1.0 if the beach was searched. Then the probability that a dead or incapacitated bird launched from i will be recovered in j is: P.. = C. Lj (1 - S) Model Calibration The mortality rate of birds encountering the oil slick was estimated using the actual observations of oiled birds found on the beaches. Let 0. be the number of birds found along coastal segment j, let E. be the number of birds which encountered the 1 slick in region i, and let )be the mortality rate. An estimate of ~ based on the observed number of beached birds found in coastal segment j, is given by: A~~~ = 0. / 7 (PijEi) In other words, the estimated mortality rate is the ratio of the number of birds observed in a given segment of coastline to 1/ Hope Jones et al., 1970. 4-301 the number which we estimate would have been beached if every3 bird which passed through the slick died or was incapacitated. Note that this estimate includes the loss of oiled birds while in transit from the point of oiling to the beac h which is subsumed in 3 the value of the P j Since the value of .,')j varies from beach to beach, we estimated the overall value of FR as the slope of3 regression of the 0.- on the (P. .E. Because the number of J ijJ 1 beached birds must be zero when the number of birds contactingI the slick is zero, the regression line was forced through the3 origin. Model Analysis and Results3 Sensitivity Analysis The values of some input parameters and the validity of some of the assumptions used in the model involve varying degrees of3 uncertainty. When the uncertainty is large, the degree to which model results might be affected by that uncertainty should be5 closely examined. Using data for the most common species, Common Murres, we examined the sensitivity of the model to the following factors:3 o The value of the mean distance moved per 3-hour time step o The assumption that the bird groups move in a random walk fashion, changing direction randomly at each time step5 o The variability in the distribution of seabirds - i.e., the natural randomness involved in how many birds were present in an area at the time when it was affect-I ed by the slick o The value of the rate at which oiled birds are lost at3 sea 4-31 3 a~~~ The assumption that an oiled bird has an equal likeli- hood of being lost each day it is at sea 3 ~We examined the effect of varying these factors on two model outputs: 1 a~~ The estimate of the total number of birds killed by the spill 1 o~~ The estimate of the percent of the birds which became debilitated or died as a result of being present in the area of the slick Distance Moved Per Time Step 5 ~~The distribution of distance moved per 3-hour time step was extrapolated from data for Xantus' Murreleti/ , and it is not U ~known how appropriate these data are for other alcid species. We examined the sensitivity of the model to this input by making model runs using values for the distances moved which were two U- ~times as great as those measured for Cassin's Auklet, and one- half as great. Decreasing the movement rate has the effect of I ~increasing the mortality rate, and increasing the movement rate has the effect of decreasing the mortality rate. This occurs because the number of birds believed to have encountered the slick increases with increasing movement rates, and to account for the observed beachings, fewer of them could have been killed I ~or debilitated. Doubling the movement rate lowered the estimated mortality rate by 21%; halving the movement rate increased the estimated mortality rate by 19%. Deleting the movement algorithm 3 ~entirely increased the estimated mortality rate by 44%. Esti- mates of total mortality, however, were only slightly affected by 1/ Hunt et al., 1976. 4-323 variation in this parameter, changing by less than 2% in any3 case. Random Turning Angles of Moving Birds It has been assumed that the movement path of birds can be3 described as a random walk process, and that model bird groups changed direction randomly at each time step. We examined the3 effect of this assumption on model results by changing the model so that model bird groups continued in the same direction inde- finitely. This change increased the rate at which bird groups3 encounter the slick, however it had little quantitative effect on model results. The estimated mortality rate was unchanged, and3 the estimated number of birds oiled increased by less than 2%. Variability in Bird Distribution Although distributional data were collected at the time of3 the oil spill, we do not know how many birds actually encountered the slick. The distributional data provide a statistical3 description of the numbers of groups per square kilometer and the relative frequency of group sizes in the general area at the time of the incident, but these data are averaged over large regions.3 Even if the mean density of groups in a zone were 0.5 groups per square kin, a given square km of slick at a given instant of time3 might have contained 0, 1, 2, or more groups. Similarly, if the mean group size were 1.5 birds, a given group might actually contain 1, 2, 3 or more individuals. We estimated the effect of3 this variability on the model results by permitting densities and group sizes to vary randomly based on their observed patterns.3 Groups were assumed to be distributed uniform random through 4-33 space, implying that the number of groups within a given area was Poisson distributed. Thus, if the area swept out by the slick during a given time interval is A, the mean density of birds in the region is b, and the number of bird groups in the area which passively encounter the slick is np, then np is a random variable with the probability density function: P[n pI = (Ab)np exp(-Ab) / np! For each time interval, the number of groups making passive contact with the slick was simulated by randomly sampling from this distribution. The size of each group was selected by randomly sampling from the known distribution of groups sizes. The number of groups making active contact -- i.e. flying into the slick -- was simulated by assuming that the probability per until time of a group entering the slick was constant, so that the number of groups actively encountering the slick would also be a Poisson process. If R is the average rate at which new groups encounter the slick (the turnover rate), T is the length of a time interval, and Na is the number of birds flying into the slick, then the na is a random variable with the probability density function: P[na] = (RT) na exp(-RT) / na The model was run 25 times using different random number sequences to evaluate the extent to which model results varied due to this form of stochasticity. The resultant coefficient of variation of the estimated mortality rate was 2.1%, and the coefficient of variation for the total number of birds killed was 1.3%. The stochastic nature of the bird distributions, 4-341 therefore, does not appear to have an important ef fect on model results. At-Sea Loss Rate Although data are available which may be used to estimate3 the rate at which dead or debilitated birds are lost at sea -- presumably due to sinking or scavenging -- these data are not conclusive. The loss rate of oiled auks following the Hamilton Trader incident-I provides the best estimate of the loss rate, 15% per day, but other data suggest that the value could be quite3 different under other circumstances. For example, Common Murre carcasses released nearshore by PRBO biologists resulted in onlyI 4% recover, implying a much higher loss rate. We therefore examined the sensitivity of model results to the entire range of possible at-sea loss rates.3 The mortality rate of birds encountering a slick and the rate at which affected birds are lost at sea both influence theI number of beached birds predicted by the model. An increased loss rate results in fewer beached birds predicted; an increased mortality rate results in more beached birds predicted. For a3 given value of the at-sea loss rate, the model uses observed beached bird data to select the corresponding mortality rateI which best fits the observed distribution of beached birds. if the at-sea loss rate is fixed at 0% per day --i.e., no birds sank or were scavenged before -reaching shore --the estimated3 mortality rate is 23% and the resulting mortality 1,029 birds. 1/ Hope Jones et al,. 1970. 1 ~~~~~~~~~~~~4-35 3 ~If the at-sea loss rate is fixed at 35%, the corresponding mortality -rate must be 100% and the- resultant mortality 3,979 I ~birds. Sinking rates of greater than 35 % per day cannot be fitted by the model because they would require mortality rates of greater than 100% to account for the observed pattern of beachings. The loss rate of oiled birds at sea is clearly the most I ~sensitive model parameter. Based on the range of possible values for the at-sea loss rate, estimates of mortality rates for birds encountering the slick varied between 23% and 100%. Our best 3 ~estimate of the sinking rate, 15%, falls roughly in the midpoint of this range. The PRBO data discussed earlier suggests that * ~this value may err in the direction of underestimating the at-sea g ~loss rate, and by implication, underestimating total mortality. Exponential Sinking Rate 3 ~~We have assumed that the loss of oiled birds at sea proceeds in an exponential fashion, in other words a constant proportion I ~of them disappear each day. This is a simple and logical assump- * ~tion, but it is not the only conceivable model. An alternative model is that at sea loss is not time dependent, but that the * ~proportion disappearing is a constant regardless of how long an oiled bird is at sea. As with the exponential sinking rate model, I ~estimates of both the mortality rate and the number of birds * ~oiled are highly dependent on the value chosen for the proportion lost at sea. If the proportion lost at sea is 0%, these esti- 3 ~mates are the same as for the exponential loss model-23 mortality. The greatest loss rate possible for this model is 4-363 77%, which corresponds to 100% mortality. This indicates thatU the use of a time-dependent loss rate is probably inappropriate for this context. The loss rate of oiled auks in the study conducted by Hope Jones et al1I' was 80%, and of Common Murres in PRBO's study was 96%. Thus, when a constant rather than an exponential at sea loss model is used, both values of the lossI rate are higher than could have occurred in this incident even if every bird which encountered the slick died. While this does-not "prove" that the loss of oiled birds at sea is time dependent,3 such an assumption seems more appropriate in this case. Model ValidationI We tested the accuracy of the model by comparing the observ- ed distribution of beached birds with that predicted by the model. The entire model was run 6 times, once for each coastal3 segment which was searched, each time excluding that segment from the calibration process. The model prediction of the number ofI oiled birds in a given coastal segment and the observed number of oiled birds were therefore independent, and the predicted number of beached birds occurring in a given coastal segment did not3 entail circular computation. The relationship between predicted and observed numbers ofU beached birds is shown in Figure 14 Linear regression of the predicted number of beached birds on the observed number of beached birds for all species indicates that the model accounts3 for 81% (r=.908) of the variance in the numbers of beached birds. I/ Hope Jones et al, 1970.3 4-37 Figure 14 Model predicted and observed numbers of beached birds on six segments of coastline. Predictions of the numbers of beached birds in a given segment of coastline were made independently of the numbers of beached birds observed in the segment. 250 . 3~ OBSERVED 200 - N PREDICTED 150 - LIJ m501 ARCTIC LOMM ON MURRE E: 0 o uo o > > <: il ii0 O0 - 50' ]CASSIN'S AUKLET I 50 Z >" m - _1Z a. < Z o z >-w C3 a0 4-383 The slope of the relationship is .906 indicating a tendencyI toward the underestimation of the numbers of beachings of about 9%. The tendency toward underestimation is most consistent in two areas: south of Bolinas and Bolinas to Point Reyes. These3 beachings cannot be accounted for by assuming passive transport of oiled birds. HAZMAT model results indicated that no beachedI birds should have come ashore south of Bolinas, and very few between Point Reyes and Bolinas. This prediction is strongly supported by the actual observations of the track of the slick.3 Although we postulated that oil slicks and oiled birds move at somewhat different proportions of the wind speed, 2.2% and 3.0%I respectively, these different values' should result in very similar patterns of beachings. The fact that no beached oil was observed south of Drakes Bay therefore indicates that passively3 floating birds should not have come ashore either. It is likely that this discrepancy resulted from oiled birds flying or swim-I ming toward the coast. Even a landward displacement of 5 or 10 km during the first several days following the spill would have been enough to have carried birds into the region influenced by the tidal inhalation of San Francisco Bay. Active movement toward land by oiled birds may also have been the cause of theI model underestimate of the number of beachings observed on the3 Farallon Islands. Model Estimates of Mortality Rates3 The mortality rate is the fraction of the birds encountering the spilled oil which were injured by the encounter to the extentI that, without intervention, they would ultimately have died. The 4-39 mortality rate for Common Murres was estimated to be 42%, for Arctic Loons 10%, and for Cassin's Auklets 30%. The lower estimated mortality rate for Cassin's Auklets than for Common Murres probably results from the assumption that both of these species are lost at the same rate while at sea. It is probable that the small body size of Cassin's Auklets makes them more likely to become waterlogged and sink, or to be scavenged while floating. If the at sea loss rate for Cassin's Auklets were in fact greater than the rate for Common Murres, the estimated mortality rate for Cassin's Auklet would be relatively' higher. We cannot demonstrate this at this time because there are no data available for at-sea loss rates of small alcids. Model Estimates of Numbers and Fates of Oiled Birds Model estimates of the number of birds which contacted the spill and their subsequent fates are summarized in Table 5 Based on the observed path of the oil slick and the simulations of the distribution and behavior of the various seabird species, we estimate that 4,255 Common Murres, 1,670 Arctic Loons, and 210 Cassin's Auklets encountered the spilled oil from the Puerto Rican. Of these encounters, we estimate that 1,787 Common Murres, 167 Arctic Loons, and 63 Cassin's Auklets were injured by the oil to the extent that, in the absence of rehabilitation, they would have died. The most likely fate of oiled birds was probably loss at sea due to sinking or scavenging, accounting for 48% of the oiled Common Murres, 49% of the Arctic Loons, and 59% of the Cassin's Auklets. The remaining oiled birds washed ashore along the central California coast, some areas of which were 4-40 Table5 3 Model Results: Model Estimates of Numbers of Three Seabird Species Encountering the Oil Slick from the Puerto Rican and Their Subsequent Fates I I Seriously Affected by Oil1 Encountering Not Seriously Found on Sunk or Floating Beached But TOTAL DEAD BUT Oil Slick Affected Beaches Scavenged Out to Sea Not Recovered NOT RECOVERED Common Murres Number of Birds 4,255 2,468 487 861 61 378 1,300 Percentage 100.0% 58.0% 11.4% 20.3% 1.4% 8.9% 30.6% Arctic Loons Number of Birds 1,670 1,503 53 82 0 32 114 Percentage 100.0% 90.0% 3.2% 4.9% 0.0% 1.9% 6.8% Cassin's Auklet Number of Birds 210 147 16 37 0 10 47 Percentage 100.0% 70.0% 7.6% 17.6% 0.0% 4.8% 22.4% I 1 These are birds that would have died without rehabilitation. Some birds found on beaches were successfully rehabilitated (see previous section). 4-41 searched for oiled birds and some were not. Trajectory simula- tions indicate that 21% of the oiled Common Murres, 32% of the Arctic Loons, and 16% of the Cassin's Auklets were beached on unsearched or inaccessible stretches of coastline between Jenner and Point Arena. The total numbers of birds which are believed to have died but were never recovered are 1,300 Common Murres, 114 Arctic Loons, and 47 Cassin's Auklets. 4-42 Additional Input Data for Hindcast and for Analysis of Chronic Release U Initial Incident 3 In addition to the analyses carried out for Common Murres, Cassin's Auklet, and Arctic Loons, estimates were also made of the numbers of birds killed by both the spill resulting from the oil released during the 3 breakup of the T/V Puerto Rican and leakage from the sunken stern section. Species of birds found oiled on the beaches were divided into two categories based on their general distributional pattern. Birds typically found in nearshore waters were assumed to have swum ashore on nearby beaches if oiled, or to have been washed ashore if oiled and dead. These species include: Common Loon Black Scoter Red-throated Loon Surf Scoter Western Grebe White-winged Scoter Hmrned Grebe American Coot - Eared Grebe Pegeon Guillemot Cormorant spp. Marbled Murrelet Ruddy Duck Rhinoceros Auklet Scaup spp. Ancient Murrelet In all cases, these species are much more common in the inshore waters near the beaches that were searched than in the waters adjacent to beaches that were not searched (primarily north of Jenner). We used data fron a Univeristy of California, Santa Cruz aerial survey conducted in October, 1980 to examine this assumption in more detail for one of the I more heavily affected species, Western Grebes. The number of Western 3 Grebes along the coast north of Jenner represents about 12% of the Western Grebes observed between 370 32' N and 380 53' N. By the time the slick had 3 reached as far north as Jenner, its real extent had greatly decreased, and the coverage within that region had also decreased from approximately I 4-43 10% to approximately 1% (J. Galt, pers. ca, m.). It is therefore unlikely that the estimates of total mortality would be substantially affected by the addition of mortality which may have occurred in waters adjacent to unsearched beaches. Birds with more offshore distributions were modeled as with the Ccamon Murres, Cassin's Auklets, and Arctic Loons, but the results and methodology for these species are reported in less detail. These species include: Red Phalarope Sooty Shearwater Western Gull Model results were used to derive correction factors relating the observed numbers of beached birds to the true mortality. Data used in carrying out these Analyses are described in Table 6. Chronic Release The effects of the chronic release on five species were examined. Ccmnon Murres, Cassin's Auklets, and Western Gulls were chosed because oiled birds of these species were reported on the Farallon Islands during the period modeled. The Brown Pelican and the northern fur seal were also examined because they are present in the area and are considered sensitive species. Biological data used as inputs to the model are shown in Tables 7-8 for the bird species and Table 9 for fur seals. Since the number of birds beached on the Farallones was not sufficient to warrant the model calibration used for the November, 1984 hindcast, we used mortality rate estimates based on that incident in analyzing the effects of chronic release. The original estimates of the proportion of the birds contacting the slick which became lethally oiled, however, are 4-44I probably boo high to be applied to the slicks resulting from the releaseI from the sunken stern since those slicks were of much smaller size. We therefore assumed that the mortality rate fran these smaller slicks was proportional to the amount of time spent in the area of the slick. For example, if the estimated time spent in the area of the slick during the initial incident were 1 hour and resulted in a mortality rate of 50%, andI the estimated time spent in the area of a slick from the sunken stern were 15 minutes," the mortality rate parameter for the sunken stern slick would be (15/60) X (0.5) = .125. We found it impossible to estimate a3 mortality rate for fur seals contacting the slick fran the sunken stern because very little is known about the behavior of fur seals in theI presence of oil. Seals are generally very dependent on their sense of smell, and there is little doubt that they canasense the presence of spilled oil. Data presented by Geraci and Smith (1982) also suggest that if it is possible to avoid the slick-i.e., if there is a nearby area of clean water-seals will do so. it is therefore likely that many or evenI all of the seals encountering the slick from the sunken stern successfully avoided it; however, it is impossible to rule out the possibility that some of the contacts between the fur seals and the slick had fatal consequences. For seabirds, we carried out these analyses for two seasons, December-I February, and March-May. These seasons were chosen to correspond to3 seasons used MMS' s Oil Spill Risk Analysis Model which was used to estimate the probability that an oil slick originating near the lcoation of the3 sunken stern section would contact the Farallon islands. We used these data as estimates of the probability that an animal encountering oil inI 4-45 4 this vicinity would be recovered on the Farallones. The wind drift factor used by the MMS model to describe oil slick movement is 3.5% which is probably somewhat high for bird carcasses, our best estimate being 2.2%. We used the HAZMAT oil slick model to examine the sensitivity of carcass trajectories to the value of the wind drift factor using the hindcast of the original incident in November, 1984. We tried wind drift factors of 2.2%, 2.8%, and 3.5% to see if the distribution of beachings varied with the value chosen. Although the beaching pattern did vary somewhat, the variation was not statistically significant, indicating that the use of the MMS model to approximate carcass trajectories is appropriate. The bird species which appeared most frequently in oiled condition on the Farallones were Common Murres. Biologists from PRBO counted 2 oiled murres during the period December-February and 4 oiled murres during the period March-May which were either dead or sufficiently oiled that it was believed that they subsequently died (G. Page, pers. comm.). Oil samples were taken from three of these birds, and all were determined to have been affected by oil of the type contained in the sunken stern section. MMS estimates of the likelihood of a trajectory launched from the region of 'the sunken stern reaching the Farallones was 17% for the winter, and 3% for the spring. Using model estimates of mortality, we would therefore estimate that 20 oiled murres would have arrived on the Farallones during the winter, and 4 during the spring which corresponds well with the observations. It should be recognized, however, that there are two counteracting sources of bias involved in these calculations. Some live oiled birds which were not aimed directly at the Farallones undoubtedly arrived at least partially under their own power. This would imply that 4-46 TABLE 6. Input values for three seabird species, T/V Puerto Rican hindcast. Species Sooty Western Red Shearwater Gull Phalarope DENSITY (bird/km2) Slope (> 200 m depth) 1.34 .81 0 Outer Shelf (100-200 m depth) .05 7.39 3.74 Inner Shelf -- North (< 100 m) 0 1.87 .95 Inner Shelf -- South (< 100 m) 0 8.29 1.52 DAILY ACTIVITY PATTERN (% on water) 11 hours active 10% 50% 50% 13 hours quiescent 100% 0% 0% MOVEMENT DATA 50% in Mean distance moved per 9 25.4 migra- 3-hour time step (km) tion at any time 4-47 TABLE 7. Analysis of chronic release: Input values for four species of seabirds. Winter. Species Common Cassin's Western Brown Murre Auklet Gull Pelican DENSITY (birds/km2) Slope (> 200 m depth) 4.22 8.87 .16 0 Shelf (< 200 m depth) 45.91 5.29 2.80 .05 DAILY ACTIVITY PATTERN (% on water) 10 hours active 79% 65% 50% 15% 14 hours quiescent 100% 100% 0% 0% MOVEMENT DATA Mean distance moved per 4.5 4.5 25.4 25 3-hour time step (km) ESTIMATED MORTALITY RATE 6.4% 4.1% 0.9% 0.7% 4-48 TABLE 8, Analysis of chronic release: Input values for four species of seabirds. Spring. Species Common Cassin's Western Brown Murre Auklet Gull Pelican DENSITY (birds/km2) Slope (> 200 m depth) 1.62 7.12 .16 0 Shelf (< 200 m depth) 16.53 5.55 .97 .05 I DAILY ACTIVITY PATTERN (% on water) 12 hours active 79% 65% 50% 15% 12 hours quiescent 100% 100% 0% 0% MOVEMENT DATA Mean distance moved per 4.5 4.5 25.4 25 3-hour time step (km) ESTIMATED MORTALITY RATE 10.8% 9.6% 4.5% 0.7% 4-49 TABLE 9. Analysis of chronic release: Input values for northern fur seal. Season December- February- April- January March May DENSITY (animals/km2 Slope (> 200 m depth) .0532 .1038 .1038 Shelf (< 200 m depth) .0339 .0631 .0631 DAILY ACTIVITY PATTERN (% on water) 12 hours active 100% 100% 100% 12 hours quiescent ' 100% 100% 100% MOVEMENT DATA Mean distance moved per 5 1 5 3-hour time step (km) 4-51 The Oil Spill Risk Analysis Model of the Minerals Management Service U.S. Department of the Interior The Oil Spill Risk Analysis (OSRA) model was developed in 1975 by the Department of the Interior (DOI) to aid in the estimation of environmental hazards of developing oil resources in Outer Continental Shelf (OCS) lease areas. The model is maintained and operated by the Branch of Environmental Modeling (BEM), with research and data collection supported by a multi-million dollar study program. Information requests regarding the OSRA Model should be directed to Robert LaBelle, Chief SBE. The large, computerized model analyzes the probability of spill occurrence, as well as the likely paths or trajectories of spills in relation to the locations of recreational and biological resources which may be vulnerable to spilled oil. The probability of spill occurrence is estimated from historical accident rates and information on the anticipated level of oil production and method and route of transport. Spill movement is modeled in Monte Carlo fashion with a sample of S00 spills per season, each transported by seasonal surface-current vectors and wind drift data sampled from 3-hour seasonal wind-transition matrices. Transition matrices are based on historic wind records grouped in 41 wind velocity classes, and are constructed seasonally for up to six wind stations. Locations and monthly vulnerabilities of up to 31 categories of environmental resources are digitized within the modeled study area. Model output includes tables of conditional contact probabilities (that is, the probability of hitting a resource, given that a spill has occurred), as well as probability distributions for oil spills occurring and contacting environmental resources within preselected vulnerability time intervals. The model provides the DOI with a method for realistically assessing oil spill risks associated with OCS development. To date, it has been used in oil spill risk assessments for 47 OCS lease sales with the results reported in Federal environmental impact statements. Additional runs have been performed to analyze OCS development plans in California. Hindcasts of actual spills using real-time data have given good results in the past (Argo Merchant, Ixtoc, and Santa Barbara Spills, see Amstutz and Samuels, 1984) and have lent credibility to the use and projections of the OSRA model. The results of these analyses are compiled in a series of the DOI Open File Reports. The OSRA model's methodology and results have been published in the scientific literature and presented at professional conferences. 4-52 The DOI performed a series of special runs of the Puerto Rican oil spill incident for the U.S. Department of Comnerce, Nationa Oceanic and Atmospheric Administration, Sanctuary Programs Division (SPD). - Sensitive areas of concern were supplied by the SPD and were prepared as targets for the OSRA. The tow path of the Puerto Rican, and the site where the stern sank were analyzed as launch sites (sites where an oil spill occurs) and the probabilities of contacting the designated targets were assessed. Historical wind transition information from the Farallon Islands was used in the analysis rather than real-time wind data. The model results provided probabilities of contact to targets within 30 days for spills launched during each of four seasons, and 3-, 10- and 30- day probabilities of contact on an annual basis. The OSRA projections for the Puerto Rican oil spill were by design calculated using a stochastic approach with 2,000 hypothetical oil spill trajectories being used to represent both the general trend and variability of winds and surface currents in the area. This is in contrast to other more deterministic approaches utilizing real-time wind, tide and current information. Real-time data input is the most likely to provide the optimum sbort-term at-site forecast of where a given slick will move in the next few hours. During a spill response, such a trajectory analysis should be continuous, with a constant modification of results being made available to containment and cleanup personnel. Both approaches to oil spill modeling are specialized tools which each have different uses, Limits, and assumptions. Any comparison should take the above factors into account. 4-53 Marine Environmental Research 13 (1984) 303-319 Offshore Oil Spills: Analysis of Risks David E. Amstutz* & William B. Samuels* Minerals Management Service, US Deparutment of the Interior, Washington, DC 20240, USA (Received: I October, 1984) ABSTRACT The methods used by the US Department of the Interior to estimate risks of accidental oil spills attendant with offshore oil production and transportation are described. Both the likelihood of spill occurrence and hypothetical spill trajectories are considered. Five separate applications of the risk assessment work are sutqmarized. INTRODUCTION The work presented summarizes efforts to quantify risks of accidental oil spills within the marine environment. This risk analysis treats both the incidence of oil spills, separately from offshore platforms (production sites), from pipelines, and from tankers, and the movement (trajectories) of these hypothetical spills. The work has been carried out within agencies of the US Department of the Interior as a part of the Outer Continental Shelf (OCS) oil and gas program. Oil spill analyses are used in the preparation of Environmental Impact Statements (required by the National Environmental Policy Act) and other Departmental decision documents. The Outer Continental Shelf Lands Act, as amended in I978, charges the Department of the Interior with expeditious development of offshore oil and gas resources. The National Environmental Policy Act of 1969, as amended (NEPA) and the regulations which implement that Act require Present address: Science Applications, Inc., McLean, VA 22102, USA. 303 4-54 304 David E. Amstutz, William B. Samuels r\ AN F-MANCM.SCO LUMBER ; 38e %. . Fig. . Mashowing tractsconsidered for leasing,takrotssaotragadh location of the lumber spill site off the California coast. quantification of risks posed to the environment through Federal undertakings. These laws and the work presented here reflect society's goals to protect the environment while pursuing reasonable development of non-renewable resources. Oil spill analyses are controversial, reflecting not only differences of opinion within the modeling community but also AN\\- % % the abundance of emotions and intuitive notions surrounding incidents of oil pollution in the marine environment. The described modeling work was initiated in 1975. Since then, more than 30 oil spill risk and trajectory analyses have been completed. All of the analyses are contained in the US Geological Survey Open-file Report Series. Each analysis has been associated with specific offerings of offshore oil leases (OCS lease offerings). For example, Fig. I shows the location of tracts considered for a lease offering in central California. Nearly all of the US offshore regions (including Alaska) have been examined. Each analysis also includes examination of other sources of �P-~A C ''F"IC Ol:EANJ � a * & 4 .II Nd N N ,- Fill. l . Map showing tracts considered for leasing, tanker routes, sea otter range and the location of she lumber spill site off the California coast. ,quantification of risks posed to the environment through Federal undertak.ings. These laws and the work presented here reflect society's goals to protect the environment while pursuing reasonable development of'non-renewable resources. Oil spill analyses are controversial, reflecting not only differences of opinion with~in the modeling community but also the abundance of'emotions and intuitive notions surrounding incidents of' oil pollution in the marine environment. T~he described modeling work: was initiated in 1975. Since then, more than 30 oil spill risk and trajectory analyses have been completed. All of the analyses are contained in the US Geological Survey Open-File Report Series. Each analysis has been associated with specific offerings of offshore oil leases (OCS lease offerings). For example, Fig. 1 shows the location of tracts considered for a lease offering in central California. Nearly all of the US offshore regions (including Alaska) have been examined. Each analysis also includes examination of other sources of. ':"'i-ii'~~ 4-55 Offshore oil spills: analysis of risks 305 potential oil spills in the area, such as those associated with tankering of imported oil. Our purposes in presenting this example of applied science are to describe techniques and to summarize results and applications. Of perhaps equal importance, is our purpose to inform the general scientific community of one means by which very large and diverse environmental data sets have been put to use to help achieve a societal goal. DESCRIPTION OF MODEL The oil spill risk analysis model represents the study area using an orthogonal grid. The grid allows input of spatial information from any map projection. The study area for each analysis includes potential oil production sites as well as tanker and pipeline transportation routes. Additionally, the area includes adjacent regions which might be contacted by spills within approximately 30 days travel time (see Fig. 1). The coastline of the study area is digitized on the model grid and divided into segments (usually two sets) which allow examination of where marine oil spill trajectories contact the coastline. 'The first set of coastal segments are approximately of equal length, thus eliminating any length-related bias. Coastal segments of approximately equal length are generally 20 to 30 miles long, thus allowing some implicit spatial extent to the simulated oil slick. A second set of coastal segments representing political boundaries (counties and/or states) is frequently used. Various at-sea resources, potentially vulnerable to offshore oil spills, are digitized on the model grid in the same manner as the coastline. The spatial portrayals of these at-sea resources may be changed each month. While some at-sea resources are potentially vulnerable year round, others may be present or vulnerable to oil spills only during specific months. Thus, the model is capable of addressing resources such as migrating birds or whales. The model computations are summarized on a seasonal basis to aid in later determinations of environmental impacts. Analysts can therefore address potential impacts to summer tourism, seasonal fisheries, etc. The two main driving forces influencing the advection of oil spills on the ocean are surface currents and local wind stress. Although the astronomical tides induce horizontal motions in the ocean, these motions generally do not contribute to the net advection of oil spills. Instead, the 4-56 A~~~ 306 David E. Amstutz, William B. Samuels 3 tidal motions contribute to lateral dispersion of an oil slick. The astronomical tides can contribute to net advection in regions of relatively shallow water such as found in portions of the Bering Sea (Liu & Leendertse, 198 1) and in some relatively confined embayments such as Cook Inlet, Alaska. The oil spill risk analysis being discussed treats hypothetical oil spills and not actual oil spills. Clearly, those concerned with actual oil spill movement in near-shore waters during real time circumstances would require knowledge of local tidal conditions to establish the approximate time during the day of initial contact and to refine estimates of land fall sites to scales less than the length of a coastal segment. The oil spill model is designed in a modular fashion such that other trajectory movement algorithms (Samuels et al., 1983) can be incorporated as needed. In the oil spill trajectory model, surface currents and local wind stress are assumed to act independently. This implies that the surface water velocity field is free of local wind effects. To satisfy this relationship, seasonal or monthly geostrophic surface water velocities (Kantha et ae., 1982) are provided to the oil spill trajectory model (see Fig. 2). Geostrophic currents are divided from the distribution of sea water density. The assumptions in the derivation are that accelerations are negligible, that the horizontal pressure gradient and the Coriolis effect are in balance, and that other forces are negligible. The contribution of local wind stress to oil spill movement is assumed to be 3-5 % of the wind speed (Smith, 1968; Stolzenbach et al., 1977) and rotated according to a variable wind deflection angle function as described by Samuels et aL (1982). The oil spill model simulates a large number of oil spill trajectories in a Monte Carlo fashion from potential and existing platform locations, pipeline routes and tanker lanes in an OCS leasing area. These trajectories are summarized as probabilities that if an oilspill occurs at a particular location, it will contact a vulnerable resource (e.g. seabird foraging area) or section of the coastline within certain defined time limits (usually 3, 10 and 30 days). The probabilistic nature of the trajectory model stems from the treatment of the variation in the wind as a first-order Markov process and the assumption that spills occur randomly throughout the year. As a first-order Markov process, the wind state (speed and direction) at any time step is considered to be solely dependent upon the wind state during the prior time step. Time series of wind velocity, collected at coastal as well as offshore (buoy) weather stations over several years, are used to construct seasonal wind transition matrices (see Fig. 3) (Smith et al, IS~ ~4-57 - Offshore oil spills: analysis of risks 30-7 ~Se~t a'e (. ........ ,* , ; 1% . I~ ~~~~Fg2 Mpfesfaceaevlctfedsmesao~fteoutsernU ii,7//. , � o ~. '' *, ti, r/ / oIV/- I~~~~~~~~-�:: ..... :~'.,.~ s ;; i ...... 1 U ..t'..~~~~....'"/' "'....*...... * a ;. * * S * * * ,* t * > e::os~ostK ' * .....;. >Q I Fig.-2. Map of-the sra oae veoci # c ( u m � e .o. off. - (Kantha et al., 1982)., , . 1982). These stations usually record o wf i eight observations per day; thus, the time step for tracking oil spills in the* model is usually 3 h. At each time step, th e wind transition matrix is sampled to yield the wtind velocity imparting motion to the oil spill. in 3 h: there is a 2 % chance the wind will be calm; a 14 % chance the wind in~~~~ 3 h : thr i . a 2 ~ chnc th win wil be cam a 14 ~ c hane Bhe w i nd X will be from the north at 5 knots; a 21 % chance the wind will persist from the north at 10 knots, etc. In making the trajectory comnputations, the transition matrix is sampled at 3-h intervals in Monte Carlo fashion to yield a 'next wind state' which is in direct proportion to the observed states. After selecting the next wind state, the locally induced wind drift I~~ ~~~ape oyedtewn eoiyiprigmto oteolsil I~~~~~een oFg ,cnie h rsn idt efo h ot t1 I~~ ~~~nt.Fo h esrdtm ere idvcos ehv bevdta -* :' 4-58 308 Darid E. Amstutz, William B. Samuels Next wind (direction, speed in knots) X0 N N N N N NE NENENENE E NE N E.. 8 calm 5 10 15 20 25 5 10 15 20 25 5 caim 44 9 1 - -- 6 2 - - - 5 N 5 23 26 9 - - 13 3 - - 4 N 10 2 14 21 8 3- 10 13 - I - 2 N 15 - 3 26 23 13- 3 - 6 10 - o N20 - - 8 15 23 - -8 23 15 -- N25 - - - - - - - - - - XNE 5 25 25 2 - I - 12 8 I - - 9 NE lo 4 8 5 2 - - 15 29 21 3 - 4 NE 15 - 1 6 5 - - 4 20 31 24 - 0 NE20 - - - - 2 1 1 5 11 49 11 S NE25 - - - - - - - - - 48 48 X- E5 26 9 1-- I 2 5 - - 17 Matrix elements in per cent:- indicates <0-5o, � indicates > 99-5'% . Fig. 3. A portion of the wind transition matrix (winter season) for the San Nicolas island, California, weather station. The matrix is constructed from 27 years of observations taken at 3-h intervals. velocity is calculated using the 3-5 % rule and variable deflection angle function described above. The locally induced wind drift velocity is then added to the surface water velocity to yield the net advection of the center of mass of an oil spill at each time step. MODEL APPLICATIONS Evidence, justifying the use of the oil spill trajectory model, has been obtained by comparing trajectories calculated by the model with the paths of actual spills. Wyant & Smith (1978) showed that the model's predicted oil spill trajectory agreed quite closely with the path of the Argo Merchant spill, which occurred in the coastal waters off Massachusetts (see Fig. 4). LaBelle et al. (1982) showed that the spatial distribution of hypothetical oil spill contacts to the southern California coast, as calculated by the oil spill trajectory model, agreed for the most part (see Fig. 5), with the spatial distribution of drift bottle land falls. We believe that the differences in the distributions are due in part to the fact that not 4-59 '~:xY 4 - 5 9:;:' Offshore oil spills: analysis of risks -309 M A S AT L ANT ::C: ECA AN I Hi AR~GOMR ERCHArT agOBSERVED - MC DCL 40Z FoEDzCT~s:1iA | ,, , , *,.o=~= I Jt L * I 3NAUIZrCAL MLQES Fig. 4. Comparison of the model's predicted trajectory with the observed movement of the Argo Merchant spill. all drift bottles are recovered. Some bottles leak or are broken, some are buried on the shore, and some are washed back to sea after a short time ashore. To the extent that bottle recovery is related to beach use, one would expect a bias towards the California coast and away from the island locations (sites 22-30 in Fig. 5). The drift bottle release study was conducted - by the California Cooperative Fisheries Investigations Program (Schwartzlose & Reid, 1972). Comparisons of this type are valid ofcourse only if the drift bottles contain ballast and move with the surface waters. VanBlaricom & Jameson (1982) reported on the movement of lumber spilled off central California on February 12, 1978, and the implications of this event for oil spills and sea otters. These authors reported major floating patches of lumber within the sea otter range as early as February 14, 1978 (within 3 days of the spill) and as late as March I11, 1978 (within 30 days of the spill). We matched the lumber spill site to the nearest tanker transportation route (see Fig. 1) considered as a potential source of oil spills in an oil spill trajectory analysis (OSTA) for the central California 4-60 1 310 Darid E. Amstutz, William B. Samuels s O-- :NCZCATES LAIJNCH-I S:RE 30" it II, : s -3 z, ,'., .Jlls I X I g is ,11,TIS.. � SMl~tTIONS 15t 15 32 ~10 S20 25 I3~ S. + SIEGMIN' NUMI1I I! N - (Y~~~~q Fig. 5. Map of the southern California coast showing coincident drift-bottle release point and trajectory model launch point. Inserts show distributions of drift-bottle landfalls and predicted oil spill contacts to the coastal and island segments (expressed as per cent chance). area (LaBelle et aL., 1983). The sea otter range was considered a target for this trajectory analysis. Given that an oil spill could occur anywhere along this transportation route and at anytime during the year, the probability that it would contact the sea otter range within 3, 10 or 30 days was calculated to be 11, 19 and 30 %o, respectively (LaBelle et aL., 1983). A further comparison was then made to simulate oil spills from the lumber spill site itself. This analysis showed that the probabilities of oil spill contact to the sea otter range for 3-, 10- and 30-day travel times increased to 19, 29 and 44%, respectively (LaBelle, 1983). Because of the uncertainty associated with the time of oil spill occurrence, these probabilities are based on the assumption that oil spills could occur anytime during the year. We decided to take this investigation one step further by simulating oil spill trajectories from the lumber spill site during the months of February and March. The oil spill contact probabilities to the sea otter range during these months for 3-, 10- and 30-day travel times were 15, 51 and 70o%, respectively. The increasing probabilities are a r')~ ~4-61 Offshore oil spills: analysis of risks 311 XND::CATE3 LOCAr=QON OF.- TAP:E'R S::NI'N{ a �t nM'RTH , " . Z ] J f/ = RCL= 14A 5 -> >2 as'/ X~ S ,x-7lt P- s~ ~ ~~~~~~~~~~~~~~~o PYIt :~~" . =^ . .~''"", . .-.' -5 03 GE ~ ~ A TATCR CrEA \1 - 'a % % . % Fig. 6. Map of the Cape Hatteras, North Carolina, area showing tanker sinking locations, lease tracts (PI-P7), transportation routes (TI-TS) and coastal segments. consequence of focusing the model spatially and temporally to a specific event. Since the oil spill modeling work presented in this report is stochastic in nature, total verification of the probability distribution of oil spill contacts to a coastline requires observations on a number of oil spills in order to draw statistical inferences. Such a data base rarely exists for a given study area. The Cape Hatteras area of North Carolina (see Fig. 6), however, was the site of intense submarine warfare during World War II, and tanker sinkings during that period provide a data base for verifying the model. Campbell el al. ( 1977) examined tanker sinkings by submarine activity during the first six months of 1942, and were able to identify 15 sinkings of laden tankers just south of Cape Hatteras, between latitudes 33�N and 35�10'N. During that same period, at least one oil spill was confirmed to have washed ashore on the Outer Banks south of Cape Hatteras and it is believed that as many as three oil spills may have come ashore on this portion of the coastline (Campbell et al., 1977). The locations of the 15 tanker sinkings were matched to three oil spill launch 4-62 312 David E. Amstutz, William B. Samuels 3 toa. -A~ RS=NCQN MVr ImAsi i? R CAUZft 3 RmA NAiAr ZAL. MZ . � ~I Fig. 7. Map of Santa Barbara Channel showing observed slick outline after 8 days, and the sequence of 64 simulated trajectories launched from Platform A at 3-h intervals. sites (lease tract group P3, and transportation routes T2 and T5 shown in Fig. 6) considered in the OSTA for OCS Lease Sale 56 (Samuels & Lanfear, 1980). The model's results were used to predict how many of the slicks were likely to come ashore on the Outer Banks (coastal segments 3, 4 or 5 shown in Fig. 6). Of the 15 spills, the model estimated that there was only a 9 % chance that none would come ashore, but that there was a 75 % chance that between I and 3 oil spills would contact the Outer Banks - (Lanfear & Amstutz, 1981). This is in good agreement with the observations of Campbell et al. (1977). The4ast model application described is the 1969 Santa Barbara oil spill. On January 28, 1969, Union Oirs platform A, located in the Santa Barbara Channel (see Fig. 7), blew out spilling approximately 5000 barrels of oil per day over the first 8 days of the spill (Allen, 1969). The size and trajectory of this slick were observed at different time intervals using aerial photography, and diagrams of the slick outline are reported in Allen (1969). To see how well our model would reproduce the slick size and trajectory, we simulated the release of oil from platform A over an 8- I 4-63 Offshore oil spills: analysis of risks - 313 day period. We began by selecting the wind records (minimum of eight observations per day) for January and February, 1969, from a weather station located on San Nicolas Island, California. The San Nicolas Island weather station is located approximately 70 nautical miles south of platform A; however, it has been shown that these wind measurements can be used to represent the offshore wind patterns observed off the California coast south of 350 N latitude (Williams et al., 1980). This station has been used in all of the OSTAs performed for the California OCS. We simulated the release of oil over an 8-day period by calculating a sequence of 64 oil trajectories. We began this sequence with the first wind measurement made on January 28, 1969. Successive trajectories were initiated at 3-h intervals in accordance with the sequence of wind measurements. The wind factor and deflection angle were applied, as previously mentioned, and this local wind-induced velocity was added to the geostrophic surface water velocity calculated by a diagnostic circulation model for this region (Blumberg et al., 1983). In one sense, then, we were not only comparing our model with an actual event but were also investigating whether the winds measured at San Nicolas Island were representative of the winds observed in the Santa Barbara Channel. This comparison also allowed us to test the appropriateness of the geostrophic currents used in this model run. The results of this comparison are shown in Fig. 7. The agreement between model results and the slick outline reported by Alien (1969) is good except that Allen's observations show a tongue of oil moving in the direction of Santa Rosa Island. This movement of oil remains so far, unexplained by the model. The movement of this slick, as predicted by the model, appears to agree quite well with a description of the slick as reported by the California Department of Fish and Game (Anon., 1969) as follows: 'On January 28, 1969, an offshore oil well being drilled offshore 62 miles south of Santa Barbara, California, ruptured.... The next day, aerial surveillance indicated a heavy concentration of oil 4 miles wide and 12 miles long stretching in a south- southeasterly direction from the site of the rupture, oil platform A. A light film of oil tapered off in a northeasterly direction and intersected the beach west of Carpinteria, then ran south along the beach to Rincon Point. Numerous patches of oil stretched 14 miles south of the main concentration and east along the coast. On February 1, heavy oil slicks were *, 0 4-64 : 314, David E. Amstutz, William B. Samuels reported I to 2 miles offshore from Summerland to Port Hueneme, and some oil was washing ashore on various beaches between Rincon Point and Pitas Point. By February 3, a large heavy oil slick surrounded Anacapa Island and was around the eastern end of Santa Cruz Island. During the night of February 4, the oil washed ashore west of Santa Barbara and into Santa Barbara Harbor; so that on February 5, oil extended along approximately 10 miles of shoreline west and east of the town. The rocks and shoreline of Anacapa Island were covered with oil. About the same situation existed on February 6. A report issued on February 7, indicated oil was on the beaches in varying amounts running from light to heavy along a 28 mile stretch of coastline from Ventura west to Hope Ranch Beach.' It is interesting to note that this description does not report oil contacts to Santa Rosa Island, in contrast with Allen's observations. Furthermore, diving surveys (conducted by the California Department of Fish and Game) were only done at Anacapa Island, and the northeast and southern portions of Santa Cruz Island. No diving surveys were conducted at Santa Rosa Island. OIL SPILL OCCURRENCE ESTIMATES The probabilities that hypothetical oil spills, originating at a particular location, will contact various resources at sea and/Qr the coastline are conditioned on the assumption that an oil spill has occurred. The two remaining steps then in modeling oil spill risk are to determine the likelihood of spill occurrence and to calculate the joint probabilities that spills will both occur and come in contact with the various resources at sea and/or the coastline. Spill occurrence is considered to follow a Poisson process (Smith et al., 1982). Thus, the probabilities of 0, 1, ..., n spills are given through knowledge of the mean (expected) number of spills. The mean number of spills is the product of the spill rate (spills per billion barrels) and the volumes of oil anticipated to be produced and transported. Thus volume of oil produced and/or transported is chosen for the exposure variable. Because of the very small numbers of platform (12 since 1964) and pipeline (8 since 1967) accidents, analysis of various exposure variables is 4-65 Offshore oil spills: analysis of risks 315 limited. In the case of pipeline-related spills, we do know that volume throughput is a better exposure variable than length of pipeline. Additionally, many alternative exposure variables (number of wells, platform years, etc.) are dependent upon the expected production volume. Volume of oil lends itself not only for risk analysis but also for benefit analysis. Spill rates, for spills _> 1000 barrels (bbls) and a 10 000 bbls, are determined through analysis of historical accident, production and transportation records. The minimum spill size of 1000 bbls is chosen for use in the risk analysis, as a 1I000-bbl spill is considered to be large enough to lend itself to trajectory analysis and of sufficient size to affect the environment. By using the historical records we assume that past experience is a suitable index of future experience. The time span over which the oil spill risks are estimated for each offshore lease offering is determined by the expected time required to produce the economically recoverable volumes of oil contained in the tracts offered for lease. Although production lifetimes vary from region to region, they generally fall between 20 and 30 years. The historical accident records have been subjected to trend analysis (Nakassis, 1982) to assure proper adjustments for recent experience. Separate oil spill data bases are maintained for platforms, pipelines and tankers. Only those platform and pipeline spills associated with US Federal offshore activities are used in projecting future spill rates. This restriction of the data base implicitly allows for any effects on spill rates that might be associated with Federal supervision and regulation of offshore exploration and production. We have used spill rates derived from platform and pipeline experience on the Federal OCS to project spills for Cook Inlet, Alaska (state offshore leases) and Prudhoe Bay, Alaska (state onshore leases). The agreement between projections based on Federil OCS experience and what has occurred in each of the Alaska areas is excellent. For example, as of March 1983, 0.8 billion barrels of oil have been produced and transported from Cook Inlet with three spills of 1000 barrels and greater being observed. Using the rates reported here, we would expect 3.45 spills in this size category. Ten spills have occurred in producing and transporting 2-8 billion barrels of oil from Prudhoe Bay; our oil spill rates predict.the occurrence of 9.1 spills. The number of oil spills > 1000 bbls and the volume of oil produced on separate areas of the OCS are too small to justify separate spill rates. Also, there is no evidence in the available data to support the notion of ) 4-66 316 David E. Amstutz, William B. Samuels TABLE I Oil Spill Occurrence Rates for Platforms, Pipelines and Tankers Source Oil spill rates (spills per 109 barrels)' a 1000 bbls 10 000 bbis Platforms !.00 0-44 Pipelines ! .60 0.67 Tankers, total 1-30 0.65 at sea 0-90 0.50 in portb 0.40 0.15 Source: Lanfear & Amstutz (1983). Includes bays, estuaries, harbors and piers. high local variability in the spill rates. Using the assumption that accidental oil spill occurrence follows a Poisson process, we have addressed the question of local variability in spill rates. Accounting for the variation in production from the separate areas, our analysis suggests that accident-free (or accident-prone) production in any specific area will not be detectable with statistical confidence for another 2 to 5 years. Tanker spill rates are projected using the worldwide record of accidental spills. We have not been able to reject the hypothesis that tanker spill rates for US waters are the same as elsewhere in the world. Also, in many instances, the areas included in an oil spill analysis extend beyond what might be formally considered as US waters. The oil spill data bases and the analyses used for determining spill rates used in the risk assessment are presented by Lanfear & Amstutz (1983) and Nakassis (1982). The spill rates currently used in our work are presented in Table 1. CONCLUSIONS The Department of the Interior oil spill analysis model has been utilized for each offshore oil and gas lease offering held in the past 8 years. The model is probabilistic and treats hypothetical spills which might result from the production and transportation of offshore oil as well as spills resulting from tankering of imported oil. The model also serves to analyze various deletion alternatives to the total proposed lease offering. The model provides a framework for synthesis of enormous amounts of 4-67 Ojishore oil spills: analysis of risks 317 environmental data (winds, currents, and the location and temporal sensitivities of various marine resources) as well as historical oil production, transportation and accident records. Results from this work are utilized in the assessment of environmental impacts and in preparing various decision documents associated with the management of offshore oil resources. Thus the modeling work conveys technical information in a useful way to resource managers. Five separate applications of the model have been carried out. Two of these applications were hindcasts of actual oil spills: Argo Merchant tanker spill and Santa Barbara platform blowout. Three other applications compared the moders predicted oil spill contact probabilities with drift bottle landfalls, spilled lumber contacts to the sea otter range in central California, and oil spillage from World War II tanker sinkings off Cape Hatteras. In all cases good agreement between model predictions and observations were shown. Because the model has been designed to treat hypothetical spills, as opposed to spills in real time, these applications of the model are not strictly verifications. The applications do provide corroborating evidence which justifies use of the model. ACKNOWLEDGEMENTS This work was supported by'the US Geological Survey, Bureau of Land Management and Minerals Management Service. The authors wish to acknowledge the contributions of R. Davis, J. R. Slack, R. A. Smith, T. Wyant, K. J. Lanfear, A. Nakassis and R. P. LaBelle to the oil spill modeling efforts. In addition we wish to acknowledge the expert technical assistance-of C. Schoen, D. Hopkins, L. Yost, B. McConville and D. Banks. We also thank 3. Poczik for typing the manuscript. Appreciation is extended to an anonymous referee for helpful suggestions. REFERENCES Allen, A. A. (1969). Hearings before the Subcommittee on Minerals, Materials, and Fluids of the Committee on Interior and Insular Affairs, United States Senate, Washington, D.C., 146-53. Anon. (1969). Santa Barbara oil leak. California Department of Fish and Game Interim Rep., Sacramento (December). 4-68 3: 318 Darid E. Amstutz, William B. Samuels Blumberg, A. F., Herring, H. J., Kantha, L. H. & Mellor, G. L. (1983). California shelf physical oceanography circulation model. First progress Report, Rep. No. 67, Dynalysis of Princeton, Princeton. Campbell, B., Kern, E. & Horn, A. (1977). Impact of oilspillagefrom World War 11 tanker sinkings. D. Rep. No. MITSG 77-4, Massachusetts Institute of Technology, Cambridge. Kantha, L. H., Mellor, G. L. & Blumberg, A. F. (1982). A diagnostic calculation of the general circulation in the South Atlantic Bight. J. Phys. Oceanogr., 12, 805-19. LaBelle, R. P. (1983). Unpublished results. La.Belle, R. P., Samuels, W. B. & Amstutz, D. E. (1982). A comparative analysis of surface water velocity fields off the Southern California coast-- implications for oil spill trajectory modelling. Paper presented at the American Geophysical Union Conference, San Francisco (December). LaBelle, R. P., Lanfear, K. J. & Karpas, R. M. (1983). An oilspill risk analysis for the central California outer continental shelf lease offering (October, 1983), Open-File Rep. US Geological Survey, 83-117. Lanfear, K. J. & Amstutz, D. E. (1981). Environmental studies for oil spill trajectory modelling in the southeastern U.S. outer continental shelf leasing area. Proc. of the 13th Annual Offshore Technology Conference No. 4063, Houston, 487-90. Lanfear, K. J. & Amstutz, D. A. (1983). A reexamination of occurrence rates for accidental oilspills on the U.S. outer continental shelf. Proc. of Eighth Conference on the Prevention, Behavior, Control, and Cleanup o' Oilspills, San Antonio, 355-9. Liu, S. K. & Leendertse, J. J. (1981). A three-dimensional oilspill model with and without ice cover. Proc. International Symposium on Mechanics of Oil Slicks. International Association of Hydraulic Research, Paris, 249-65. Nakassis, A. (1982). Has offshore oil production become safer? Open-File Rep., US Geological Survey, 82-232. Samuels, W. B. & Lanfear, K. J. (1980). An analysis for the South Atlantic (Proposed Sale 56) outer continental shelf lease area. Open-File Rep., US Geological Survey, 80-650. Samuels, W. B., Huang, N. E. & Amstutz, D. A. (1982). An oilspill trajectory analysis model with a variable wind deflection angle. Ocean. Eng., 9, 347-60. Samuels, W. B., LaBelle, R. P. & Amstutz, D. A. (1983). Applications of oilspill trajectory models to the Alaskan outer continental shelf. Ocean Manage., 8, 233-50. Schwartzlose, R. A. & Reid, J. L., Jr. (1972). Nearshore circulation in the California current. California Cooperative Fisheries Investigation Rep. 16, 57-65. Smith, J. E. (Ed.) (1968). Torrey Canyon pollution and marine life. Cambridge University Press, Cambridge. Smith, R. A., Slack, J. R., Wyant, T. & Lanfear, K. J. (1982). The oilspill risk analysis model of the U.S. Geological Survey. Prof. Pap. US Geological Survey, Number 1227. v '~~~~~~~~~~~ ?I.>~ ~ 4-69 Offshore oil spills: analysis of risks 319 Stolzenbach, K. D., Madsen, O. S., Adams, E. E., Pollack, A. M. & Cooper, C. K. (1977). A review and evaluation of basic techniques for predicting the behavior of surface oil slicks. Ralph M. Parsons Laboratory, Rep. No. 222, Massachusetts Institute of Technology, Cambridge. VanBlaricom, G. R. & Jameson, R. J. (1982). Lumber spill in Central California waters: implications for oil spills and sea otters. Science, 215, 1503-5. Williams, R. G. et al. (1980)..4 A climatology and oceanographic analysis of the Califjrnia Pacific outer continental shelf region. US Department of Commerce, NOAA, Washington, DC. Wyant, T. & Smith, R. A. (1978). Risk forecasting for the Argo Merchant spill. Proc. of a Symposium held January 11-13, at the Center for Ocean Management Studies, University of Rhode Island, Providence. 4-70 -3 U List of Publications Related to the Oil Spill Risk Analysis Model of the Minerals Management Service "The Application of Oceanography to Oil-Spill Modeling for the Outer P-ll Continental Shelf Oil and Gas Leasing Program," Robert P. LaBelle and Cheryl M. Anderson, Marine Technology Society Journal, 1985, Vol. 19(2), pp. 19-26. "Analysis of the Sensitivity to Variations in Wind Deflection Angle of Oil Spill Trajectories Modeled Off the California Coast," Robert P. LaBelle, Transactions, American Geophysical Union, EOS 65(45):940, November 6, 1984. "Offshore Oil Spills: Analysis of Risks," David E. Amstutz and William B. P-IO Samuels, Marine Environmental Research, 1984, Vol. 13, pp. 303-319. "!'An Examination of the Argo Merchant Oil Spill Incident Using a Probabilistic Oil Spill Model," Robert P. LaBelle, William B. Samuels, and David E. Amstutz, Presented at the 47th Annual Meeting of the American Society of Limnology and Oceanography, Vancouver, B.C., June 11-14, 1984. "Potential Oil Spill Risks in the Chukchi Sea," William B. Samuels and David E. Amstutz, Presented at the 47th Annual Meeting of the American Society-of Limnology and Oceanography, Vancouver, B.C., June 11-14, 1984. "An Analysis of Surface Winds Off the California Coast," David E. Amstutz, I William B. Samuels, and Kenneth J. Lanfear, Presented at the 47th Annual Meeting of the American Society of Limnology and Oceanography, Vancouver, B.C., June 11-14, 1984. "An Oil-Spill Risk Analysis for the Gulf of Mexico (Proposed Sales 94, 98, 0-33 and 102) Outer Continental Shelf Lease Area," Robert P. LaBelle, Anastase Nakassis, and A. Doreen Lucas, Minerals Management Service OCS Report MMS- 84-0066, April 1984, 172 p. "An Oil-Spill Risk Analysis for the South Atlantic Lease Sale 90," David 0-32 E. Amstutz, William B. Samuels, and A. Doreen Banks, Minerals Management Service OCS Report MMS-84-0037, 1984, 127 p. "Comment on 'Steady Wind- and Wave-Induced Currents in the Open Ocean'," P-9 David E. Amstutz and William B. Samuels, Journal of Physical Oceanography, 1984, Vol. 14(2), pp. 484-485. "An Oilspill Risk Analysis for the St. George Basin (December 1984) and 0-31 North Aleutian Basin (April 1985) Lease Offerings," William B. Samuels, Minerals Management Service OCS Report MMS-84-0004, February 1984, 71 p. "Oilspill Trajectory Modelling in the Beaufort Sea," William B. Samuels, Transactions, American Geophysical Union, EOS 64(52):1049, December 27, 1983. 4-71 "Coupling of Oil Spill Modeling in the Gulf of Mexico with a Method for Ranking Biological Resources," Robert P. LaBelle and J. Kenneth Adams, Transactions, American Geophysical Union, EOS 64(52):1056, December 27, 1983. "Oilspill Trajectories and Beaufort Sea Ice," Richard T. Prentki and William B. Samuels, Transactions, American Geophysical Union, EOS 64(52):1049, December 27, 1983. "Calculations of Seabird Population Recovery from Potential Oilspills in P-8 the Mid-Atlantic Region of the United States," William B. Samuels and Anthony Ladino, Ecological Modelling, 1983, Vol. 21, pp. 63-84. "An Oilspill Risk Analysis for the Diapir Field (June 1984) Lease 0-30 Offering," William B. Samuels, Doreen Banks, Dorothy Hopkins, U.S. Geological Survey Open-File Report 83-570, September 1983, 138 p. "An Oilspill Risk Analysis for the Gulf of Alaska/Cook Inlet Lease Offering 0-29 (October 9lg84)," Robert P. LaBelle, Anastase Nakassis, and Kenneth J. Lanfear, U.S. Geological Survey Open-File Report 83-882, August 1983, 74 p. "Applications of Oilspill Trajectory Models to the Alaskan Outer P-7 Continental Shelf," William B. Samuels, Robert P. LaBelle and David E. Amstutz, Ocean Management, 1983, Vol. 8(3), pp. 233-250. "Small Scale Oil Slick Modeling in the Santa Barbara Channel California," P-6 Kenneth J. Lanfear and William B. Samuels, In: Waste Disposal in the Oceans: Issues for the 80's (D.F. Soule & D. Walsh, Eds.), Westview Press, Boulder, Colo., 1983, pp. 63-75. "An Oilspill Risk Analysis for the Southern California Lease Offering 0-28 (February 1984)," Robert P. LaBelle, Kenneth J. Lanfear, A. Doreen Banks, and Robert M. Karpas, U.S. Geological Survey Open-File Report 83-563, July 1983, 115 p. "Simulation of Spreading and Weathering of Oilspills in the Southern California Bight," William B. Samuels, Presented at the 46th Annual Meeting of the American Society of Limnology and Oceanography, St. Johns, Newfoundland, June 13-16, 1983. "An Oilspill Risk Analysis for the North Atlantic (February 1984) Lease 0-27 Offering," David E. Amstutz and William B. Samuels, U.S. Geological Survey Open-File Report 83-567, May 1983, 177 p. "An Oilspill Risk Analysis for the Navarin Basin Lease Offering (March 0-26 1984)," William 8. Samuels, Kenneth J. Lanfear, and Doreen Banks, U.S. Geological Survey Open-File Report 83-120, April 1983, 34 p. "An Oilspill Risk Analysis for the Central California Outer Continental 0-25 Shelf Lease Offering (October 1983)," Robert P. LaBelle, Kenneth J. Lanfear, and Robert M. Karpas, U.S. Geological Survey Open-File Report 83- 117, March 1983, 67 p. 4-72 - "An Oilspill Risk Analysis for the Central Gulf (April 1984) and 0-241 Western Gulf of Mexico (July 1984) Lease Offerings," Robert P. LaBelle, U.S. Geological Survey Open-File Report 83-119, February 1983, 63 p. "A Reexamination of Occurrence Rates for Accidental Oilspills on the P-5 U.S. Outer Continental Shelf," Kenneth J. Lanfear and David E. Amstutz, Proceedings of the Eighth Conference on the Prevention, Behavior, Control, and Cleanup of Oil Spills, American Petroleum Institute Publ. No. 4356, 1983, pp. 355-359. "The Relative Contributions of Local Wind Drift and Geostrophic Surface Currents to the Movement of Potential Oilspills on the U.S. Pacific Outer Continental Shelf," William B. Samuels, Robert P. LaBelle, and David E. m Amstutz, Transactions, American Geophysical Union, EOS 63(45):1003, November 9, 1982. "A Comparative Analysis of Surface Water Velocity Fields off the Southern California Coast - Implications for Oil Spill Trajectory Modeling," Robert P. LaBelle, William 8. Samuels, and David E. Amstutz, Transactions, American Geophysical Union, EOS 63(45):1003, November 9, 1982. "An Oilspill Risk Analysis for the South Altantic (Proposed Sale 78) Outer 0-23 Continental Shelf Lease Area," William B. Samuels, U.S. Geological Survey Open File Report 82-807, September, 1982, 161 p. "An Application of a Vulnerability Index to Oil Spill Modeling in the Gulf P-4 I of Mexico," Robert P. LaBelle, Gail Rainey, and Kenneth J. Lanfear, U.S. Geological Survey Open File Report 82-804, August 1982, 14 p. "An Oilspill Risk Analysis for the Gulf of Mexico Outer Continental Shelf 0-22 Lease Area Regional Environmental Impact Statement," Robert P.LaBelle, U.S. Geological Survey Open File Report 82-238, April 1982, 210 p. "An Oilspill Risk Analysis for the Mid-Atlantic (Proposed Sale 76) Outer 0-21 Continental Shelf Lease Area," William B. Samuels and Dorothy Hopkins, U.S. Geological Survey Open File Report 82-27, April 1982, 163 p. "Has Offshore Oil Production become Safer?," Anastase Nakassis, U.S. P-3 I Geological Survey Open File Report 82-232, April 1982, 26 p. "An Oilspill Trajectory Analysis Model with a Variable Deflection Aigle," 0-20 William B. Samuels, Norden E. Huang and David E. Amstutz, Oce-ni Engineering, 1982, Vol. 9, No. 4, pp. 347-360. "Oilspill Trajectory Modeling in Alaskan Waters," William 3. c uels, Robert P. LaBelle and David E. Amstutz. Ocean Sciences AGU/ASLC Joint Meeting, 16 - 19 February, 1982, San Antonio, Texas. "An Oilspill Risk Analysis for the Beaufort Sea, Alaska (Proposed Sale 71) 0-19 Outer Continental Shelf Lease Area," William B. Samuels, Oorothy Hopkins and Kenneth J. Lanfear, U.S. Geological Survey Open File Report 82-13, January 1982, 102 p. 4-73 "The Oilspill Risk Analysis Model of the U.S. Geological Survey,'t Richard P-2 A. Smith, James R. Slack, Timothy Wyant and Kenneth J. Lanfear, U.S. Geological Survey Professional Paper 1227, 1982, 40 p. "Simulations of Seabird Damage and Recovery from Oilspills in the Northern P-1 Gulf of Alaska," William B. Samuels and Kenneth J. Lanfear, Journal of Environmental Management, 1982, Vol. 15, pp. 169-182. "An Oilspill Risk Analysis for the North Atlantic (Proposed Sale 52) 0-18 Outer Continental Shelf Lease Area," Robert P. LaBelle, U.S. Geological Survey Open-File Report 81-865, September, 1981, 120 p. "An Oilspill Risk Analysis for the St. George Basin, Alaska, (Proposed Sale 0-17 70) Outer Continental Shelf Lease Area," William B. Samuels and Dorothy Hopkins, U.S. Geological Survey Open-File Report 81-864, 1981, 89 p. "An Oilspill Risk Analysis for the Southern California (Proposed Sale 0-16 68) Outer Continental Shelf Lease Area," William B. Samuels, Dorothy Hopkins and Kenneth J. Lanfear, U.S. Geological Survey Open-File Report 81- 605, May, 1981, 206 p. "Documentation and User's Guide to the U.S. Geological Survey Oilspill Risk D-2 Analysis Model: Oilspill Trajectories and Calculation of Conditional Probabilities," Kenneth J. Lanfear and William B. Samuels, U.S. Geological Survey Open-File Report 81-316, May 29, 1981, 95 p. *~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ "An Oilspill Risk Analysis for the Norton Sound, Alaska, Proposed Sale 57) 0-15 Outer Continental Shelf Lease Area," William B. Samuels and Kenneth J. Lanfear, U.S. Geological Survey Open-File Report 81-320, May 29, 1931, 114 p. "Sensitivity of an Oilspill Trajectory Analysis Model to Variation in Wind Deflection Angle," William 8. Samuels, Norden Huang and David E. Amstutz. Presented at the 44th annual meeting of the American Society of Limnology and Oceanography, Milwaukee, Wisconsin, June 15-18, 1981. "Environmental Studies for Oilspill Trajectory Modeling in the Southeastern U.S. Outer Continental Shelf Lease Area," Kenneth J. Lanfear and David E. Amstutz. Presented at the 13th Annual Offshore Technology Conference, Houston, Texas, May 4-7, 1981. "An Oilspill Risk Analysis for the Gulf of Mexico (Proposed Sales 67 and 0-14 69) Outer Continental Shelf Lease Area," Robert P. LaBelle and Kenneth J. Lanfear, U.S. Geological Survey Open-File Report 81-306, 1980, 118 p. "An Oilspill Risk Analysis for the Mid-Atlantic (Proposed Sale 59) Outer 0-13 Continental Shelf Lease Area," William B. Samuels and Kenneth J. Lanfear, U.S. Geological Survey Open-File Report 80-2026, October, 1980, 82 p. "Documentation and User's Guide to the Spatial Environmental Data 3- Digitizing System, SEDOS," Kenneth J. Lanfear and Anastase Nakassis, U.S. Geological Survey Open-File Report 80-871, August, 1980, 129 p. L) U4-74 3 U "An Oilspill Risk Analysis for the Cook Inlet and Shelikof Strait 0-12 (Proposed Sale 60) Outer Continental Shelf Lease Area," Robert P. LaBelle, William B. Samuels and Kenneth J. Lanfear, Open-File Report 80-863, August, 1980, 129 p. "An Oilspill Risk Analysis for the South Atlantic (Proposed Sale 56) Outer 0-11 Continental Shelf Lease Area," William B. Samuels and Kenneth J. Lanfear, U.S. Geological Survey Open-File Report 80-650, May, 1980, 76 p. "An Oilspill Risk Analysis for the Central and Northern California 0-10 (Proposed Sale 53) Outer Continental Shelf Lease Area," William B. Samuels and Kenneth J. Lanfear, U.S. Geological Survey Open-File Report 80- 211, January, 1980, 44 p. "An Oilspill Risk Analysis for the Kodiak Island (Proposed Sale 46) Outer 0-9 Continental Shelf Lease Area," William B. Samuels, Kenneth J. Lanfear and Anastase Nakassis, U.S. Geological Survey Open-File Report 80-175, January, 1980, 92 p. "Applications of the USGS Oilspill Trajectory Analysis (OSTA) Model to Decisions Regarding OCS Oil Development," Kenneth J. Lanfear, Proc. of the Workshop on Government Oilspill Modeling, Wallops Island, Virginia, November 7-9, 1979, pp. 13-14. "The USGS Oilspill Trajectory Analysis Model," William B. Samuels, Proc. of I the Workshop on Government Oilspill Modeling, Wallops Island, Virginia, November 7-9, 1979, 20 p. "An Oilspill Risk Analysis for the Northern Gulf of Alaska (Proposed Sale 0-8 55) Outer Continental Shelf Lease Area," Kenneth J. Lanfear, Anastase Nakassis, William B. Samuels and Christina T. Schoen, U.S. Geological Survey Open-File Report 79-1284, July, 1979, 79 p. "Oilspill Risk Minimization Through Optimal Tract Selection," Richard A. Smith, Kenneth J. Lanfear and Ivan C. James, II, presented at the National Weather Service Conference, "Physical Behavior of Oil in the Marine Environment," at Princeton University, May 8-9, 1979. "An Introduction to the Oilspill Risk Analysis Model," K.J. Lanfear, R.A. Smith and J.R. Slack, presented at the 11th Annual Offshore Technology Conference, Houston, Texas, April 30 - May 3, 1979. "An Oilspill Risk Analysis for the Southern California (Proposed Sale 48) 0-7 Outer Continental Shelf Lease Area," James R. Slack, Timothy Wyant and Kenneth J. Lanfear, U.S. Geological Survey Water-Resources Investigation 78-80, October, 1978. "An Oilspill Risk Analysis for the Mid-Atlantic (Proposed Sale 49) Outer 0-6 Continental Shelf Lease Area," James R. Slack and Timothy Wyant,-U.S. Geological Survey Water-Resources Investigations 78-56; 1978; 79 p. I.:/ 4-75 "An Oilspill Risk Analysis for the Eastern Gulf of Mexico (Proposed Sale 0-5 65) Outer Continental Shelf Lease Area," James R. Slack and Timothy Wyant, U.S. Geological Survey Open-File Report 78-132; 1978; 72 p. "An Oilspill Risk Analysis for the Western Gulf of Alaska (Kodiak Island) 0-4 Outer Continental Shelf Lease Area," James R. Slack, Richard A. Smith-and Timothy Wyant; U.S. Geological Survey Open-File Report 77-212, 1977, 57 p. "An Oilspill Risk Analysis for the South Atlantic Outer Continental Shelf 0-3 Lease Area," James R. Slack and Richard A. Smith, U.S. Geological Survey Open-File Report 76=653, 1976, 54 p. "An Oilspill Risk Analysis for the Mid-Atlantic Outer Continental Shelf 0-2 Lease Area," Richard A.e Smith, James R. Slack and Robert K. Davis; U.S. Geological Survey Open-File Report 76-451, 1976a, 24 p. "An Oilspill Risk Analysis for the North-Atlantic Outer Continental Shelf 0-1 Lease Area," Richard A. Smith, James R. Slack and Robert K. Davis, U.S. Geological Survey Open-File Report 76-620, 1976b, 25 p. I I I I ~~~~~~~~~APPENDIX 5 DISCUSSION OF' FACTORS INFLUENCING MARINE BIRD MORTALITY I COUNTS RES.ULTING FROM 'THE T/V PUERTO RICAN OIL SPILL INCIDENT I I I I I I I I I I I I I I Appendix 5: Marine bird mortality resulting from oil spilled by T/V Puerto Rican Of the many adverse environmental impacts that resulted from the T/V Puerto Rican oil spill incident, oiled dead birds observed during beach surveys were among the most visible. In their assessment of bird mortality caused by the oil spill, the Point Reyes Bird Observatory reported that about 1,000 oiled birds died during the period 6-19 November 1984.1 The direct effects of oil contamination on populations of birds are determined by counting and examining oiled birds. Thus estimates of mortality and of the effects of sublethal doses of oil on the ability to see, and collect affected individuals. The probability of detecting affects individuals is a function of many environmental, behavorial, and physiological factors, and as a results of oil contamination are more likely to be underestimated for scme species and populations than others. For example, different species exhibit differing degrees of buoyancy subsequent to having their feathers coated with oil. Those species that remain relatively buoyant after exposure to oil, such as qulls, are more likely to remain afloat and wash ashore than are species that are relatively non-buoyant, such as the alcids (auks, murres, and puffins). The species that are less buoyant after exposure to oil are mrove likely to sink rather than wash ashore, thus they are less likely to be counted in beach surveys, and the number of birds of that species affected by oil will probably be underestimated. Therefore, the probability that the effects of the oil spill are underestimated increases from one group to the next in the following list of birds, ranked in approximate order of decreasing gulls, shearwaters, northern fulmars, grebes, loons, alcids (auks, murres, puffins), and cormorants.2 I1 Point Reyes Bird Observatory, 1985 5-2 Probability of detection also decreases as the distance offshore increases for the location where the bird is exposed to oil. Birds that inhabit areas further offshore and which are farther from the beach at the time of expoosure are more likely to sink or suffer predation than those birds that are oiled relatively close to shore. Thus the probability of underestimating the number of individuals adversely affected by oil increases for the following list of birds, ranked in approximate order of increasing distance offshore: coots, grebes, ducks, phalarlopes, alcids, scoters, fulmrs, loons.3 Some species tend to cane ashore soon after being exposed to even minimal amounts of oil, while other species treavel relatively great distances from the site before finally coming ashore. For example, a murre with only a very small patch of oil on its plumage will sometimes ccae ashore, while a-gull will tend to fly some distance, if able, before making its way to the beach.4 Thus the bird's behavior affects the probability that estimates of ill affects due to oil may be too low. The probabilityof underestimating the number of individuals affected increases from one group to the next in the following list of birds ranked in approximate order of tendency to disperse from the exposure site prior to coming ashore: alcids, grebes, gulls, fulmars, loons. Other biological factors that affect the determination of the effects of an oil spill on bird populations include the overall state of health of the population, the response of individuals to sub-lethal doses of oil, and the degree of contact with oil. For example, 362 scoters and all 19 northern fulmars counted in the initial beach surveys were subtracted from the total number of mortalities caused by the Puerto Rican on spill. This was done 2. Glenn Ford, Pers. Comm., 1985 3. Sterling Hobe Corporation, 1985 4. ibid. I ~~~~~~~~~~~~5-3 I ~because many of these birds were found in poor health prior to the oil I ~spill incident, thus their deaths could not be attributed strictly to exposure to oil. Sane birds, especially gulls, acquired sublethal doses I ~or oil. These birds were not systematically recorded during the oil spill incident and could not be included in the analysis. Environmental factors prevented some affected birds from being I ~included in the total number oiled. Wind direction and speed and current direction and speed acted in conjunction with biological factos (i.e. I ~response to sublethal doses, degree of contact with oil, degree of oil saturation, distance from shore, bouyancyl scavenging of carcasses prior to reaching shore or being counted on the beach, and behavior) to reduce I ~the probability that affected individuals were detected and included in the total number of mortalities. Furthermore, not all areas where I ~carcasses could have washed ashore were systematically surveyed, due to constraints of beach access, time, and personnel. For example, the beaches north of Bodega Head were not included in the survey, although there were sanme occasional visits, and it is likely that some bird carcasses did drift northward and come ashore in these areas. I ~~Thus despite the fact that a total of 378 dead birds was subtracted from the initial beach survey count of total mortalities, the factors discussed above suggest that the numbers in Figure 17 underestimate the total of birds adversely affected by the oil spill of the T/V Puerto Rican. The following is a brief discussion of the species most likely most likely to I ~have been underestimated by the beached bird counts. 5-4 Loons. None of the three loon species normally come ashore, beaching only when sick or oiled. Because Arctic Loons remain further offshore than Common or Red-throated loons, they would be less likely to reach land, and thus an underestimation of this loon species mortality is more likely on beach counts. The buoyancy of all loons is probably low compared to gulls and shearwaters, thus more oiled loons would sink and remain unreported. During the spill, large numbers of Arctic Loons were migrating south. Individuals that landed on the oily water may have become lightly to moderately oiled before moving on. Such birds might have flown a considerable distance before becoming incapacitated, and would have remained unrecorded. Thus the number of Arctic Loons reported as oiled in Figure 17 is likely underestimated to a greater degree than the numbers recorded for Ccmmon and Red-throated Loons. Western Grebes. Western Grebes are a near-shore species that swim ashore when they are lightly to moderately oiled. They are likely less buyoyant than alcids and possibly loons. Because of their proximity to shore, most, if not all, of the lightly to moderately oiled individuals probably beached; however, a large number were already dead and heavily oiled when found. It is likely they died before reaching shore. We expect sane additional oiled Western Grebes sank and never rached shore, but cannot estimate the numbers of these birds. Northern Fulmars. Fulmars tend to occur further out at sea than grebes and scoters and do not swim ashore in numbers when lightly to moderately oiled. They ar probably more buoyant than loons, grebes, or alcids. Because large numbers appeared to be dying of non-oil-related causes during the spill, and because most carcasses had relatively small amounts of oil on them (when oil could be found at all), we were unable to assign an oil-related impact for this species. It is likely that the impact on northern fulmars was minimal compared to the impact on common murres. Cormorants. There is little evidence of an impact on cormorants. Figure 17 indicates 7 dead of two species. In addition, observers on the Farallones reported seeing 8 oiled Brandt's Cormorants and 10 oiled Pelagic Cormorants. Aerial surveys reported in the original PRBO report indicate cormorants were seen over both the inner and outer shelf. Given the distances cormorants range offshore, and their relatively poor buoyancy, it can be assumed the numbers shown in Figure 17 substantially underestimate the casualties. Brown Pelican. Despite a census total of 2,583 pelicans at the Farallon Islands roosts on 6 November 1985, the day the oil first collided with the island shoreline, and subsequent polican counts up to 1,000 until 13 November, only 2 pelicans with oiled plumage were found in the roots. Rehabilitation centers reported receiving only 1 oiled pelican for treatment during the spill period. Thus, in light of this information, it appears that pelicans suffered little impact from the spill. Waterfowl and Coots. Coots and the species of ducks shown in Figure 17 reside close to shore. Most of the waterfowl that were oiled were recovered in the Bodega Harbor region where the recovery of beached oiled birds was was very complete. Therefore, we suspect the numbers reported in Figure 17 are close to the actual numbers of birds affected by the spill. Red Phalaropes. Aerial surveys showed phalaropes migrating over a broad front between the shore and the continental slope. The 3 oiled Red Phalarope carcasses found and reported in Figure 17 probably greatly under- estimate mortality of this species. There are two reasons for this: 5-6 phalaropes dying far from shore are likely to be scavenged before beaching and after beaching they are difficult to detect because of their small size. Gulls. Figure 17 reports only 5 Western Gull casualties although obvservers on the Farallon Islands reported seeing 60 oiled Western and 2 oiled California gulls. Many additional oiled gulls were seen between Bodega Harbor and Point Arena (Page, pers. obs.). Gulls probably disperse farther fram the place of oil contact than other species and are therefore less likely to be found. However, gull carcasses are highly buoyant, and live gulls tend to move to shore. Many moe carcasses would have been found had large numbers of gulls died from their initial contact with the oil. Because gulls may come ashore and preen the oil fram their plumage, we suspect many oiled gulls survived and therefore, assume a limited effect on gulls from the spill. Alcids. Comnon Murre and Cassin's Auklet mortalities from the spill may be the most underestimated of any species shown in Figure 17. Both species are widely distributed from the inner continental shelf seaward to the continental slope. Carcass buoyancy of these birds is low compared to gulls or shearwaters. The alcids seem to be detrimentally affected by even small amounts of oil on their plumage; a murre with only a half-dollar sized patch of oil on its breast will sometimes come ashore. Rhinoceros Auklets are much less abundant than the other two species and after an absence of 100 years, the Rhinoceros Auklet has gradually been re-establishing a breeding population on the Farallon Islands. The population was estimated at 300 birds prior to the spill. The level of reduction from the spill may beccme apparent in the 1985 breeding season. Until then, the impact on this species will remain unknown. Only 4 Rhinoceros Auklet casualties were reported. 5-7 The other species of alcids listed as casualties (Marbled Murrelet and Ancient Murrelet) are scarce in the Farallon Islands area and the limited casualty estimates reported in Figure 17 reflect this scarcity. In summary, the estimates given in Figure 17 indicate that scoters, Ccmmon Murres, Cassin's Auklet, Western grebes, and loons were the most severely affected species in the T/V Puerto Rican incident. Of these species, three were included in the group of species selected for computer modeling of the initial spill incident: common murres, Arctic loons, and Cassin's Auklet. Computer modeling of the chronic release of oil from the sunken stern was done for: Common Murres, Cassin's Auklet, Western Gulls, and the Brown Pelican. The computer model was used to derive an estimate of the total number of bird mortalities resulting from the initial spill of 35,000 barrels of oil and the chronic release of the 8,500 barrels of bunker fuel oil in the sunken stern. I I I I ~~~~~~~~~APPENDIX 6 1 ~~~ DAMAGE TO PLANKTCN CAUSED BY THE T/V PUERTIO RICAN INCIDENT I I I I I I I I I I I I I I 6-1 Appendix 6: Damage caused to the plankton of the Gulf of the Farallones by the T/V Puerto Rican oil spills with emphasis on the most economically valuable type such as Dungeness crab larvae (prepared by A. Horne, 1986). Laboratory studies mimicing the effects of the November spill and the year long diesel fuel leak (Horne et. al. 1985) provide a firm scientific basis for estimating the damage to the ocean biota. Literature and field observations provide the other data for quantifying the number of organisms killed or otherwise damaged. The main assumptions are as follows: O The toxic water-soluble fraction (WSF) of oil separates frcm the surface slick within two days. o Vertical diffusion of the WSF (in fact a mixture of tiny droplet suspension and true solution) is restricted by droplet buoyancy and is much less than would be predicted from vertical eddy diffusion in the ocean. 0 Horizontal eddy diffusion can be predicted from normal ocean coefficients. o The numbers of dungeness crab zoeae (young stages) can be estimated from sampling near the time of the spill and from long-term averages for the Gulf of the Farallones. o The oil concentration measured by Woodward-Clyde two days after the spill is typical of actual values. Alkylbenzene (Alkane 60) levels measured in the afternoon of Novermber 5, 1984 showed a 0-11 m water column to contain 87 ppb (total alkanls). Alkylbenzenes comprised about 40% of spilled so the toxic WFS "disolved" 2 to 3 days after the spill was 218 ppb. All individual oil levels (Alkylbenzene, OLOA, WITCO) at 40-80 ppb and mixtures of all three oils at 200 ppb showed severe and statistically significant mortality in the most sensitive and economically most valuable organisms - shrimp and crab larvae - within 8 days and possibly sooner. Also very large and statistically significant declines had occurred at 6-2 these low oil levels for the most abundant zooplankton group - the nauplii or young stages. Mortality became obvious between 4 and 8 days after the spill, became severe after 10 days and literature review shows that these effects are not necessarily reversible. The Woodward-Clyde data on alkane concentrations shows that the WSF was underdispersed by about a factor of 60. This assumes a vertical eddy diffusity of 100 cm2 sec -1 and is probably due to the slight buoyancy of the minute droplets. Although not part of the outface slick they also do not behave like a true solution. Horizontal expansion of the toxic WFS can be modelled by assuming a patch of water at the end of day 2 (all WSF separated from surface slick). Assuming a conservative mixing depth of 12 m by extrapolating the Woodward-Clyde data the patch can be -expanded using a horizontal eddy diffusion coefficient of 104 cm2 sec-1. Oil droplet buoyancy will not affect horizontal motion. The results can be used to dilute the measured 218 ppb total oil present at day 2-3. Two end points are reached. First after 7-8 days 40 ppb WSF remains in a volume of 39 km3 (12 m deep). This is close to the levels and exposure times used in laboratory toxicity tests. Second, after 48 days, 1 ppb WSF remains in a volume of 3;00 km2 (25 m deep since the ocean bed restricts further mixing). The ppb level is likely near the minimum which will kill substantial amounts of the plankton after 10- 20 days exposure as determined by extrapolation of the bunker oil studies. A comparison of bunker oil and alkylbenzene, witco and oloa shows that bunker oil is about twice as toxic (determined from median survival times). Mortality was detected with bunker oil at 0.4 ppb after 10 days (WSF) and 6-3 I assume that this is about 50% of the measurable minimum for the other oils (see above). Laboratory studies using the entire natural plankton community show that for the first 7-8 days about two-thirds of the most sensitive plankton (including dungeness crab larvae) will die. These studies also show that about half of all 300 plankton (including dungeness crab larvae) will die in the next forty days. After this lingering mortality and non-lethal damage will persist but is not considered here. Measured crab densities in winter 1985 indicate that all larval crabs were present at a density of 0.1 per liter. Studies over several years by the California Department of Fish and Game (Hatfield et al., 1983; Wild and Tasto (eds.) 1983) show that 2.3% of crab zoea are dungeness crabs. Thus, the density of dungeness crabs affected by the winter 1984 oil spill was 0.0023 per liter (2.3 x109 km3). For the shorter, higher oil exposure (7-8 days, >1 ppb) 39 x 2.3 x109 dungeness crab larvae were present in 39 km3 of water. 2/3 of these were killed (= 6 x 1010 crabs). Assuming the most conservative survival rate for a stable population, 2 out of 106 will become adults. Thus 1.2 x 105 adults are available. Assuming in this well-fished water area (San Francisco - Bodega) 10% are harvested (a conservative assumption), 1.2 x 104 or 12,000 harvestable adult crabs were potentially killed by the Puerto Rican oil spill. For the longer, lower oil exposure (48 days > lppb) 3,700 x 2.3 x 109 dungeness crab larvae were present in 3,700 km3 of water 50% of these were killed (=4.3 x 1012 crab larvae). Assuming (as above) 6-4 that only 2 out of 106 become adults. This leaves 0.9 x 107 adults available. Assuming a very conservative 1% harvest of the crop (in the larger, less populated Bodega-Crescent City coastline) this suggest that 90,000 potentially harvestable adult crabs were lost by the oil spill. The annual catch for the SF-Bodega fishery is about 106 of crab and that of the Bodega-Crescent City about 107 lbs. Thus, the predicted losses represent a small fraction (0.2-4%) of the annual yield. Those losses due to oil would be thus realistic. Other Organisms Dungeness crab larvae, although economically valuable, were among the least commnon planktonic species. Shrimps were 98 times more ccmmon and are bought for aquarium food. Small Zooplankton form the food of all ocean water organisms, including young fish of commercial and recreational value. These species were much more numerous and were also killed by the oil. In addition phytoplankton, the "grass of the sea" was destroyed to a large extent by the higher oil levels used experimentally. No estimate of these losses is given but can easily be calculated using the above procedures. t~~~~~~~~ Bunker-Fuel Oil Spills The loss of dungeness crabs and other organisms by fuel oil can be approximated in the same way as was used for the November 1985 oils. Unlike the catastrophic losses in winter, the bunker oil leaked slowly using the method described above the volumne of toxic WSF can be estimated and the toxic concentration found. Assuming a loss of 8,500 barrels of oil (WSF = 1.3 mn3) per year, a mixed depth of 10m and a measured average 6-5 slick width of 81 m (G. Ford, per. ccmm.), and an average advective water flow of 10 anm sec -1 the WSF of bunker oil will be about 0.6 ppb. This is greater than the level shown to kill zooplankton after 10 days exposure. However, assuming a horizontal eddy diffusion of 104 cm2 sec-1 no organisms will be exposed to toxic fuel oil level for long enough to be certain that a kill will occur. Thus, the dispersion of the toxic WSF from the sunken stern of the Puerto Rican will not produce demonstrable mortality in the zooplankton or phytoplankton. I I I APPENDIX 7 I I I I) I I I I I I I I I I I I I I I I BIOLOGICAL HINDCAST MODELING AND SPILL TRAJECTOlRY ANALYSIS OF TH~E TAV PUERTO RICAN. OIL SPILL I I I I I I I I I I I I I I 7-1 BIOLOGICAL HINDCAST OF THE EFFECTS OF THE T/V PUERTO RICAN OIL SPILL ON THREE SEABIRD SPECIES R. Glenn Ford ECOLOGICAL CONSULTING 2735 N.E. Weidler Street Portland, Oregon 97232 September 1985 7-21 INTRODUCTI ON Following the explosion and breakup of the tankship PuertoI Rican,, numbers of oiled seabirds were recovered along the coast of central California and the Farallon islands. It is virtually certain that these beached birds represented only a part of theU total mortality resulting from the spill (see discussion of actual versus observed mortalities in previous section). Dead or dying birds in unknown numbers were probably lost at sea orU beached on inaccessible or unsearched sections of coastline. A biological model originally designed for analyzing oil spill risks to seabirds was adapted to carry out a "hindcast" of the actual spill events (Ecological Consulting, 1985). This model was coupled with a new model which used trajectory simulations of the movement of oiled birds to predict the fate of the seabirds which had encountered the slick. The modelling effort focussed onI three, species for which observed mortalities were known to under- estimate total mortality. These species were: (1) Common Murre, (2) Arctic Loon, and (3) Cassin's Auklet. The methodology and results of the analysis are described below. MODEL DESCRIPTION AND INPUT DATA SUMMARY The model for estimating bird mortality is divided into three sections:U 1) Oil Slick Movement Model: A detailed description of the history of the slick including the location, size, shape, and. percent coverage at different points in time. Descriptions of the slick are derived either from observational data, or from hindcasts made using HAZMAT's On Scene Spill Model (OSSM). 2) Bird Contact model: A simulation of bird distribution' and movement in the area of and around the slick. This section5 estimates the number of birds which actually encountered the oil. 3) Bird Fate Model: A model which tracks the fate of birds which encountered the slick. These birds may or may not becomeU seriously oiled. Of those which were, some sank or were sca- venged, some were carried out to sea, some were washed ashore on beaches which were searched, and some were was hed ashore onI beaches which were not searched. The output of the f ate section of the model provides esti-3 mates of the number of birds which died but were not counted. The overall model structure is diagrammed in Figure 1. The fol- lowing are more detailed accounts of the three sections of the3 model1. 21 0I) ~ 7-3 oil Slick I Observations OVEMENT MODEL I~~~~~~~~~~~~~~~~~~~~~~~ I ~ ~ ~ Digital Maps of Slick Movement *Behavioral Data BIRD CON Distributional Data Locations and Numbers of Birds I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Encountering Slick e Bird Trajectory Model IRD FATE *At Sea Loss Rates * Beached Bird Data MODEL Estimates of Total Mortality Figure 1. Flow diagram showing the relationship among the three sections of the model used in this analysis. Ovals indicate programs, rectangles indicate input or out- put data. 3 0 ~~~7-4I Oil Slick Movement Model A detailed description of the location, shape, and movement of an oil slick is required in order to estimate the number of birds encountering spilled oil. This description was obtained using maps of the observed slick at various points in time andI interpolating between successive observations to describe the behavior of the slick during intervening time periods. Description of oil Slick Data Observations -of the position of the slick from the T/V3 Puerto Rican were summarized by James Dobbins Associates (see Appendix C). The observations consisted of the position and shape of the slick at each point in time. Twenty-one such obser- vations were used, spanning the period from the vessel breakup onU 6 November, to the last sighting of a slick north of Bodega Head on 12 Novemrber. These data were input to the model as sets of ordered pairs of latitudes and longitudes describing the shapeU and position of the slick at a given time. in addition to these data, we also used estimates of the percentage of the slick which was actually covered by oil -- since typically so-called slicks . contain large areas of open water, and these open areas increase in proportion with the age of the slick. Estimates of coverage were provided by the MAS group of NOAA, Seattle.3 An alternative approach was als~o used based on a HAZMAT hindcast of the spill using the On Scene Spill Model, OSSM. The hindcast was based on real time wind data collected during theU period of 6-11 November at weather stations in and around the Gulf of the Farallones. For this analysis, OSSM was conditioned to include the response of the slick to the shelf-edge jetU which occurred on 5 and'6 November, causing the slick to reverse direction and move rapidly northward toward the Farallon islands. The results of the HAZMAT hindcast were recorded at 6 or 24 hour3 intervals depending on the rate of movement of the slick. These results consisted of estimates of the shape and position of the slick at each point in time, and were input to the bird contact3 model in the same way as the real observations. The results of the hindcast are generally very similar to the observations. There are two ways in which the observed and hindcast slicks differ. (1) The hindcast shows the main body of the slick remaining east of the 200 meter depth line (which defines the edge of the continental shelf) prior to the reversalI of 5 and 6 November. The observations show the main body moving west of the shelf break prior to this time. (2) The sizes of the slicks predicted from the HAZMAT results are about 66% higher3 than those observed. it is unclear whether the hindcast or the observations are more accurate in this case, since some portions of the actual slick probably went unreported.5 41 0D 7-5 Interpolation of Slick Movement In order to completely describe the path of the slick, it was necessary to "fill in" the parts of the path between observa- tions. This entailed pairing each observed piece of the slick with a piece observed at a later time and interpolating between the two irregular shapes. Paired observations of the shape of the slick were input in the form of two sets of ordered latitudes and longitudes defining two closed polygons. Lines forming all possible connections between the two sets of vertices were cons- tructed, and the two lines containing no intersections with other lines were chosen to represent the boundaries of the path of the slick during the period between the observations. If an oil slick was observed at times T-1 and T-2, the area swept out during the period T-1 to T-2 was defined as the area between the two non- intersected lines and the edges of the two polygons formed by the slick at T-l1 and T-2 plus the area of the slick at time T-2. The slick area at time T-1 was not included in the estimate because its area would have been included in calculating the area swept out between times T-0 and T-1. The method used to estimate the area swept out by the slick is illustrated in Figure 2. A map showing the passage of the slick is shown in Figure 3. I ~~CK AT SLICK AT between times T-1 and T-2. The new area affected c" INTERPOLATED "/''�"~ ; acREGION -.".."ii SLICK AT " TIME T-I Figure 2. The interpolation of the passage of an oil slick between times T-l and T-2. The new area affected during the period T-l to T-2 would be the stipled region plus the crosshatched region. o . 7-6 ,]/ ' ' Fort Bragg 39' --- ---- t.Arena . e . ....... . .. .. ....... * ~ ~ I I ISnaCu � ; r 124 123 Figure 3. interpolated passage of the oill slickebased on ober ~~~Factalslic bevtonIs.. Stp% areas.-, are antesro p~~ - oltd ar eas hr h lc sasmd to av reported cntinuing*northward pa st Pt. Aren a 37' 124� ~~~~~~Fagraeo 3. I Snterpltdasageofthcislckbsdonosr reprte c o t n i n 2othadpat. Arna 7-7 Bird Contact Model We assume that birds resting or feeding on the water within the region where the passage of the slick was observed or where Ithe passage of the slick was interpolated were at risk of oil contact. A "risked" bird, however, does not necessarily become oiled for several reasons. Areas reported as being part of an Ioil slick are generally only partially covered, birds may delib- erately avoid oil contact, or the oiling of a bird may be so light as to have little immediate consequence. Estimates of I mortality are based on the assumption that some percentage of the birds on the water in the region of the slick will become oiled sufficiently that they will become debilitated and die.. The method used to estimate this percentage is described in Model UCalibration. Distributional and Behavioral Data U ~~Bird density estimates were derived f rom data collected by personnel of Point Reyes Bird Observatory (PRBO) on 14 and 17 I November 1984 (PRBO 1984). Bird counts were made from a fixed wing aircraft f lying at 185 km/hr at an altitude of 65 m. Ob- servers scanned a 50 m wide corridor on the glare-free side of I the aircraft and recorded the species (or lowest recognizable taxon), the behavior, and the numbers of all birds within the strip. On 14 November eight east-west transects totalling about U610 km were flown, extending from 37 degrees 30 minutes to 39 degrees 5 minutes north latitude. on 17 November six transects totalling about 510 km were flown, extending from 37 degrees 20 minutes to 38 degrees 25 minutes north latitude. I ~~Three biological zones were identified as being potentially different in terms of bird densities (K. Briggs pers. comm.). IBiological zones and transect lines are shown in Figure 4. The zones were defined as: North Shelf: That area between 38'N and 39"N, west of the California coast and east of the 200 meter depth line. This corresponds generally to the area of the continental shelf north of Pt. Reyes and south of Point Arena. I ~~Central Shelf: That area between 37 015'N and 380N, west of the California coast and east of the 200 meter depth line. This Icorresponds generally to the area of the continental shelf north of Bean Hollow State Beach and south of Pt. Reyes. Pelagic: That area south of 39"N and north of 370 15'N and I eastward from the 200 m depth line to the limits of our transect data at about the 2,000 m depth line. mean dens~ity of groups of birds was calculated separately for each spe c~ies in each biological zone and was expressed as I groups per km . Mean group size and the distribution of group sizes was determined separately for each species. Group size I ~~~~~~~~~~~~7 9S ~ 7-8 ) ' I Fort Bragg I I 39* -- - --- Pt. Arena I ?T~~~~~~ .. .eel ?5" I U, :Transect 52 Transect 49 odega Head *NORTH SHELF -4 - - - - - --- I T .ansect 46 Pt. Reyes iransect 44 I PELAGIC * h ZONE ... CENTRAL SHELF - ~~~~~~~OE Farallon Is. * - ZONE San Francisco Transect 41I 4~~~~~ Transect 39 s, * ~~~~~I '~~*~9a***e**e4*~~......,...... 37 ' , , Santa Cruz I 124' 123' Figure 4. Aerial transect lines and biological zones used in compiling data on seabird distributions. Dotted lines denote boundaries of biological zones, dashed lines denote positions of transcts. lines denote positions of transects. 7-9 distributions did not vary between biological zones and zones were combined to generate a single value of mean group size and a 3 ~single group 2size distribution for each species. Densi~y of birds per km is the product of density of groups per km and mean group size. Results of transect surveys are shown in I ~Figures 5-10. The likelihood that a bird will contact oil depends on I ~whether or not it is on the water when in the vicinity of the slick. Members of the species considered here were assumed to be resting on the water during the hours of darkness, and to be in I~ flight for part of the daylight hours -- about I11 hours in Novem- ber at the latitude of'the Farallones. Percentages of birds on the water during the daylight hours were calculated from un- published data provided by Dr. Kenneth Briggs. These data were U ~collected during November 1984 in the same geographic area using the aerial strip census protocol described previously. Common Murres spent 97.4% of their time on the water, Arctic Loans I~60.,8%, and Cassin's Auklets 64.2% Estimates of movement rates for Common Murres and Cassin's Auklets are based on data for radiotagged Xantus' Murrelets (Hunt et. al. 1976). The mean distance moved per three-hour time interval was 4.5 km. Arctic Loons are engaged in southward E ~migration at this time of year, but periodically stop to rest on the water. No data are available describing short term (i.e on the order of hours) movement patterns of migratory loons or any migrating seabird species. However, since migratory distances I tend to be very large-relative to the size of the slick, we assumed that Loons either did not move at all -- i.e. were resting -- or moved a distance large enough to carry them beyond the range of the slick during each three-hour time step. input values for the Bird Contact Model are given for each I ~species in Table 1. Estimating Numbers of Birds oiled I~~The total area swept out by an oil slick multiplied by the the density of birds in the region through which it passed provides a minimal estimate of the number of birds at risk of oil 3 ~contact. The area swept out by the slick during a given time interval was estimated using the methodology described in Inter- polation of Slick Movement. Bird density estimates were based on the observed densities in the three zones describe-! above. If the slick crossed between zones during a time in~terval, the average density was computed based on the relative .-rea which lay * ~in one zone or the other. Although the density of birds on the water provides a par- tial estimate of the numbe'r of birds at risk, the actual number I ~at risk may in fact be much greater. Seabirds are highly mobile, and although the population of birds present in a particular area of ocean might remain relatively constant, the individuals cont- prising that population could change completely during a given I ~~~~~~~~~~~~~~9 i72 7-10 Fort Bragg AERIAL TRANSECT DATA For each transect, the top line of numbers represents the number k\~~~~~ ~~of birds sighted per km2 in each 5' block; the second line repre- )\~~~~ ~~sents the number of Iroups of _ )~~~~ ~~birds sighted per km in that 39' --- ;-- block. Numbers are typically :i (~Pt. Arena fractional because they are coar- :~~~~~~~~~~~~~~~~~~~~~~~~ a 0 rected to km from a narrower � * ,,.... effective transect width. ~~~~~~~~~~~P. R e ye , o *, 4 - * ....... . :. o o o o:a o oa o o o ) oea9a - 4 49944: : S~~~ 4 ,._ .7 4~ S * 9 o o a O � � � � Bodega Head : * 13 a na Cru 38~~~~ ~~: a ao : a 2m.7 I. I 0 : o,0 o O � ,. ,.A, }:~ ~ ''., F, ; . t~~~~~~~ Pt. Reyes _ o o * ��o o o z-7 8 .ula8. 16.2 a I~~~o./~ ~~~~~~~: * .7 5.4 )S., ,o.8~ i, *. .~~'.............. '--),.: � 3: Farallon Is. : . S an Francisco F o5. 5.4tibt o C.r o 1642 o O v 1" ,,~~~~~~i. ,.,e. J ~ 2.7 tr o6 2.9 70II.2 >< ,~~~1 3~,,0e 37 , .' ' , , , , , , Santa Cruz 12o ~~~~~~~~123� Figure 5 . Distribution of Common Murres on l14 November 19 84 . 10 7-11 Fort Bragg AERIAL TRANSECT DATA For each transect, the top line of numbers represents the number of birds sighted per km2 in each 5' block; the second line repre- sents the number of Broups of birds sighted per km in that 3980 .. .. block. Numbers are typically : ~ / Pt. Arena fractional because they are cor- rected to km from a narrower :.... ; sreffective transect width. *�8a O Oa e c a ea0 a * o * o a 0 - -a a aBOdeaa a 3I� 12 Fiur O Dst0 o � b 4 eo2.7 1:a.a 156.6 116.1 24.. o o e---------e ee-_ � n o a 0: 0 ~ 8& a.19.)o0.0 2.7 10.O I . oS,.l. 21.6 2 oJ a O a a a a g 2 a a o, O a 0 0 e a Ha 2.7 16oa 1*Xe@****,���*******,,,O,,^,,,*^^@eZ*:: ...................... , 124 I , ,23 i , , , | ,SantaCruz * 124 11 36 Pi~Fae 6. Distribution of CoIuon Murres on 17 November 1984. 7-12 Fort Bragg AERIAL TRANSECT DATA For each transect, the top line of numbers represents the number of birds sighted per kmZ in each 5' block; the second line repre- sents the number of groups of birds sighted per km in that block. Numbers are typically 3 li ~Pt Arena fractional because they are cor- rected to km from a narrower ~a o ,o o o0 o effective transect width. * 4 I * � * a aH - * ..... ....� 2.7 ,�' Q u 0 2 7 Oa ., O, O 7 0 *-. , a 0270 � 2.7 � o, 2.7'] , 5 . BodegaHead i. . eyes : * C 4 a a a o o * o o ���' 4o a d'cOX' I : c,__� Farallon Is.! % Farallon Is. * * _ San Francisco 2.7 , 0 00 a 0 0: a0 a Boe H 0 0 e ............ ........ .......... ;3; ~ , . , X . , . , , \ ~~~~~Santa Cruz 124 123' Figure 7. Distribution of Arctic Loons on 14 November 1984. 12 7-13 ; Fort Bragg Fort*~~ Bragg ~~AERIAL TRANSECT DATA For each transect, the top line of numbers represents the number of birds sighted per km in each 5' block; the second line repre- sents the number of Iroups of birds sighted per km in that :zg ,. ,, Arena block. Numbers are typically : , ~ <Pt. ~Arena fractional because they are coar- I: ; rected to km2 from a narrower . ...... ' effective transect width. ' * * 6t = o o o ~~~P. Ree F eI. Sa Gm * - ......0.. � *o a a 2 7 * o al:2 a 0 2 7 Bodega Head , � � . � a � �a27 � i4 10.8. - a a~a a O I2 a 0 0 : 07 a O a 0 ~~~~~2* a 2.a Farallon Is. 1 San Francisco a a � � * 4 2 7 2 A 2720 �:,.4 242(: . _ � � � � � O o 0 O o 0 a a a Figure 8 . Distribution of Arctic Loons on l17 November 1984 . 13 '~~~~~ 16.2 � 0:0 a .72 8. * 2 .7 I 7 I.... F ..... F . , ~~~~~~~~~Santa Cruz 124e 123� Figure 8o Distribution of Arctic Loons on 17 November 1984. I~~~~~~I :8 ~ 7-14 I Fort Bragg AERIAL TRANSECT DATA For each transect, the top line of numbers represents the number of birds sighted per km2 in each 5' block; the second line repre- sents the number of Iroups of birds sighted per km in that 39 - ------..-..-.- block. Numbers are typically *o (\ C1Pt. Arena fractional because. they are cor- *i � � i4 � �rected to km2 from a narrower 2:* , efFective transect width. * . 2.3 0540 0.a 0 o. a 0 0 2.7.7 * * * I. 2 7 � � � ''27 � � � � Bodega Head - ' o-lt5s.l7l52.7 2.7: 4 2.7 2.7 a a a 0 0 2 a 0 0 0 0 0 a 0 0 a Farallon I an isco 124' 123* Figure 9. Distribution of Cassin's Auklets on 14 November 1984. 14 ~~9 ~ 7-15 Fort Bragg AERIAL TRANSECT DATA For each transect, the top line of numbers represents the number of birds sighted per km2 in each �\~~~~~~ ~~5' block; the second line repre- sents the number of ~roups of birds sighted per km in that block. Numbers are typically fractional because they are cor- P.Arena2 rected to km from a narrower oe\~~~ , + ~~~~~~~effective transect width. I * o o 0 0 o e 0 e e o I e c 0e 2.7 5.4 �j~ ~ 27 ��� BodegaHead I 0:0 0 0 0 0 0 0 7 Pt. Reyes 5.4 � O 0 a OO 0 e~~~~~~ � Farallo n Is. San Francisco i* 4 a a 0 a : 0 a a a a a 0 a 0. a 0 0 a a a a 0 �.7 0.1 18.9 5 .1 . SdY.1 10.82.l 5 0 � � 0 � � O O 1 . . B . . , , | , \ ~~~~~~Santa Cruz 1124 123- I Figure 10. Distribution of Cassin's Auklets on 17 November 1984. I * 5 7-16 TABLE 1. Input values for three seabird species, T/V Puerto Rican hindcast. Data for November 1984. Mean Density of Groups (Groups/Km2) Biolouical Zone Common Murre Arctic Loon Cassin's Auklet North Shelf 2.28 1.456 0.96 Central Shelf 5.929 8.966 0.53 Pelagic 0.11 0.324 1.19 Distribution of Group Sizes (% Frequency) Group Size Common Murre Arctic Loon Cassin's Auklet 1 39.5 61.0 64.0 2 16.1 22.2 18.0 3 7.3 2.8 10.0 4 2.2 2.8 6.0 5 8.0 2.8 2.0 6 4.4 0 7 1.5 2.8 8 0.7 2.8 9' 1.5 0 10 5.1 2.8 11-15 5.8 16-20 2.2 21-30 2.1 31-40 0.7 41-50 0.7 51-60 0.7 61-70 1.5 Mean Group Size: 6.066 2.083 1.64 Daily Activity Pattern (Percent On the Water/Percent in the Air) Common Murre Arctic Loon Cassin's Auklet 11 hours active 97.4 / 2.6 60.8 / 39.2 64.2 / 35.8 13 hours quiescent 100.0 / 0.0 100.0 / 0.0 100.0 / 0.0 Movement Data Common Murre Arctic Loon Cassin's Auklet Mean distance moved 40% in flight in southward per 3-hour time step 4.5 km migration at any given time 4.5 km during the active period 16 7-17 time interval. The turnover rate within the region of an oil slick was estimated based on the distribution of distances moved per three-hour time step, assuming that the movement of an indi- vidual or group could be modelled as a random-walk process. One thousand simulated bird groups were distributed uniformly across a region the shape of the slick. At three-hour intervals, each group was moved a random distance in a randomly selected direc- tion. The distance moved was selected by sampling from a distri- bution of distances moved per three hours. Each group was moved until it had passed out of the region and the number of time steps required to do so was noted. This process is illustrated in Figure 11. The average number of time steps required to leave the region from a random starting position represents an estimate of one-half the average time spent in an area of ocean the size and shape of the slick (since groups were already within the region when they were placed in motion, this time is multiplied by two). The turnover rate is therefore the time step divided by the mean length of time spent in the region. For example, if a group spends on average 5 hours in the region before exiting, and the time step is 3 hours, the estimated turnover rate would be 0.6 -- i.e., there would be 60% replacement of old animals with new animals during one time step. The number of birds encoun- tering the slick would then be 1.6 times the number of birds on the water. Note that this is an estimate of the turnover in an unpolluted region of ocean. We assume that the encounter rate would remain the same if the region were covered by the slick, although in some cases flocks might turn aside after initially encountering it. Figure 11. Illustration of the method used to estimate the length of time a flock of birds would spend in a region the size and shape of an oil slick. Circled num- bers indicate the posi- tion of a hypothetical flock at three-hour in- tervals. The distance moved is based on the relative frequency of distances moved; the turning angle is random. In this case, the flock required about 6.2 inter- vals or 18.6 hours to leave the region. 17 7-18 Bird Fate Model Birds which were killed or incapacitated by oil are assumed to have had one of four fates: either (1) they were carried out to sea by winds and currents, (2) they sank or were scavenged before reaching shore, (3) they came ashore along stretches of coastline which were not searched, or (4) they washed up or swam ashore on beaches which were regularly searched. Only in the case of (4) would the birds have been enumerated. The purpose of the bird fate model is to partition birds among these various fates. Trajectories of Oiled Birds The HAZMAT oil spill trajectory model was used to compute families of possible trajectories assuming that dead or incapaci- tated birds move at 2.2% of the wind velocity. This value is based on published experiments with dead, oiled auks in the Irish Sea (Hope Jones et al. 1970). The value is lower than the 3.0% typically used for surface films of oil, probably because of the relatively large subsurface resistance of the carcasses. Eight typical regions were selected from which to launch circu- lar clusters of model bird bodies (J. Galt pers. comm.). Groups of 100 birds were launched from each region and followed until they either came ashore along one of six coastal segments, or were washed out to sea (see Figure 12 for a map of launch regions and coastal segments). A matrix was constructed containing the probability that a bird which was oiled in a given region would come ashore along a given stretch of coast and the length of t.ime that would be req-aired to do so (see Table 2). TABLE 2. Probability (expressed as percent chance) that a bird carcass launched in a given area will come ashore on a particular stretch of beach (for launch areas and recovery areas shown in Figure 12). Percent chance of bird carcass recovery if launched from a given launch area Launch Area Recovery Area 1 2 3 4 5 6 7 8 A 0 0 0 0 0 0 0 0 B 2 0 0 0 100 0 0 0 C 44 3 21 23 0 0 0 0 D 53 4 4 17 0 0 0 0 E 1 93 75 60 0 100 100 29 F 0 0 0 0 0 0 0 0 G 0 0 0 0 0 0 0 26 H 0 0 0 0 0 0 0 45 18 7-19 OSSM LAUNCH POINTS FOR SIMULATED BIRD CARCASSES = Fort Bragg ( 37�23.3 122057.4 3 Nov 1200 �37018.0 122a49.5 5 Nov 1200 @)37043.4 122053.6 7 Nov 1200 )37�56.7 122" 55.1 9 Nov 1200 � 380 0.7 122055.1 9 Nov 1200 Q\ �3 �37059.6 1230 4.9 9 Nov 1200 )3B0 14.5 1230 7.0 10 Nov 1200 @37�16.7 1230 5.2 5 Nov 1200 39 ------ ;rnaRECOVERY AREAS FOR : <Ft. Arena SIMULATED BIRD CARCASSES (Z)Sauth of Bolinas �Bolinas to Pt. Reyes I e � ~)Pto Reyes to Tomales Pt. :e~ :>t~ '~@~~ (Bodega Area (Tomales Pt. to _-~~~~~~~~ e .X~~~eo�X > 'Jenner area) )Jenner to Pt. Arena : , *�_ (@)North of Pt. Arena : ........ .: ....... (OFarallones Ou t to Sea (Never Beached) Bodega Head Reyes '�-�------ "--- --- Farallon Is. I t San Francisco I.@*s~' ) I i , I , 2 ' t ' @ * * t \Santa Cruz 1240 123' Figure 12. Simulated' bird carcass launch and recovery areas used in Bird Fate Model. 19 7-20 The number of birds estimated by the bird contact model to have been oiled within a given launch region were partitioned ac- cording to the probability that they would come ashore in one of the six coastal segments. Of these coastal segments, all were carefully searched during the period of the spill except the coastline north of Jenner and South of Pt. Arena. This area is composed primarily of steep cliffs with little or no access to the intertidal, and except for one location was,never searched for beached birds (G. Page pers. comm.). Loss of Oiled Birds at Sea Although an oiled bird-might drift in the direction of a particular beach and ultimately come ashore there, it might also be lost to natural processes before reaching shore. Of 106 murre carcasses released by PRBO biologists in 1980 and 1981 within 1,500 m of shore, only 4 were ever recovered (PRBO unpublished data). However, a large proportion of the weighted drift bottles and gull carcasses released at the same time were eventually recovered, indicating that the disappearance of the Murres did not result from being carried out to sea, but rather that the carcasses sank or were scavenged along the way. The experiment carried out by Hope Jones et al. (1970) resulted in 20% recovery even though bird carcasses were at sea ten or more days and is the source of the loss rate estimate used here. We assume that the loss of bird carcasses is a function of how long they are at sea, and that the fraction lost each day is constant. For example, if one-half of the carcasses were lost each day, there would be one-half remaining after one day, one- quarter remaining after 2 days, one-eighth remaining after 3 days, etc.. If 20% were still floating after 10 days as Hope Jones et al. (1970) found, the estimated loss rate would be 15% per day. We use these data to generate a matrix of probabilities that a bird oiled in a given region would be recovered, along a given segment of coastline. Let Ll be the probability that a carcass launched from region i will ome ashore in segment j, S be the loss rate (sinking or scavenging) per day, andD be the number of days required to float from i to j based on the results of the HAZMAT trajectory model. In addition, let C . be a vari- able which assumes the value 0.0 if a beach was not searched, and 1.0 if the beach was searched. Then the probability that a dead or incapacitated bird launched from i will be recovered in j is: Pi j= Cj Lij (1 - S) D Model Calibration The mortality rate of birds encountering the oil slick was estimated using the actual observations of oiled birds found on the beaches. Let O. be the number of birds found along coastal segment j, Iet Ei be the number of birds which en- countered the slick in region i, and let 9 be the mortality rate. An estimate of ~2 based on the observed number of beached 20 7-21 birds found in coastal segment j, oi, is given by: i j = Oj /(PijEi) In other words, the estimated mortality rate is the ratio of the number of birds observed in a given segment of coastline to the number which we estimate would have been beached if every bird which passed through the slick died or was incapacitated. Note that this estimate includes the loss of oiled birds while in transit from the point of oiling to the beach which is subsumed in the value of the P. . Since the value of 0. varies from beach to beach, we estiAted the overall value of as the slope of the regression of the 0 . on the (Pi Ei). Because the number of beached birds must be ze~o when the ndmber of birds contacting the slick is zero, the regression line was forced through the origin. MODEL ANALYSIS AND RESULTS Sensitivity Analysis The values of some input parameters and the validity of some of the assumptions used in the model involve varying degrees of uncertainty. When the uncertainty is large, the degree to which model results might be affected by that uncertainty should be closely examined. Using data for the most common species, Common Murres, we examined the sensitivity of the model to the following factors: *The value of the mean distance moved per 3-hour time step �The assumption that the bird groups move in a random walk fashion, changing direction randomly at each time step �The variability in the distribution of seabirds -- i.e. the natural randomness involved in how many birds were present in an area at the time when it was affected by the slick a The value of the rate at which oiled birds are lost at sea �The assumption that an oiled bird has an equal likelihood of being lost each day it is at sea We examined the effect of varying these factors on two model outputs: �The estimate of the total number of birds killed by the spill �The estimate of the percent of the birds which became debilitated or died as a result of being present in the area of the slick 7-22 Distance moved Per Time Step The distribution of distance moved per 3-hour time step was extrapolated from data for Xantus ,Murrelet (Hunt et al. 1976), and it is not known how appropriate these data are for other alcid species. We examined the sensitivity of the model to this input by making model runs using values for the distances moved which were two times as great as those measured for Cassin's Aukiet, and one-half as great. Decreasing the movement rate hasI the effect of increasing the mortality rate, and increasing the movement rate has the effect of decreasing the mortality rate. This occurs because the number of birds believed to have encoun- tered the slick increases with increasing movement rates, and to account for the observed beachings, fewer of them could have been killed or debilitated. Doubling the movement rate lowered the estimated mortality rate by 21%; halving the movement rate in- creased the estimated mortality rate by 19%. Deleting the move- ment algorithm entirely increased the estimated mortality rate by 44%. Estimates of total mortality, however, were only slightlyI affected by variation in this parameter, changing by less than 2% in. any case. Random Turning Angles of Moving Birds it has been assumed that the movement path of birds can beI described as a random walk process, and that model bird groups changed direction randomly at each time step. We examined the effect of this assumption on model results by changing the model so that model bird groups continued in the same direction indef- initely. This change increased the rate at which bird groups encounter the slick, however it had little quantitative effect on model results. The estimated mortality rate was unchanged,I and the estimated number of birds oiled increased by less than 2%. Variability in Bird Distribution Although distributional data were collected at the time of the oil spill, we do not know how many birds actually encountered the slick. The distributional data provide a statistical description of the numbers of groups per square kilometer and the relative frequency of group sizes in the general area at the time of the incident, but these data are averaged over large regions. Even if the mean density of groups in a zone were 0.5 groups per square kin, a given square km of slick at a given instant of timeI might have contained 0, 1, 2, or more groups. Similarly, if the mean group size were 1.5 birc-11, a given group might actually contain 1, 2, 3 or more individuals. We estimated the effect ofI this variability on the model results by permitting densities and group sizes to vary randomly based on their observed patterns. Groups were assumed to be distributed uniform random throughI space, implying that the number of groups within a given area was Poisson distributed. Thus, if the area swept out by the slick during a given time interval is A, the mean density of birds in 22 iJ 7-23 the region is b, and the number of bird groups in the area which passively encounter the slick is n , then np is a random variable with the probability density function: P[np] = (Ab)np exp(-Ab) / np For each time interval, the number of groups making passive contact with the slick was simulated by randomly sampling from this distribution. The size of each group was selected by ran- domly sampling from the known distribution of group sizes. The number of groups making active contact -- i.e. flying into the slick -- was simulated by assuming that the probability per unit time of a group entering the slick was constant, so that the number of groups actively encountering the slick would also be a Poisson process. If R is the average rate at which new groups encounter the slick (the turnover rate), T is the length of a time interval, and na is the number of birds flying into the slick, then the na is a random variable with the probability density function: P[na] = (RT) na exp(-RT) / na! The model was run 25 times using different random number sequences to evaluate the extent to which model results varied due to this form of stochasticity. The resultant coefficient of variation of the estimated mortality rate was 2.1%, and the coefficient of variation for the total number of birds killed was 1.3%. The stochastic nature of the bird distributions, therefore, does not appear to hgve an important effect on model results. At-Sea Loss Rate Although data are available which may be used to estimate the rate at which dead or debilitated birds are lost at sea-- presumably due to sinking or scavenging-- these data are not conclusive. The loss rate of oiled auks following the Hamilton Trader incident (Hope Jones et al. 1970) provides the best estimate of the loss rate, 15% per day, but other data suggest that the value could be quite different under other circumstances. For example, Common Murre carcasses released nearshore by PRBO biologists resulted in only 4% recovery, implying a much higher loss rate. We therefore examined the sensitivity of model results to the entire range of possible at- sea loss rates. The mortality rate of birds encountering a slick and the rate at which affected birds are lost at sea both influence the number of beached birds predicted by the model. An increased loss rate results in fewer beached birds predicted; an increased mortality rate results in more beached birds predicted. For a given value of the at-sea loss rate, the model uses observed beached bird data to select the corresponding mortality rate which best fits the observed distribution of beached birds. If the at sea loss rate is fixed at 0% per day -- i.e. no birds sank 23 /- * ~~~~~~~~~~~~7-24 or were scavenged before reaching share -- the estimated mortal- ity rate is 23% and the resultant mortality 1,029 birds. Tf the at-sea loss rate is fixed at 35%, the corresponding mortalityI rate must be 100% and the resultant mortality 3,979 birds. Sinking rates of greater than 35% per day cannot be fitted by the model because they would require mortality rates of greater than 100% to account for the observed pattern of beachings. The loss rate of oiled birds at sea is clearly the most sensitive model parameter. Based on the range of possible valuesI for the at-sea loss rate, estimates of the number of oiled birds varied from 1,480 to 3,979, and estimates of mortality rates for birds encountering the slick varied between 23% and 100%. OurI best estimate of the sinking rate, 15%, falls roughly in the midpoint of this range. The PRBO data discussed earlier suggests that this value may err in the direction of underestimating theI at-sea loss rate, and by implication, underestimating total mor- tality. Exponential Sinking Rate We have assumed that~the loss of oiled birds at sea proceeds in an exponential fashion, in other words a constant proportionI of them disappear each day. This is a simple and logical assumption, but it is not the only conceivable model. An alternative model is that at sea loss is not time dependent, butI that the proportion disappear-ing is a constant regardless of how long an oiled bird is at sea. As with the exponential sinking of birds oiled are highly-dependent on the value chosen for the proportion lost at sea. if the proportion lost at sea is 0%, these estimates are the same as for the exponential loss model -- 23% mortality. The greatest loss rate possible for this model is 77%, which corresponds to 100% mortality. This indicates that the use of a time-independent loss rate is probably inappropriate for this context. The loss rate of oiled auks in the study conducted by Hope Jones et al. (1970) was 80%, and of Common Murres in PRBO's study was 96%. Thus, when a constant rather than an exponential at sea loss model is used, both values of theI loss rate are higher than could have occurred in this 'incident even if every bird which encountered the slick died. While this does not "prove" that the loss of oiled birds at sea'is time dependent, such an assumption seems more appropriate in this case. Model Validation We tested the accuracy of the model by comparing theI observed distribution of beached birds with that predicted by the model. The entire model was run 6 times, once for each coastal segment which was searched,each time excluding that segment fromI the calibration process. The model prediction bf the number of oiled birds in a given coastal segment and the observed number of oiled birds were therefore independent, and the predicted number 24 l- j7-25 of beached birds occurring in a given coastal segment did not entail circular computation. The relationship between predicted and observed numbers of beached birds is shown in Figure 13. Linear regression of the predicted number of beached birds on the observed number of beached birds for all species indicates that the model accounts for 81% (r=.908) of the variance in the numbers of beached birds. The slope of the relationship is .906 indicating a tendency toward the underestimation of the numbers of beachings of about 9%. The tendency toward underestimation is most consistent in two areas: south of Bolinas and Bolinas to Pt. Reyes. These beachings cannot be accounted for by assuming passive transport of oiled birds. HAZMAT model results indicated that no beached birds should have come ashore south of Bolinas, and very few between Pt. Reyes and Bolinas. This prediction is strongly supported by the actual observations of the track of the slick. Although we postulated that oil slicks and oiled birds move at somewhat different proportions of the wind speed, 2.2% and 3.0% respectively, these different values should result in very similar patterns of beachings. The fact that no beached oil was observed south of Drakes Bay therefore indicates that passively floating birds should not have come ashore either. It is likely that this discrepancy resulted from oiled birds flying or swim- ming toward the coast. Even a landward displacement of 5 or 10 km during the first several days following the spill would.have been enough to have carried birds into the region influenced by the tidal inhalation of San Francisco Bay. Active movement toward land by oiled birds may also have been the cause of the model underestimate of the number of beachings observed on the Farallon Islands. Model Estimates of Mortality Rates The mortality rate is the fraction of the birds encountering the spilled oil which were injured by the encounter to the extent that, without intervention, they would ultimately have died. The mortality rate for Common Murres was estimated to be 42%, for Arctic Loons 10%, and for Cassin's Auklets 30%. The lower esti- mated mortality rate for Cassin's Auklets than for Common Murres probably results from the assumption that both of these species are lost at the same rate while at sea. It is probable that the small body size of Cassin's Auklets makes them more likely to become waterlogged and sink, or to be scavenged while floating. If the at sea loss rate for Cassin's Auklets were in fact greater than the rate for Common Murres, the estimated mortality rate for Cassin's Auklet would be relatively higher. We cannot demonstrate this at this time because there are no data available for at-sea loss rates of small alcids. Model Estimates of Numbers and Fates of Oiled Birds Model estimates of the number of birds which contacted the spill and their subsequent fates are summarized in Table 3. Based on the observed path of the oil slick and the simulations 25 J 7~~~-26 2 250- ..... OBSERVED 200- - ....~~ PREDICTED 150 - 5 1oo - COMMON MURRE ......... 50 - . .. LLJLz. �0 50 CASSIN'S AUKLET 2 >-�0UL ..4Z W (n 4I- �0 iZa.U Figure 13. Model predicted and observed numbers of beached birds on six segments of coastline. Predictions of theI numbers of beached birds in a given segment of coast- line were made independently of the numbers of beached birds observed in the segment.3 26 TABLE 3. Model results: Model estimates of numbers of three seabird species encountering the oil slick from the Puerto Rican and their subsequent fates. Seriously Affected by Oill Encountering Not Seriously Found on Sunk or Floating Beached But TOTAL DEAD BUT Oil Slick Affected Beaches Scavenged Out to Sea Not Recovered NOT RECOVERED Common Murres Number of Birds 4,255 2,468 487 861 61 378 1,300 Percentage 100.0% 58.0% 11.4% 20.3% 1.4% 8.9% 30.6% Arctic Loons Number of Birds 1,670 1,503 53 82 0 32 114 Percentage 100.0% 90.0% 302% 4.9% 0.0% 1.9X 6.8% Cassin's Auklet Number of Birds 210 147 16 37 0 10 47 Percentage 100.0% 70.0% 7.6% 17.6% 0.0% 4.8% 22.4% These are birds that would have died without rehabilitation. Some birds found on beaches were successfully rehabilitated (see previous section). � ~!~.7-28 1 of the distribution and behavior of the various seabird species, we estimate that 4,255 Common Murres, 1,670 Arctic Loons, and 210 Cassin's Auklets encountered the spilled oil from the Puerto Rican. Of these encounters, we estimate that 1,787 Common Murres, 167 Arctic Loons, and 63 Cassin's Auklets were injured by the oil to the extent that, in the absence of rehabilitation, they would have died. The most likely fate of oiled birds was probably loss at sea due to sinking or scavenging, accounting for 48% of the oiled Common Murres, 49% of the Arctic Loons, and 59% of the Cassin's Auklets. The remaining oiled birds washed ashore along the central California coast, some areas of which were searched for oiled birds and some were not. Trajectory simula- tions indicate that 21% of the oiled Common Murres, 32% of the Arctic Loons, and 16% of the Cassin's Auklets were beached on unsearched or inaccessible stretches of coastline between Jenner and Pt. Arena. The total numbers of birds which are believed to have died but were never recovered are 1,300 Common Murres, 114 Arctic Loons, and 47 Cassin's Auklets. REFERENCES Ecological Consulting. 1985. A Risk Analysis Model for Marine Mammals and Seabirds: A Southern California Bight Scenario. Final Report. Prepared for Minerals Management Service, Pacific OCS Region. Contract No. 14-12-0001-30224. Hope Jones, P., G. Howells, E.I.S. Reese and J. Wilson. 1970. Effect of 'Hamilton Trader' oil on birds in the Irish Sea in May 1969. British Birds 63(3): 97-110. Hunt, G.L.,Jr., M. Quammen, and K.T. Briggs. 1976. Foods and foraging range of nesting seabirds. pp. 681-726. Vol. 3, Book 2. Marine mammal and seabird survey of the Southern California Bight. Univ. of California, Santa Cruz. 28 7-29 SPECIAL OIL SPILL RISK ANAYLSIS (OSRA) OF TANK VESSEL PUERTO RICAN INCIDENT by Lawrence J. Hannon and Cheryl M. Anderson Minerals Management Service Branch of Environmental Modeling November 1985 7-30 PT. ARENA I Pi1 PI: ' PT. REYES P$ Figr . MI to I . II BAY 3 N N t Figure 1. -- Map showing the location of oil spill launch points P1 to P19 and launch point S (S = sunken stern of PUERTO RICAN). 7-31 rPT. ARENAt I I WT. REYE AI AN FRANCISCO BAY * Dg F I H I 2 3'4 5 6 7 i~~ 37- ~~~MONTEREY l*~~~~ ~~BAY I I I " N N0 Figure 2.--Target grid matrix for special OSRA run of PUERTO RICAN oil spill incident. U <?'p 7-32 - 'PT. ARENA A~~~~~~~~~~ .REYES \g AN FRANCISCO BAY ~~~~~11 -S7~ MONTEREY BAY -~~~~~~~~~~~~~~ _ N Figure 3.--Map showing the location of the target "Farallon Islands": crosshatching indicates areal extent. TABLE I1.- PROOABILITIES (EXPRESSED AS PERCENT CHANCE) TIJAT AN oil. SPILL STARTING AT A PARTICULAR L-OCATION WILL CONTACT A CERTAIN TARGET WITHIN 3 DAYS. TARGET STEM H~~~~~~~~YPOTHIETICAL. SPILL LOCATION IS) PI P2 P3 P4 PS P6 P7 Pe p9 pi( p11 p12~ jp3 P14 Pis P16 pi? pie PIS LAND 1 4 14 I 5 34 N S 7 67 N N 22 59 N N 3 27 A-1 N 16 a 40 se 19 N 20 1 2 N N I N A-2 1 25 20 11 41 63 19 3 it 1 3 N I N A-3 I I11 21 1 8 34 49 N *A N a N N m A-4 1 2 5 N 1 7 22 N la N 10 I N N 2 1 N N N N A-S N N N N N I N a N 3 5 N N 2 2 N N N N 0-1 N 6 5 33 30 10 N 33 N 4 N N I N N a I N N N 9-2 1 12 7 13 34 35 11 6 it 3 3 N I I N N N N N 11-3 2 8 12 2 11 29 28 1 95 I is N H I N I N I N 6-4 1 3 6 N 2 11 22 N SO N 20 2 N N A I N N I N B-S N H 4 N 4 a N N 4 3 N N N N 9-6 N I N N 13 N N 3 4 N N N N C-i N ~~~~~~~~3 2 24 16 7 N 39 1 NN 3 1 N N 2 1 N N C-2 2 5 3 15 25 19 7 12 11 11 3 N 2 3 N H 2 1 N N C-3 3 4 5 3 12 20 15 1 75 2 N N 1 2 N I I I C-4 3 2 5 N 3 12 la N 70 N ~' 2 N N 7 2 N N 2 N C-S I N I N 1 2 6 *N 14 9 914 N N 13 7 N N I 1 C-6 N N N N N N I N 2 N I *' N N a 13 N I I D-1 I I 1 12 6 3 N 31 N 99 N N *a a N 3 1 NN D-2 4 1 1 10 12 7 4 14 8 33 2 N 4 It* N 3 2 1 N D-3 6 1 1 3 8 10 6 2 33 4 48 N 1 2 1 N 2 1 2 N D-4 5 I I N 3 a 9 N se 1 72 1 N I 11 2 N N 3 1 D-S 2 N N N N 2 4 N 20 N la 19 N N 11 12 N N 2 1 D-6 ~~ ~~~~~N N N N 1 1 N 4 3 390 ff N 14 H N 2 1 E-1 N N N 6 3 2 N 20 N 03 N N 9 N N 5 2 N N 3-2 4 N N 7 6 3 2~~~ ~~~~ ~~~~214 6 67 1 N 13 At N 6 3 1 N 3-3 12 N N 2 5 4 3 3 19 10 26 N 2 7 1 N 3 2 3 N 3-4 10 N I 2 5 4 N 44 2 58 I N 1. 20 1 1 1 6 1 E-S 2 N N N N 2 2 N 22 N 24 17 N N ** 20 N 3 E-6 I N N I N 6 N 5 67 N N 28 -~ N 2 2 F-I I 3 2 1 N 12 N 53 N N 90 10 N N 12 5 NN F-2 S N N 4 3 1 1 10 s 68 2 N 52 98 N N 19 7 1 N F-3 A N 1 1 3 13191 laN 6 31 1 N6 3 4 N F-4 41 N N N 1 2 1 1 29 5 43 1 1 621 1 12 1 F-S ~~~~ ~~ ~~ ~~ ~~ ~~ ~~~~~ ~~~~~~~~ ~~ ~ ~~4 N N I I N 22 N 28 14 N 1 79 21 N N 9 4 F-6 1 N N N N N 7 N 9 42 N N 45 77 N N 3 7 G-I N N I 1 m 4 N 20 N N 73 9 N N 38 9 N N 0- 2 4 N II N I 5 3 49 2 N 68 so N N ** 14 1 N 0-3 B e N N I N N 2 6 24 11 N 12 56 N N 10 5 4 N v- 4 59 N N N N N N 1 12 6 22 N 2 1 2 15 I I 1I' 2 j- 5 0 H N N N 13 1 22 9 N 2 40 15 N N 22 7 0-6 1 N N N N N N N S N 9 14 N N 39 33 N N 5 16 6-7 N ~~ ~~ ~~N N N N N N N I N 1 4 N N 9 12 N N I a H-1 N N N N N N I N 11 N N 42 a N 30 38 N N 11-2 3 N N N N N N 2 2 36 3 N 64 56 N N 87 *4 N N 11-3 53 14 N N N 1 3 25 7 N 22 59 N N 26 12 4 N H1-4 66 N N N N N N N 5 12 1 6 22 11 N 4 2 1 51-5 ~~~ ~~~20 N N N N N N N 7 1 16 6 1 4 29 11 N N * 11-6 2 N N N N N N N 4 N a 9 N H2 2 If-? N N N~~~~~ N N N N N I N 2 4 N N 14 1 3 N N 1 22 FARALLON ISLANDS 3 2 2 1 5 10 6 N 32 1 49 N N11 N 1 1 2 N NOTE.' GREATER THAN 99.5 PERCENT; N =LESS MhAN 0.5 PERCE21. TABLE 2. -- PROBABILITIES (EXPRESSED AS PERCENT CHANCE) THAT AN OIL SPILL STARTING AT A PARTICULAR LOCATION WILL CONTACT A CERTAIN TARGET WITHIN 10 DAYS. HYPOTHETICAL SPILL LOCATION TARGET STERN (S) P1 P2 P3 P4 PS P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 PIO P19 LAND 1 4 14 H 1 S 34 H 5 N 7 67 N H 22 59 N N 3 27 A-i N 16 8 40 58 19 N 20 1 2 N H 1 N H N N H N H H A-2 1 25 20 11 41 63 19 3 11 ' 1 3 N 1 N H N N N N H A-3 1 11 21 1 8 34 49 N ** N 8 N H H 1 N N N N N A-4 1 2 S N 1 7 22 N 18 H 10 1 N H 2 1 N N N N A-5 NH NH H H 1 H 2 H 3 5 HN 2 2 H H N N 8-1 N 6 5 33 30 10 N 33 .N 4 H H 1 N N N 1 N N N B-2 1 12 7 13 34 35 11 5 11. 3 3 NH 1 H N N N N N 8-3 2 8 12 2 11 29 28 1 95 1 18 N N N 1 N 1 N 1 N B-4 1 3 6 N 2 11 22 N 50 H. 20 2 N N 4 1 N N 1 N B-5 N H H N N 1 3 N 4 HN 4 8 N 4 3 H H H 8-6 N N N N H H N N 1 N N 13 N N 3 4 N N N NH C-1 N 3 2 24 16 7 N 39 1 ** N 3 1 N 2 1 N N C-2 2 5 3 15 25 19 7 12 11 11 3 N 2 3 H H 2 1 H N C-3 3 4 5 3 12 20 15 1 75 2 ** H HN 1 2 H 1 1 1 N C-4 3 2 5 H 3 12 18 N 70 H ** 2 N H 7 2 H N 2 N C-5 1 N 1 N 1 2 6 N 14 N 9 14 N N 13 7 N H 1 1 C-6 N H N H H H I H 2 N 1 a** N N 8 13 N N 1 1 D-1 1 1 1 12 6 3 N 31 N 98 H N ** a N H 3 1 N N 0-2 4 1 1 10 12 7 4 14 8 33 2 N 4 ** N N 3 2 1 N D-3 6 1 1 3 8 10 6 2 33 4 48 N 1 2 1 H 2 1 2 N D-4 S 1 1 H 3 8 9 N 58 1 72 1 N 1 11 2 N N 3 1 D-5 2 H N N H 2 4 N 20 N 18 19 N N ** 12 N N 2 1 0-6 N H H N H 1 1 N 4 N 3 90 N N 14 ** H N 2 1 - E-1 N N N 6 3 2 N 20 N 83 N N *^ 9 N N 5 2 N NH E-2 4 H H 7 6 3 2 14 6 67 1 N 13 * N H 6 3 1 N 4 E-3 ' 12 H H 2 S 4 3 3 19 10 26 N 2 7 1 N 3 2 3 N E-4 10 H 1 H 2 S 4 N 44 2 58 1 N 1 20 1 1 1 6 1 E-5 2 H N N H 2 2 H 22 N 24 17 N N ** 20 N H 5 3 E-6 1 N N N N N 1 NH 6 N 5 67 N N 28 ** H H 2 2 F-1 1 N N 3 2 1 H 12 H 53 N H 98 10 N H 12 5 H N P-2 5 H N 4 3 1 1 10 5 68 2 H 52 98 N H 19 7 1 N F-3 H* N H 1 2 1 1 3 13 19 18 H 6 31 1 H 6 3 4 N F-4 41 H H N 1 2 1 1 29 5 43 1 1 6 21 1 1 1 12 1 F-5 4 N N H N 1 1 N 22 N 28 14 N 1 79 21 N N 9 4 F-6 1 H H H H H H N 7 N 9 42 H N 45 77 N H 3 7 G-1 H N N 1 1 N N 4 N 20 N N 73 9 N N 38 9 H N G-2 4 H 1 H N 1 5, 3 49 2 N 68 80 H N ^* 14 1 N 6-3 80 H N N 1 N N 2 6 24 11 N 12 56 N N 10 5 4 H G-4 69 N N N NH H 1 12 6 22 N 2 12 15 1 1 1 2 G-5 8 H N H N N N N 13 1 22 9 N 2 48 15 N N 22 7 G-6 1 N N H N N N N 5 N 9 14 N N 39 33 H N 5 16 G-7 N N N N N N N N 1 N 1 4 N N 9 12 N H 1 8 H-l N N N N H H N 1 H 11 N H 42 8 N H 30 38 N N H11-2 3 N N N N N N 2 2 36 3 N 64 56 N H 897 ** H N 11-3 53 N H N NH H N N 1 3 25 7 N 22 59 N N 26 12 4 N 11-4 66 N N N N N H 5 8 12 1 6 22 11 N 4 2 ** I H-5 20 N N N N N N N 7 1 16 6 1 4 29 11 N N ** 11 11-6 2 N N NH N H H H 4 H 8 9 N 1 32 21 H H 9 ^^ H-7 N H N H N N N N 1 N 2 4 N H 14 13 H N 1 22 FARALLON ISLANDS 3 2 2 1 5 10 6 N 32 1 49 N N 1 1 N 1 1 2 H WOTE: = GREATER THAN 99.5 PERCENT; N = LESS THAN 0.5 PERCENT. NROBARPIES RR SS") P~ERCENIT CHANCE) IIIAT AN OIL SPIM~~TN AT A PARTICULAR LOCATION WILL CONTACT A CERTAIN TARGET WITHIN 30 DAYS. HYPOTHETICAL SPILL LOCATION TARGET STERM is) Pi P2 P3 P4 PS P6 P7 Ps P9 Pie P11 P12 P13 P14 PIS P16 PI? P19 P19 LAND 1 4 It N I 5 34 N 5 N 7 67 N N 22 59 N N 3 27 A-i N 16 3 40 58 19 N 20 1 2 N I N N N N A-? 1 25 20 11 41 63 19 3 11 1 3 N I N N N N N N N A-3 1 it 21 1 a 34 49 N ** 8 N N N I 11 N N N N A-4 I 2 5 I 7 22 N is N 10 I N 11 2 1 N N N N A-S a N N I Nf 2 N 3 5 N N 2 2 N N N N 0-1 N 6 5 33 30 10 N 33 N 4 N I N N I N N N 9-2 1 12 7 13 34 35 11 S .11 3 3 N I I N N N N N 9-3 2 9 12 2 11 29 28 1 95 1 toN N I I N I N 9-4 1 3 6 N 2 11 22 N150 N 20 2 N N 4 1 N N I N B-S N4 N N N 1 3 N 4 N 4 8 N N 4 3 N N N N ,FA 9-6 N N N N N I N N 13 N N 3 A Nf N N C-i 3 2 24 16 7 N 39 1 a* N 3 1 N 2 1 N N k C-2 2 S 3 Is 25 19 7 12 11 11 3 N 2 3 N m 2 1 N c-3 3 4 5 3 12 20 15 1 75 2 N N 1 2 N I I I N C-4 3 2 5 N 3 12 19 N 70 N " 2 N N 7 2 N N 2 N C-S I N 1 N 1 2 6 N 14 N 9 14 N N 13 7 N I 1 C-6 N Nf N N N I N 2 N I N N 9 13 N N I I D-1 I I 1 12 6 3 N1 31 N 90 Nf N B aN N 3 1 N N D-2 4 1 1 10 12 7 4 14 S 33 2 N 4 it N N 3 2 1 N D-3 6 1 1 3 la 10 6 2 33 A 48 N 1 2 1 N 2 1 2 N D-4 S I I N 3 a 9 N so 1 72 1 N I I11 2 N N 3 1 D-5 2 N N N N 2 4 N 20 N is 19 N N 12 N N 2 1 D-6 N N N N I I N A Nf 3 90 N N 14 ** NN 2 1 E-1 N N N 6 3 2 N 20 N 83 N N * 9 N 5 2 N N 9-2 4 N N 7 6 3 2 14 6 67 I N 13 N N 6 3 1 N E-3 ~~~ 12 N N 2 5 A 3 3 19 10 26 N 2 7 I N 3 2 3 N E-4 ~~~10 N I N 2 5 4 N 44 2 58 1 N 1 20 1 1 1 6 1 E-S 2 N m 2 2 N 22 N 24 17 N N ** 20 N N 5 3 E-6 1 11 N N I N 1 N 6 N 67 N N 28 N N 2 2 F-I I N Nf 3 2 1 N 12 N 53 N N 99 10 Nf N 12 5 N N P-2 S N N 4 3 1 1 10 5 68 2 N 52 98 N N 19 7 I N P-3 * I N 1 2 1 1 3 13 19 18 6 31 1 N 6 3 4 N F-4 41 N N 1 2 1 1 29 5 43 1 1 6 21 1 1 1 12 1 F-S 4 N N N N I I N 22 1120 14 N 1 79 21 N N 9 4 F-6 1 N N N N N N 7 N 9 42 N N 45 77 N N 3 7 G-IN N N 1 1 N N 4 N 20 N N 73 9 N N 38 9 N G-2 4 N N I Nf N I 5 3 49 2 N fie so N ** 14 1 N 0- 3 a s I N N 1 N N 2 6 24 11 N 12 56 N N la 5 4 N 0- 4 69 N N N N N N 1 12 6 22 N 2 12 15 I I I1 0 2 0- S 0 N N N N N 13 1 22 9 N 2 48 IS N 22 7 0-6~~~~~~~~~~~ 1f N N N N S N 9 14 11 1139 33 N N 5 16 G-7 N N N N N N Nf I N 1 4 N N 9 12 N N I a "-IN N N N N N I N 11 N N 42 B N N 30 38 N N H1-2 3 N N N N N N 2 2 36 3 N 64 56 N N 87 t* N N 11-3 53 N N N N N N 1 3 25 7 N 22 59 N N 26 12 4 N 11-4 66 N N N N N N 5 9 12 1 6 22 It N 4 2 1kI 11-5 20 N N NN N N N 7 1 16 6 1 4 29 11 N N 11 i 11-6 2 N N N 4 N 9 N 1 32 21 11 N 9 it, 11-7 N N N N I 2 4 N N 14 13 N N 1 22 FARALON ISLANDS 3 2 2 1 5 10 6 N 32 1 49 N I I I 1 2 N NOTE: GREATER THAN 99.5 PERCENT,' N =LESS THAN 0.5 PERCENT. TABLE 4. -- PROBABILITIES (EXPRESSED AS PERCENT CHANCE) THAT AN OIL SPILL STARTING In THE WINTER SEASON AT A PARTICULAR IfLCATION WILL CONTACT A CERTAIN TARGET WITHIN 30 DAYS. HYPOTHETICAL SPILL LOCATION TARGET STERN (S) P1 P2 P3 P4 PS P6 P7 PS P9 P10 P11 P12 P13 P14 PIS P16 PO P18 P19 LAND 36 48 65 31 43 52 69 24 58 25 61 92 21 26 69 86 22 23 47 61 A-I 4 39 25 40 61 43 1I 27 9 12 8 1 8 7 2 1 6 4 4 1 A-2 9 40 39 23 43 64 40 14 36 13 17 2 8 10 5 2 9 6 5 1 A-3 10 22 30 13 21 35 51 10 *6 10 26 4 7 8 6 3 8 6 6 I A-4 10 9 13 8 9 13 24 6 32 7 28 10 5 8 11 6 6 5 5 2 A-S 6 3 3 3 3 7 41 11 3 15 172410 9 3 3 5 3 8-1 5 29 19 41 45 32 14 32 10 17 7 1 11 9 3 1 9 5 5 1 0-2 9 33 31 25 41 50 27 16 36 15 19 2 12 11 5 3 12 8 5 1 B-3 13 22 28 13 23 35 370 11 95 10 46 5. 16 12 8 4 11 8 7 2 0-4 12 12 15 10 13 17 27 8 54 8 42 12 7 10 16 9 8 6 7 3 0-5 7444 5 6 9 3 1441 15 27 3 4 1413 4 4 6 3 8-6 3 2 N 1 3 2 3 16 2 430 1 2 12 14 1 2 4 4 C-I 8 26 17 41 35 28 12 40 11 ** 8 1 15 12 3 1 14 9 5 1 C-2 16 32 27 32 39 44 23 22 33 26 22 3 18 18 6 3 18 14 7 1 C-3 18 24 28 17 26 38 30 14 79 16 * 6 13 17 11 5 15 14 11 3 C-4 161718 13I15 24 29 9 63 11 *1410 14 26 11 10 913 5 C-S 10 6 6 6 8 10 14 4 20 8 23 41 5 7 30 23 5 4 9 7 C-6 6 3 2 1 4 5 4 9 5 8 *f 4 3 22 29 3 3 9 7 D-1 10 21 14 36 28 22 10 42 11 98 8 1 ** 29 2 1 19 13 4 1 D-2 19 26 22 33 34 36 19 24 26 47 17 3 22 A* 6 2 22 18 9 2 D-3 24 24 23 16 24 34 25 16 48 18 54 5 17 20 11 4 18 16 14 3 D-4 22 18 18 11 18 24 27 11 57 12 66 12 11 16 36 13 12 11 17 7 D-5 15 7 7 7 9 12 13 5 24 10 20 33 5 9 1* 36 7 7 16 10 D-6 8 3 4 3 7 6 3 11 5 11 80 4 427* 14 11 E-1 10 16 13 32 27 19 8 39 8 87 0 1 *6 34 3 1 25 16 5 1 8-2 22 24 20 31 32 30 16 29 25 67 14 2 31 ** 6 1 29 21112 E-3 35 24 20 18 25 31 21 17 38 23 37 4 20 25 10 4 23 19 18 4 E-4 27 19 17 11 20 27 23 11 51 15 56 12 14 20 43 12 16 14 24 S R-5 15 9 9 9 10 14 14 6 30 11 29 25 9 11 ** 39 9 8 23 13 E-6 84 4 5 9 8 3 15 7 14 52 5 5 30 ** S 7 15 12 P-I 12 18 12 31 25 15 7 39 8 65 9 1 97 36 3 1 43 25 5 1 F-2 28 24 17 30 30 25 14 34 22 69 16 2 58 97 6 1 47 32 11 2 F-3 9 23 18 22 26 29 19 21 34 33 30 5 26 41 11 4 28 26 24 4 F-4 45 21 10 13 22 29 22 12 43 22 49 11 17 27 35 11 18 21 40 11 F-5 21 11 10 9 13 17 15 7 34 13 35 19 11 15 70 30 10 12 32 20 F-6 13 5 7 7 7110 10 4 19 9 17 36 7 9 41 63 8 9 20 27 r--I 13 16 11 26 20 11 5 31 6 43 6 1 79 32 2 1 64 33 4 1 0-2 30 20 15 26 24 21 12 32 17 58 13 1 67 82 5 1 4* 41 11 2 0-3 83 19 16 21 24 22 16 21 26 36 24 3 28 57 9 3 30 29 27 5 G-4 63 20 18 13 20 26 19 14 38 25 40 9 18 32 27 9 20 25 ** 13 0-S 24 13 11 8 15 18I15 7 33 14 35 14 11 19 48 21 11 11 41 26 0-6 13 6 6 6 9 10 10 5 19 10 19 17 6 9 37 34 7 9 23 39 G-7 5 2 3 3 3 3 2 6 4 7 74 1 5 10 55 3 410 18 H-1 12 14 8 22 21 10 6 28 6 39 5 I 55 28 2 1 49 61 4 2 H-2 26 18 13 27 23 19 9 31 16 53 11 1 69 67 5 178 46 11 4 H-3 65 18 13 22 22 21 15 24 23 43 20 3 35 60 0 3 37 36 27 6 11-4 65 18 15 IS 20 24 19 16 33 29 33 7 23 39 26 8 26 27 1* 14 11-5 30 15 12 10 15 20 16 8 34 19 36 13 16 23 39 16 15 18 *9 37 H-6 15 7 7 6 11 12 9 5 21 12 22 15 10 13 38 26 9 12 26 *9 H-7 10 4 3 4 5 5 441 10 6 9 10 6 7 1717 5 5 15 30 FARALLON ISLANDS 17 19 19 10 18 28 21 12 44 12 5t 4411 16 10 4 12 11 11 3 NOTE: ** GREATER ThAN 99.5 PERCENT; N = LESS THAN 0.5 PERCENT. TABLE S. -- PRODABILITIES (EXPRESSED AS PERCENT CHANCE) THAT AN OIL SPILL STARTING IN THE SPRING SEASON AT A PARTICULAR LOCATIOM WILL CONTACr A CERTAIN TARGET WITHIN 30 DAYS. HYP(IRTICAL SPILL LOCATION TARGRT STERN (S) P1 P2 P3 P4 PS P6 P7 PS P9 PIO P1l P12 P13 P14 PIS P16 PI7 P19 P19 LAND 21 57 70 28 40 54 79 17 54 16 55 92 6 12 68 90 7 4298 57 A-I H25 9 65 67 13 N 31 N 2 N N N N N N M I N N A-2 2 58 34 34 6476 13 15 5 21 N NI N I Ni I N A-3 2 44 49 13 26f61S55 6** 27 N 112 N N IlIKN A-4 1 21 22 7 12 22 36 3 19 2 NIt N 3 11 Ni N A-S N 3 3 1 3 5 5 N A N 5 3 N N 3 1 N N N N 0-1 H IS 7 5938? HS1 N A N N MN N N IHI N N 8-2 2 46520 42 62 49 10 21 5 4 NIl N NIlIN N a-3 2 42 40 19 33 59 34 10 96 2 13 M. 1 1 2 H I I I N 0-4 2 27 30 10 18 34 38 4 65 2 19 N 1 1 A N I N I M B-s 179 8 3 6 10 12 2 8175 N MN6 2 1 H I M 8-6 N 1 3 1 2 3 2 1 3 M 1I10 N N 2 2 N N I N C-i N 12 5 51 27 5 NM66 N ** N N 2 1 N M 2 N N C-2 2 36 14 54 54 32 8 35 5 IS M N 12 NH i 1 2 N N C-3 3 42 33 30 45 53 21 15 77 6 ** N 1 2 1 N 2 1 1 N C-4 3 36 33 17 27 51 34 8 81 3*N* NIl 5 N 2 2 1 C-5 2 14 16 6 12 20 21 4 24 1 18 7 N 1 12 3 1 N 2 1 C-6 1 3 8 3 4 8 9 2 H M 5 ** N M 996 M NI N D-1 N 7 4 34 174 NM58 SM98 N N ** 4 NHN 3 1 NH D-2 3 27 9 50 38 22 5 45 4 35 M N 4 *4N H 4 2 1 M D-3 6 33 21 34 44 39 15 20 26 9 43 N 2 4 N N 3 1 2 N D-4 6 37 31 20 33 50 27 12 71 4 84 M 1 3 6 M 2 1 3 1 D-S 4 19 18 10 17 28 2, 6 36 1 30 7 1 2 ** 4 A N 2 2 0-6 298 11 4 8 It112 3 14 l 993 Ni3e I IS I M 11 2 E-1 M 6 3 25 IS 3 H49 N 93 N n** 4 MH 9 4 1 NHN E-2 2 20 8 47 30 15 4 52 4 82 M N 15 N N 6 2 H N E-3 9 31 18 39 40 28 13 26 20 20 21 N 5 10 N N 4 2 2 H E-4 11 38 32 28 36 47 23 16 63 9 74 H 3 5 9 N 4. 2 4 1 E-s 6 25 22 13 19 34 23, 8 47 3 42 11 2 3 A* 7 2 1 5 2 E-6 3 13 13 7 10 16 14 3 24 2 16 73 1 2 36 *6 1 M 3 2 F-i M S 2 21 13 2 H 38 N 50 N HM97 4 NH? 7 3 M M F-2 2 17 5 38 24 12 4 54 3 83 M N 62 99 N N 16 5 H N PF3 * 27 IS 4236 23 11 34 18 3716 M 13 37 1 N 10 4 2 H F-4 44 36 30 36 41 43 21 22 53 17 63 N 6 13 12 N 6 2 9 1 F-S 9 28 24 20 26 41 24 12 55 6 59 10 3 6 83 13 A 2 10 3 F-6 6 19 16 9 16 23 19 6 35 3 30 52 2 3 63 83 2 1 6 6 G-1 N 4 2 16I10n2 HN25 N 1B H N 75 3 N N27 S N N G-2 1 12 4 30 17 8 3 45 2 67 1 H 79 83 N MH** 10 M N G-3 91 20 10 35 26 17 7 36 12 41 10 N 22 73 1 N 16 7 1 N G-4 79 33 25 37 38 34 IS 26 35 25 39 M 9 22 10 N 7 4 *4 I G-5 16 28 23 23 28 39 21 15 48 10 52 7 5 10 53 12 5 2 22 4 0-6 10 20 15 12 18 27 18 7 37 4 36 p19 3 5 58 36 3 2 9 15 G-7 3 8 4 3 5 8 6 2 13 1 12 8 2 2 20 18 1 2 9 11-1 N 4 2 13 9 H N 20 N 13 N 40 3 N N 21 27 NMH U-2 1 10 3 25 15 6 2 38 2 57 H N 79 61 N N 94 " N N U-3 53 17 8 32 24 15 6 42 0 59 8 N 41 78 1 N 39 16 1 H 11-4 79 28 20 38 38 30 13 34 25 36 29 1 18 38 8 N IA a A* N 1U-5 35 30 22 28 34 36 18 19 43 17 49 7 7 15 34 9 7 4 at 4 11-6 13 25 17 16 23 30 17 9 42 6 41 16 4 8 56 26 4 2 17 *' H-7 8 14 9 6 12 16 10 S 23 3 20 11 2 4 34 21 3 1 8 28 FARALLON ISLANDS 3 23 18 20 30 33 13 10 23 5 42 N 1 2 1 N 1 I I M NOTE: 14 x GREATER THAN 99.5 PERCENT; MLESMS "IAN 0.5 PERCENT. TABLE 6. -- PROBABILITIES (EXPRESSED AS PERCENT CHANCE) THIAT AN OIL SPILL STARTING IN THE SUMMERf SEASON AT A PARTICULAR LOCATION WILL CONTACT A CERTAIN TARGET WITHIN 30 DAYS. HYP(foTITICAL SPILL LOCATION TARGET STERN (S) P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18 P19 LAND 20 47 67 19 34 52 71 17 58 12 52 90 7 12 68 83 5 2 29 61 A-i N 26 14 66 63 17 N 2 N N N N N N N N N N N N A-2 N 57 32 30 69 77 17 7 2 N N N N N N N N N N N A-3 N 45 47 8 27 58 55 2 ** N N N N N N N N N N N A-4 N 19 22 3 7 23 32 N 19 N 2 N N N N N N N N N A-S N 3 3 1 1 2 3 N1 2 N 1 N N N N H N N NH 0-1 N 18 13 61 34 12 N 52 N N N N N N N N H N N N B-2 N 40 20 41 62 46 15 16 3 N N N N N N N N N N N B-3 N 43 34 13 39 58 36 4 S95 1 3 N N N N N N N N N B-4 N 29 29 5 15 38 37 1 56 N 9 N N N i N N N N N 0-5 N 8 9 1 4 9 11 N 8 N 3 N N NH N N N N NH 9-6 N 2 1 N 1 1 1 N 2 HN 1 2 N H 1 N N N N N C-l N 18 12 48 25 13 N 64 N ** N N N N N N N N N N. C-2 N 32 17 52 52 33 14 35 6 10 N N N N N N N N N N C-3 N 41 27 28 47 50 22 11 72 2 ** N N N N N N N N 14 C-4 N 37 34 12 29 50 36 4 83 1 ** N N N N N N N N N C-5 N 16 17 3 9 23 22 1 30 1 13 N N N 3 N N N N N C-6 N 5 6 1 3 8 8 N 8 N 5 ** N N 2 N N N N D-1 N 14 10 31 19 12 1 53 N 99 N N ** 3 N N N N N N 0-2 N 25 14 46 36 21 12 45 6 41 N N 1 ** N N N N N N D-3 N 32 20 33 44 34 15 18 32 6 45 N 1 1 N N N N N N D-4 N 35 29 18 37 49 28 8 69 3 82 N N 1 N N N N N N D-5 N 21 19 5 15 30 25 2 43 1 36 16 N N ** N N N N N D-6 N 10 10 2 5 14 14 N 18 N 9 97 1 N 11 0 N N N N E-1 N 13 10 25 16 11 1 39 N 79 N N ** 3 N N N N N N co E-2 N 22 12 40 28 10 12 50 7 79 N N 11 ** N H N N N N E-3 1 27 16 35 37 271 13 28 21 20 26 N 2 6 N N 1 N N N E-4 2 35 27 23 42 44 23 13 55 5 69 N N 1 16 N N N N N E-5 1 24 22 9 21 34 25 4 50 2 49 20 N 1 A 414 N N N N 2-6 N 16 13 3 9 21 10 1 27 1 21 75 N N 39 ** N N 1 N F-I N 11 8 23 16 9 2 34 N 49 N N 99 4 N N N N N N P-2 N 20 11 34 26 16 10 49 8 90 2 N 60 99 N N 8 N N N F-3 . * 26 15 37 33 21 13 36 19 40 24 N 10 43 H N 4 N N N F-4 46 36 23 31 43 39 19 22 40 16 49 N 3 8 22 N 1 N 1 N F-5 4 29 23 15 30 37 24 8 51 5 56 18 1 2 87 21 N N 2 N P-6 1 20 16 7 15 27 20 4 41 2 37 41 N 1 62 90 N N 2 1 G-1 N 8 6 20 11 7 2 25 N 1B N N 69 5 N N 21 N N N G-2 N 17 10 27 19 12 9 42 7 61 6 N 83 79 N N ** 2 N N s-3 94 21 12 33 27 17 12 35 16 52 19 N 24 66 N N 12 2 N N -4 82 26 19 34 37 30 14 27 28 24 30 N 5 23 20 N 3 N * H N G-S 12 28 19 20 32 32 19 12 41 8 47 15 2 4 51 20 1 N 15 N G-6 3 18 14 8 18 26 15 5 38 2 39 15 1 1 53 35 N N 3 4 G-7 1 6 5 2 6 10 5 2 16 1 12 5 N N 22 15 N H 1 4 11-1 N -7 5 19 11 5 2 18 1 14 N N 37 5 N N 23 24 N N 11-2 N 15 10 26 17 11 8 37 0 52 7 N 79 51 N N 97 ** N N 11-3 50 19 12 31 24 18 11 36 14 56 18 N 47 69 N N 44 11 N N 11-4 03 25 17 38 36 2H 13 32 25 39 24 2 21 44 18 N 11 3 ** N H-5 40 29 20 29 31 30 17 18 33 15 38 13 5 15 33 10 3 1 ** N 11-6 8 21 15 12 22 28 16 B 39 4 42 13 2 4 47 27 1 N 10 ** H-7 3 11 9 6 13 18 10 4 26 2 25 8 1 2 37 18 N N 3 22 FARALN ISLANDS N 24 16 19 30 20 12 6 30 2 43 N N N N N N N N N NOTE: * GREATER THAN 99_5 PERCEN T - AW m so w M ift ow~~w"PI~~E m OW LE AO PRJILITIMXPREI AS M6 CM, JRB oiL, STARTING W IN llE AtUIM SEASON AT A PARTICULAR LOICATION ILL WMACTI A CETAIN TARGET WITUIN 30 DAYS. HYPOlIHETICAL SPILL. LOCATION TARGET STERN (S) P1 P2 P3 P4 PS P6 P7 PO P9 P10 Pll P12 P13 P14 P1S P16 P17 P1 P19 LAND 12 27 46 12 18 a27 53 7 31 7 30 81 3 9 42 73 5 4 19 30 A-1 1 46 20 50 7633 4 24 2 3 1 N 2 1 N N 1 N N N A-2 1 54 52 22 49 79 35 10 13 4 3 N 3 1 N N 1 N N N A-3 3 23 47 0 17 43 62 6 ** 2 9 N 2 a 1 N 2 1 N N A-4 3 1019 5 7 13 27 3 20 2 14 1 1 1 3 N I I N N A-S 1 3 5 2 1 2 5 N 2 1 7 5 1 1 5 2 N N 1 N 9-1 1 34 17 51 57 24 4 35 2 9 1 N 32 N N N N 8-2 2 51 40 27 57 66 30 12 15 7 4 N 3 2 N N a 1 N N 9-3 3 27 45 14 21 51 49 8 96 4 23 1 3 2 1 N 2 1 N N 9-4 3 12 21 7 9 19 34 5 43 a 29 2.2 a 5 1 1 I 1 1 0-5 1 4 5 2 3 a3 N 5 2 10 9 N 1 8 5 N 2 N 0-6 N 2 1 1 1 N 1 N 21216 N I 6 6 N N I N C-1 1 33 16 50 48 21 4 49 3 h 1 N 711 N 2 1 N N C-2 2 51 37 34 60 55 24 19 17 16 5 N 7 6 N N 4 2 N N C-3 5 31 43 18 32 52 40 10 85 7 1 4 4 2 N 4 2 1 N C-4 7 16 27 11 13 29 40 6 71 5 ** 3 3 3 9 2 2 2 3 1 c-5 3 8 9 4 5 6 12 2 12 2 16 162 1 19 11 1 2 3 1 C-6 2 3 4 2 2 3 4 N 4 1 4 ** N 1 11 1B N N 2 1 D-1 N 26 13 42 37 17 5 51 3 99 1 N ** 7 NN 3 2 N N D-2 3 44 30 35 53 41 20 25 15 34 5 1 11 ** 1 N 6 4 N N D-3 8 33 36 19 37 47 30 12 52 11' 63 1 7 5 2 N 6 3 2 N D-4 9 20 29 12 15 36 37 8 71 6 78 3 4 4 13 3 3 2 5 1 D-5 6 11 12 5 6 11 17 3 22 4 24 26 3 3 ** 13 1 1 4 2 D-6 2 4 4 3 2 4 6 1 7 1 6 92 2 2 20** N N 2 2 E-1 1 23 12 41 32 16 4 51 3 91 1 N *k 8 N N 5 4 N N E-2 3 41 28 38 47 34 18 34 16 61 6 1 21 ** 1 N 10 6 N N E,3 16 35 33 23 38 44 27 13 37 15 40 1 11 12 3 1 9 5 3 N e-4 17 22 29 14 19 41 35 9 69 8 74 2 6 7 26 3 4 3 7 a E-5 8 13 14 5 7 15 20 4 32 4 32 26 3 3 ** 27 1 1 7 3 E-6 4 5 6 4 3 6 8 3 9 2 10 752 2 29 I' 1 N 4 3 F-i 1 21 10 43 31 15 4 52 3 73 1 N 99 11 N 1 13 7 N N P-2 7 37 25 39 45 32 16 40 14 76 8 1 52 99 N 1 23 13 N N F-3 * 36 31 23 42 39 27 17 33 25 34 1 16 33 3 1 13 8 4 1 P-4. 50 25 31 17 26 43 35 11 63 12 70 2 9 12 33 5 8 5 16 3 P-5 12 15 18 8 11 20 23 5 42 4 41 23 4 7 88 30 2 3 s15 7 F-6 6 8 9 4 5 0 12 4 15 3 18 57 2 5 50 82 1 1 10 10 G01 1 19 9 34 26 13 4 44 3 47 1 N 84 11 N N 45 12 N N G-2 7 31 21 37 40 27 12 42 11 70 8 N 71 91 N 1 ** 23 N N G-3 92 31 27 25 37 35 20 21 27 32 27 N 19 57 3 1 19 12 3 1 G-4 75 27 29 18 .29 39 32 13 51 14 57 3 11 10 26 5 11 6** 3 0-5 18 16 19 10 13 25 23 6 47 7 46 17 4 8 69 24 5 4 33 9 0-6 8 10 11 5 6 10 11 4 21 4 21 28 3 5 51 46 3 2 15 19 G-7 2 5 3 2 2 2 4 1 6 1 8 9 1 1 12 18 N 5 11 11H-1 2 17 9 31 25 11 4 42 2 37 1 N 62 14 N N 39 4e N N 11-2 10 30 19 38 40 24 11 46 10 67 8 N 77 73 N N 93 aA N N H-3 65 34 24 29 39 34 19 26 23 42 24 1 31 70 2 1 33 20 5 1 11-4 76 29 29 23 34 41 30 16 45 21 50 -3 15 28 24 6 16 10 * 3 H-5 27 17 21 14 17 28 24 9 48 9 49 15 8 11 57 20 8 5 * 4 H-6 11 11 13 7 7 13 12 5 27 6 25 25 3 7 54 38 3 2 23 ** 14-7 3 7 7 3 3 6 6 2 12 2 1 3 13 2 3 23 23 2 1 8 26 FARALLON ISLANDS 6 23 29 13 23 37 27 8 51 5 63 N 4 3 2 N 3 2 2 N NOTE: * = GREATER THAN 99.5 PERCENT; N LESS TIAN 0.5 PERCENT. 7-41 Trajectory Anal7sis and Modeling Support for the Puerto Rican Oil Spill J. A. Gait Seattle, Washington December 6, 1984 INTRODUCTION The maritime -ilosion on :he 7essei ?Puerto li=n, :he ;ubseauen: release of cargo, breakup of the ship, and the resulting pollutant release problems have provided a subject of considerable discussion. These discussions have taken place both in and out of the press, and have spanned the community of spill responders, responsible parties, special interest groups, and interested bystanders. The wide variety of opinions on both the event itself and the adequacy of various response efforts is a clear testimony to-the complex and many-faceted nature of major environmental spills. This will be a review of the trajectory analysis support for the 2uerto Rican spill event. This task began on October 31, 1984, and required continuous trajectory analysis and forecasting efforts until November 10. After that time, occasional consultations have been initiated due to the intermittent leaks from the sunken stern secticn of. ithe vessel. During this period, the coastal currents off the San Francisco 3ight went through a major reversal associated with the onset of a typical winter regime, or the so-called "Davidson Current:'. The details of this frontal passage and the comiLes eddy pat:er= in :he San Francisco 3ight contributed to the difficulty in carrying out =rajectl'r' anal7sis and making acc4orace forecasts of pollucant movement. The objec:ive of this discussion is to outline :he avaliabie information 7-42 INTRODUCTION, cont. that was used for the forecasts and to discuss the controlling oceanographic processes as they are understood. The second section of this paper presents a chronology of the trajectory analysis response to the spill evenc. T-he third section covers a discussion of oceanographic processes and how rno3wledge : these processes was used to make specific recommendations. The final section of this paper presents conclusions and a discussion of alternate trajectory analysis, or forecasting, procedures that could have been used for support during the Puerto Rican pollution event. CHRONOLOGY October 31, 1984 0600 Initial notification of the Puerto Rican explosion and ship casualty was reported to Jerry Gait of the Modeling and Simulation Studies (A.SS) group of the NOAA Eazardous Materials Response Branch (HMRB), Seattle. At that time, a partial list of the ship's cargo was provided, and an approximate location of the ship was given. An immediate check was made of available forecasted tidal infor-ation and a trajectory response was initiated. Debra Payton and Glen Watabayashi were called to the Seattle wIMRB office for immediate trajectory analysis support. 06J0 MASS personnel called the National weather Ser-ice (NW-S) office in San Francisco (Bill Hackle) and identified :wo offshore weather data buoys, (buov Ii26, just off the Goiden gace, and buov -!12, just off Half Moon Bay), as concinuous real-ti- e wind inrcr-mation sit:s which would be available during the crajec:or-7 analysis resoonse. At :hat time, 3uoy ,256 repor:ed norvhwest winds at four k:nosc, and 3uoys :1- 7-43 October 31, 1984, cont. reported north winds at ten to twelve knots. Afternoon and evening �winds for October 31 were forecasted for speeds of 15 to 25 knots from the northwest. The extended forecast for November i was for light and variable winds. Based on these data and expected tidai currents, :he initial trajectory f any oil that may liave spilled from The 3hi s present location (37045'N, 122�50'W) would have been to the south and nearly parallel to the coast. This information was passed to the Scientific Support Coordinator (SSC) on-scene at the U.S. Coast Guard Marine Safety Office in San Francisco. 0700 MASS personnel in Seattle immediately began to compile and enter data into the computerized trajectory analysis routines used for spill support. This information included tidal current data, forecasted wind data, digitized trajectory maps for the San Francisco Bight region, and a hydrodynamic circulation model for the San Francisco 3ight area which included geostrophic and Ekman dynamics to represent the surface flow. 0820 A call was placed to the National Marine Fisheries ?acific Environmental Group (PEG) group, located in Monterey. This group studies nearshore winds and upwelling processes along the California coast. A discussion with PEG researchers indicated that their best information showed that upwelling had ceased in the Monterev area and that northerly flow was expected in the Monterey Bay region (i.2., the Davidson current regime had advanced as far north as this region). 0835 Discussions were carried out with the NOAA contract chemist (Dr. Ed Overton) at Louisiana State Universit- (LSU) co ascertain :he nature of the ?uerto Rican's cargo componenrs. From these discussions, 7-44 October 31, 1984, cont. it was decided that the major components of the cargo were hydrocarbon additives and should be treated as hydrocarbons with regard to trajectory analysis. 0840 MASS personnel in Seattle obtained satellite analysis data of infrared sea surface temperature mans for the Califoarnia zoast. This imagery clearly delineated a band of lower temperature water along the California coast. This band of water indicated a residual of the coastal California system and highlighted the fact that the Davidson Current was not yet strongly established in the San Francisco Bight region. Based on this satellite imagery, the results of the 3 hydrodynamic circulation modeling for the region, discussions -with the PEG group, and on-scene reports of southerly currents, it appeared that the advance of the.Davidson Current front was located somewhere between Monterey and the San Francisco Bight region itself. This meant that flow in the San Francisco Bight region itself was to the south and that flow in the Monterey, and possibly Santa Cruz area, was to the north. 0845 Based on the available current information and wind forecasts, MASS passed a recommendation to the SSC on-scene that the ship should be towed offshore and south of the Farallon Islands. This recommendation was based on three specific factors. The first factor in this decision was the reported condition of the ship was extremely unstable with an active fire in progress. This suggested that a perhaps catascrophic release of oil was imminenc and that a large fraction of the cargo could be spilled at any Moment. All available initial trajector/ estimates indicated a southerly flow and the oniv aiternarive to taking the ship offshore and south of the Farallon Islands -would have 7-45 October 31, 1984, cont. been to leave it in place or to try and route the ship north of the Farallon Island. In either case, the potential of the spill would have been significantly more hazardous for the Farallon Islands region and the mainland coast area of the San Francisco Bight. The second major factor in :his iecision was that :he vydrcdynan-_ flow modeling of the San Francisco Bight area, literature, discussions, and personal experience indicated that a region of extremely complex current eddies lay inshore of the Farallon Islands. This region extended over a major section of the Bight region. A second area of eddy current formation was south of the Pt. Reyves area. It was felt that in it would be extremely difficult both to estimate trajectory movement in either of these areas and to determine the the ultimate region of impact if cargo were spilled in this region. The third factor in this decision was the convergence, or coming together, of the southerly-flowing currents which were present in the San Francisco Bight area, and the northerly-flowing currents which were suggested off the Monterey-Santa Cruz area. The joining of these currents would force a general offshore flow to the water, and this would tend to carry floating pollutants offshore away from the coast in the area south of the Farallon Islands. 0859 MASS received a follow-up call from the chemists at LSU. LSU had previously contacted toxicological snecialists at Chevron and obtained an updated, corrected description of the shiD's zargo. This description confirmed the previous conclusicr that the materials involved should be treated as various classes of hydrocarbons with regard to trajectory analysis. 7-46 October 31, 1984, cant. 0925 MASS was informed by the SSC on-scene that the fire on the Puerto Rican appeared to be somewhat better controlled, and that the ship was being towed on a course of 2680 at 3.3 knots. After receiving this information, MASS reiterated the suggestion that the ship's track be moved south of the Tarallon :siands, and pointed out :hat :he ihin '3 present course did not fulfill those requirements. At that time, MASS suggested that the ship fcllow a course between 2300 and 240� to maximize the distance between the ship and the Farallon Islands as the vessel passed out of the San Francisco Bight region. 1120 MASS personnel contacted the Pt. Reyes Bird Observatory (Sarah Aikin) and requested that hourly wind data observations, taken routinely by the biologists on the Farallon islands, be oassed to MASS on a regular basis. To accomplish this, MASS set up a three times daily transmission schedule from the Farallon Islands directly to MASS personnel in Seattle. MASS also discussed present circulation conditions around the Farallon Islands with observers there. These observers indicated that the general trend of the currents in the region was still south, and that the extension of the Davidson Current had apparently not reached the San Francisco Bight region. Once again, this indicated that the Davidson Current advance and its subsequent expected reversal was stalled somewhere between the Monterey B3a region and the San Francisco Sight area. 1211 The SSC on-scene asked for a specific recommendation of how far the ship would have to be towed. Since the ship was on fire and being towed direct.r away :rom firefigh:ing support euipment and logistic bases, there vas some concern about how :ar -c would have to 0 7-47 October 31, 1984, cont. before it would outrun all possible relief efforts. MASS personnel recommended that the ship should be towed towards the dump site located at the continental shelf break (37�32'N, 122�59'W) at a minimum, and that this course would optimize its movement away from the Farallon Islands -nd the California mainland with regard to anv threacs associated with catastrophic spills during the next few hours. During this discussion, contact by other MASS personnel was made with biologists on the Farallon Islands and updated on-scene wind information was collected and factored into the analysis. 1500 Computer maps and printouts of trajectory analysis estimates were forwarded to the U.S. Coast Guard On-Scene Coordinator (OSC) via NOAA's electronic mail system. These estimates indicated that spills that occurred during the initial explosion would move south with no landfall anticipated for the initial forecast period. At this time, the N-WS Office in San Francisco presented an updated weather forecast which basically extended and corroborated the morning's forecast. This forecast information was used to extend and update the computer analysis trajectory estimates. At this time, on-scene modeling and crajectory analysis support was requested, and Jerry Gait prepared to depart for San Francisco with an estimated arrival time on-scene of 2200. 2210 At this time, the final evening radio contact was made with the Farallon Islands to receive the upated on-scene weather. These observations, forecasted wind information received from NWS, and the observations from the offshore data buovs were used to updace the trajectory analysis forecast. 7-48 October 31, 1984, cont. 2300 Jerry Galt arrived at the U.S. Coast Guard Marine Safety Office (MSO) in Oakland, obtained a briefing of the current situation for the Puerto Rican, and presented MASS's latest trajectory analysis estimates that had been prepared by Debra Payton and Glen Watabayashi in :he Seattle office. These estimates indicatad :hac am- pocancfai sni`ls would have a southerly and slightly onshore movement, but that no landfall would be anticipated during a two-day forecast period. November 1, 1984 0800 On-scene MASS personnel forwarded the morning's briefing to the MASS group in Seattle, and provided an updated position of 37"21.1'N, 122�56.9'W for continuous trajectory analysis. This position put the ship at the outer edge of the submarine operation area, some 20 miles south of the Farallon Islands. A4SS recommended that the Puerto Rican remain in this area or move slightly cifshore, but that it should not return north, or closer, to the Farallonl Islands. This recommendation was based on the persistent northerly winds and the fact that the advance of the Davidson Current had been stalled south of the ship's location. Wind forecasts predicted weak or variable winds, with occasional stronger winds from the north or northwest. In addition, the ship was positioned over the mouth of a large submarine canyon, and thus over particularly deep water. This meant that if the ship were to sink at its present location, it would be below the region of trawl fisheries, and it would also provide some additional protection for the Farallon Islands because currents in the region tended :o follow bathymetric contours. Taus, northerlv or southerly flcw would follow 7-49 November 1, 1984, cont. the slope of thle submarine canyon and tend to be offshore, regardless in which alongshore direction the flow of the spill initially moved. All of these arguments indicated that the present ship position would optimize spill threats, or sunken vessel threats, to 1) the .arallon Islands; 2) the mainland coast area of :he San 7rancisco 3ay region; and 3) the fisheries potential of the shelf region. 0910 MASS personnel in Seattle were requested to put in additional digitized computer maps to cover the larger area of the extended San Francisco Bay Bight region. Throughout the day of November 1, additional wind information was gathered from the NWS office in San Francisco, and from direct contact with observers on the Farallon Islands. This data and on-scene reports were used to provide continual updates for trajectory estimates, and these were forwarded to the U.S. Coast Guard through the-SSC. These trajectory analysis results indicated a southerly flow for any spilled pollutants during this period. In addition, on-scene MASS personnel worked with other HMRB personnel to gather background information to factor into planning for the possible use of chemical dispersants if a major oil release occurred. 1700 A general debriefing meeting was held at the U.S. Coast Guard MSO in Oakland, and at that time, MASS recommended that the present position of the ship should be maintained in the outer segment of :he submarine operation area, over the submarine canyon. Thlis recommendation was mutually agreed upon between everyone participating in the meeting, including :he U. S. Coast Guard and 'TOAA scientrflc support personnel. 7-50 November 2, 1984 During the morning, estimates were made of the area covered by spilled oil from the previous day. This information was incorporated into the possible dispersant use plan. MASS and U.S. Coast Guard personnel participated in an overflight of the ship's site, and mapped the position of spilled oil. n add.i-on, '-L4SS perlormed iffaranr:ai oil-water velocity movements for the yellow, or light brown, component of the spilled oil (tentatively identified as lube oil additives). These measurements indicated that the radiation wave pressure on this component of the oil was minimal, and, as a result this fraction of the spilled cargo may have anomalous results with regard to typical hydrocarbon spills. In particular, there should be less of a tendency for wind drift to affect the movement of this component, and there would be less of a tendency for this material to beach under the influence of wind. Throughout the day, continual weather information was gathered from both the Farallon Islands and NWS sources. Trajectory analyses were continued, and hindcasts were made to comoare co the previous day's sightings, and to upgrade and correct the hydrodynamic flow model estimates. All of the observations and trajectory analysis information, combined with available wind data, indicated the presence of southerly flowing currents in the area of the ship and the spilled material from the previous day. This information also indicated the absence at that time of the advancing Davidson Current front for the same area. November 3, 1984 0100 On-scene !4ASS personnel vera informed that :he shiD had broken in tcwo, and that three floating pieces had been identified. :n 7-51 November 3, 1984, cont. addition, during the night the ship had been moved off-station and was north of the recommended area over the submarine canyon. Personnel in Seattle were immediately alerted to begin work on trajectory analysis updates, and both Seattle and on-scene MSS personnel worked on updacing the computer informacion and trajecror- inalysis estimatas. 'hese estimates were routinely passed on to the U.S. Coast Guard via the SSC. During the morning of November 3, oil was reported offshore of the Golden Gate region, and during a morning overflight by U.S. Coast Guard and MASS personnel, these locations were checked and it was dtermeined that no oil was present in these areas. During the overflight, a detailed map of the spilled oil was made with the distribution of oil plotted from both the sunken stern section and the bow section which was under tow. Throughout the day, continual weather updates and trajectory analysis results were presented, and indicated that the spilled oil was predicted to move nearer to shore and to the south over the next two-day period. November 4, 1984 Overflights and mapping of the oil continued with U.S. Coast Guard personnel and members of the NOAA scientific support team. Trajectory analysis results were presented at a press briefing with an estimate hat the oil would continue to move east during the day, and then south. Throughout the forecast period, the oil was not expected =o make anv landfalls. On the afternoon of November 4, Jerer Gait, the on-scene MASS group member, returned to Seactte. 7-52 November 5, 1984 Throughout the day, overflight information, trajectory analysis estimates, and additional weather information were used to forecast oil movement. A continued southerly path was predicted, and observations indicated that the oil was in fact moving south and slightly offshore. During :he morning of November 5, on-scene rports or il on 3 Montaro Beach were investigated and determined to be negative sightings. Also on the morning of November 5, strong northwesterly winds were reported, with the afternoon forecast indicating a shift to strong winds from the south. Trajectory analysis results anticipated that the strong southerly winds would slow down the south-flowing currents with the possibility of causing a weak current reversal. Trajectory analysis estimates based on this data indicated that the oil could stop its southerly motion and could possibly move slightly north. November 6, 1984 On the morning of November 6, oil was reported around the Farallon Islands, some 20 miles north of where it was reported the previous evening. This indicated that a current reversal and extremely strong current event had occurred during the night of November 5-6. This had the effect of moving a major portion of the spilled oil to the vicinity of the Farallon Islands and into the San Francisco Bight area itself. By the afternoon of November 6, the oil had moved beyond the southern Farallon Islands and into the region inshore from the Farallon Islands where the current patterns were dominated by intermediate-si:ed eddies, and :he complex nature of the flow made it extramely difficuit to predict detailed motion. 7-53 November 7-November 9, 1984 During this period, a weak northerly flow was present throughout the San Francisco Bight region, and the floating oil between the Farallon Islands and Pt. Reyes was subject to a weak and complex current pattern that moved it generally north and northwest up around the Pt. Reyes region. The major components of the floaing pollutant appeared to be the lube oil additive which had previously been observed to have a minimal wind drift factor. This material also showed a reduced tendencyv to come ashore, and as such, the shoreline impacts were considerably less than could have been anticipated from a heavier form of hydrocarbon. During this period, Jerry Galt returned on-scene to the U.S. Coast Guard MSO, San Francisco. November 10-November 11, 1984 After the floating oil residual from the spill moved north around Pt. Reyes, it continued to move north under the influence of the wind and the advancing Davidson Current. Once beyond the Farallon Islands and Pt. Reyes region, this current formed a more persistent flow and the remainder of the oil moved north past Bodega Bay, where some oil -vent ashore, on to the Russian River area, and eventually off Pt. Arenas. During this period, trajectory analysis once again was able to indicate the expected movement of the oil and the slight offshore tendency associated with the convergence seen along the advancing Davidson Current front. After November !2. 1984 After the initial major spill was tracked to the ?'. Arenas reg�on, only occasional trajectory anal-sis support -was requested for intcerittent spills that were released from the sunken stern sec:tion 7-54 After November 12, 1984, cont. south of the Farallon Islands. Each of these relatively small releases were subject to a weak northerly flow that tended to follow isobathic, or constant depth, contours. These current movements and local wind effects in each case tended to move the spilled oil to the northwest for several miles before it was dissipated to :he point -here -; was not recognizable as an observable surface sheen, and was below the level of concentration where it could be cleaned up. OCEANOGRAPHIC PROCESSES AFFECTING TRAJECTORY ANALYSIS ESTIMATES DURING THE PUERTO RICAN SPILL EVENT The California coastal current system is notoriously complex and has been the subject of intense study for many years. Particular long-term studies have been carried out by Scripps Institute of Oceanography and the California Coastal Fisheries Project. A comprehensive review is presented by Dr. Hickey in Progress of Oceanogravhv, Volume 8, published by Pergamon Press. SMore recent detailed data has been available from various satellite imagery projects and National Ocean Survey and National Weather Service investigations. In a general sense, the coastal currents off San Francisco can be described as falling into one of two general regimes. The first regime is southerly flow, or the so-called "upwelling" period which is typically present during summer months. The second regime is northerly flow, or the so-called "Davidson Current" which is typically present during the late fall and winter. Based on this simpliscic view of the offshore currents, one might conclude that during the Puerto Rican spill event, a persistent northerly trajectorv would be expected. During the first few hours of the ?uerto Rican spill event, it was established that 7-55 OCEANOGRAPHIC PROCESSES, cont. this was not a typical or climatologically average fall period, and that the Davidson Current had not vet established itself as dominant in the San Francisco Bight area. Rather than a dominant Davidson Current period, it appeared that the frontal advance of the Davidson Current was slowly working izs way up the coast and was .siuared somewhere between the Monterey Bay area and the San Francisco Bay region as of October 31. In coastal regions where a seasonal reversal of the currents is present, it is typical that the reversal does not occur simultaneously along the entire coast. Mfore commonly, the current reversal works its way up the coast as a progressive front. Behind this front, currents will be in one direction; ahead of the front, currents will be in the other direction. The progress of this front does not move uniformly; it will typically be affected by the local winds and can be seen to advance under favorable wind conditions, stop or stagnate during contrary winds, and under extreme wind events, may actually reverse and move back down the coast in the opposite direction of its advance for short periods. This was the situation that was anticipated for the Puerto Rican spill period. It was assumed that the Davidson Current front would continue to move north, and at some stage the south-flowing regime in the Bight area would stop and then show a reversal with possibly some period of ambiguity and oscillatory currents as the front passed through the San Francisco Bight region. Based on this conception, the computer analysis, the observational data that was available, and the theoretical descriptions of the seasonal nature of the currents, :rajectory analysis correctly described the pollutant movement -rom the initial accident on October 31 through the evening of November ;. 7-56 OCEANOGRAPHIC PROCESSES, cont. During the late afternoon and evening of November 5, an ex:r-emei- rapid northerly movement of the pollutant indicated the passage of the Davidson Current front as a very energetic event which is atypical of the vast majority of the available observations for this type o frontal passage. To tCr- and explain this unusually energetic fr-ontal system, a number of factors must be considered. First of all, there were strong sustained winds during the week previous to the frontal passage in the region of the Farallon Islands. These winds appeared to actually retard the advance of the Davidson front more than usual. In addition, the northerly and northwesterly winds tended to actually move water out of the San Francisco Bight region and the shelf area just south of the Farallon Islands. During the afternoon of November 5, the reversal of .the wind direction and the intensity of the flow not only caused the reversal of the current, but also forced a significant on-shelf movement of offshore water. As this water moved onto the shelf and encountered the rapid variations in depth along the shelf break, an intense current jet was formed. This jet-like phenomenon was a strong and narrow current that was present right along the edge of the shelf break, which happened to be where the major patch of oil was on the afternoon of November S. The duration of this current was between 12 and 18 hours, and it was probably no wider than a few kilometers. Following this et-1like relaxation and readjustment of the frontal movement, the Davidson Current advance continued in a more or less predictable manner into the San Francisco Bight and Farallon islands region. From this point on, the area was dominated by a relatively weak northerly flow throughout the region. 7-57 OCEANOGRAPHIC PROCESSES, cont. A complete understanding and description of the relaxation and jet-type current that appears to be responsible for the movement of oil to the vicinity of the Farallon Islands cannot be obtained. Some theoretical discussions of this type of flow are presented in the oceanographic !iteratur3, ': no detailed observational documentation has been available from preazious studies. The effects of one suca jet, however, have been observed during an intense 10-month observational program that was carried out during the IXTOC 1 well blowout in the Gulf of Mexico. 'At that time, a seasonal reversal along the Texas coast was being tracked using a number of radio-tracked buoys deployed in a line perpendicular to the shore. During this jet relaxation event, one of the buovs in the line was displaced some 20 miles over a 12 to 18 hour period during which time the remaining buoys in a. line perpendicular to the shore moved only slightly in the same direction. This type of phenomenon is extremely complex and beyond the ability of hydrodynamic modeling to predict at the present time. Frontal passages are always difficult to predict and small-scale rapid movement such as this Jet cannot be properly included in any preseutly available forecasting techniques. Since they are rare, this is generally not a problem. Within the San Francisco Bight region itself, the general alongshore pattern of the currents is disrupted by the complex geometry7. Modeling studies indicated that the Farallon islands'cau.se a major disrurcion of :he alongshore flow with the developmenc of strong eddy patterns or small-scale :ircualtion features behind and inshore from the islands. Similar small-scale features are seen to the south of Pt. I) 7-58 OCEANOGRAPHIC PROCESSES, cont. Reyes. Pollutants spilled in this inner area are expected to generally mill around and move slowly in the direction of the dominant current regime. Once again, during the first few hours of the spill event, this region was recognized and identified as one where decailed trajector- analysis would be difficult or impossible. As was noticed early in the observational studies of the spill, the differential oil-water motion associated with the lube oil additives appeared to be much Xess than would typically be expected for hydrocarbon products. This meant that the spilled pollutant had a tendency to follow the water more closely and not be affected by the local winds as much as a normal oil. This had the advantageous effect that the oil would not be as likely to go ashore, and this was in fact observed as the oil went around the Farallon Islands and later as it rounded Pt. Reyes and moved north towards Bodega Bay. This would definitely be considered as a positive factor with regard to the beaching of the oil. However, it had a disadvantage with regard to the dispersive effects that the wind can have on an oil slick. The patches of lube oil were seen to remain together as consistent, coherent patches well beyond the period that would normally be expected, and that has been observed in many previous spills. The movement of the major spilled oil along with the front of the Davidson Current had an additional advantageous effec:t. In particular, as the Davidson Current moved north it encountered the residual of the southerly flowing currents along the coastline. Where these two currents net, the water was forced offshore. This presented a weaK offshore tendency as the flow moved north along the coast, and had :he 7-59 OCEANOGRAPHIC PROCESSES, cont. effect of holding the spilled pollutant somewhat away from the coastline, thus reducing potential beach impacts. This effect was recognized during the initial trajectory analysis studies and, in fact, was considered as one of the factors in initially requesting that the ship move south towards this suspected convergence and of-snor flow condition. CONCLUSIONS The Puerto Rican spill event required continuous spill trajector7 analysis for a period of about 12 days. Within the first two to three hours of notification, the trajectory analysis team located evidence that the typical Davidson Current, or winter current regime, was not present. This resulted in correct trajectory estimates which were counter to the climatologically-averaged expectations or "conventional wisdom" for the region during this time of the year. Initial dynamic modeling results also available within the first few hours indicated troublesome and uncertain flow regimes inshore from the Farallon Islands and south of Pt. Reves. The advancing front of the Davidson Current regime was estimated to be north of the Monterey Bay area, and its arrival in the San Francisco Bight region was anticipated, but short of actual observations of the frontal passage, no detailed theoretical procedures were available to predict its arrival at any particular site. Prior to the frontal arrival (six days into the spill), standard trajectory analysis procedures gave basically correct descriptions of the direction and speed of spilled pollutant movement. The derailed arrival time of the front associated with the Davidson Current, and more specifically, the intense nature of the shelf break 7-60 CONCLUSIONS, cont. jet, were not anticipated; thus, for the overnight period of November 5-6, the trajectory analysis predictions were thus wrong. 3v the evening of November 6, the frontal passage of the Davidson Current was recognized, modeling procedures were updated, and once again =rajectory techniques were :esolving generally correct stimatas of oollutant movement. This situation then continued throughout the remainder of the spill incident. Reviewing this performance of the modeling and trajectory analysis techniques that were used, we might consider improvements or alternate procedures that might have been possible. First of all, we can categorically state that the use of climatological data to estimate pollutant movement would have been a significant error. In particular, this would have led to incorrect results for almost exactly half of the spill event duration. More significantly, this error would have occurred during the early part of the spill when the largest amounts of undispersed oil were being found in concentrated areas. As a second point, we may note chat the early observations of southerly flow and the evidence of a stalled Davidson Current front were critical for obtaining accurate trajectory estimates during the first six days of the spill. An obvious question would be, "'What kind of additional observations would be necessary to track the advance of the Davidson Current front as it approached the San Francisco Bight region?" Some experience during the IXTOC spill and studies of current reversals along the Texas coast indicate that frontal regimes can be cracked using air-deplyvable current probes and helicopter mapping techniques. ln order to suppor: this kind of activity, two or more specially trained 7-61 CONCLUSIONS, cont. personnel and the logistics support for multiple helicopter overfligh:s must be available. With this kind of added support, it might be possible to track the advance of the frontal system itself and be able to more closely predict the time of the reversal at any particular location along the coast. As a final point, one might consider what would be necessary to observe the intense current jet which was seen along the shelf break during the night and morning of November 5-6. In order to delineate this feature, it would be necessary to have in place many hundreds of thousands of dollars' worth of experimental.current imagery equipment. Under these conditions, it might be possible to actually document the details of the jet. One might note, however, that even under optimum conditions when this current could be documented, there is still no theoretical basis on which to base predictions for a trajectory. The only advantage that could be obtained by this kind of data is that its passage and currents could be documented. There would still be no -wa,'r to incorporate this into ant a priori forecast. It is hoped that this chronology and discussion will provide useful information and details for the ongoing review and critique of the response efforts associated with the Puerto Rican oil spil. R47 1986