[From the U.S. Government Printing Office, www.gpo.gov]
Coastal Zone Information Center OCT 30 1975 JAN 13 1998 LONG ISLAND SPILL TRAJECTORY STUDY by J. W. Devanney III Robery J. Stewert Massechusetts Institute of Technology Report to Regional Marine Resources Council Nassau-Suffolk Planning Board COASTAL ZONE INFORMATION CENTER 28 February 1974 U.S. DEPARTMENT OF COMMERCE NOAA FINAL COPY COASTAL SERVICES CENTER 2334 SOUTH HORRON AVENUE CHARLESTON, SC 29405-2413 TD Property of COE Library 427 .P4 D48 1974 Nassau Suffolk Regional Planning Board Leonard W. Hall, Esq. Veterans Memorial Highway Hauppauge, L.L, N.Y. 11787 Chairman Area Code (516) 724-1919 Seth A. Hubbard, Esq. Vice Chairman OUCT 3 0 1975 Vincent R. Balletta, Jr. H. Lee Dennison Walter G. Michaelis April 9, 1975 Thomas Halsey Lee E. Koppelman Executive Director 15 APR 1975 CzM InformationO @67/., Mr. Richard Gardner Office of Coastal Zone Management NOAA Rockville, Md. 20852 Dear Mr. Gardner: As a result of our Federal Project Advisory Committee Meeting on HUD Contract H-205OR held in Hauppauge on April 3, 1975, 1 am sending you under separate cover copies of the following reports for your information and review: 1. Long Island Spill Trajectory Study; 2. Potential Biological Effects of Hypothetical Oil Spills Occurring in the Nearshore Waters of Long Island's South Shore; and 3. Probabilistic Trajectory Assessments for Offshore Oil Spills Impacting Long Island. These reports were produced as part of the Nassau-Suffolk Regional Planning Board's study on the implications and impacts associated with the development of potential oil and gas reserves on the Atlantic Outer Continental Shelf. Should you have any comments on these reports, I would appreciate hearing from you. Sincerely, LEK:er Lee E. Koppe n Boa. Executive Director Purpose oil spills can be transported many miles.from the site of an accident by the action of the winds, waves, and currents found in the offshore region. Any proposal to develop petro- leum deposits in the offshore region must therefore be viewed from the standpoint of the possibility of spills originating at the development site and eventually impacting the shores of nearby coastal communities. Such a possibility is of great importance to coastal communities because, as was pointed out in "The Georges Bank Petroleum Study," (Offshore Oil Task Group, 1973), these communities are in the unenviable position of bearing a potentially significant portion of the cost of the development, while the benefits will be distributed rather uniformly over a much larger group. Furthermore, the costs incident on coastal communities are typically associated with the disruption of esthetic values, or perhaps in the impaired viability of the local ecology. Schemes for rewarding due compensation for these effects are highly controversial. Many wou ld hold that no compensation is adequate. In view of the uncertainties and inequities involved, a key issue for coastal zone planners is therefore what are the risks involved in any particular offshore development and where are they centralized. With respect to the possible discovery of petroleum deposits in the region lying to the south of Long Island, thekey questions for Long Island planners are therefore which offshore sites are likely to expose the beaches of Long Island to spills and which are not. In order .to answer this question fully, we need both a complete description of the mechanism by which oil is trans- ported, and a statistical description of the phenomena driving this (transport) mechanism. Unfortunately, we do not at present understand either well enough to address the question except in a very approximate form. The answer to the question therefore must be phrased in light of what we do know, and how we might improve the answer if we should find the uncer- tainties associated with the analysis of critical importance. Without going into great detail, this report will there- fore discuss the uncertainties, present a reasonably simple model that would appear to represent all the important effects, and make some predictions of which regions in the offshore area are of interest to Long Island planners from the stand- point of oil spill exposure. As we shall see, a critical variable in the simple model proposed is the average transport imparted to an oil spill by those motions of the underlying water that are essentially uncorrelated with the wind. The sensitivi.ty of the results to this parameter will be discu@sed. 3 Uncertainties Despite the ten or fifteen papers available on the subject of oil spill transport on the ocean, it is fairly clear that we do not understand how the waves passing underneath an oil slick, the wind blowing over an oil slick, and the gross motions of the underlying water combine to move the oil. In fact, we find that the motions of the water lying right at the air-sea interface in the absence of oil are still the subject of much current research (Lee, 1972, and Dorman, 1971). Some of our ignorance with respect to oil spills is no doubt attributable to the novelty of our concern about oil spillage on the seas. It wasn't until the"Torrey Canyon" grounding and subsequent sinking (1967) that oil spills became important to the world at large. Since that time the number of tests involving the planned release of oil in the offshore region has been limited to no more than twenty, and these tests have usually had very specific goals associated with immediate operational problems, e.g., can we spot the oil on the surface using remote sensing devices (infrared, ultra- violet,and microwave scanners). The available literature has tended to attribute the velocity imparted to the slick by the wind to the formation of a simple wind-induced surface boundary layer. A number of things seem to be responsible for this. First of all, an after-the-fact analysis of the trajectories of the major oil slicks of the "Torrey Canyon" disaster showed that the path of the oil at any instant could best be estimated by 4 taking the vectorial sum of the underlying current velocity and 3.4% of the surface wind velocity (P. 150,.Smith, 1969). Secondly, Wuls (1968) laboratory studies indicate that,wind blowing over a clean water surface generated surface currents ranging from 3% to 5% of the wind speed, depending on the wind speed. Moreover, Van'Dorn's (1953) study of pond set-up included some data indicating that even if we suppress some of the wave motion with a surface film, we still get surface drift velo- cities similar to 3% of the wind speed. Finally, Hoult (1972) has presented a very simple argument that if logarithmic, constant stress boundary layer profiles are formed in the air and in the water simultaneously, then the two profiles will differ only by.a scaling factor equal to the square root of the ratio of the densities of air and water. This value is also approximately 3%. Unfortunately, this conjunction of similar values may amount to little more than happy coincidence. There can be little doubt that Hoult's argument does indeed explain a major portion of Wu's observations. Further- more, the existence of logarithmic profiles in surface wind boundary layers and in the underlying water have been verified in field observations reported by Dorman. This is about all that is required to validate the argument as.it applies to water with a clean surface. However, these results do not apply to regions in which oil films cover the surface simply because it is known that the logarithmic behavior of the surface wind boundary layer collapses (see Ruggles, 1969, p. 40). Furthermore, Van Dorn's study also demonstrated that the 5 shear force exerted on the surface of a pond having a thin surface film is only about half that observed on a pond having a clean surface. This indicates that Wuls results probably have only a qualitative bearing on our problem. Finally, Van Dorn's observation of the surface drift may be explained by invoking the arguments of Phillips (p. 38, 1969) regarding oil slick drift as induced by the action of suppressing waves. This leaves us with only one really hard piece of infor- mation and that is the "Torrey Canyon" analysis. This,, how- ever, is a highly empirical observation. Judging by the comparison of observed and predicted trajectories in Figure 37 of Smith (1968)t we can see that on some days a wind drift factor of 2.5% might have yielded a better fit, while on others, 4.5% might have been appropriate. Without a better understanding of the transport mechanism it is speculative to choose any particular value. In short, it is not at all clear that the present literature explains oil slick drift properly. In add ition to the uncertainties surrounding our under- standing of oil spill transport, we also have the problem of specifying the motions of the waters in the offshore region. A brief listing of the type of motions we should like to consider would include tidal motions, geostrophic motions, and wavelike motions of either the inertial type or the Kelvin (or Shelfwave) type. Unfortunately, we'are presently just at the point of being able to identify these motions. The creation of a model in which we coupled all of them together 4 i I - 6 I and attempted to relate them to atmospheric driving would be I an almost hopelessly speculative task. I I I I I I I I I I I I I I I I 7 Our approach In view of these problems, it is clear that any sort of model we might conjure up for estimating spill trajectory probabilities in the offshore region must necessarily be a fairly humble creature. Its results must likewise be accepted with an amount of reservation commensurate with its uncer- tainties. Moreover, the model should at best be fairly simple so that it is possible to understand the sensitivity of the output to variations in the parameters governing the model's behavior. An obvious candidate for the job is the simple observation by Smith that oil on the surface tends to move at a velocity approximately equal to the vectorial BUM of 3% of the surface wind's velocity and the velocity of the residual currents, Uoil *2Uresidual + .03d surface current wind Clearly, the definition of the residual velocity is some- what fuzzy, but for our purposes its properties are readily understandable and we can usually determine the sort of current that makes the most sense for a given locale. With respect to the Long Island region, a study of the geostrophy of the New York Bight reveals that this current may reasonably be expected to travel from east to,west down the Long Island coast and then proceed south down the New Jersey coast. While there may be some variation in the strength of the current during the year, it is simplest to consider the current as steady, and explore a variety of current speeds. 8 If the residual current is held steady, then all varia- bility in the mode 1 must come from.the surface wind component. Thus, the modeling of the surface wind becomes of some impor- tance in properly simulating the oil spill motion. Under these circumstances we desire a simulation that induces the proper mean motion and introduces a dispersive component that is of approximately the right size. Due to the approximate nature of the model, the need to get a completely analytical solution (which is possible but difficult) may be dispensed with in favor of using a Monte Carlo technique that compromises the quality of the answer only slightly. In short, we end up implementing the simple spill trajectory simulation technique first used in the Georges Bank study. The basis for accepting the results of this simple model rests on two points. First, if we apply the 3% rule to oil spills other than the "Torrey Canyon"Is, we again find reason- able agreement between observed and predicted trajectories (Stewart, 1973). Secondly, we have applied the technique extensively up and down the East Coast and compared the results for our best-guess current pattern to drift bottle launch and, recovery statistics (Stewart, 1974). The quanti- tative agreement has been.sufficiently good that we can usually rationalize mismatches, and from a quantitative standpoint, we have so far always managed to duplicate seasonal and locational trends. 9 Results Figure 1 is a sketch of both the rectangular outline used to represent Long Island and the current pattern we selected for our best guess of the offshore residual current. In addition to this hypothesized current pattern, we investi- gated a null current hypothesis and a current hypothesis using a drift speed of .1 knot versus the .25 knot of our best-guess pattern. The winds were modeled using the simple 9 x 9 Markov model representation developed for the Georges Bank study. The wind data was reduced on a seasonal basis, so different matrices were used for each of the four seasons. in view of the potential difference between the properties of the winds nearshore and offshore# the region was broken into three areas, and the wind. properties were determined for each area based on representative wind data for the region acquired from the National Climatic Center (NCC). The matrices used to simulate the surface wind in the Long Island region were determined from a ten-year record from-JFK Airport; the matrices used along the New Jersey coast were determined from ten years of weather records from Atlantic City; and the matrices used in the offshore region (more than thirty miles offshore) we determined from weather records acquired by ships stationed at Ocean Station Hotel.* *The use of Ocean Station Hotel wind data to represent the winds in the area lying off Long Island is somewhat of a compromise because Hotel is located at 36'N, 70*W, which is about 230 nautical miles (nm) from Long Island. We did inves- tigate using local lightship data,,.but we found the data unsuitable'for our purposes due to the irregular sampling scheme adopted on lightships. 740 W 73OW 72'oW ,e %-.- CONNECTICUT LONGISLAND 41ON SOUND v 1@00 - @,@q .25 LONGISLAND .25 .25 p .25 .25 SANDY HOOK .25 .25 40ON S@l 100" .25 FIGURE I. GRID REPRESENTATION OF LONG ISLAND AND OUR **%BEST GUESS" RESIDUA CURRENT PATTERN. 74OW 73OW 720W We were motivated to investigate the best-guess current pattern through a discussion with Dr. E. R. Baylor of SUNY at Stony Brook-in which he described the preliminary results of a current study performed by EG&G in the region just to the south of Long Island. It was found that the principal difference between the behavior of the model when using the .25 knot current value versus the lower value of .1 knot lay in the spatial distri- bution of impact zones. There was very little difference in the total percentage ashore. The explanation is straight- forward. Lower current values reduce the av erage westerly drift and allow a larger proportion of the simulated spills to be transported to the more easterly areas. Since the flow is not offshore, the total number of spills hitting shore remains about the same. It was found that the best-guess current gave us a better qualitative fit to drift bottle records and this observation in conjunction with the fact that the EG&G data represented our best hard information regarding offshore currents, led us to select it. The following figures present a more detailed description of the trajectory probabilities for this best-guess current. Figure 2a-d summarizes the seasonal dependency of the proba- bility of impacting Long Island's shores upon launch point location in the offshore region. Notice that the contours of equal probability are rather similar in all seasons except the summer. In summer it appears that there is a large region lying southeast of Shinnecock Inlet where the probability is higher than .9 that a spill will come ashore on Long Island. 740W 73OW 720W or CONNECTICUT B LONGISLAND Is SOUND 94 41'*N .7 LONGISLAND 0 r@ @s; -@lrn- %SAND .4 HOOK .3 '40'ON.. FIGURE 2A. CONTOURS OF PROBABILITY THAT SPILL RELEASED IN THE OFFSHORE REGION - el, DURING WINTER WILL IMPACT LONG ISLAND. 740 W 73'OW 72'OW 74* W 73OW 72OW CONNECTICUT LONGISLAND SOUND 41ON LONGISLAND .9 -T SANDY HOOK .6 .5 .4 .2 40ON FIGURE 2B. CONTOURS OF PROBABILITY THAT A SPILL RELEASED IN THE OFFSHORE REGIO "00 DURING SPRING WILL IMPACT LONG ISLAND. 74OW 73OW 720W 740 W 73OW 720W CONNECTICUT B LONGISLAND I SOUND 41ON .6 .5 LONGISLAND amp 4v AeV .9 SANDY HOOK .8 .7 .2 .6 I1.3 .4 .5 40ON I FIGURE 2C. CONTOURS OF PROBABILITY THAT SPILL RELEASED IN THE OFFSHORE REGION '000@- I r DURING THE SUMMER WILL IMPACT LONG ISL 74ow 73'OW 72OW 740 W 730W- 720W CONNECTICUT B LONGISLAND Is SOUND 41*N LONGISLAND .8 jp Ago @.5 SANDY .4 HOOK .3 .2 40ON FIGURE 2D. CONTOURS OF PROBABILITY THAT SPILL RELEASED IN THE OFFSHORE REGION DURING AUTUMN WILL IMPACT LONG ISLAND. r 74OW 73'OW 72'oW 16 Figure 3a-d shows the variation in the average time to shore with distance from shore. Figures 4a-d, 5a-d, and 6a-d show the specific launch regions in the offshore area that have a high probability of impacting the Amityville region, the eastern portion of Fire Island, and the Montauk Point area respectively. Notice that the Amityville and eastern Fire Island areas are threatened primarily by spills lying to the east, while the Montauk Point region is threatened by spills to the southeast. One explanation for this behavior is that spills in the waters lying south of western Long Island will be subjected to both beaching on the New Jersey shore and to the southerly transport of our hypothesized current in this area. .The generation of these figures was quite expensive due to the number of points involved, so it was judged inadvisable to create similar plots for the other current hypotheses. However, based on our preliminary results we can be reasonably confident that the principal changes would be to shift the regions of high probability slightly to the southwest., This would be of rather little consequence with two major exceptions: 1. In the region centered about 400N, 72*W we might find a substantial change in the summer trajectory behavior. Specifically, the chance of impact might increase from .5 or .6 to .8 or .9. 2. A review of Figures 2a-d reveals that there is a sharp dropoff in the probability of a spill impacting Long Island along a line running about southeast 740 W 730W 720W of CONNECTICUT E LONGISLAND SOUND 41'ON LONGISLAND \0 @",\SAND HOOK ca 40'ON FIGURE 3A. CONTOURS OF EQUAL AVERAGE TIME TO IMPACT LONG ISLAND IS COASTLIN DURING WINTER. 74OW 73ow 72OW 74OW 73"W 72'oW CONNECTICUT E LONG ISLAND SO UND MONTAUK 41ON r FIRE ISLE -j I LONGISLAND AMITYVILLE r - - SANDY \10 HOOK 0 40ON FIGURE 3B. CONTOURS OF EQUAL AVERAGE TIME TO IMPACT LONG ISLAND'S, COASTLINE DURING SPRING. 74OW 73OW 720W m m m m = m M = = 74OW 73ow 72'OW CONNECTICUT B L NG ISLAND I 41ON SO UND MONTAUK r-_ m r FIRE. ISLE LONGISLAND AMITYVILLE SANDY HOOK %0 40'ON FIGURE 3C. CONTOURS OF EQUAL AVERAGE TIME TO IMPACT LONG ISLAND it S COASTLINE DURING - SUMMER. SIR 74OW 730W 720W = M M m m M M M M 740 W 73'DW 720W CONNECTICUT B LONG ISLAND IE S UND MONTAUK 41'ON -13 FIRE ISLE LONGISLAND AMITYVILLE r - - A @\SANDY HOOK 40ON FIGURE 3D. CONTOURS OF EQUAL AVERAGE TIME TO IMPACT LONG ISLAND )S COASTLIN DURING AUTUMN. 74OW 73OW 72OW 74OW 73OW 720W of CONNECTICUT LONG ISLAND SO UND MONTAUK 41c'N FIRE ISLE LONGISLAND AMITYVILLE \SANDY HOOK 40ON FIGURE 4A. CONTOURS OF PROBABILITY THAT SPILL RELEASED IN THE OFFSHORE REGION DURING WINTER WILL IMPACT THE AMITYVIL AREA.. I I 74OW 73OW 720W 74OW 73OW 72OW CONNECTICUT (5 B ONGISLAND SOUND -MONTAUK 41'ON 5p r FIRE ISLE I LONGISLAND AMITYVILLE r - SANDY HOOK .02 40ON FIGURE 413. CONTOURS OF PROBABILITY THAT A SPILL RELEASED IN THE OFFSHORE REGION @-@ A- DURING SPRING WILL IMPACT THE AMITYVILL AREA. 74ow 730W 720W 74OW 73OW 720W CONNECTICUT E LONG ISLAND I SO UND MONTAUK 41ON "%A r FIRE ISLE I LONGISLAND AMITYVILLE r SANDY HOOK .02 40ON FIGURE 4C. CONTOURS OF PROBABILITY THAT SPILL RELEASED IN THE OFFSHORE REGION DURING SUMMER WILL IMPACT THE AMITYVIL AREA. FF �r 74OW 73OW 72OW 74OW 73*W 72'oW CONNECTICUT LONG ISLAND 41ON SO UND MONTAUK FIRE ISLE LONGISLAND AMITYVILLE r SANDY HOOK .02 40ON FIGURE 4D. CONTOURS OF PROBABILITY THAT SPILL RELEASED IN THE OFFSHORE REGION DURING AUTUMN WILL IMPACT THE AMITYVIL AREA. FT 74ow 73OW 720W 74OW 73OW 72OW CONNECTICUT B LONG ISLAND Is SO UND MONTAUK 41ON r FIRE IS E LONGISLAND AMITYVILLE SANDY HOOK .02 40ON FIGURE 5A. CONTOURS OF PROBABILITY THAT A SPILL RELEASED IN THE OFFSHORE REGION F7 7:@@l DURING WINTER WILL IMPACT EASTERN FIRE ISLAND. 740 W 730W 720W 74OW 73OW 72'oW CONNECTICUT E LONG ISLAND SO UND MONTAUK _jc 41ON i r FIRE ISLE LONGISLAND AMITYVILLE 01) SANDY HOOK .02 40ON 11 1 FIGURE 5B. CONTOURS OF PROBABILITY THAT A SPILL RELEASED IN THE OFFSHORE REGION DURING SPRING WILL IMPACT EASTERN FIRE ISLAND. 740 W 73OW 720W 74OW 73OW 720W CONNECTICUT LONG ISLAND SO UND 41'ON 5;?-MONTAUK r FIRE ISLE I LONGISLAND AMITYVILLE r - - .2 SANDY HOOK .02 40ON FIGURE 5C. CONTOURS OF PROBABILITY THAT A SPILL RELEASED IN THE OFFSHORE REGION DURING SUMMER WILL IMPACT EASTERN FIRE ISLAND. 740 W 73OW 720W 74OW 73'*W 72'OW CONNECTICUT E LONG ISLAND I SO UND MONTAUK 41*N r FIRE ISLE LONGISLAND AMITYVILLE r SANDY HOOK .02 40'ON FIGURE 5D. CONTOURS OF PROBABILITY THAT SPILL RELEASED IN THE OFFSHORE REGION DURING AUTUMN WILL IMPACT EASTERN FIRE ISLAND. I - I @r-- n 740 w 730W 720W 74OW 73OW 720W CONNECTICUT B LONG ISLAND Is SO UND MONTAUK 41'ON r FIRE ISLE I LONGISLAND AMITYVILLE .02 SANDY HOOK 40ON FIGURE 6A. CONTOURS OF PROBABILITY THAT .4 SPILL RELEASED IN THE OFFSHORE REGION " -, 0 A- DURING WINTER WILL.IMPACT THE MONTAUK POIN T REGION 74OW 73OW 72OW 74OW 73OW 72OW 7c=" CONNECTICUT B LO ISLAND O-UND MONTAUK 41'ON r FIRE ISLE LONGISLAND AMITYVILLE r SANDY HOOK 40ON FIGURE 6B. CONTOURS OF PROBABILITY THAT i SPILL RELEASED IN THE OFFSHORE REGION FT IG 74@ e.0-10 -4 DURING SPRING WILL IMPACT THE MONTAUK POINT REGION. 74OW 73OW 72OW 74OW 73OW 72OW CONNECTICUT 15E LONG ISLAND eo-o I S UND MONTAUK 41ON r FIRE IS LONGISLAND LE .3 --_j AMITYVILLE r SANDY HOOK .02 40ON FIGURE 6C. CONTOURS OF PROBABILITY THAT 4 SPILL RELEASED IN THE OFFSHORE REGION DURING SUMMER WILL IMPACT THE MONTAUK POINT REGION. 740W 730W 720W 740 W 75OW 72OW CONNECTICUT B LONGISLAND SOUND 41ON LONGISLAND .001 amp 4p SANDY HOOK 40ON I FIGURE 6D. CONTOURS OF PROBABILITY THAT A SPILL RELEASED IN THE OFFSHORE REGION DURING AUTUMN WILL IMPACT THE MONTAUK POINT REGION. 0 XA-1 r ;-@- - '-J%N 74OW 73OW 72'oW low 33 from Long Beach* Under lower*reiidual current hypotheses we can expect this line to swing to the south. Another point of uncertainty in the model is the wind drift coefficient (here set at .03) as mentioned above. Based on our prior experiehce'with the model, we know that the prin- cipal effects in changes to this coefficient show up in two places. First, increasing the coefficient will increase the wind-induced dispersion. Secondlyr this increase will also amplify.that portion of the average velocity attributable to the wind. Decreasing the coefficient's value will cause the controverse effects. Thus, this coefficient determines both the average direction of drift and the dispersion. The effects with respect to Long Island of increasing th is coefficient will be to rotate the basic patterns of the contours of equiproba- bility to the south# and to smear out the regions where there is now a strong gradient in this probability. The former effect may be shown to be rather smallp while the latter is large. Decreasing the value of the.coefficient will swing the pattern.in the reVerse direction and increase the proba- bility gradients. 34 Conclusion Despite the uncertainties, severa 1 features emerge from the analysis. First of all, in the offshore area bounded approximately by 400N latitude in the southr 71OW longitude on the east, and a line running about southeast from Long Beach on the west, we can expect a high percentage of all oil spills to come ashore during at least one season. The season offering.the greatest exposure is quite clearly summer. Additionally, the time to shore averages around 10 to 20 days over the whole region. This is sufficient time so that most-of the low boiling fractions will have been lost to evaporation. However, it is not sufficient time so that we would expect a great deal of turbulent dispersion to have taken place. It is likely that the oil would still be in the form of large contiguous patches. These results are reasonably independent of the current presumption or the speci- fication of the wind drift coefficient. If we believe in the validity of selecting .03 as the wind drift coefficient and in the validity of our best-guess current hypothesist then the contours of Figure 2a-d may be used to further spedify the regions of greatest exposure. 35 References Dorman, Craig E. 1973. The relationship between microscales and wind-wave spatial development. Cambridge, Mass.: MIT Department of Hoult, D. D. 1972. Oil spreading on the sea. Ann. Rev. Fluid Mechanics. 59-64 Lee, F. A. 1972. Some nonlinears aspects of wind-wave inter- action. J. Physical Oceanography. 2:432-438. Offshore Oil Task Group. 1973. The Georges Bank petroleum study. Cambridge, Mass.: MIT Sea Grant Project Office. Phillips, O. M. 1969. The Dynamics of the upper ocean. 2nd ed. Cambridge, England: University Press. Ruggles, Kenneth Warren. 1969. Observation of the wind field in the first ten meters of the atmosphere above the ocean. Cambridge, Mass.: MIT Department of Meteorology. Smith, J. E., ed. 1968. "Torrey Canyon"-- Pollution and marine life. Cambridge, England: University Press. Stewart, R. J. 1973. A survey of oil spill transport on the surface of the ocean. Paper presented to the New England Regional Society of Naval Architects and Marine Engineers. DATE DUE GAYLORD No 2333 i 14T Fr, 14 3 6668 14107 9774