Some climatological characteristics of hurricanes and tropical storms, gulf and east coasts of the United States

[From the U.S. Government Printing Office, www.gpo.gov]

NOAA Technical Report NWS 15

Some Climatological
Characteristics of Hurricanes
and Tropical Storms, Gulf and
East Coasts of the
United States
Francis P. Ho, Richard W. Schwerdt, and Hugo V. Goodyear
WASHINGTON, D.C.
MAY 1975

Property of CSC Library

C Z I C collection

U. S. DEPARTMENT OF COMMERCE NOAA
COASTAL SERVICES CENTER
2234 SOUTH HOBSON AVENUE
CHARLESTON, SC 29405-2413

UNITED STATES NATIONAL OCEANIC AND National Weather
DEPARTMENT OF COMMERCE / ATMOSPHERIC ADMINISTRATION Service
Rogers C. B. Morton, Secretary / Robert M. White, Administrator / George P. Cressman, Director
%ENTr OF C
COASTAL ZONE
INFORMATION CETER

Contents

Page
1. Introduction ...........................

1.1 Purpose............................. 1

1.2 Scope of report..........................2

1.3 Sources of data......................... 4

1.4 Previous studies .........................4

1.5 Funding..............................13

2. Frequency of hurricane and tropical storm occurrences.......13

2.1 Classification of hurricanes and data...............13

2.2 Frequency of landfalling hurricanes and tropical storms......13

2.3 Frequency of exiting hurricanes and tropical storms........23

2.4 Frequency of alongshore hurricanes and tropical storms ......23

2.5 Ratio of hurricane to total storm occurrences...........25

3. Probability distribution of central pressure ...........30

3.1 Data ...............................30

3.2 Analysis .............................33

3.3 Results..............................35

3.4 Evaluation of results.......................39

4. Probability distribution of radius of maximum winds........41

4.1 Data ...............................41

4.2 Analysis .............................48

4.3 Reasonableness of analysis ....................51

4.4 The radius of maximum winds of Hurricane Camille .........52

5. Probability distribution of speed and direction of storm motion .55

5.1 Speed of storm motion ......................

iii

Page

5.2 Forward speed probability distribution for landf ailing hurricanes 55

5.3 Forward speed probability distribution for alongshore hurricanes . 56

5.4 Probability distribution of direction of storm motion for
landf ailing tropical storms and hurricanes.............59

6. The joint probability question...................67

6.1 Central pressure, p., vs. radius of maximum wind, R ........67

6.2 Forward speed, T, vs. direction of motion, e............70

6.3 Other joint probability questions .................71

7. Summary ..............................71

7.1 Highlights.............................71

References.............................74

Acknowledgments ..........................76

Appendix - Track density method for storm frequency ........77

Track frequency..........................77

Direction probability.......................78

Landfalling frequency .......................78

Necessity of octagons.......................79

Tables
Page
Table l.--Hurricanes with central pressure < 982 mb (29.00 in)
ranked in chronological order from 1900-73,
gulf coast United States ..................................... 5

Table 2.--Hurricanes with central pressure < 982 mb (29.00 in)
ranked in chronological order from 1900-73, east
coast United States .......................................... 9

Table 3.--Count of hurricane and tropical storm tracks crossing
lines normal to the coast -- 6871-1973) ....................... 26

Table 4.--Frequency of hurricane and tropical storm tracks
crossing lines normal to the coast -- 6871-1973) .............. 27

Table 5.--Data for probability analysis ................................ 72

Illustrations

Figure

1.--Locator map with coastal distance intervals marked (n.mi.) ......... 3

2.--Count of landfalling tropical storms and hurricanes
(1871-1973) by 50-n.mi. segments of a smoothed coastline
(points plotted and connected by dashed lines). Solid
line denotes the entry frequency curve obtained by
applying the objective smoothing function .......................... 15

3.--Smoothed coastline obtained by applying the objective
smoothing function of figure 2.;..................................... 17

4.--Coastal segments selected for frequency analysis by
track density method (result shown in fig. 5) ...................... 18

5.--Frequency of landfalling tropical storms and hurricanes
C871-1973) by track density method ................................. 19

6.--Adopted frequency of landfalling tropical storms and
hurricanes (1871-1973) for the gulf and east coasts of
the United States .................................................. 21

7.--Frequency of exiting hurricanes and tropical storms
(1871-1973). Curve fitted subjectively ............................ 24

8.--Accumulative count of hurricane and tropical storm tracks
passing the coast at sea (1871-1973). Based on counts
along heavy dashed lines shown projected normal to coast ........... 28

Figure Page

9.--Ratio of landfalling hurricanes to total number of landfalling
hurricanes and tropical storms (1886-1973). Based on counts
in overlapping zones centered at 50-n.mi. intervals and objective
smoothing along the coast ................................... 29

10.--Graphs of central pressure vs. cumulative percent of
occurrences (a) Gulf of Mexico (milepost 750), near
Bay St. Louis, Miss., where hurricane Camille went
ashore and (b) east coast (milepost 1950) along central
South Carolina coast...................................... 34

ll.--Probability distribution of central pressure of hurricanes,
gulf and east coasts (1900-73). Numbered lines denote
the percent of storms with po equal to or lower than the
value indicated along the ordinate. Plotted points (a)
are taken from frequency analyses at 50-n.mi. intervals
for the 5th percentile (sec. 3.2) .................................. 36

12.--One-percentile chart of minimum central pressure of
hurricanes and tropical storms for the Gulf of Mexico
and surrounding areas. Pressure values plotted at the
centroid of data reference points for each 50-n.mi.
increment of coast .............................................. 37

13.--Fifty-percentile chart of minimum central pressure of
hurricanes and tropical storms for the Gulf of Mexico
and surrounding areas. Pressure values plotted at the
centroid of data reference points for each 50-n.mi.
increment of coast ................................................. 38

14.--Graph of wind speed and station-to-hurricane-center
distance vs. time at Miami, Fla., September 15-16, 1945.
The radius of maximum winds determined by the relation
of peaks in the wind curve to distance from Miami to
the storm center is 24 n.mi ............... 43

15.--Radius of maximum winds (RMW) vs. inner radar eye radius
(IRR). Points falling on the 45� line are those where
the RMW and IRR coincide. The curved line indicates the
best fit curve (from Shea 1972) .................................... 44

16.--Difference between the radius of maximum winds (RMW) and
the inner radar eye radius (IRR) vs. maximum wind speed.
The best fit curve is indicated by the heavy line (from
Shea 1972) .........................................................

17.--Variation of the radius of maximum winds with elevation
for (a) storms with simultaneous lower and upper tropospheric
reconnaissance data and (b) storms in which two or more
simultaneous reconnaissance missions were flown in the
lower troposphere only (from Shea 1972) ............................45
vi

Figure Page
18.--Graphs of radi-l of maximum winds vs. cumulative percent
of occurrences (a) Gulf of Mexico (milepost 750), near
Bay St. Louis, Miss., where hurricane Camille went ashore
and (b) east coast (milepost 1950), along central South
Carolina coast ..................................................... 49

19.--Probability distribution of radius of maximum winds of
hurricanes, gulf and east coasts (1900-73). Numbered
lines denote the percent of storms with R equal to or
less than the value indicated along the ordinate.
Plotted points (A) are taken from frequency analyses
at 50-n.mi. intervals for the 16-2/3 percentile (sec.4.2) .......... 50

20.--Hand-drawn sketch of New Orleans (Moisant Field) radar
image, August 18, 1969, at 0431 GMT. Note locations
of the eye center (x), inner wall cloud radius (a), wall
cloud center radius (b), and outer wall cloud radius (c) ........... 53

21.--Composite graph of inner wall cloud radius, wall cloud
center radius, and outer wall cloud radius vs. time for
Camille. Storm made landfall about 0430 GMT ....................... 54

22.--Landfalling hurricanes, forward speed vs. cumulative
percent of occurrences, (a) Gulf of Mexico (milepost
750) near Bay St. Louis, Miss., (b) east coast (mile-
post 1950) along central South Carolina coast ...................... 57

23.--Probability distribution of forward speed for landfalling
hurricanes, 1886-1973. Numbered lines denote the percent
of storms with forward speed equal to or less than the
value indicated along the ordinate. Plotted points (A)
are taken from frequency analyses at 50-n.mi. intervals
for the 80th percentile (par. 5.2.1) ............................... 58

24.--Probability distribution of forward speed for alongshore
hurricanes, 1886-1973. Numbered lines denote the percent
of storms with forward speed equal to or less than the
value indicated along the ordinate. Plotted points (A)
are taken from frequency analyses for the 80th percentile
(par. 5.3.) ....................................................... 60

25.--Scatter diagram of direction of storm motion measured at
the point of landfall, 1871-1973. Circled dots denote
two events ......................................................... 61

26.--Landfalling hurricanes and tropical storms, direction of motion
vs. cumulative precent'of occurrences, (a) Gulf of Mexico
(milepost 750) near Bay St. Louis, Miss., (b) east coast
(milepost 1950) along central South Carolina coast ................. 62

vii

Figure Page

27.--Probability distribution of direction of landfalling
storm motion for gulf coast. Numbered lines denote
the percent of storms approaching the coast at an
angle equal to or less than the value indicated along
the ordinate. Plotted points (A) are taken from
frequency analyses at 50-n.mi. intervals for the
50th percentile (sec. 5.4) ....................................... 63

28a.-Probability distribution of direction of landfalling
storm motion for east coast, south of Cape Hatteras,
N.C. Notations are the same as figure 27 ........................ 64

28b.-Probability distribution of direction of landfalling
storm motion for east coast, north of Cape Hatteras,
N.C. Notations are the same as figure 27 ........................ 65

29.--R vs. po, Gulf of Mexico hurricanes. Data from table 1.
Numbered lines denote the percentile of storm occurrences ........ 68

30.--Same as figure 29 except for east coast hurricanes;
data from table 2 ................................................ 69

Appendix Illustrations

A-l.--Octagon for counting hurricane tracks ........................... 79

A-2.--Hurricane and tropical storm track count for 2.5�
octagons in the Gulf of Mexico. 1871-1973 ...................... 80

A-3 to A-16.--Hurricane and tropical storm track direction
histograms and probability distributions for gulf
coast octagons. 1871-1973 ...................................... 81-87

viii

Some Climatological Characteristics of Hurricanes
and Tropical Storms, Gulf and East Coasts
of the United States

Francis P. Ho, Richard W. Schwerdt, and Hugo V. Goodyear

ABSTRACT

A climatology of hurricane factors important to storm
surges is presented for the U.S. gulf and east coasts.
A smoothed frequency of tropical storms and hurricanes
entering and exiting the coast and storms passing within
150 n.mi. of the coast during the period 1871-1973 is
given. The central pressure for hurricanes and tropical
storms and the radius of maximum winds and speed of
forward motion for hurricanes were obtained from data
analysis. Directions of landfalling hurricanes and
tropical storms at the time they crossed the coast at
selected points were also analyzed. The probability
distribution of each factor was plotted and analyzed for
each 50-n.mi. interval along the coast. Selected prob-
ability levels of each distribution were then summarized,
and smoothed variations along the coast were obta ined by
analysis. The speeds of motion for two classes of hur-
ricanes (those that entered the coast and those that
passed within 150 n.mi. of the coast) were studied sep-
arately and a smooth speed analysis determined for each.
The question of joint probability among the various
factors and with latitude is discussed qualitatively.

1. INTRODUCTION

1.1 Purpose

This report assembles in one volume hurricane climatological data that
have been developed in studies by the National Weather Service (NWS) of the
National Oceanic and Atmospheric Administration (NOAA) for the Federal
Insurance Administration (FIA) of the Department of Housing and Urban Devel-
opment (HUD) and the Corps of Engineers (CoE), U.S. Army.

The FIA is the executive agency for the "National Flood Insurance Act of
1968" (Public Law 448, 90th Congress, Title XIII). This act provides for a
national Flood Insurance Program for insuring residences and small businesses
against the hazard of damage or destruction by floods. A flood frequency
analysis is essential to establishing this program for a given community.
The Flood Insurance Act directs other federal agencies to cooperate with HUD
in developing the frequency analyses required by the Flood Insurance Program.
HUD has solicited the assistance of NOAA in these technical assessments for

coastal regions. Coastal tidal inundations on the gulf and Atlantic coasts
on the United States are primarily caused by hurricanes.1 Therefore, the
characteristics of these storms are the beginning point in making tidal flood-
ing frequency analyses. This report is a climatological assessment of the
central pressures, radius of maximum winds, forward speed, and other charac-
teristics of hurricanes along the U.S. east and gulf coasts in a manner
suitable for determining the frequency of storm surge levels. The report in-
cludes only the atmospheric characteristics of hurricanes and does not
include surge levels that are contained in other reports.

The Flood Control Act of 1936 gave the Corps of Engineers responsibility
for constructing flood control projects throughout the United States. This
includes coastal protective works against high storm tides. Public Law 71,
84th Congress, 1st session, 1955, directs the Chief of Engineers to determine
areas of potential damage from hurricanes and to propose remedial measures.
This "shall include the securing of data on the behavior and frequency of
hurricanes, ... and possible means of preventing loss of human lives and
damages to property, with due consideration to the economics of proposed
breakwaters, seawalls, dikes, dams, and other structures, warning services
or other measures which might be required." Under reimbursable funds the
Hydrometeorological Branch, NWS, since 1955 has assisted the Corps of
Engineers by carrying out studies determining meteorological factors impor-
tant to storm surges, reconstructing wind fields of historical and hypothet-
ical storms, and developing criteria for a Standard Project Hurricane,
defined as the "most severe hurricane considered reasonably characteristic
of a region" for the gulf and Atlantic Coasts.

The present report is an update of similar hurricane climatological data
published in the first Standard Project Hurricane bulletin (Graham and Nunn
1959). It includes hurricanes influencing the coast through 1973 and more
detpl-led regional analysis than the earlier report. It is the first step in
a proposed revision of the Standard Project Hurricane criteria.

1.2 Scope of Report

The geographical region covered by the report is the U.S. gulf and Atlantic
coasts from Texas to Maine (fig. 1). The first objective is to define cli-
matologically the frequency of hurricanes and tropical storms influencing
each coastal reach. This is done in three classes -- storms entering the
coast from the sea (entering or landfalling), storms having entered one coast
and then proceeding from land to sea at another coastal point (exiting), and,
thirdly, storms skirting the coast close to shore bw~ with the center remain-
ing at sea within 150 n.mi. of the point under consicderation (alongshore or
bypassing). It is possible for the same storm to be considered in each of
the three classes at different times. Probability diarributions are devel-
oped and the along-coast variation depicted of hurricanie central pressures,

Of course, extratropical storms such as that of Marc4 1962 have also
wrought severe damage.

300

55.~~~~~~~~~~~~~~1- I ESHSIAR0

000270~

G1.~500

LA. "RI�J'I 6I AL

TEX. PENSA"COLA SASILL -700
LAECHARLES ST5510 I P CIII

~ELAS09O00-A. 1600
8 0 10~~~~~~~~~~0 II

CORPUS~~~ KISOD 6 0 \700 200- E` s~UOS

NISABEL
ARPOWTISALLE 2000
C 00'\00

Figure l.--Locator map with coastal distance intervals marked (n.mi.).

an index to storm intensity; the radius of maximum wind, an index to the
storm lateral extent; forward speed; and direction of motion. Each of these
factors influences the capability of the storms to produce storm tides. The
degree of statistical independence of these four parameters is discussed in
chapter 6 of this report.

Hurricanes are a threat to life and property not only from storm tides, but
from the wind, and from rain-induced floods. These factors are not included
in the present report. Thom (1968) discusses extreme fastest-mile wind
speeds and illustrates this with 2-, 10-, 25-, 50-, and 100-yr mean recur-
rence interval maps for the United States. The great majority of extreme
fastest-mile wind speeds along the gulf and Atlantic Coasts south of Cape Cod
have occurred during hurricanes. T~he frequency and areal distributions of
tropical storm rainfalls in a form suitable for use in engineergi~g design
criteria along the gulf coast is'the subject of a report by Goodyear (1968).

1.3 Sources of Data

Tables I and 2 list the factors discussed in the various chapters of this
report for hurricanes during the years 1900-73. These data are an update,
revision, and extension of table A in National Hurricane Research Project
Report No. 33 (Graham and Nunn 1959). The original sources of the data are
barograph traces from land stations and ships, wind records from National
Weather Service and military stations', aircraft reconnaissance flight data,
radar data, miscellaneous pressure and wind reports, and textual descriptions
in scientific literature. These descriptions have appeared in the periodi-
cals Monthly Weather Review (published since June 1872), and Climatological
Data National Summary (since 1950), National Hurricane Research Project
Report No. 39 (Graham and Hudson 1960), NOAA Technical Memorandum NWS SR-56
(Sugg et al. 1971), the book Tropical Cyclones (Cline 1926), and a few other
sources.

Tropical cyclone track information, was used to determine the frequency of
entering, exiting, and alongshore tropical storms and hurricanes, direction
of forward motion, and in some cases forward speed. Tracks from 1871-1963
are from Cry (1965), and from the Monthly Weather Review beginning with 1964.

1.4 Previous Studies

One of the first systematic compilations of the characteristics of hurri-
canes affecting the coast of the United States is Tropical Cyclones (Cline
1926). Table I in Hydrometeorological Report No. 32 (Myers 1954) provides
the first compilation of all hurricane central pressures and Rs (radius of
maximum winds) during a definite period of years. National Hurricane
Research Project Report No. 33 (Graham and Nunn 1959) updates the list and
systematizes the geographical distribution of the factors. Technical Paper
No. 55 (Cry 1965) describes all the hurricane tracks during a definite period
of time and cites the earlier works of this kind. BUR 7-97, Interim Report -
Meteorological Characteristics of the Probable Maximum Hurricane, Atlantic
and Gulf Coasts of the United States (NOAA 1968) updates and revises the data
in NHRP No. 33 and gives the geographical distribution of the characteristics
of the hypothetical hurricane having that combination of characteristics

Table l.--Hurricanes with central pressure c 982 mb (29.00 in.) ranked in chronological order from 1900-73

Gulf Coast United States

Approx. coastal Storm direction P p Station(s) Station(s)
D vatue a where P was Remarks
Date* Storm reference point (clockwise from Po value a where P was R whe re R was T
(GMT) name t (see fig. 1) north). (mb) (in.) applied to (mb) observed (n.mi.) observed (kt)s

Sept. 9, 1900 373 130' 936.0 27.64 Coast 964.4 Galveston, Tex. 14a 10

Aug. 15, 1901 763 195' 972.6 28.72 Coast 992.6 Mobile, Ala. 33a 14

June 17, 1906 1393 185� 979.0 28.91 Coast 997.6 Jupiter, Fla. 26a 10

Sept. 27, 1906 795 160' 965.1 28.50 Coast 965.1 SS Winona 43 Mobile, Ala. 16 SS Winona in eye of storm while
anchored off anchored off Scranton, Miss.
Scranton, Miss.

Oct. 18, 1906 1419 230' 976.6 28.84 Coast 990.9 Jupiter, Fla. 35a 6

July 21, 1909 360 115� 958.7 28.31 Coast 982.1 Bay City, Tex. 19 12

Sept. 20, 1909 657 150� 980.0 28.94 Coast 989.8 New Orleans, La. MSG 11

Oct. 11, 1909 1419by 235' 957.0 28.26C Knights Key,957.0 Knights Key,Fla. 22b Key West, Fla. 10
Fla.
200' 941.4 270d' 111 SS Jean in eye of storm (12 n.mi.
Oct. - 17, 1910 1343 200� 941.4 .27.80 12 n.mi. S. 941.4 SS Jean in eye of storm (12 n.mi.
Dry TortuRas, south of Dry Tortugas, Fla.)
U,
Fla. b
Aug. 17, 1915 373 130' 948.5 28.01a Coast 952.9 Velasco, Tex. 29 Galveston and 11
Houston, Tex.

Sept. 29, 1915 671 170' 932.3 27.53a' 27.0'N. 935.0 HMS Hermione 26a New Orleans, La. 10 HMS Hermione experienced some eye
89.3'W & other stations effects at an unknown distance from

July 5, 1916 810 160� 961.1 28.38a Coast 979.3 Mobile, Ala. 45b Mobile, Ala. 25of minimum pressure

Aug. 18, 1916 184 115' 948.2 28.00 Coast 948.2 Santa Gertrudis, 25a 11
Tex.

Oct. 18, 1916 842 220' 973.9 28.76 Coast 973.9 Pensacola, Fla. 19 Pensacola, Fla. 21

Sept. 29, 1917 892 230' 964.4 28.48a' Coast 965.5 Pensacola, Fla. 33b Pensacola, Fla. 13

Sept. 10, 1919 135by 110 929.2 27.44' d' 929.2 See Remarks a 8 Lowest pressure obtained from mean of
Tortugas,Fla. two ships (Lake Winona, Fred W. Weller)
and Dry Tortugas, Ela.
5 Sept. 14, 1919 213 105' 947.9 27.99c' Coast 947.9 PortAransas, MSG 20
Tex.

Sept. 21, 1920 630 155' 979.7 28.93 Coast 981.7 Houma, La. 28a

June 22, 1921 309 175' 953.9 28.17b' Cas 994.6 HoustonTex. 17' 11

Oct. 25, 1921 1207 235' 952.3 28.12 Coast 952.3 Tarpon Springs, 18 10
Fla.

Oct. 20, 1924 - 1355 250' 971.9 28.70 Dry Tortugas, - See Remarks 19a 8 Parameters obtained by interpolation bet-
Fla. ween SS Toledo (off western end of Cuba)
and Miami, Fla., and applied to the vicin-
ity of Dry Tortugas, Fla.
See Legend at end ot' table 2.

Table l.--Continued

Gulf Coast United States

Approx. coastal Storm direction P p p Station(s) Station(s)
Date* Storm reference point (clockwise from o o value a where P was R where Rk es T
(GMT) name t (see fig. 1) north) (mb) (in.) applied to (mb) obsered (n.mi.) observed (kt) Remarks

Aug. 26, 1926 603 180� 958.7 28.31a'c' Coast 958.7 Houma, La. 27a 10

Sept. 20, 1926 842 140' 955.0 28.20c Coast 955.0 Perdido Beach, 17b Pensacola, Fla. 7

Oct. 21, 1926 1419by 220 931.9 2752al Ala.
Oct. 21, 1926 119 220 931.9 27.52 60 n.mi. So. 987.5 Key West, Fla. 21 16
Key WestFla.
Sept. 17, 1928 1552 120- 958.3 28.30a 50 nmi East West Palm Beach MSG 12 Lowest pressure for the gulf coast
inland Coast and Everglades occurred as the storm was filling
from coast 935.3 Exp. Sta., Fla. about 9 n.mi. west of Avon Park,
Fla., or about 50 n.mi. east-
southeast of Tampa Bay.
June 28, 1929 296 130' 969.2 28.62at Coast 986.1 Port O'Connor, 13a 15
Texa.

Sept. 30, 1929 966 160' 975.3 28-80c Coast 975.3 Panama City, 55b Pensacola, Fla. 6 Storm becoming extratropical.
Fla.

Aug. 14, 1932 373 135' 942.4 27.83 Coast 942.4 East Columbia, 12 15
Texa.

Aug. 5, 1933 109 070' 975.3 28.80a Coast 981.4 Brownsville,Tex. 25 Brownsville,Tex. 10

Sept. 4, 1933 1525 120' 964.4 28.48at 50 n.mi. East Jupiter, Fla. 29 Tampa, Fla. 11 Lowest pressure for the gulf coast
inland from Coast occurred as the storm was filling
coast 947.5 just west of Avon Park, Fla., or

Sept. 5, 1933 139 090' 948.9 28.02a' Coast 950.6 Brownsville,Tex. 20b Brownsville,Tea. 8

June 16, 1934 617 180' 965.8 28.52 Coast 967.8 Jeanerette, La. 37 16

Sept. 3, 1935 1393 130' 892.3 26.35c' Long Key, 892.3 Long Key, Fla. 6a 9
Fla.

Nov. 5, 1935 1393ex 065' 972.9 28.73 East Coast 972.9 Miami, la. 10 d Miami, Fla, 15
ref. point
1459

July 31, 1936 904 150� 963.8 28.46a' Coast 972.9 Valpariso, Fla. 19a 9

Aug. 8, 1940 468 140' 971.9 28.70e Coast 971.9 Sabdne, Tex. 11a 8

Sept. 23, 1941 348 180' 958.7 28.31 Coast 970.5 Houston, Tex. 21a 13

Oct. 7, 1941 996 170' 981.4 28.98a Coast 982.1 Carrabelle,Fla. 18 11

Aug. 30, 1942 309 135' 950.6 28.07a' Coast 951.6 Seadrift, Tex. 18a 14

July 27, 1943 419 110' 974.6 28.78e Coast 974.6 Ellington Field, 16b Houston, Tex. 8
Tex.

Oct. 19, 1944 1292 195' 948.9 28.02c Dry 948.9 Dry Tortugas,Fla.27a 13
Tortugas,Fla.

Aug 27, 1945 309 185' 967.5 28.57c Coast 967.5 Palacios, Tex. 18 4

See Legend at end of table 2.

Table 1.--Continued

Gulf Coast United States

Approx. coastal Storm direction P P p Station(s) Station(s)
Date* Storm reference point (clockwise from o o value a where Pa was R where R was T
(GMT) name t (see fig. 1) north) (mb) (in.) applied to (mb) observed (n.mi.) observed (kt) Remarks

Sept. 15, 1945 1433 130� 951.2 28.09' Coast 951.2 Homestead, Fla. 24b Miami, Fla. 10 Lowest pressure observed by Fla.

East Coast Railroad personnel at
Homestead, Fla.
Sept. 18, 1947 1330ex 085 949.2 28.03a 50 n.mi. East Hillsboro, Pla. 34 Miami, Fla. o
inland from Co Miami, Fla. 7 Lowest pressure for the gulf coast
coast 947.2 occurred some 50 n. mi. east-north-
east of the Ten Thousand Island
area of southwest Florida as the
storm was weakening.
Sept. 19, 1947 716 115- 966.5 28.54a' 9 n.mi. SW 967.5 New Orleans WHO, 23b New Orleans, La. 16
New Orleans La.
Sept. 21, 1948 1380 210� a' WBO,La.
WEO, La.
~Sept. 21, 1948 1380 210' ~935.3 27.62 8 n.mi. east 963.4 Boca Chica 7a
Boca Chica Airport, Fla.
Airport,Fla.
Oct. 5, 1948 1446 230' 977.0 28.85 Coast 979.3 Miami, Fla. 31b Miami, Fla. 13

Aug. 27, 1949 1525 130' 960.7 28.37a 50 n.mi. East West Palm Beach, 23 West Palm Beach, 14 Lowest pressure for the gulf coast
inland from Coast Fla. Fla. occurred as the storm was filling
coast 954.0 about 10 n.mi. east-southeast of
Lake Placid, Fla., or 50 n.mi. north-
east of Charlotte Harbor (Gulf of
Mexico).

Oct. 4, 1949 360- 190� 963.4 28.45a Coast 978.0 5 Miles SW of 20b Composite of many 11
Freeport, Tex. - Texas stations

Aug. 31, 1950 (Baker) 810 190' 979.3 28.92a' Coast 979.3 Ft. Morgan, Ala. 21a 23

Sept. 5, 1950 (Easy) 1162 230' 958.3 28.30e' Cedar Key, 958.3 Cedar Key, Fla. 15 d 3
Fla.

Oct. 18, 1950 (King) 1459 150' 978.0 28.88,a 50 n.mi. East Miami, Fla. MSG 17 Lowest pressure for the gulf coast
inland from Coast occurred as the storm was filling
coast 955.0 about 12 n.mi. east-southeast of
Haines City, Fla., or 50 n.mi.
east-northeast of Tampa Bay.
Sept. 24, 1956 (Flossy) 904 250' 973.9 28.76 Coast 973.9 See Remarks 22b Burrwood, La. 10 Lowest pressure taken from the barom-
eter of a dredge within the eye at
Destin, Fla., and from a reconnaissance
plane just off the coast at Pensacola,
Fla.
June 27, 1957 (Audrey) 451 200' 946.5 27.95a' Coast 958.4 Hackberry, La. 19a 14

Sept. 10, 1960 (Donna) 1330 170' 933.0 27.55c' Conch Key, 933.0 Conch Key, Fla. 20e Near Conch Key, 9
Fla. Fla.

Sept. 15, 1960 (Ethel) 747 175' 972.0 28.70 150 n.mi. S.972.0 Aircraft 18b Keesler AFB,Miss. 10
off Miss. Reconnaissance
Delta
See Legend at end of table 2.

Table l.-Continued

Gulf Coast United States

Approx. coastal Storm direction P P P Station(s) Station(s)
Date* Storm reference point (clockwise from o o value a where Pa was R where R was T
(GMT) name t(see fig. 1) north) (mb) (in.) applied to (rob) observed (n.mi.) observed (kt) Remarks

Sept. 11, 1961 (Carla) 296 170' 930.9 27.49 Coast 930.9 Aircraft 20 6 Lowest pressure indicated by air-
~~~Reconnaissance ~craft reconnaissance off Port
O'Connor, Tex. A recently cali-
brated barometer at Port Lavaca,
Tex., read 27.62 in. (935.3 .jb)
for 1 hr, 50 min. Available
information indicated the needle
was below scale during that period.

Oct. 4, 1964 (Hilda) 590 175' 959.4 28.33 Coast 961.7 Franklin, La. 21e Near 26'N, 92'W 7

Oct. 14, 1964 (Isbell) 1368 220' 964.1 28.47e' 24.3'N, 964.1 Aircraft 10d 0Near 24'N, 83'W 15
82.7�W Reconnaissance

Sept. 8, 1965 (Betsy) 1419 090� 947.9 27.99 25.2'N, 947.9 Aircraft 19a e West of Cape 15
82.1"W Reconnaissance Sable, Fla.

� Sept. 10, 1965 (Betsy) 657 135' 941.1 27.79a' 28.2'N, 946.2 Aircraft 32a b c Port Sulphur,La. 17
89.2�W Reconnaissance
at 27.9'N,
88.8�W

re Dry Tortugas Dry Tortugas, 23 d Near 30N
June 9, 1966 (Alma) 1026 200' 970.2 28.65 & 60 n.mi. W 970.2 Fla. & Air- 84'W
of Cedar Key, craft Recon.
Fla.
Oct. 4, 1966 (Inez) 1419by 065' 977.0 28.85e 24.1N 977.0 Aircraft 19b Ke West, Fla.
84.2'W Reconnaissancey es, a. 7 Lowest pressure 135 n.mi. w-sw
Key West, Fla.

Sept. 20, 1967 (Beulah) 169 155' 923.1 27.26e' 24.8'N, 923.1 Aircraft 25b BrownsvilleTex 8
96.3'W Reconnaissance
Oct. 19, 1968 (Gladys) 1162 235' 977.0 28.85e Coast 977.0 Aircraft 21a 10
Reconnaissance

Aug. 18, 1969 (Camille) 747 160' 907.9 26.81 28.2'N 907.9 A ircraft 8 Near 28'N, 89'W 16 See pages 32, 52, 55
88.8ws Reconnaissance

Aug. 3, 1970 (Celia) 243 115' 944.5 27.89 Coast 944.5 Ingleside, Tex. 9 Corpus Christi, 14
Tex.

Sept. 12, 1970 (Ella) 11 100 , 966.8 28.55' Coast 966.8 Aircraft 21a 7
Reconnaissance

Sept. 10, 1971 (Fern) 243 050' 979.0 28.91p Near coastal 979.0 Aircraft 26b Palacios and 5 Aircraft reconnaissance observed
Ref. Point Reconnaissance Point Comfort,Tex. lowest pressure just off the Tex.
340 Coast south of Matagorda, Tex.
Sept. 16, 1971 (Edith) 500 230' 978.0 28.88' Coast 978.0 Aircraft 27 Lake Charles, La. 15
Reconnaissance
June 19, 1972 (Agnes) 966 1950 978.0 288e 285 978.0 Arcraft 20 Near 28', 86 11
85.7'W Reconnaissance

See Legend at end of table 2.

Table 2.--Hurricanes with central pressure < 982mb (29.00 in.) ranked in chronological order from 1900-73

East Coast United States

Approx. coastal Storm direction Po Station(s) Station(s)
Date* Storm reference point (clockwise from o value P where p was R where R was T
(GMT) name t (see fig. 1) north) (mb) (in.) applied to (mb) observed (n.mi.) observed (kt) Remarks

Sept. 12, 1903 1499 120' 976.6 28.84 Coast 998.0 Tampa, Fla. 43a 8

June 17, 1906 1552ex 220' 979.0 28.91 Gulf Coast 997.6 Jupiter, Fla. 26a 12
ief. point
1393
b. b
Sept. 17, 1906 2011 105' 981.4 28.98 Coast 999.0 Columbia, SC 44b Charleston, SC 16

Oct. 18, 1906 1459ex 220' 976.6 28.84 Coastal ref. 990.9 Jupiter, Fla. 35a 6#
point 1419

Oct. 11, 1909, 1419by 230' 957.0 28.26c' Knights Key,957.0 Knights Key, Fla. 22b Key West, Fla. 10
Fla.

Aug. 28, 1911 1886 100' 979.3 28.92 Coast 982.7 Savannah, Ga. 27b Savannah, Ga. 8

Sept._ 3, 1913 2157 115' 975.6 28.81 Coast 994.2 Raleigh, NC 38 Hatteras, NC 16

Sept. 10, 1919 1355by 120 929.2 27.44 d' Dry 929.2 See Remarks 15a 8 Lowest pressure obtained
Tortugas, Fla. from mean of two ships
(Lake Winona, Fred W.
Weller) and Dry Tortugas,
Fla.
Oct. 26, 1921 1659ex 260' 979.0 28.91 50 n.mi. Gulf Tarpon Springs,Fla. MSG 10 Lowest pressure for the

inland from Coast East Coast occurred as
coast 952.3 the storm was filling a
few miles north of Cler-
mon tFla., or about 50
n.mi. from the Atlantic
Ocean (north of Titus-
ville, Fla.)
Aug. 26, 1924 2182by 210' 971.9 28.70a' 25 to 30 n.mi975.3 Hatteras, NC 34b Cape Hatteras,NC 22
SE of Cape
Hatteras, NC

S Aug. 26, 1924 2731by 220' 971.9 28.70 12 n.mi. SE 972.2 Nantucket, Mass 66b Nantucket, Mass. 29 Storm becoming extra-
Nantucket, tropical.
Mass.

Dec. 2, 1925 2130 220� 980.4 28.95 Coast 987.8 Wilmington, NC 54b Wilmington, NC 14 W.B. Technical Paper

No. 55 implies that this
storm was becoming extra-
tropical and did not have
hurricane-force winds when
July 28, 1926 1619 150� 959.7 28.34a Coast 975.3 Meritt Island,Fla. 14ck the NC coast.

Sept. 18, 1926 1433 110' 934.3 27.59a Coast 935.0 Miami, Fla. 24a 17

See Legend at end of table 2.

Table 2.--Continued

East Coast United States

Approx. coastal Storm direction Po Station(s) Station(s)
Date* Storm reference point (clockwise from o value Pa where Pa was R where R was T
(GMT) name t (see fig. 1) north) (mb) (in.) applied to (mb) observed (n.mi.) observed (kt) Remarks

Oct. 21, 1926 1419by 220' 931.9 27.52 60 n.mi.south 987.5 Key West, Fla. 21a 16
Key West, Fla

Sept 17, 1928 1552 120� 935.3 27.62c' Coast 935.3 W.Palm Beach and 28a 13
Everglades Exp.
Sta., Fla.

Sept 28, 1929 1393 100� 948.2 28.00c Key Largo,Fla 948.2 Key Largo, Fla. 28a 10

Aug. 23, 1933 2219 145' 969.5 28.63a' Coast 970.5 Cape Henry, Va. 36b Hatteras, NC 18

Sept. 4, 1933 �525 120" 947.5 27.98c Coast 947.5 Jupiter, Fla. MSG 11
c C b
Sept 16, 1933 2194 220' 956.7 28.25 Coast 956.7 Hatteras, NC 40b Hatteras, NC 9

Sept. 3, 1935 1393 130� 892.3 26.35c' Long Key,Fla 892.3 Long Key, Fla. 6a 9

Nov. 4, 1935 1459 060' 972.9 28.73' Coast 972.9 Mimi, la. 10c dMiami, Fla. 12

Sept. 18, 1936 2219by 180� 965.8 28.52 38 n.mi E Cape 965.8 See Remarks 34a 16 Lowest pressure is mean of
Hatteras, NC two ships (El Occidente and
Limon) off Cape Hatteras, NC.
C- D Sept 21, 1938 2576 180� 9397 278.75N, 949.5 Hartford, Conn. 50a 47 Storm becoming extratropical.
72.5�W
Aug. 11, 1940 1899 100' 974.6 28.78 Coast 974.6 Savannah, Ga. 27b Savannah, Ga 9

Sept. 14, 1944 2194 195' 944.1 27.88a' Coast 947.2 Hatteras, NC 17b Hatteras, NC 23

� Sept. 15, 1944 2601 220' 958.7 28/31c' Coast 958 7 Pt. Judith, RI 36b Providence, RI 30 Storm becoming extratropical.

Sept. 15, 1945 1433 130' 951.2 28.09 Coast 951.2 Homestead, Fla. 24b Miami, Fla. 10 Lowest pressure observed
by Fla. East Coast Rail-
road personnel at Home-
stead, PFla.
Sept. 17, 1947 1485 080' 940.1 27.76a' Coast 947.2 Hillsboro, Fla. 34b Miami, la. 10

Oct. 15, 1947 1858 080� 968.2 28.59 Coast 973.9 Savannah, Ga. 13a 17

Sept. 22, 1948 1538ex 230' 962.1 28.41 50 n.mi, 963.4 Boca Chica Airport, 16a 11 Lowest pressure for the East
inland from Fla. Coast occurred some 50 n.mi.
coast west of the Atlantic Ocean
north of Boca Ration or 12
n.mi. s-sw of Clewiston, Fla.
The storm filled only an addi-
tional 2-3 mb before entering
the Atlantic near Jenson Beach,
-Fla.
Oct. 5, 1948 1446ex 230' 977.0 28.85 Coast 979.3 Miami, Fla. 31b Miami, Fla. 13

Aug. 24, 1949 2182by 220' 977.3 28.86d' Diamond 977.3 Diamond Shoals 24a 22
Shoals Light- Lightship, NC
ship, NC
See Legend at end of table 2.

Table 2.--Continued

East Coast United States

Approx. coastal Storm direction p Po Station(s) Station(s)
Date* Storm reference point (clockwise from o value Pa where Pa was R where R was T
(GMT) name t (see fig. 1) north) (mb) (in.) applied to (mb) observed (n.mi.) observed (kt) Remarks

Aug. 27, 1949 1525 130' 953.6 28.16 Coast 954.0 W. Palm Beach,Fla. 23 . Palm Beach, 14
Fla.
Oct. 18, 1950 (King) 1459 C9 cd
150' 955.0 28.20 Coast 955.0 Miami, Fla. 6 Miami, Fla. 6

Aug. 31, 1954 (Carol) 2169 210' 960.0 28.35 33.4'N, 960.0 Aircraft MSG 10
76.8'W Reconnaissance

Aug. 31, 1954 (Carol) 2614 200' 961.1 28.38 Coast 962 4 Suffolk Co. AFB,NY 22b See Remarks 33 R was obtained from a cot-
posite of many New England
and middle Atlantic coastal
stations. Three stations
which were heavily relied
upon are Block Island,and
Quonset Point, R.I, and
Suffolk Co. AFB, N.Y.
Sept. 11, 1954 (Edna) 2718 210 947.2 27.97e' 39.7N, 949.9 Chatham, Mass. 18b Nantucket, Mass. 40
71.3'W

Oct. 15, 1954 (Hazel) 2045 190' 936.7 27.66 Coast 938.0 Tilgham Point, NC, 21b "vrtle 'eac}, rC 26
"~L by fishing boat
H-'L Judyv "'Tn'a

Aug. 12, 1955 (Connie) 2182 180' 961.7 28.40 Coast 961.7 Fort Macon, NC 45a 7

Sept. 19, 1955 (Ione) 2145 175' 960.0 28.35 Coast 960.0 Morehead City, NC 42a 9

Aug. 28, 1958 (Daisy) 2182by 195' 957.0 28.26e' 65 n.mi. 957.0 Aircraft 25de Near 35'N,74'W 17
east of Cape Reconnaissance
Hatteras, NC

5 Aug. 29, 1958 (Daisy) 2718by 240' 979.0 28.91 60 n mi 979.0 Aircraft 50e 60 mi. SE 21
SE Nantucket, Reconnaissance Ilantucket, Mass
Mass.

Sept. 27, 1958 (Helene) 2132by 240' 932.0 27.52 80 n.mi. 932.0 Aircraft 21a 14
ESE Charles- Reconnaissance
ton, SC
32.2'N, Aircraft
Sept. 29, 1959 (Gracie) 1913 150� 950 9 28.08e 80.2'W 950.9 Reconnaissance 102

Sept. 10, 1960 (Donna) 1330 170' 933.0 27.55c' Conch Key,PFla.933.0 Conch Key,Fla. 20e Near Conch Key, 9
vla.

I Sept. 12, 1960 (Donna) 2132 215' 958.0 28 29 33.9N, 77.9W 958 0 Aircraft 34b Wilmington, NC 26
34.6'N, 77.7'W Reconnaissance

S Sept. 12, 1960 (Donna) 2601 205' 961.1 28.38c Coast 961.1 Brookhaven, N.Y. 48b Suffolk Co., N.Y 32 Storm becoming extratropical
AFBPE --East-West radius of eye was
50 n mi.

See Legend at end of table 2.

Table 2.--Continued

East Coast United States

Aug. 27, 1964 (Cleo) 1499 160 967.5 28.57c Coase 967.5 . Mimi, Fla. 7 Mimi, Fla. 9
160� 9 67,5 28.57 Coast 967.5 N. ~~~~~~Miami, Fa

Sept. 10, 1964 (Dora) 1724 100� 965.8 28.52 Coast 965.8 St. Augustine,Fla 20 Near 30'N,80'W 7

Sept. 8, 1965 (Betsy) 1419 090� 951.9 28.11a' Few mi. W 952.3 Tavernier, Fla. 22ab e Plantation Key, 11
Tavernier,Fla. Fla.
e'
Sept. 17, 1967 (Doria) 2262 020� 981.0 28.97 e38.0N, 981.0 Aircraft 20 Near 38�N, 74�W 9 Lowest pressure 150 n.mi.
71.9'W Reconnaissance east of Delmarva Peninsula.
Sept. 10, 1969 (Gerda) 3060 195� 979.0 28.91d 65 n mi. 979.0 Nantucket MSG 40
SE Cape Cod Lightship

General Legend Legend Legend

P- lowest pressure detected by barometer or dropsonde Source of Radius of Maximum Winds Data Source of Minimum Central Pressure Data

P- minimum central pressure (for either the East or a - computed from pressure profile a' - computed from pressure profile along
Gulf Co ast) or near coast
Gulf Coast)
b - observed from wind speed record
R - radius of maximum winds b' - computed from pressure profile and
c - extracted from Monthly Weather Review adjusted to the coast
T - forward speed of storm
d - approximation (about 5 or 6 n. mi. added c' - observed by land barometer
to eye radius as observed by aircraft
by - bypassing storm or radar) d' - observed by ship barometer

ex - exiting storm e - aircraft reconnaissance wind data a' - observed by reconnaissance plane
dropsonde
MSG- missing

* Date applies to approximate coastal reference point

t Point at which storm entered, exited, or came closest to the coast

T Lower central pressures at distances greater than 150 n. mi. from the coast were not considered

I Same hurricane as previous line

which will make it the most severe that can probably occur in the particular
region involved.

1.5 Funding

Preparation of this report was funded by the Corps of Engineers, U.S. Army,
and the Federal Insurance Administration, HUD, under reimbursable agreements.
Annual agreements between the Office of Chief of Engineers, Department of the
Army, and the National Weather Service, NOAA, have provided for hydrometeoro-
logical studies by the NWS in support of Corps of Engineers requirements.
Annual agreements between the FIA and NOAA since 1970 have provided general
authority for NOAA to carry out flood insurance studies for coastal regions.
Specific project orders authorized studies for specific locations. This re-
port is a collection and synthesis of the hurricane climatological studies
made over the last 4 years under these authorities.

2. FREQUENCY OF HURRICANE AND TROPICAL STORM OCCURRENCES

2.1 Classification of Hurricanes and Data

The frequency that the coastal area under study experiences tropical storms
and hurricanes based on the period 1871-1973 is analyzed in this chapter.
There are three categories of storms that affect the coast in different ways.
It is therefore logical to examine the frequency of hurricane and t~ropical
storm occurrences according to these three categories: 1) landfalling
storms, 2) exiting storms, and 3) alongshore storms. These 'classes of storms
are defined in the previous chapter. The frequency of the storm occurrences
is defined as the number of tracks of each class of storms per year per
nautical mile of a smoothed coast.

The statistics on the frequency of hurricane and tropical storm occurrences
are based on the yearly storm track charts by Cry (1965) from 187 '1-1963 and
from Monthly Weather Review articles between 1964-73. Following the criteria
used in the track charts, tropical storms are defined as storms with maximum
winds 34 to 63 knots, and hurricanes as storms with winds 64 knots or greater.
For conciseness we use the term "tropical cyclone" in this report to include
both. Storms classified as "tropical depressions" (less than 34 knots) are
not included in the statistics.

2.2 Frequency of Landfalling Hurricanes and Tropical Storms

Determination of the frequency of landfalling storms in a given area would
be relatively simple if an infinite sample were available. However, data are
available for only about 100 years. During the period 1871 to 1973, 281
tropical cyclones entered the gulf coast of the United States and 131 the
east coast, a total of 412. Inspection of this sample reveals variations
within short coastal strips which may be chance occurrences due to small
sample size. A goal of this report is to smooth. out such variations. In the
vocabulary of statistics it is desired to describe ttte population, not the
sample. Special effort was made to take into account the effect of coastal
orientation on the frequency of storms landfalling on the coast of the Gulf
of Mexico.

2.2.1 Direct Count Method

The most direct method of assessing the frequency of landfalling hurricanes
and tropical storms is to count the tracks, then smooth along the coast. The
number of entries was totaled for each 50-n.mi. segment of the smoothed
coastline from a point some 250 n.mi. south of the Texas-Mexico border to the
Maine-Canada border. The 50-n.mi. segment counts are then smoothed by an
objective.technique. Figure 2 shows the frequency plot of these discrete
storm entry values at 50-n.mi. intervals (points joined by a dashed line) and
the smoothing obtained as described in the next paragraph. These frequencies
do not take into consideration the lateral extent of coast affected by an
individual hurricane, but depict the frequency of tracks of storm centers.

2.2.1.1 Objective Smoothing Procedure. The 50-n.mi. segment counts were
smoothed by weighted averaging of each successive 11 data points. These dis-
crete values (A) may be considered as a continuous input series. The
smoothed frequency values (Fi) are obtained from the equation:

5
F = W n i+n/EW (1)
n=--5 nn

If the weights (Wn) were set to l,the above equation would yield a smoothed
curve of 11-interval running means. Here we used a weight function in the
same manner as in low-pass filtering in time series analysis. The adopted
function has the following assigned weights (after Craddock 1969):

W = 0.300, 0.252, 0.140, 0.028, -0.040, -0.030; for

n = 0, �1, �2, �3, �4, �5, respectively.

An alternate smoothing procedure often applied in climatological analyses
uses a running mean approach (Wn = 1). The results thus obtained may have
distortions in phase angle variation (shifting of maximum or minimum posi-
tions) and in the total area under the curve. The weighting function adopted
here is designed to maintain the frequency and phase angle of the original
input series. These weights were applied to all successive discrete values
from Texas to Maine, yielding a weighted mean number of landfalling storms
for each 50-n.mi. interval of the smoothed coastline. The tail end of the
input series was extended as a mirror image of the original series to permit
application of equation (1) all the way to the Canadian border. These values
were connected to give a continuous smoothed curve of the frequency of land-
falling tropical cyclones (solid curve of fig. 2).

2.2.1.2 Evaluation of Procedure. The direct count method derives its data
from counting of tropical cyclones at the coast and not out over the water.
It gives the best measure of the variation along a smooth coastline of the
frequency of landfalling storms. However, it tends to obscure real varia-
tions due to coastal shape. A stretch of the coast that turns sharply in a
direction almost parallel to that of the predominant storm motion is less
exposed than adjacent coastal segments more nearly normal to the track

-~~ 0i~- W -V C') 0n C)Z
o > ~> - > > H - >
~~~~~~~~~~< 0 >
3 6- C' m m n z m > Z M
> 0 M -- > E n ) :r
m > r-C CD~ -- >
- ~~~~~~ - Cn -n mn z z
F H r- r- > -f co,
rI M> n >d

0~~~~~~~~~~~~~~ 0
028- '
* 0 :1

Cd,

CD2.0-9

10
0

04-0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

DISTANCE (n mi x 102 )

Figure 2.--Count of landfalling tropical storms and hurricanes (1871-1973) by 50-n.mi.
segments of a smoothed coastline (points plotted and connected by dashed lines).
Solid line denotes the entry frequency curve obtained by applying the objective
smoothing function.

direction. We have implicitly smoothed out the coast while smoothing out the
accidental landf ailing points of storms.

To identify areas where the implied smooth. coastal direction differs sig-
nificantly from the actual coastline, a smoothed coastline was constructed.
Coastal locations at 50-n.mi. intervals along the entire gulf and east *coasts
were smoothed (separately by latitude and then by longitude) using the same
smoothing function mentioned earlier. These points were plotted and a con-
tinuous line joining these points was drawn as shown in figure 3. This
diagram reveals that this smopth line cuts across the actual coastline at
several places -- specifically, along the west coast of Florida and across
the Mississippi Delta. For the most part, the smoothed coastline approxi-
mates quite well the orientation of the actual coast.

2.2.2 Track Density Method

To provide a tool for varying storm landfalling frequency where the coast
turns abruptly on a scale of less than 50 n.mi., for example at Apalachee
Bay, the track density method was developed. This method enables us to
smooth storm behavior without regard to the coast. It must assume, however,
homogeneity of storm behavior over a fairly sizable map area, including both
land and sea. This method was carried out for only the Gulf of Mexico.

2.2.2.1 Basic Data. Tropical cyclonie tracks on the charts by Cry and suc-
cessors (sec. 2.1) over the Gulf of Mexico and adjacent areas were counted
passing through each 2.50 latitude-longitude square, with the corners cut off
to form an octagon, approximating a circle. Frequencies of track directions
by 150 class intervals were also tabulated for each octagon. The track count
through each octagon is construed as the number of storms passing through a
circle of 142-n.mi. diameter per 103 years and is converted to point track
density. A detailed description of procedures is given in the appendix.

2.2.2.2 Landfalling Frequency from Basic Data. Figure 4 shows straight-line
segments representing the gulf coast to which the track density method was
applied to obtain a separate estimate of the frequency of landfalling
tropical cyclones. 'The direction of each segment was used to separate storms
that strike the coast from those that exit the coast, a 1800 direction span
for each. To get the landfalling frequency the 1800 landfalling direction
span is further subdivided into 150 class intervals (par. 2.2.2.1). The
frequency with which storms enter the coast for a particular direction class
is the product of the point track density of the region, the fraction of
total storms within the class interval, and the sine of the angle between the
coast and the storm direction class interval mean. Summing over class inter-
vals gives the total frequency. Since (by definition) alongshore storms move
at a small angle or parallel to the coast, the sine of the angle between the
coast and the alongshore storm direct-ton approaches zero. Alongshore storm
counts automatically disappear in the resulting frequency value.. Further
details on this procedure are given in the appendix, Figure 5 shows the
frequency values of landfalling storms on the stylized gulf coast of figure
4 from application of this. method.

EASTPORT

EW PORTLAND

ATLANTIC CITY

sJ\ A . O R FOLK 0

l X , , , st \,~~~~~~~~F

t.- ------rS.C. Nf WILMINGTON

C\ HAR LESTON

r- -) -l ---| ALA G SAVANNAH

TEX. LA. PENSACOLA - MAYPORT
NEW ORLEANS DAYTONA BEACH
! APALACHICOLA

TAMPA VERO BEACH

CORPUS CHRISTI FT. MYER
MIAMI

PORT ISABEL

G U F M KEY WEST

Figure 3.--Smoothed coastline obtained by applying the objective
smoothing function of figure 2.

Figure 4.--Coastal segments selected for frequency analysis by track density method
(result shown in fig. 5).

I t I l > I i t t
w' w U) J- -I4
Ui LUL-i
j~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~L U. LL LL U LL
z LUl 6CwL
c~~~~~~~~o ox w
U) -J~ ~~~~~~~~~~~~~~ 0 0 C L
U) W
I-~~~~~~~~~~L <. U)
-i w z <-
Q. -j <
3.0-<

0
2.5-
0 -

w 2.0-

co~o
0~ --

0 5- 0J

I- 1.5-� -

0.5 -

1 2 3 4 - 5 6 7 8 9 10 11 12 13 14 15

DISTANCE In mi x 1021

Figure 5.--Frequency of landfalling tropical storms and hurricanes (1871-1973) by track
density method.

2.2.2.3 Evaluation of Method. Areas where a smoothed coastal direction
differs substantially from the actual coastal direction may be detected in
figure 3. These areas may either be sheltered from or exposed to the pre-
vailing direction of storm motion more than the smoothed coastal direction
would predict; the track density method was designed to tackle this problem.
The effect of the coastal orientation on frequency count is illustrated in
figure 5 by differences in frequency values between north-south segment 6 and
adjacent east-west segments 5 and 7.

2.2.3 Combination of Methods

Figure 6 is a combined depiction of the frequency of landfalling tropical
cyclones. The east coast portion of the frequency curve is the same as that
shown in figure 2, obtained from applying the smoothing function cited
earlier. On the gulf coast, differences between the direct count method
(fig. 2) and the track density method (fig. 5) were treated as follows: At
the southern tip of the Florida peninsula~ the curve branches. The upper
branch applies to the Florida Keys, smoothly joining Florida east coast
values. The lower branch pertains to the mainland coast of Florida Bay, con-
tinuing from Florida west coast values, and is a smoothed version of figure
5. The northeast gulf coastal region, mile 1,000 to mile 1,250, is also a
smoothed version of figure 5. In the eastern Mississippi Delta region no
attempt was made to compromise the two curves and figure 2 is replicated
here. The user is advised to consider fully both figure 2 and figure 5 for
applications in this area. On the remainder of the gulf coast the two
approaches give substantially the same result and we adopt the figure 2
smoothed curve.

2.2.4 Discussion of Results

2.2.4.1 Overall View. Figure 6 reveals that the range of occurrence of
landfalling tropical cyclones over a 100-yr period varies from a minimum of
0.1 storms per 10 n.mi. of smoothed coastline near Boston, Mass., to a maxi-
mum of 2.2 in the middle of the gulf coast of northwest Florida and in the
Florida Keys. A frequency of close to 2.0 storms per 10 n.mi. per 100 years
appears to the south of Galveston, Tex. Highest frequency of landfalling
tropical cyclones on the east coast is in southern Florida, and a compara-
tively high frequency appears to the south of Cape Hatteras, N. C. The fre-
quency of entries drops off rapidly from Miami to Daytona Beach, Fla., and
from Cape Hatteras northward to Maine.

2.2.4.2 Areas of High Entry Frequencies

a. Northwest Florida. The high frequency of storm entries along the
northwest Florida coast near Pensacola (fig. 6) suggests that this stretch
of the coast is a favorable crossroad for tropical cyclones which passed east
of the Yucatan Peninsula and recurved in the Gulf of Mexico. This coastal
region is also vulnerable to Atlantic storms that cross the Florida peninsula.

b. South Florida. A maximum in landfalling storm frequency appears near
the tip of the Florida peninsula and along the Florida Keys. The southern-
most portion of this area is exposed to both Atlantic and Caribbean hurri-

If I 4' I 44 4' I 4'Ij't I 4''4I (I ' I '4 $1 14
>m - > > m 0 >
I- ;r :-I > > < -u )
-I <m -Cm 4
Cl) Cl -< :- I z 0
:-- m0 0 0
En M -n >
>r ~ 4 0 > (n -
a >I- C) --4
m z > - v" m
1- - I~~~~~~~~ n ii m M l
* -~~ riM -n - - > ?! cn
mC/ 1 r, > CrnC)
m- x 1- ~E
x > > w
I- > -n z

LO
o 2.8 -

U-
L
2.4-
Cl)

2.0 -

~-1.6-

LL -
a:
cc

w 1.2-

0.8 -

0.4 - LANDFALLING STORMS

o 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

DISTANCE (n mi x 102)

Figure 6.--Adopted frequency of landfalling tropical storms and hurricanes (1871-1973) for
the gulf and east coasts of the United States.

canes. Generally, tropical cyclones strike the east coast of South Florida
from an east-southeasterly direction -- a predominant direction for Atlantic
hurricanes before recurvature. The west coast of South Florida is vulnerable
to late-season tropical cyclones moving in a northeastward direction after
recurvature. The most frequent areas of recurvature in the month of October
have been near Bermuda and over the northwestern Caribbean (Cry 1965).

c. Upper Texas Coast. The comparatively high frequency along the upper
Texas coast is partially caused by the predominantly westward-moving storms
in the Gulf of Mexico during the early hurricane season. Only six storms
have recurved and moved northeastward (away from the Texas coast) during the
months of June, July, and August since 1901. These early season storms
accounted for more than half the total number of storms that struck the Texas
coast.

d. Cape Hatteras. The high frequency of storm entries just south of Cape
Hatteras, N.C. (1.6 storms per 10 n.mi. per 100 years) is the combined result
of the number of storms that reentered the North Carolina coast after exiting
the east coast of Florida and Georgia in addition to hurricanes of Atlantic
origin that move in a northerly direction after recurvature. Almost 90% of
the storms entered the North Carolina coast, south of Cape Hatteras, in a
northwesterly to a northeasterly direction.

2.2.4.3 Areas of Low Entry Frequencies. The frequency of storm entries is
less than 1 per 10 n.mi. of coastline per 100 years over the northern section
of the east coast from a point some 50 n.mi. north of Cape Hatteras northward
to the Canadian border and also in the vicinity of Daytona Beach, Fla. The
rapidly decreased frequency of entries north of Cape Hatteras, N.C., is
easily understandable. With a few exceptions, hurricanes recurving south of
Cape Hatteras either enter the North Carolina coast or move northeastward
farther from the U.S. mainland.

a. East Coast. Colon (1953) has shown the locus of points of highest
frequency of recurvature for different months of the hurricane season.
Hurricanes off the east coast of the United States frequently recurve between
latitudes 270 and 29'N during the months July through September. For the
other months of the hurricane season, recurvatures occur in latitudes farther
south, following the shift of the subtropical ridge (Alaka 1968). The
northern limit of hurricane recurvature at about 29�N appears to coincide
with an area of minimum frequency of landfalling hurricanes along the east
coast. Hurricane Dora of September 1964 was the only hurricane that struck
the northeastern Florida coast in recent years.

b. Gulf Coast. The relative minimum in storm entry frequency along the
west coast of Florida (compared to the mid-gulf coast and the southern tip of
the Florida peninsula) can be explained by the prevailing westward motion of
hurricanes of Atlantic origin. The relatively low frequency of storm entries
along the Louisiana Coast west of the Mississippi Delta does not have a ready
explanation.

2.3 Frequency of Exiting Hurricanes and Tropical Storms

2.3.1 Analysis

The frequency of exiting tropical cyclones was defined by a subjective
smoothing of 50-n.mi. segment coastal crossings. These counts were obtained
from the storm track information previously cited. A total of 141 tropical
cyclones exited the east coast and 20 the gulf coast during the period 1871-
1973. The shape of the coast relative to storm tracks and meteorological
considerations were taken into account in the smoothing. For storms exiting
the coasts of Florida, consistency in frequency and direction of movement was
maintained with the frequency of landfalling storms on the opposite coast.
The objective smoothing technique was not used in this analysis because the
observed data are closely related to the geographical features of the coasts
and.because of physical considerations. For these reasons, the high fre-
quencies of exiting storms that concentrated in two areas of the east coast
were not smoothed out by averaging with lower frequencies of adjacent coastal
areas.

2.3.2 Results and Discussion

Figure 7 shows the smoothed frequency distribution of exiting tropical
cyclones. This curve indicates high frequencies along the coasts of northern
Florida and Georgia and along the North Carolina coast north of Cape Hatteras.

2.3.2.1 Gulf Coast. The comparatively few exiting storms along the northern
portion of the west coast of Florida agrees with the decrease of landfalling
storms northward along the Atlantic coast of Florida. A maximum of exiting
storm frequency occurred near Fort Myers, Fla.

2.3.2.2 East Coast. The maximum frequency of exiting storm occurrence
appears near Jacksonville, Fla. with 3 storms per 100 years per 10 n.mi. of
the smoothed coastline (see fig. 7). The frequencies decrease southward with
2 storms/100 years/10 n.mi. near Daytona Beach, 1 storm/100 years/10 n.mi.
near West Palm Beach,.and 0.4 storm/100 years/10 n.mi. near Miami, Fla. The
frequency diminishes rapidly north of Jacksonville. Higher values appear be-
tween Cape Hatteras, N.C., and Cape Henry, Va.

Many exiting storms along the Atlantic coast originally were eastward-
moving storms in the Gulf of Mexico. They can also be traced to storms that
recurved over the gulf or over the Florida peninsula south of the 29th para-
llel and moved northeastward north of the subtropical ridge. This last group
accounts for the high frequency of exiting storms over the northeastern por-
tion of the Florida peninsula. The concentration of exiting storms just north
of Cape Hatteras and Cape Cod reflects the orientation of the coastline and
the comparatively high counts of entering storms south of these capes.

2.4 Frequency of Alongshore Hurricanes and Tropical Storms

2.4.1 Analysis

The frequency of tropical cyclones that bypassed the coast is based on the

'I I 4! ji I 41 ' Ii I ' I 4 '' 4 I I t 4 mI
-v 0- w' -m cn -n 7 o o >
o - A -g - r m >
-4 m >
M~~ ~ > z:
> m cs m -n z m > ZE
M > -I r--
Cn cn -1 ?;-w
* ~ ~~~~~ ~ ~ 3 v, >v im > n
m ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ x z
P~ r- > > >Y

x r - -I

O 2.8 -
UA.

~ 2.4-
o 0
E,

~ 2.0- 09
20 -
0~~~~~~~~~~~~~~~~~~

1.6-

o 0 ~~~~~~~~~~~~~~~~~~~~~0

w 1.2 -

0.8 - 0 0 0 0

0.4 - 0 0
EXITING STORMS

0 , I I * I I i I * i.~ L I I I ,-rCk *n,
0 2 4 6 8 10 12 14 16 18 20' -22 24 26 28 30

DISTANCE (I mi x 102)

Figure 7.--Frequency of exiting hurricanes and tropical storms (1871-1973). Curve fitted
subjectively.

same maps and data period used before (sec. 2.2). A count was made of storms
intersecting 5-n.mi. intervals along lines drawn perpendicular to a smoothed
coastline centered at each of the coastal locations (A to Z in fig. 8). The
same storm may be counted several times as it moves parallel to the coast.
These track counts are summarized in table 3. The cumulative track counts
along each of the 26 lines normal to the coast were plotted against the dis-
tance from the coast. A smooth curve was then fitted by eye to the data on
each of these frequency plots. Storm track frequencies were then scaled from
the smooth curves for certain distance intervals from the coast. These fre-
quencies are listed in table 4 with units of tracks per nautical mile per
100 years.

The frequency distributions were also plotted on a map and smoothed subjec-
tively both along the coast and perpendicularly outward. This distribution
is shown in figure 8 by isolines of accumulated number of storm tracks by-
passing the coast at sea for the period 1871-1973.

2.4.2 Results and Discussion

Figure 8 reveals that the riaximum concentration of alongshore storms
occurred off Cape Hatteras, N.C. Fewer than 5 tropical cyclones bypassed
within 50 to 80 n.mi. off the coasts of northwest Florida, Alabama, and
Mississippi and within some 100 n.mi. of the Texas coast. The higher values
off the Mississippi Delta may be caused by geographic protrusion. There is a
high frequency of bypassing storms off the coast' of Cape Hatteras for the
same reason that there is a high frequency of landfalling storms' south of
Cape Hatteras. This was discussed in par. 2.2.4.2d.

2.5 Ratio of Hurricane to Total Storm Occurrences

Figure 9 shows the ratio of hurricanes to the total number of tropical
cyclones that entered the coast during the 88-yr period 1886-1973. These
ratios are needed to adjust the hurricane central pressure frequencies to
full range. A description of this application is included in chapter 3. The
source of data for this analysis is the storm track charts previously cited.
Tracks prior to 1886 were not used because the classifications of hurricanes
and tropical storms are not specified in the earlier data. The designation
of "hurricane" or "tropical storm" by Cry (1965) (based on maximum surface
wind speeds) was accepted. These criteria were mentioned in section 2.1.

2.5.1 Analysis of Data

To determine a smooth ratio of hurricane occurrences to tropical cyclones
at a set of coastal points the numbers of landfalling hurricanes (H) and
tropical storms (T) per 50-n.mi. increments were counted separately. Then
ratios of H to H+T were computed. For consistency, the sampling and smooth-
ing of the ratios were designed in the same manner as those for the hurricane
central pressure (chap. 3) since pressures are the principal application.
The count of hurricane or tropical storm tracks (centered at each of the 61
coastal points at 50-n.mi. intervals) included landfalling storm tracks
intersecting the coast within 500 n.mi. along the gulf coast and within 400
n.mi. along the east coast. These ratios were used as input to the

Table 3.--Count of hurricane and tropical storm tracks

crossing lines normal to coast --- 1871-1973

Line on Distance zntervals from coast (n.ml.)
Figure 8 0-5 5-10 10-15 15-20 20-25 25-30 30'-35 35-40 40-45 45-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-1~9

A 0 0 0 0 0 1 0 0 0 2 1 1 2 0 1 1 3 4 0 2
B 0 0 0 0 0 0 0 0 0 1 0 0 4 1 1 2 0 1 1 1
C 0 0 0 2 0 0 0 0 0 0 0 1 2 2 i 1 0 3 2 1
D 0 0 0 0 0 1 1 0 0 2 1 1 5 0 6 2 2 1 1 1
E 0 0 1 4 1 4 0 2 0 0 1 3 2 4 3 2 3 3 3 1
F 0 0 0 0 0 0 0 0 0 0 0 3 1 3 4 1 3 4 2 1
G 0 0 0 0 0 0 0 0 0 0 0 0 4 0 1 1 3 1 1 1
H 0 0 0 0 0 0 0 0 0 0 0 I 3 0 5 5 1 1 0 1
I 0 0 0 0 0 1 1 1 0 1 0 0 2 0 0 2 3 0 0 5
J 0 0 0 1 0 4 1 0 0 2 2 1 4 2 i 1 0 3 2 3
K 0 1 i 0 1 3 0 0 0 1 4 2 4 0 1 0 0 0 0 0

bo L 1 1 0 0 0 1 0 1 1 0 1 3 2 2 3 1 1 2 0 1
M 2 0 0 4 2 4 0 2 0 0 4 3 5 1 3 0 0 0 0 0
N 1 0 1 1 2 1 0 1 1 1 0 1 7 3 4 2 2 2 3 4
O 0 0 0 1 1 1 0 1 1 4 1 4 5 0 5 0 2 9 2 8
P 0 1 t 0 1 1 0 2 0 5 1 2 6 1 ,2 4 3 3 3 3
Q 0 0 2 3 1 1 0 2 0 3 6 4 4 3 7 2 3 2 0 2
R 0 1 1 0 3 2 ,1 3 0 9 0 4 6 3 6 2 10 3 4 5
S 2 1 0 0 3 2 1 2 0 0 1 4 5 3 7 3 3 6 0 5
T 2 0 0 2 2 1 3 1 0 4 3 3 4 5 6 2 7 10 6 4
U 0 0 0 0 1 0 1 1 1 0 1 2 3 3 5 2 4 11 1 0
V 0 0 0 0 0 2 0 2 1 2 4 2 6 5 3 1 5 5 3 3
W 1 1 1 4 1 2 0 1 0 3 1 2 0 2 5 0 4 5 4 4
X 0 0 0 1 0 1 0 0 0 0 0 6 3 3 3 3 5 1 2 2
Y 0 0 1 0 1 2 0 1 0 2 0 4 0 1 4 0 4 4 3 5
Z 0 0 0 0 1 1 0 3 0 2 2 4 4 1 1 2 3 7 2 4

Line on
Figure 8 0-5 5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-45 45-50 50-60 60-70 70-80 80-90 90-100 100-110 110-120 120-130 130-140 140-150

Table 4.--Frequency of hurricane and tropical storm tracks*
crossing lines normal to coast -- 1871-1973

Distance interval from coast (n.mi.)
Line on
Figure 8 0-10 10-20 20-30 30-40 40-60 60-80 80-100

A 0 .039 .098 .078 .087 .11 .12
B- 0 0 0 0 .068 .13 .14
C .025 .025 .049 .049 .082 .087 .14
D 0 .01 .078 .12 .18 .21 .28
E .17 .19 .21 .24 .25 .27 .28
F 0 0 0 0 0 .068 .27
G 0 0 0 0 .022 .11 .12
H 0 0 0 0 .05 .14 .24
I 0 .048 .087 .097 .088 .088 .097
J .13 .13 .19 .21 .23 .24 .24
K .12 .12 .16 .18 .21 .25 .28
L .16 .14 .11 .12 .12 .14 .25
M .35 .35 .33 .33 .31 .31 .29
N .12 .18 .18 .19 .20 .26 .31
0 .019 .11 .18 .19 .21 .27 .35
P .048 .14 .18 .19 .21 .27 .37
Q .16 .21 .27 .33 .34 .42 .42
R .097 .21 .23 .37 .43 .46 .48
S .27 .12 .16 .18 .23 .34 .41
T .16 .19 .25 .29 .36 .42 .43
U .01 .039 .068 .12 .14 .21 .32
V 0 .078 .14 .18 .29 .44 .51
W .27 .27 .24 .24 .24 .23 .23
X .02 .078 .097 .16 .18 .20 .27
Y .05 .087 .16 .16 .17 .18 .22
Z 0 .058 .14 .19 .27 .27 .27

0-10 10-20- 20--30 30-40 40-60 60-80 80�-100

Unit is storm tracks/n.mi./100 years. Derived by smoothing table 3.

TEXLA

CMKM
MATAGORA \ 5. .-~~Kw

CHIIS~~~~~~~~~i "~~3

WHPOnsM N

Figure 8.--Accumulative count of hurricane and tropical storm tracks
passing the coast at sea (1871-1973). Based on counts along
heavy dashed lines shown projected normal to coast.

I *' I ' ' 1I 4' 4' I 4' '1 I ' ' f '4 I 1
- - -o 0 - n o z m
o > ~ -- m -~ -4 -I 3>: m 0>
r- z ' � ~, _ > 0
>~~~
< m x > > m 0
:> --o r- Z )
M M co 0 0 0
m .-I z
Mm -n F- > >

0~~~~~~~~

Ix
, X ~~> 3>

0.3 -
cn
M ,

0.7

co 0 3> I' 03 oI -I ._ I ~ i: - - I
0~~~~~~~~~~~~~
0 r- 2 4 6 8 1 12 1 16 10 02 2 2 2
D (n m �
ticn c

0nr -
~~~~~~~~~~~~~z

0.3 -
F9o o f ofs a
~~~~~oo

0O A
~:0.4- 0 ~ ~ OP -

n.3-

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

DISTANCE (n mi x 102)

Figure 9.--Ratio of landfalling hurricanes to total number of landfalling hurricanes and
tropical storms (1886-1973). Based on counts in overlapping zones centered
at 50-n.mi. intervals and objective smoothing along the coast.

objective smoothing scheme previously described. A smoothed continuous curve
along the coast was obtained in the same manner as for landfalling hurricane
frequency distribution. This is shown in figure 9.

2.5.2 Results and Discussion

Figure 9 shows high ratios of hurricanes to total tropical cyclones for the
vicinity of Miami, Fla., (0.62), Cape Hatteras, N.C., (0.64),Long Island,
N.Y., (0.59), and in areas south of Galveston, Tex., (0.56). These local-
ities are close to areas having high frequencies of landfalling tropical
cyclones (compare figs. 6 and 9). This suggests that areas most vulnerable
to weaker tropical storms are also vulnerable to hurricanes. The one excep-
tion is the coastal area of northwest Florida. Only 42% of landfalling
storms attained hurricane intensity along the northwest Florida coast where
the frequency of landfalling tropical cyclones has the highest value of all.
A dip in the ratio along the west coast of Florida, north of Fort Myers also
indicates that this coastal area experienced more landfalling tropical storms
than hurricanes. This implies a less frequent visit of more intense hurri-
canes and is reflected in a trend of decreasing storm intensity northward
along the west coast of the Florida Peninsula as discussed further in the
next chapter.

3. PROBABILITY DISTRIBUTION OF CENTRAL PRESSURE

3.1 Data

Minimum central pressure is a universally used index of hurricane inten-
sity. Harris (1959) demonstrated that storm surge height is approximately
proportional to the central pressure depression, other factors being con-
stant. This chapter develops probability distributions of minimum hurricane
and tropical storm central pressure at the coast.

The data on which to base hurricane central pressure probability distribu-
tions for the gulf and east coasts of the United States have been collected
in tables I and 2, which are updated and revised versions of table A in
National Hurricane Research Project Report No. 33, Meteorological Considera-
tions Pertinent to Standard Project Hurricane, Atlantic and Gulf Coasts of
the United States (Graham and Nunn 1959). This paper is hereafter referred
to as "Report No. 33." Revisions were made in table A data where the authors
discovered that recent published reports (e.g., Sugg et al.1971) had un-
covered new data or had verified suspect data not accepted for Report No. 33,
and, in a very few cases, as an analyses judgment after reviewing all the
data. The tables list parameters of all hurricanes with a central pressure
less than 982 mb (29.00 in) that crossed the gulf and east coasts or passed
within 150 n.mi. on the seaward side of the coast during the 74-yr period,
1900-73. The year 1900 was chosen to initiate the central pressure study by
weighing the inaccuracies that would result from the sparse data of earlier
years against the desirability of a longer record. Those exiting storms,
still of hurricane intensity at the coast of exit and within 50 n.mi. of the
coast, are included.

The specific pressure values given for each hurricane are the lowest pres-
sure from actual observations (po) by a barometer or dropsonde and the min-
imum central pressure (po) estimated from the observations. Observed
pressures are extrapolated inward to p0 where necessary from Pa by using
visually-fitted radial pressure profiles based. on the formula

__PO -R/r
= e (2)
n o

where p is the pressure at radius r, po is the pressure at the center, Pn is
the pressure at some large distance from the center to which the profile is
asymptotic, and R is the radius at which the wind speed is greatest.
Schloemer (1954) tested 10 possible pressure profile formulas before select-
ing the one above as best fitting 9 sample hurricanes. The observing station
for Pa values and a geographical reference point for p0 values indicating
whether the lowest central pressure pertains to the coast or as far as 150
n.mi. offshore are also listed in tables 1 and 2. The superscript letters
following the po values refer the reader to notes at the end of the table
giving the source of minimum central pressures.

In some areas, barometric pressures could not be obtained near the coast.
The central pressure was determined at the location nearest the coast where
reliable observations could be obtained and adjusted downward to a coastal
value. The adjustments of this type made �n Report No. 33 were carried over
to this report. Since the 1950s the availability of reconnaissance data has
eliminated the need for this kind of adjustment.

The criterion that the central pressure be less than 982 mb was based on
the consideration that the maximum cyclostrophic wind speed, computed from
the Hydrometeorological Branch model' (Myers 1954), with a central pressure
of 29.00 in (982 mb) and an asymptotic pressure of 30.00 in (1015.9 mb), is

1 2 1R -R/r
Vc = n - Po) 'e (3)

where

V = cyclostrophic wind speed, at which the centripetal acceleration
c
exactly balances the horizontal pressure gradient force at radius, r.

p = density of air

p = asymptotic pressure

p0 central pressure

R = radius of maximum wind

73 mi/hr, or about the wind speed required for classification as a hurricane.2

A virtual absence of pressure data made it necessary to omit one storm
altogether - - the Louisiana hurricane of August 6, 1918, in which the closest
recorded pressure was some 90 n.mi. from the path of the storm center. An
estimate of the central pressure from such a distance would be so unreliable
as to be useless.

Two storms used in an earlier tabulation of hurricanes (Report No. 33,
table A), those of September 11, 1903 (gulf coast) and October 20, 1924 (east
coast), have been eliminated from the present summary. Upon reanalysis of
the data, it was decided that both had weakened to tropical storm strength
before they reached a point 50 n.mi. from where they would exit the coast.

Questions have been raised as to the minimum central pressure of Hurricane
Camille which struck the northern gulf coast in 1969. The best obtainable
value is needed because Camille had the lowest central pressure on the main-
land coast since record-keeping began dur'ing the last century, and strongly
influences the lower end of the probability distribution of central pressure.
A minimum pressure of 905 mb was measured by an Air Force reconnaissance air-
craft at 0016 GMT on August 17, 1969 near 25"2'N, 87.2'W, or 250 mi southeast
of the mouth of the Mississippi River. Eighteen hours later, and only a few
hours before the center made landfall, another reconnaissance aircraft pene-
trated the hurricane, and reported an even lower central pressure of 901 mb.
A post-audit of the dropsonde computation at the National Climatic Center
adjusted this to 908 mb. This value, which is quoted by Bradbury (1971), is
the value in table 2. The eye passed over Bay St. Louis, Miss,, at landfall
and an aneroid barometer a few blocks from the west end of the Bay St. Louis-
Pass Christian bridge read 26.85 in (909.4 mb). This barometer was later
checked and found to be accurate by the New Orleans National Weather Service
Office (DeAngelis and Nelson 1969). One may assume then that Camille re-
mained in a near steady state during its last 28-plus hours at sea.

Table A of Report No. 33 listed hurricanes by zone and in many cases a
particular storm appeared in more than one zone. Tables I and 2 of this re-
port do not list storms by zone; the tables l~ist a storm twice only if it
crosses the coastline a second time (or if a bypassing storm makes another
approach to the coast) after it has traveled a distance of 400 n.mi. (500
n.mi. along the gulf coast). These duplicate storms are identified by a
section mark (� in the two tables.

2We realize that there have been storms with hurricane-force winds and
central pressures as high as 990 mb south of 35'N. A recent example-is
hurricane Alma (May 1970, 993 mb, 70-kt winds). There have also been recent
tropical storms whose central pressure was less than 990 mb (Delia, September
1973, 986 mb, 60-kt winds). The 982-mb criterion is to put definite bounds
around a data sample. It is not intended to be used as a forecasting
criterion to distinguish hurricanes from tropical storms.

With central pressure available for an average of less than one hurricane
per year for the period of record for each coast, the data in tables 1 and 2
are a limited sample. Additional data will change the results of the study.

3.2 Analysis

Cumulative frequencies of hurricane central pressures were determined from
the po's in tables 1 and 2 for overlapping zone centered 50 n.mi. apart
(fig. 3). These increments were 400 n.mi. long on the east coast, and
500 n.mi. on the gulf coast, approximately comparable to the zones in
Report No. 33.

On the east coast, a new overlapping zone was used each 50 n.mi. as far
north as the mouth of Chesapeake Bay, each 100 n.mi. from there to eastern
Long Island, and a single zone from there to the Canadian border. Gulf coast
data were grouped into overlapping zones at 50-n.mi. intervals, except at
twice this interval in south Texas, because of sparse data.

Tables 1 and 2 include only hurricanes with central pressures below 982 mb.
The track counts, by contrast, on which the storm frequency count (chap. 2)
is based, include tropical cyclones of hurricane and tropical storm intensi-
ties. In order to enforce consistency at this point, we must either expand
the central pressure probability distribution to statistically include the
weaker storms, or else adjust the storm track count to storms with central
pressures less than 982 mb. We must make this choice because the frequency
of a representative climatologically specified hurricane of given character-
istics is the product of the frequency of all storms and the probability of a
storm having those particular characteristics. We chose to expand the
central pressure frequency distribution rather than contract the storm track
count, in the manner to be described.

For each 400- or 500-mile zone the po's from tables 1 or 2 were plotted on
a graph of central pressure vs. cumulative percent of storms by the usual
formula

R-0.5 (100) (4)
P - N

where p is the probability expressed as a percent of the total number of
storms, N, and R is the rank from lowest to highest. To get N for all trop-
ical cyclones, the count of po's (up to 982 mb) is divided by the ratio of
hurricanes to total storms from figure 9. The upper part of the eye-fitted
curve for each graph is extended smoothly to 1003 mb at the 100% level to
arbitrarily represent the tropical cyclones with central pressures > 982 mb.
Examples of cumulative frequency curves for two coast zones are shown in
figure 10. The first is centered near Biloxi, Miss. and the second along the
central South Carolina coast.

Using the smoothed set of cumulative frequency of minimum central pressure
graphs, we then read off the 1, 5, 15, 30, 50, 70, and 90th percentiles for
each increment and plotted as alongshore profiles. Raw data obtained from
strict adherence to best-fit cumulative frequency curves (without alongshore
33

G I
z I
u- I

uL 60- I
O I
z _
I
c) I

. 40-

D
0
_ os
20- -
o~~~~~20-~~~~~~~~~~~~~~ (a)
COASTAL REFERENCE
o MILEPOST 750 -
(NEAR BAY ST. LOUIS, MISS.)
0 � I I , I
900 920 940 960 980 1000
CENTRAL PRESSURE (mb)

100

I
t J
z 80 -
w I

c
0 II

O 60- o
0 - I
z
G --

w
40 -

o 20 -
u (b)
COASTAL REFERENCE
MILEPOST 1950 -
0 I(CENTRAL SOUTH CAROLINA COAST)

900 920 940 960 980 1000
CENTRAL PRESSURE (mb)

Figure 10.--Graphs of central pressure vs. cumulative percent of occurrences
(a) Gulf of Mexico (milepost 750), near Bay St. Louis, Miss.,
where hurricane Camille went ashore and (b) east coast (milepost
1950) along central South Carolina coast.
34

smoothing) are shown at the 5% level in figure 11. Analysis was then under-
taken with the object of obtaining a set of curves representing a consistent
view of the probability distribution of central pressure for the gulf and
east coasts of the United States. The first attempt at the coastal analysis
of the data was not acceptable to the authors in many areas because the best
fit data of the cumulative frequency graphs were not always the most con-
sistent solution for successive 50-n.mi. increments. The cumulative fre-
quency graphs were then reexamined and those needing adjustment away from a
best fit to enforce consistency were reanalyzed. A new set of percentiles
for each increment was plotted and then analyzed smoothly using the same
weighting function employed in chapter 2. The 5% curve from mile 600 to
1;200 is drawn below the data to fit the 1% curve, which in turn is drawn to
Camille.

The relative infrequency of hurricanes near the Canadian border and of po
data near the Mexican border forced us to further smooth these end areas by
eye. A discontinuity in the analysis with respect to all but the uppermost
class interval was found to exist between the chain of Florida Keys and Cape
Sable because of the physical separation of some 0.5� of latitude (30 n.mi.).
This discontinuity is reflected in the graphs.

For additional control, maps of the gulf and the lower portion of the east
coast were analyzed at various percentages using a central location for the
pressure data from each overlapping 400-or 500-n.mi. zone. These locations
(indicated by stars in figure 12) are the centroids of pressure-observing
land stations, ships, or reconnaissance planes supplying observed pressures
(Pa's) for the zone; they were used to approximate the centroids of po's.
The central pressure for the respective zone at the given probability level
was assigned to each centroid location and the data were then analyzed on
constant percentile charts. The resulting analyses were studied along broad.
lines to see if there was any basis for further alteration of the coastal
analyses cited in the paragraph above. Some minor changes were made.

The 1% and 50% control maps are illustrated in figures 12 and 13.

3.3 Results

An inspection of figure 11 reveals that there is an overall increase in
central pressure from south to north, a well-known fact primarily caused by
decreasing water temperature toward the north. Distinct minima ranked in
order of lowest-pressure at the 5% level are found on 1) the Florida Penin-
sula south of about 280N; 2) at the Texas-Mexico border; 3) at the South
Carolina-North Carolina border; 4) near Louisiana's Mississippi Delta; 5) and
over the southern New England coast.

The primary maximum 1) rests near the (until-recently) sparsely populated
coastal area west of Cross City, Fla. (mile 1,080 in fig. 11). Secondary
maxima lie 2) near the mouth of Delaware Bay; and 3) near Jacksonville, Fla.
The 'Jacksonville maximum exceeds the Delaware Bay maximum when the three
lowermost class intervals are ignored. Pressures also rise northward along
the upper New England coast.

'+1 ~~ 'I14' I I4$ 1$ I ' I' I$' I I$ '$ I $I f$ I fItIf
C) r- m - n :z 0 z co m
>~~~r >~ m >0

rn n > m
M M E~~~~~~~~~n >
z~~~~I > co >/
1000-~~~~r -n M

?I>980 -
.00

to AA5A5

w~~~~~~~~~~~~~~~~~~~~ ~LRD STRIT DAT

0 294 6 8 10 -2 1 6 1 0 2 4 2 8 3

DSANC nmx12

Figure~~~~~~~~~ ll PoAbit ditibto %-cnrlpesr fhurcns ufades
coastsA (10-7) Nubrdlnsdnt h ecn fsom ihp qAA
to or loe tha th vauAniaeAln h rdnt.Potdpit
94~aetkn rmfeuny nlssa 0nmi itrAlsfrte5hpretl
(sec. 3.2).~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

2 ~~~~N.C. 3

950 850~~~~~~~~~ 3500
~~~~~~~~~~~ 350 5
~~~~~3500-

/ ~~~~~~~~~~~~~~~~~S. C. (9
N"~~~~~~ N

>~~~~~ms LAGA

TEX~~~ ~~~ A 940L (9593
TEX ~~~~~~935? - 80400
0~~~~~~

9.-r3 ;, 92~16300

(92 .0)'~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~(93250)0 2

4"~~~~~~~'Y ~ ~ ~ ~ ~ ~ 1 (910.05 --N /
2 ~~~~~~~~~~~~~~~~~(0.3) 89A209300
905~~~~~~~~~~~~~~~~~9
(6-25.2- 9250

!-~~~~~~~~~25

0 PRESSURE (mb)

Figure 12.--One-percentile chart of minimum central pressure of hurricanes and tropical
storms for the Gulf of Mexico and surrounding areas. Pressure values plotted
at the centroid of data reference points for each 50-n.mi. increment of-coast.

~iJ N.C. -'

J 8- 5,� ~50 18
950 850
,-,350 9;002 o o -35�( -
-'35"

j < 1 MISS I ALA GA 985
I / (984.9)

TEX " LA L9 9-0 0o

00 -!300 \ \ 900 i3 /(9 85.4)

975o ~ ~ ~ ~ ~ ~ ~ ~ ~~97 I 9.-)

0<3~~ PRESSURE91~~'I (mb(975.7)

(970.) ( .(969.8)

Rigure 13.--Fifty-percentile chart of minimum central pressure of hurricanes and tropical
storms for the Gulf of Mexico and surrounding areas. Pressure values plotted
at the centroid of data reference points for each 50-n.mi. increment of coast.
O~~~~~~~~~~~ ,9~iSUe- ~ 97

Fg75 1.8)lt-eretl chr fmnmmcnrlpes-eo urcnsadtoia
(97rm ,n ~ 970fMxioadsrrudn aes rssr aue lt

98 0 ~ ttecetodo at eeec /ont .or _a 50nm. iceetof o..'

3.4 Evaluation of Results

3.4.1 Minima

Reasons for the increase in central pressure from south to north include:
the inability of hurricanes to carry their warm, moist, tropical atmosphere
into temperate latitudes and the entrance of colder and drier air at low
levels, which destroys the upward slope of the isotherms from outside to in-
side the circulation and decreases the amount of energy available to the
storm. Finally, jet streams at high levels, which according to Riehl (1954)
are detrimental to tropical storms, are more common in temperate latitudes.
Riehl,'states that the "arrival of the equatorward margin of a westerly jet
stream at high levels will destroy a (tropical cyclone) circulation rapidly
since it favors upper convergence, entrance of cold air aloft, subsidence,
and drying."

a. South Florida Minimum--The lowest accepted sea level barometer reading
(892.3 mb) not including tornadoes, in the Western Hemisphere occurred at
Long Key, Fla., in the hurricane of September 2, 1935. This implies a South
Florida minimum for the United States.

b. South Texas Minimum--Hurricane Beulah (923 mb), the third most intense
storm (in terms of central pressure) included in this study, struck the Port
Isabel area of Texas in September 1967. Hurricane Carla (931 mb) and the
Galveston hurricane (936 mb), two other notably severe hurricanes struck the
Texas coast between Matagorda and Galveston Islands. There is no reason why
Carla and the Galveston storm would not have been at least as strong if they
had struck the South Texas coast. If we look at storms outside the bounds of
this report, Hurricane Janet (1955) also lends strong support for the South
Texas minimum. Janet brought a minimum central pressure of 914 mb to
Chetumal, Mexico (18�N) in September 1955 (Dunn et al. 1955).

c. and d. Carolinas and Southern New England Minima--The two lowest trop-
ical cyclone central pressures observed along the coast of Georgia, South
Carolina, North Carolina, and Virginia were in Hurricane Hazel (Oct. 1954)
and Helene (Sept. 1958). Hazel struck the coast near the border of the two
Carolinas. Helene aimed her win'ds at the same area but turned away to the
northeast a few hours before the center made landfall. The eastward projec-
tion of the coast gives it a high exposure to north-northeastward-moving
cyclones, some of which like Hazel and Helene are of great intensity. Over
southern New England, the same reasoning holds true.

e. Mississippi Delta Minimum--This minimum was induced principally-by
Hurricane Camille (1969), and its effect is most prominent in the lowermost
percentiles. Even though Camille passed east of Louisiana on her way to the
Mississippi coast, the minimum appears near the mouth of the Mississippi
River because of the lower latitude. The raw data near the Mississippi Delta
do not show a minimum. The data at the 1% level (now shown) do show a well-
defined minimum; the 5% analysis in figure 11 was lowered to provide
continuity with the 1% curve.

3.4.2 Maxima

a. Cross City, Fla. Maximum---The lowest central pressure recorded in a
hurricane entering the northern gulf coast of the Florida Peninsula was
952 mb in the storm of October 1921, which entered the coast near Tarpon
Springs. This is not nearly as low as hurricane central pressures observed
on the Mid-gulf coast (Mississippi, Alabama, and the Pensacola area) and on
the Florida Peninsula gulf coast to the south. Is an extreme low p0 here
less likely climatologically or is this simply sampling variation during the
period of record? Present indications are for a real variation and the 1% to
15% curves in figure 11 ref le~ct this. A possible explanation follows.

A good many storms have paralleled the west coast of Florida close to shore
from the Keys northward. Although 'the eye of the hurricane remains over
water, air entering the storm at the surface has to cross the Florida Penin-
sula from east to west. Miller (1963) has shown that sensible heat is lost
from a parcel as it travels westward across the Peninsula. His calculations
from Hurricane Donna (1960) show that the surface inflow over land is essen-
tially a moist adiabatic process, which leads to the hypothesis that since
the major portion of the eastern semicircle of an alongshore west Florida
hurricane is over land, a quantity of the strms surface latent and sensible
heat source is removed, the equivalent potential temperature of the surface
air is lowered, and the radial gradient of equivalent potential temperature
at the surface is weakened. Movement of the storm out of tropical waters
further weakens the gradient. The Labor Day hurricane of 1935 is a good
example of what can happen when an intense hurricane leaves the Florida Keys
and heads up the west coast of Florida. After crossing Long Key with a
central pressure of 892.3 mb (26.35 in), the hurricane brushed Cape Sable and
paralleled the west coast of Florida for about 30 hours before entering the
coast near Dead Mans Bay. By then, the storm had reduced to minimal hurri-
cane intensity. The air mass north of the hurricane and surface water tem-
peratures had remained essentially constant as the storm skirted no more than
50 n.mi. off the coast for those 30 hours.

Caution; the Cross City area is exposed to hurricanes moving in from the
southwest (e.g., appendix fig. A-12) although there has not been a severe one
in recent years. The land effect would not apply. For example, a hurricane
could develop over the Bay of Campeche, attain great strength over the
central gulf, and then aim its destructive winds directly at the area.
Figure 11 is intended to combine these possibilities.

b. Delaware Bay Maximum--Just as the shape of the coastline plays the
major role in establishing the Massachusetts minimum, the north-south
orientation of the east coast between Cape Hatteras and New York City mili-
tates against landfalls and establishes a central pressure distribution maxi-
mum near the mouth of Delaware Bay. The strongest tropical cyclone to move
inland on the New Jersey coast during this century was a minimal hurricane
(September 1903) with central pressure above 982 mb. Also, only one tropical
storm (less than hurricane intensity) has penetrated the Delmarva Peninsula
during the last 74 years - in October 1943. Storms heading north-
northeastward over Delmarva after having entered the coast at a point farther
south are more common, but these storms have usually filled to a considerable

degree by the time th~ey reach Delaware Bay.

The raw data of figure 11 have been deliberately undercut in thes Delaware
Bay area because our method of data collection is more sensitive to landfall-
ing storms than bypassing storms. Most of the hurricanes affecting this part
of the coast bypass offshore before striking or bypassing southern New
England but they could turn into the Delmarva-New Jersey coas t. These storms
have central pressures comparable with the landfalling storms of southern New
England. Therefore, the curves for Delaware Bay are tailored to reflect both
the raw De laware Bay data and the New England analysis farther up the coast.

c. Jacksonville Maximum--The central pressure probabilities achieve another
high point along the northeast coast of Florida. Again, the shape of the
coastline plays a niajor role. The direction of the coastline is about 1600
to.34Q0 (measured from north) in this region. When a storm recurves suffi-
ciently to miss the southeast coast, it usually misses the northeast coast.
Until 1964, the city of Jacksonville was unique in that it was the only large
ci'ty on the Atlantic Coast south of Connecticut that had never sustained
winds of hurricane force in modern times. Hurricane Dora spoiled this fortu-
itous record in September 1964, lashing the Jacksonville area with 82-mi/hr
winds and demonstrating that Jacksonville is not immune to a major hurricane.

Although the incidence of exiting tropical cyclones near Jacksonville is as
high as or higher than at any other place on either coast, all but a few of
these are tropical storms with central pressure higher than 982 mb and have
little impact on the lower central pressures.

d. Northern New England Coastal Maximum--The central pressure of hurricanes
rises steadily as we move from southeastern Massachusetts northward to Canada.
The "cold wall" of the Labrador Current is principally responsible for this
effect. During August, the month of warmest sea-surface temperatures, water
temperatures average between 650 and 70'F from Long Island to Cape Cod. Along
the coast of Maine during the same month, the temperature is in the upper
50's - cold enough to give any hurricane an extratropical character.

4. PROBABILITY DISTRIBUTION OF RADIUS OF MAXIMUM WINDS

4.1 Data

The data used to arrive at a distribution of radius of maximum winds
(radius, R, from the center of a hurricane to the radius at which the wind
speed is the greatest) for the gulf and east coasts of the United States are
listed in tables 1 and 2, which are described in sections 1.3 and 3.1. The
values of R are for certain locations and times. They would most likely be
different at other locations and times.

4.1.1 Source of Radius of Maximum Winds

The numerical values of R entered in the table are derived from several
sources. 1) When possible, wind speed records from land stations were used.
When wind speed records were not available, the data were 2) approximated

from eye radii gathered by aircraft or radar, or 3) drawn from aerial recon-
naissance wind reports. When the latter type of data were not available
(usually for pre-1950 storms), the R value was 4) computed from an estimate
of the pressure profile, or 5) checked against narrative or tabular data in
the Monthly Weather Review. Each R in the tables is followed by a superscript
letter or letters that r-efer to a legend at the end of the tables giving the
source of the R value.

4.1.1.1 R fro~m Wind Records. Observed winds were acquired by noting the tine
when a wind-reporting station (also identified in tables I and 2) experienced
a maximum wind speed prior to the wind slacking off in the hurricane eye.
From a knowledge of the location of the storm center at that time, one can
then make an estimate of the value of R.. Figure 14 is a graph of wind speed
at Miami and distance of the station from the hurricane?'s wind center vs.
time for the hurricane that struck the southeast Florida coast on September
15, 1945. The radius of maximum winds for the forward and rear portions of
the storm is seen to be about 24 n.mi. Hydrometeorological Report No. 32
(Myers 1954) contains a detailed discussion of this method of obtaining R
from wind records. Additional examples are given in figure 3 of that report.

4.1.1.2 R from Eye Radius. In his work, The Structure and Dynamics of the
Hurricane' s Inner Core Region, Shea (1972) states that in the mean the radius
of maximum winds occurs at radii 5 to 6 n.mi. outside the inner radar eye
radius (IRR) - assumed synonymous with the inner cloud wall. The IRR may be
obtained from land-based radar, ships at sea, or aircraft. Figure 15, taken
from Shea, shows the position of R relative to the IRR for 21 Atlantic Ocean
a-ad Gulf of Mexico hurricanes. Figure 16, also from Shea, shows the differ-
ence between R and IRR vs. the maximum wind speed for radial legs. Note that
the more intense the wind the better the agreement between R and the IRR.

4.1.1.3 R from Aerial Reconnaissance. Flight reports from reconnaissance
aircraft usually include a plain language report of the radius of maximum
winds at flight level (around 700 mb). Shea finds that "~only the weaker
storms exhibit a tendency for a slope of the radius of maximum winds with
height; more intense storms (fig. 17) do not. The vertical slope of the
radius of maximum winds is probably related to the intensity of cumulus con-
vection. Stronger eye wall convection in intense storms is more effective in
transporting horizontal momentum to upper levels. This causes the cumuli to
stand straighter and the maximum winds at upper levels to occur more directly
above those at lower levels." Thus, reliable flight reports can be used to
approximate the surface value of R in hurricanes of average or greater than
average intensity when more direct data sources do not exist.

4.1.1.4 R from Pressure Fit. Computed Rs are arrived at by fitting an expo-
nential pressure profile to a given hurricane. By their nature, computed
values of R are more subject to error than observed Rs. The procedure is:

a. Plot the observed pressure data for each particular time and quadrant
on a graph of pressure vs. distance from the Tiurricane center.

b. Draw a smooth curve to the plotted data, giving a first approximation
of the pressure profile.

100

R = 24 n mi

P'I
80 - R R 24 n mi

\/ '� I |
80 ~-~' , I I 0

J/ % %
0 0/0 /%,
60- \/ WIND SPEED (mi/hr)

0/ I //
z I

O , | , l l I 0 %
0 1 T
EE
I N.
w 40- ,

DISTANCE TO HURRICANE CENTER FROM MIAMI, FLA.

20 -

Figure 14.--Graph of wind speed and station-tohurricane-center
distance vs. time at Miami, Fla, September 15-16, 1945
The radius of maximum winds determined by the relation
of peaks in the wind curve to distance from Miami to
the storm center is 24 n.mi.

43
43

30 -

-:25-

520-

15-
43

0 5 1 0 IS 2 0 2 5 30 35 4 0 45 5 0
RADIUS OF MAXIMUM WIND

Figure 15.--Radius of maximum winds (RMW) vs. inner radar eye
radius (IRR). Points falling on the 450 line are
those where the RMW and IRR coincide. The curved
line indicates the best fit curve (from Shea 1972).

c, 30

26- W
>- 24-

_ 20- a
2 w0 -

12 -

4
m0 -NO DIFFERENCE ----0d----LS!.~~. *0 ___

o -2-- R

o -6 I I
4 40 50 60 70 80 90 100 110 120 130 140 150
MAXIMUM WIND SPEED ( knots

Figure 16.--Difference between the radius of maximum winds (RMW) and
the inner radar eye radius (IRR) vs. maximum wind speed.
The best fit curve is indicated by the heavy line
(from Shea 1972).

- (0o~/ -
It

, -
I

I -

3r- ar I /
,I Z i I
E4 i ~~~ ,
o W / w/

3- < /

I , co I
I I I

I j l I
I I
I I

E 4 -- I I
co
8I)?o II
I I i I

8-, I i

aUj
-~ a, 1 I v
"'I I I I-I /

0 10 20 30 40 50 0 0 20 30 40 50

(a) RADIUS (n.mi.)
Figure 17.--Variation of the radius of maximum winds with elevation

for (a) storms with simultaneous lower and upper
tropospheric reconnaissance data and (b) storms in
which two or more simultaneous reconnaissance missions,
were flown in the lower troposphere only (from Shea 1972).

45
I ~~~~~~~~~~~~~~~~
I I�

0 0 I0 3 0 5 1 0 3 0 5

8iue 7--Vaito Nf Ih raiso aiu id iheaio
I,~ _..I -- rswt iutneu oe n pe
LJJrpshri eonisac aaad b trsi
Z~~hc w rmr iutneu eonisnemsin
C~ r flw 0 ntelwrtoopeeol fo ha17)

Ld ~ ~ ~ 4

c. Fit a curve to each tentative profile of the family:

P -o -R/r
~~~~~~~~~~ ~~~~~~~~~(2)
Pn- Po 2

which, solved for R, reads

r Ir2 Pi - Po
12 =ln (5)
rI -r2 P2 - Po

Where p = pressure at distance r from hurricane center, p0 = central pressure,
Pn = asymptotic pressure, R = radius of maximum winds, and r = radius at
pressure p (r1 and r2 are known radii of observed pressures Pl and p2).

The value of P, is usually taken from an observation about 250 to 300 n.mi.
from the hurricane center and P2 from a point close to the storm center. The
radius of maximum winds for each quadrant is the R parameter. Reliability
depends on the quantity and consistency of data and whether there is a pres-
sure observation close to the center. At the radius of maximum winds in a
hurricane, the centrifugal force term -exceeds Coriolis force by an order of
magnitude. Thus, the wind if balanced is essentially cyclostrophic. Differ-
entiating (2) shows that the maximum cyclostrophic wind is at R (Myers 1954,
pp. 2, 3, 27). In using equation (2), Pn is taken from the outer edge of the
hurricane's sphere of influence.

d. Obtain the best overall value of R by averaging the Rs of the forward
and rear halves of the storm. Some subjectivity is involved here. Frontal
systems and polar highs distort the pressure profile around a hurricane.
Values of R should not be derived from the above equation beyond the frontal
boundary or in quadrants of the stormy that show a marked baroclinicity.

4.1.1.5 R from Monthly Weather Review. Radius of maximum winds reports
extracted from the Review usually consist of estimates of eye diameters from
the measured time interval between the slackening and resumption of hurricane-
force winds over some point near or along the east or gulf coast. In other
instances, researchers have reported their findings in the Review, and these
results (including estimates of the radius of maximum winds) have been
accepted by the authors.

4.1.1.6 R not Measured. In a few cases, R could not be obtained by any
reliable method. Storms with Rs in this category are represented in tables 1
and 2 by the abbreviation MSG (missing).

4.1.2 Limited Representativeness of Large Rs

Do the larger values of R represent a true hurricane or one that is in an
advanced stage of becoming extratropical? The answer to this question is an
integral part of a climatology because nontropical influences upon the value
of hurricane factors will limit the way in which the data may be used.

The numerical values appearing in tables 1 and 2 range from 6 to 66 n.mi.
Myers (1970) ignores all Rs larger than 45 mi (39.1 n.mi.) in his study Joint
Probability Method of Tide Frequency Analysis Applied to Atlantic City and
Long Beach Island, N.J. since the employed storm surge model is less reliable
with Rs much larger than this. Studies of radar films of hurricanes have
failed to provide visual confirmation of any hurricane having an R as large
as 50 n.mi. unless it was becoming extratropical. Keeping this in mind, the
authors were able to revise downward several of the large values of R listed
in table A of Report No. 33 (see sec. 3.1 for reasons for revising other R
values).

Six of the 124 hurricane positions in tables 1 and 2 have an R greater than
45 n.mi. One of these large Rs (September 21, 1938) is a computed value
unconfirmed by direct wind observations. This hurricane had also appeared to
develop extratropical characteristics (Myers and Jordan 1956). The R=66 n.mi.
in the August 26, 1924 hurricane (near Nantucket, Mass.) was observed at a
time-when the hurricane was becoming extratropical. The R=54 n.mi. in the
December 2, 1925 hurricane; the R=55 n.mi. in the September 30, 1929 storm;
the R=50 n.mi. in Hurricane Daisy off Nantucket; and the R=48 n.mi. in
Hurricane Donna off Long Island were also noted at a time when the hurricanes
were becoming extratropical. The highest winds in extratropical storms are
usually observed much farther from the storm center than the maximum winds
within hurricanes. Based on the above discussion, all Rs > 45 n.mi. are
eliminated from the frequency analysis to be described.

4.1.3 Determination of R in Hurricane of September 1928

The hurricane of September 16, 1928 is an exception to the rule in
par. 4.1.1 that observed values of R would be accepted in preference to
computed values. This hurricane passed directly over West Palm Beach, Fla.
The wind-time graph at Miami (no wind records are available for West Palm
Beach) appears to indicate an average R of 53 n.mi. which is only about
5 n.mi. short of the distance between the two South Florida cities.

However, in the Monthly Weather Review for that year, Charles L. Mitchell
(1928) relates, "The damage at Miami was negligible, being confined princi-
pally to a few plate-glass windows and to awnings. Hollywood and Fort
Lauderdale escaped with only slight structural damage to buildings, the most
serious losses being from water damage, resulting from broken windows and
leaking roofs. A few thousand dollars will cover the losses at both places.
From Pompano north to Jupiter, especially at Delray, Lake Worth, Palm Beach,
West Palm Beach, and Kelsey City there was serious structural and water
damage, the losses being greatest at Palm Beach and'West Palm Beach. There
has been no authentic statement as to the total losses, but they amount to
several million dollars."

Pompano is about 30 n.mi. down the coast from West Palm Beach, and it was
at Pompano where serious damage became apparent. The value of R is not 53
but 28 n.mi. when the exponential pressure profile equation is used. The
latter value agrees very well with the damage information in the Monthly
Weather Review. The lull at Miami apparently was not the eye but some other
phenomenon.
47

4.2 Analysis

Cumulative frequencies of radius of maximum wind for each storm counted (in
some cases a hurricane is counted more than once as described in chapter 3)
within 150 n.mi. seaward from the coast and 50 n.mi. inland from the coast
were determined from the Rs in tables 1 and 2 for the same overlapping zones
centered 50 n.mi. apart used for the p0 analysis. The frequencies were
analyzed according to the procedure illustrated in section 3.2. Greater
freedom was ta.ken in analyzing the cumulative frequency graphs of radius of
maximum winds for each 50-n.mi. increment and the eventual probability dis-
tribution along both coasts than for p0 because the best available R data
were deemed to be often too sparse and to be less reliable on the average.
Examples of the frequency graph analyses are given in figure 18. Three per-
centiles (16-2/3, 50, and 83-1/3) were chosen, instead of the 7 chosen for
minimum central pressure, to portray the end product of the analysis proce-
dure--the coastal trend in R (fig. 19) along the east and gulf coasts. Raw
data from initial best fit curves for the cumulative frequency of minimum
central pressure graphs are shown at the 16-2/3 percentile of figure 19.
Centroids of R data were not used for additional control.

We did not expand the radius of maximum winds distribution to enforce con-
sistency with the storm frequency count as we did with the central pressure
distribution; instead we used only those values presented in tables I and 2
(central pressure of all listed hurricanes less than 982 mb, period of record
1900-73). Tropical storms, especially the weaker ones, often have no well-
defined radius of maximum winds, and when they do it is often a hundred or
more miles from the apparent storm center. Assigning values of R to these
storms would be haphazard at best. Only hurricanes (those in tables I and 2
with R < 45 n.mi.) were considered in the frequency analysis of R along the
coast.

4.2.1 East Coast

Since it was shown in previous papers (Weather Bureau 1957; Graham and Nunn
1959) that R tends to increase with increasing latitude along the east coast,
the cumulative frequencies at the three intervals should show this relation
in figure 19. For the most part, the above concept holds true at all three
percentiles from the southern end of Florida to the mouth of the Chesapeake
Bay. An initial analysis showed a dip in the three percentile curves toward
lower R values between the northeast coast of Florida and the southeast coast
of South Carolina ranging from 2 to 3 n.mi. in the lowermost curve (16 2/3%)
to 4 n.mi. in the uppermost curve (83 1/3%). This was smoothed out in the
final analysis (fig. 19) because 1) the data in this area compared with that
over most of Florida and the rest of the Carolinas are relatively sparse, and
2) the dip could not be accounted for on physical grounds.

We encountered a particular problem when analyzing R on the southeast
Florida coast compared to the same latitude on the Texas coast. The raw data,
particularly at the 83-1/3% level (not shown in fig. 19), had significantly
larger values in Florida. This difference was reduced in South Florida such
that Rs there exceed their Texas counterparts by only about 20% at the upper-
most percentile. Also, the Florida gulf coast curve at the 16-2/3% level was

100 100

0, 80- e - 80 -
w wJ O
� 0
z z

060- - if' 0
LLI LL

o o
- o 60- -

z z

I40- 0 40-

H H

0DU OMAIUWID n (a) 0 (b)

o COASTAL REFERENCE COASTAL REFERENCE
MILEPOST 750 MILEPOST 1950
0 (NEAR BAY ST. LOUIS. MISS) (CENTRAL SOUTH CAROLINA COAST)
0 0
0 10 20 30 40 0 10 20 30 40

RADIUS OF MAXIMUM WINDS (n mi) RADIUS OF MAXIMUM WINDS (n mi)

Figure 18.--Graphs of radius of maximum winds vs. cumulative percent of occurrences (a) Gulf of
Mexico (milepost 750), near Bay St. Louis, Miss., where hurricane Camille went
ashore and (b) east coast (milepost 1950), along central South Carolina coast.
u6 u

I4!I # 414 I I14141 I4I. I4' 14'1 1 4' I 41 '1 * I I I4 I mI.itI1
o)r-0 -0 m -n - m 0 !r
<- M 0 C z - -I --
X > -< - 0
M M ) m ,-n m 0
>) I A-ir W:
co - > m-C 1 -
r- C -n 11 m Zn U
M -n r- r- > ~ ~ -n
rn~~~~~~~~~~~~~I r->

50~ ~ ~ ~ ~ ~ -n

E 40~~~~~~~~~~~~~~~~~~

0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~5 -

0 -

10-

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

DISTANCE In mi x 102)

Figure 19.--Probability distribution of radius of maximum winds of hurricanes, gulf and east
coasts (1900-73). Numbered lines denote the percent of storms with R equal to
or less than the value indicated along the ordinate. Plotted points (&) are
taken from frequency analyses at 50-n.mi. intervals for the 16-2/3 percentile
(sec. 4.2).

lowered slightly to bring it into agreement with the east coast analyses.

The raw data at the 16-2/3 percentile level of figure 19 were enveloped
from South Carolina to Virginia to bring the analysis more into line with the
upper two curves.

There are only seven known values of the parameter R for the east coast of
the United States from Virginia northward. The smoothing procedure used in
previous chapters was discarded for these latitudes, and a subjective method
was used in figure 19 to extend the curves to eastern Maine. All three
curves .have leveled off substantially upon reaching the coast of southern New
England-because here the coast is nearly east-west and the latitude effect of
R is greatly dimished.

4.2.2 Gulf of Mexico

When the three percentiles for each 50-n.mi. increment along the gulf coast
were plotted and analyzed, the resulting curves (fig. 19) depicted a trend of
larger Rs with ascending latitude, which is quite acceptable. The uppermost
percentile curve, however, was dropped farther downward along the west coast
of Florida than the data indicated. (See par. 4.3.1.)

Data proved to be too sparse to obtain cumulative frequencies of radius of
maximum winds for the central Texas coast southward. The three curves were
extended smoothly down the coast of Mexico to about 25'N, keeping in mind
that as we proceed southward along the coast the value of R should not
increase with descending latitude.

4.3 Reasonableness of Analysis

4.3.1 East Coast

The curves reflect the fact that the radius of maximum winds tends to in-
crease with latitude between the Florida Keys and Canada.

The pinching together of the 50% and 83-1/3% curves (fig. 19) occurs north
of Cape Hatteras because of the idiposed limit of R < 45 n.mi. to exclude
extratropical storms in the more advanced stages of transition.

The three percentile curves attain their greatest slope between coastal
Georgia and the Cape Hatteras area. It is in these latitudes that the hurri-
cane passes from a tropical to a temperate environment, and it is in this
region where one would expect hurricane radii of maximum winds to show -their
greatest increase for those reasons mentioned earlier. The slope of the lowEr
curve is less because of a few small radii storms in each sample.

4.3.2 Gulf Coast

4.3.2.1 Florida and Mexico Minima. Just as with the east coast, there is a
-variation of R with latitude and minima are reached on both the eastern 'and
western edges of the Gulf of Mexico portion of figure 19. This is to be
*expected. For example, with the exception of Hurricane Camille (1969), an R

less than 15 n.mi. has not been noted over the middle Gulf of Mexico, while
three hurricanes with Rs of less than 10 n.mi. have affected the western or
eastern rims of the gulf. The analysis shows moderately lower values on the
western rim of the gulf than at the same latitude on the eastern rim and
agrees with Report No. 33, which shows the same trend.

4.3.2.2 Mississippi Coastal Maximum. The northernmost extension of the gulf
coast is at Mobile Bay. From what has been said so far with regard to varia-
tion with latitude, it is reasonable to expect a maximum of R in this general
area.

4.4 The Radius of Maximum Winds of Hurricane Camille

Camille (1969), which was an intense hurricane with an R estimated at
8 n.mi. was an outlier that did not fit into the Mississippi coastal maximum
even though it swept ashore near Bay St. Louis, Miss. Two reconnaissance
missions on August 17, 1969, indicated that the eye of Camille had a radius
of 4 to 5 n.mi. and that the thickness of the wall cloud was 5 to 10 n.mi.
As mentioned previously, Shea (1972) states that, in the mean, R occurs 5 to
6 n.mi. outside the inner radar eye radius. For small eye radii of 7 n.mi.
or less, the largest R given by Shea (fig. 15), with the exception of one
outlier, is 15 n.mi. and the smallest is 5 n.mi. In our figure 16, he shows
that the more intense the wind the better the agreement between R and the
inner radar eye radius. For storms like Camille with observed maximum wind
speeds > 140 kt, the difference is usually < 2 n.mi. but could, in a rare
case, be imagined to be as large as 5 n.mi. The close agreement between R
and the radar eye radius is, perhaps, a result of the stronger eye subsidence
observed in the more intense storms.

The above interpretation of the reconnaissance data from the two missions
of August 17 suggests the value of R for Camille as within the range 4 to
15 n.mi. Shea's data in the mean would indicate an R of 10 n.mi. However,
for hurricanes with extreme winds, R would range from 4 to 10 n.mi.

Land-based radar coverage from New Orleans (fig. 20) indicates that the eye
of Camille had a radius of about 5 n.mi. and the wall cloud a width of about
4 to 5 n.mi. at the time the storm center reached the coast. These values are
in general agreement with the reports of the two reconnaissance missions.
The radius of maximum winds placed at the center of the wall cloud is about
7 n.mi. for figure 20. Based on the above information from Shea, aircraft
radar, and land-based radar, a value of 7 or 8 n.mi. should be about right
for Camille near its time of landfall.

Throughout most of its life over the Gulf of Mexico, Hurricane Camille
exhibited a double wall cloud on radar. At the time the storm made landfall,
the outer wall had a radius from the center of the storm of about 12 n.mi.
Figure 21 shows time graphs of the inner wall cloud radius, radius from the
center of the wall cloud (measured from eye center), and radius of the outer
edge of the wall cloud when Camille was near the coast on August 17-18, 1969.
As one can see, the three curves (particularly that of the inner wall cloud
radius and wall cloud center radius) show rapidly decreasing radii during the
last couple of hours before landfall. The wall cloud center radius lowers

00 LEGEND

X EYE CENTER
A INNER WALL CLOUD RADIUS
B WALL CLOUD CENTER RADIUS
C OUTER WALL CLOUD RADIUS
D ATTENUATION

270� 9

NEW ORLEANS D
FRAME NO. 3282 d a
0431 GMT
AUG. 18, 1969 1800

Figure 20.--Hand-drawn sketch of New Orleans (Moisant Field) radar image,
August 18, 1969, at 0431 GMT. Note locations of the eye center
(x), inner wall cloud radius (a), wall cloud center radius (b),
and outer wall cloud radius (c).

25 I I I I I I I I

- --- O OUTER WALL CLOUD
*-- A INNER WALL CLOUD
\ ........ 3 O WALL CLOUD AXIS
20 - O. O (RADIUS FROM
EYE CENTER)

* 0sb.c

E � o. ..

O. 1.
0r_ *] O . @00 0"\0

- 00000

0 134 3

14 16 18 20 22 24 02 04 08 08 1 12
............................... 3 3133133

._A A A A A-- ' IO

14 16 18 20 22 24 02 04 06 08 10 12
AUG. 17, 1969 TIME (GMT) AUG 18, 1969

Figure 21.--Composite graph of inner wall cloud radius, wall cloud center
radius, and outer wall cloud radius vs. time for Camille.
Storm made landfall about 0430 GMT.

from more than 10 n.mi. to 7 n.mi. in about 2-1/2 hours, or during Camille's
last 40 n.mi. at sea before landfall.

Based on what has been said in the last four paragraphs, we have chosen an
R of 8 n.mi. as our best estimate of the radius of maximum winds of Camille
during the last hour or two before landfall.

The National Weather Service dynamic model (Jelesnianski 1972) replicates a
hurricane wind field from an index R, then computes the coastal surge. With
Camille, the best replication of the surge is with an index R of 12 n.mi.
This discrepancy suggests that Camille did not fit the standardized wind pro-
file of the storm surge model. The reader is therefore alerted to the fact
that other radii of maximum winds listed in tables 1 and 2 may not be the
uniquely best values for replicating observed surges with a standardized wind
profile.

5. PROBABILITY DISTRIBUTION OF SPEED AND DIRECTION OF STORM MOTION

5.1 Speed of Storm Motion

Speed of forward motion of landfalling and alongshore hurricanes was
extracted from storm track charts. Forward speeds were carried forward from
table A of Report No. 33 for the storms listed there. Those speeds were
mostly derived from detailed track charts depicting hourly or bi-hourly posi-
tions in the vicinity of the coast (e.g., Myers 1954, Graham, and Hudson 1960).
The forward speeds pertain to the time of landfall or closest approach to the
coast. Cry's tracks and the charts published in the Monthly Weather Review
for recent years were used to update and extend table A. These charts give
12- or 24-hr positions and sometimes indicate slower forward speeds than
detailed hourly tracks because of the acceleration associated with recur-
vature. Speed data used in this analysis included all hurricanes since 1886
as defined by Cry (maximum surface winds 64 knots or greater),regardless of
central pressure.

Tropical storms are omitted from this analysis of storm speeds to eliminate
the transition of recurved storms from hurricane to tropical storm to extra-
tropical storm stages. The high speeds characteristic of the latter transi-
tion are not necessarily representative of the hurricane stage. It is not
always possible to identify these transition stages individually.

Forward speeds of the hurricanes with central pressure below 982 mb during
the period 1900-73 are listed in tables 1 and 2 for information but the speed
probabilities are derived from the larger sample just described.

5.2 Forward Speed Probability Distribution

For Landfalling Hurricanes

5.2.1 Analysis

From the data summarized in the preceding paragraph, cumulative frequencies
of speed of forward motion for landfalling hurricanes were determined for

locations spaced at 50-n-mi. intervals along the gulf and east coasts. The
cumulative frequency curve for each location was based on 30 data points,
i.e., speed data for the 15 hurricanes landfalling on each side of the loca-
tion. (This more generous subsample than for p0 and R is appropriate to the
larger total sample.) The curves were determined in this manner for locations
on the east coast as far north as Cape Charles, Va. A lack of data to the
north -necessitated analyzing the frequencies with a smaller data sample and
at larger intervals. Seven data points were used in the frequency curves for
locations near New York, N.Y., and Boston, Mass., and six for Eastport, Me.

The cumulative probability curves were constructed using eq. (4) for the
plotting position. Figure 22 shows examples of these curves for two coastal
locations, near Bay St. Louis, Miss., and on the central South Carolina coast.
Values at the 5, 20, 40, 60, 80, and 95th percentiles from each probability
curve were then smoothed along the coast by the weighted mean procedure
described in par. 2.2.1.1, except that north of Cape Hatteras on the east
coast subjective smoothing was used. The resultant smoothed profiles are
shown in figure 23. The 80%-level points from the curves before this final
smoothing step are plotted on the figure.

5.2.2 Results and Discussion

Figure 23 shows the probability distributions of forward speed for land-
falling hurricanes. The speed of hurricane forward motion generally increases
with their northward progression, especially after recurvature to a northerly
or northeasterly direction. The diagram reveals no significant variation in
the speed of forward motion of hurricanes along either coast of the Florida
peninsula, but north of Daytona Beach, Fla., on the east coast the forward
speed of hurricanes increases. The upper 50% of forward speeds increases
from 10-17 knots near Jacksonville, Fla., to 30-48 knots at the northern U.S.
boundary. A latitudinal variation is also found in the Gulf of Mexico. Hur-
ricanes striking the mid-gulf coast have higher speeds, especially the upper
10%.

5.3 Forward Speed Probability Distribution

For Alongshore Hurricanes

5.3.1 Analysis

The probability distributions of forward speed for alongshore storms were
based on data grouped into zones extending 250 n.mi. along the coast and from
the coast to 150 n.mi. at sea. A probability curve constructed for each
250-n.mi.-long zone was assumed to be representative of hurricanes crossing a
line drawn perpendicular to the coast at the center of the zone. Speeds at
the same selected probability levels (par. 5.2.1) were used to construct
smoothed profiles along the coasts. This was accomplished by subjective
analysis, using figure 23 as a background guide. (Analysis at 50-n.mi. inter-
vals of speeds of alongshore hurricanes based on 12- and 24-hour movements
would be redundant and would result in repeated counts of the same storms.)
The resultant probability distributions are shown in figure 24.

100.

LU
0
Z 80-
L 5

0
0
LL 60-
0
z
/wo

LU.
" 40-

�

, 20- (a) o
< COASTAL REFERENCE
_- IMILEPOST 750 -
(NEAR BAY ST. LOUIS. MISS.

0 5 10 15 20 25 30
FORWARD SPEED(kt)

100

LU0o
o
z 80-

W
LU
-I

I- o

0 - (b)
o-
O 7" COASTAL REFERENCE
MILEPOST 1950 _
( CENTRAL SOUTH CAROLINA COAST)

0 5 10 15 20 25
FORWARD SPEED(kt)

Figure 22.--Landfalling hurricanes, forward speed vs. cumulative percent
of occurrences, (a) Gulf of Mexico (milepost 750) near Bay
St. Louis, Miss., (b) east coast (milepost 1950) along
central South Carolina coast.

' I I 41 ' I ' 1 I ' ' I 4 '' I ' I ' I ' ' w
- C) r w - Cn -n r z m r
o > - m H - m >
<- - -CA :
D < i� , x i) > > Z D
- 50 - _ m
>-I 0 2 >
a>o > r- r- >H
111Y- 11 o m m
- H nm -n- r r- > cn m
m 40"H>- r- > >
m >

0, I -I .

50-

w 80%

40-

DISTANCE (n mi x 102)

Figure 23.--Probability distribution of forward speed for landfalling hurricanes, 1886-1973.
Numbered lines denote the percent of storms with forward speed equal to or less
than the value indicated along the ordinate. Plotted points (a) are taken from
frequency analyses at 50-n.mi. intervals for the 80th percentile (par. 5.2.1).

5.3.2 Results and Discussion

A comparison of the two distributions (figs. 23 and 24) reveals that the
forward speeds of alongshore hurricanes of the east coast are generally
higher than those of landfalling hurricanes, while alongshore hurricanes of
the gulf coast move more slowly than landfalling hurricanes. Alongshore
hurricanes off the east coast are northward-moving storms that maintain a
faster forward motion after recurvature. Alongshore hurricanes in the
central portion of the gulf are eastward- and westward-moving storms travel-
ing at a slower speed.

The comparison of the two classes of hurricanes further reveals that along-
shore hurricanes off the coasts of southern Florida moved at a slightly
slower speed than landfalling hurricanes.

5.4 Probability Distribution of Direction of Storm Motion

for Landfalling Tropical Storms and Hurricanes

5.4.1 Data and Analysis

Hurricane and tropical storm tracks in Cry's publication since 1871 and
charts published in the Monthly Weather Review for recent years were used in
summarizing the directions of storm motion. Directions of landfalling trop-
ical cyclones were measured at the time they crossed the coast. A plot of
these entry directions against the landfalling points is shown in figure 25.
Cumulative frequencies of the entry direction for each tropical cyclone
within 75 n.mi. on each side of a location were counted at 50-n.mi. increments
along the east and gulf coasts, except that where the coastline turns
abruptly, frequency counts over a shorter distance were used. Because of
insufficient data north of Cape Hatteras, analyses were made at larger dis-
tance increments. Using the plotting position formula given in eq. (4),
cumulative probability curves were plotted and analyzed for each of the
selected points. Examples of these diagrams are shown in figure 26. Values
at the 16-2/3, 50 and 83-1/3% levels of each probability curve were grouped
into three sections along the coast: the gulf coast, and the east coast north
and south of Cape Hatteras. The values were connected with continuous curves
by subjective analyses.

5.4.2 Results and Discussion

Figures 27, 28(a), and 28(b) show the smoothed profiles along sections of
the coast for the probability distributions of directions of landfalling hur-
ricane and tropical storm motion mentioned in the preceding paragraph. The
direction of landfalling storm motion curves parallel the coastal orientation
curve because the definition of "landfalling" restricts the storm direction
data selection, "exiting" and "alongshore" storm motions being excluded.
Under the influence of the easterly circulation of the lower latitudes the
general movement of storms in the tropics is westward. There is a tendency
for these low latitude storms to drift slowly northward. As the storms drift
toward higher latitude, they come under the influence of westerly winds and
-recurve northeastward.
59

I ' I 1 4I I I I I' I '' i ' I
Q r- o '" M
- m m >
En < m z -I -0 0
> , 23 -- > o z
. -0 -n z m ->
m Z : > r- .nE 3> c
-Mn >
r- O m > 'U
U) - -n - w ' m
-n i-- r:r
--I m S~ r -) ]> S>~t >-
m X ' �
coA> 2
X r - >- o
rn~~~~~ > m3
>~~~~~~~~~~

60-

95 %

50-
w
U)
LU j 80 %

CO
0
cc 60 %
< 40

O 40 %
u.

30 -
30 - / 20 %

20-

10-

0 I I I I I I I I I I I I I I
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

DISTANCE (n mi x 102)

Figure 24.--Probability distribution of forward speed for alongshore hurricanes, 1886-1973.
Numbered lines denote the percent of storms with forward speed equal to or less
than the value indicated along the ordinate. Plotted points (A) are taken from
frequency analyses for the 80th percentile (par. 5.3.1).

i- v I l I' I W I ' I ' I I' I 0 m
-I~~~~ { I I ' II I I ~I toI' I~
o >r> F m : > H, > m
rx 7Z~
-- m ~ z -< > -D -< rr

H . . O ; I .-I C rI
~< ~ ~m E n ~ ~ ~ ~ ~ H
m z > > 3>-D
,- r " -- O "< 0 m"o
m zz 0
-- m - n m r - Ca

x~~~~~~~~~~~~~~~~~~~~~~~~~~~" r r
m~~~~~~~~~~~~
>~~~~~~~~~~~~~~~~~~~~~~~r : ;> (n--

co-r -

280- -280
-n 03 0 - -I >q M -

260 -
260-::> . ~~~~~~~~~~~~~~ ~- 260

240- . ....n ".n - 240
SI..

-I 22:0 - . 2
o20- . . .. .~'
cr ..4. - 220
0
z200 � � . . - . 200
O ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~

0 180 -.
L. 10. .-180
6 - .*. ..
W'UJ 160- . *-160
CD~~~~~~~~~~~~~~~~~ *

140. .. . . -140

z 120 .0 ' : * . 2
o

20 - .*
"0- 200

200- 2 O'-0
40~ ~ ~~~~~~~~:C ' � . )�'-. -40
20~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - -200
0 ' '
0'

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

DISTANCE In mi x 102)

Figure 25.--Scatter diagram of direction of storm motion measured at the point of landfall,
1871-1973. Circled dots denote two events.

100 , I I I ' I I 100 I ' I ' I ' I ' i

W 80- - 80- -
o 0 0
z Z
w w

0
U

O 60- - 60- .

00
w- /0 w

0 o /40 o
o 40- - 40-
w

U 0t~~~~~~~~~~~~~0

o 0 /
20- / - <20- o
- �(a) 0/ (b)
0_ � COASTAL REFERENCE - COASTAL REFERENCE _
MILEPOST 750 MILEPOST 1950
�/ (NEAR BAY ST. LOUIS. MISS.) / (CENTRAL SOUTH CAROLINA COAST)
0 . I . I , I , I . I 0 I . I . I I I I I
100 120 140 160 180 200 220 240 100 120 140 160 180 200 220 240

DIRECTION (DEGREES FROM NORTH) DIRECTION (DEGREES FROM NORTH)

Figure 26.--Landfalling hurricanes and tropical storms, direction of motion vs.
cumulative percent of occurrences, (a) Gulf of Mexico (milepost 750)
near Bay St. Louis, Miss., (b) east coast (milepost 1950) along
central South Carolina coast.

~~~~~~~~Ei- U. U- LL LL LL

<l < Iw )

I-~~~~ ~~~~ -J w U-<

360-C <

320 -

0 280-

OC)n 240 -313%
Uw

160~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~5 -
~~~~~~~~~~~~~~~~~83 1/3 %

120 - 0

0 1 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15
DISTANCE (n mi x 102)I

Figure 27.--Probability distribution of direction of landfalling storm motion for gulf coast. -
Numbered lines denote the percent of storms approaching the coast at an angle
equal to or less than the value indicated along the ordinate. Plotted points()
are taken from' frequency analyses at 50-n.mi. intervals for the 50th percentile
(sec. 5.4).

LL~~~~~~~~~~~~~ U-

260-

240- H>-
<C

220-
z
0

U~~~~~. .....

LU W 8
wo

0A ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~16203

14 15 16 17 18 19 20 21 22

DISTANCE (n mi x 102)

Figure 28a.--Probability distribution of direction of landf ailing storm motion for east coast,
south of Cape Hatteras, N.C. Notations are the same as figure 27.

I ~ ~~~ ~ ~~~~~~~~~~~~~~~ I I I

< ~~~~~~~0.

260 W w-

Z 200~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~*~~~~~~~~~~~~~......
0~~~~~~~
24 .....T.T
0I-~~~~~~~~~~~~~... ..
180"'~~~~~~~~~~~~.......*2...........
I-0~~~~~~~~~~~~~~~...... ........: . . ..
L�~~~~~~~~~~~~~~.... .........
Ow~~~~~~~~~~~. .....'' !.......
2W ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~~~3/ 16
22w . . . . . . 8 /
ow~~~~~~~~~~.. ...
UJO~~~~~~~~~.. .....
~~~~~~~~~~~~~. . . . ...............4 0. . .. . . .

60~~~~~~~~ ~ ~ ~ ~~~~~~~~~~~~~~~~~~~~.. . . . .
22 23 24 25 26 27 28 29 30 31..
.....~ ~ ~ ~ ~~ITAC .. .....02
Figue 28.--Pobablitydistibuton o dirctio of an.......sto..moion or.est.cast
north~~~~~~~~~~~~~5 ofCp atrs C oaiosaetesm sfgr 7

5.4.2.1 Gulf Coast. Figure 27 shows the alongshore profile at selected per-
centiles for the direction of motion for landfalling tropical cyclones of the
gulf coast. As expected, the tropical cyclones striking the west coast of
Florida come from the southwest direction and those striking the Texas coast
from the southeast. Along the mid-gulf, coastal areas are vulnerable to
storms approaching from both southeast and southwest.

5.4.2.2 East Coast, South of Cape Hatteras. Figure 28(a) confirms that for
landfalling storms near Miami, Fla., the predominant direction of storm
motion is from the east or southeast. North of Daytona Beach, Fla., the
higher percentage of landfalling storms coming from the south and southeast
reflects the increasing number of recurving storms. North of Brunswick, Ga.,
the percentage of northeastward-moving storms increases gradually northward.
This group of landfalling storms includes recurving tropical cyclones of
Atlantic origin and storms that exited the Florida coast and reentered the
coast south of Cape Hatteras. More than 50% of the landfalling tropical
cyclones near Cape Hatteras are northeastward-or north-northeastward-moving
storms.

5.4.2.3 East Coast, North of Cape Hatteras. Figure 28(b) shows the along-
shore profile at selected percentiles for the direction of landfalling trop-
ical cyclones north of Cape Hatteras. These smoothed profiles are based on a
rather small sample and the data are sparse. The stretch of coast south of
Cape Henry, Va., is vulnerable to landfalling tropical cyclones coming mainly
from the easterly directions, the coastal orientation excluding the north-
eastward moving storms from the landfalling category. Tropical cyclones
striking this part of the coast from the northeast are generally weak.
Figure 28(b) also reveals that tropical cyclones striking the coast east of
New York consist mostly of northward or northeastward moving storms.

5.4.3 Areas of Discontinuous Direction Profile

The directions of landfalling storm profiles along the east coast are not
continuous in the vicinity of Cape Hatteras, N.C. and Cape Cod, Mass., because
of abrupt turning of the coast. It is advisable to use the direction distri-
butions to the south of these corner points in applications at these capes,
since the maximum wind region of a hurricane lies to the right of the hurri-
cane track. The values indicated for Cape Sable (fig. 27) may be used as
representative for hurricanes striking the mainland coast of Florida Bay.
Where the coastal orientation departs greatly from a smoothed coastline
(fig. 3) along the gulf coast, the direction of storm motion for landfalling
tropical cyclones may be obtained from the track direction over a 2.5� octa-
gon as described in the following paragraph.

5.4.4 Track Direction Frequency by 2.5� Octagons--Gulf Coast

In the frequency analysis of landfalling storms for the gulf coast using
the track density method (see chap. 2), frequencies of track directions by
15� class intervals for each 2.50 octagon were tabulated and histograms con-
structed. A smoothed probability distribution was fitted to each histogram.
More detailed description of the procedure involved is discussed in the
appendix. The resultant histograms and eye-fitted probability distributions

for the gulf coast are shown in figures A3 to A16 of the appendix. From these
diagrams the proportion of occurrences for landfalling storm directions may
be determined for various coastal orientations.

6. THE JOINT PROBABILITY QUESTION

An objective of this report has been to define climatological probability
distributions of hurricane central pressure (po), radius of maximum winds (R),
forward speed (T), and direction of motion (O)along the Atlantic and gulf
coasts of the United States. In some applications of these data--for example,
in calculating frequency distributions of hurricane-induced surges on the
coast--it would be necessary to combine two or more of the individual proba-
bility distributions and form a joint probability distribution. In such
applications, the question of statistical independence of the probability
distributions has to be dealt with. Example: Of all the hurricanes affecting
a given coastal stretch over a long period of time, consider the 10% with
lowest p0 (intensity index) and the 10% with largest R (size index). What
fraction of the storms are both this intense and this large? The answers
are: If p0 and R are independent, 1% (.10 x .10 = .01); p0 and R positively
correlated, more than 1%; p0 and R negatively correlated, less than 1%.

Establishing the joint probability of two parameters with a given degree of
reliability requires a much larger sample of data than required for the same
degree of reliability for a single probability distribution. The hurricanes
listed in tables 1 and 2 are insufficient to give unequivocal results in
joint probability analysis. The data must be supplemented by a generous
measure of deduction and judgment. This chapter discusses the joint proba-
bility question in qualitative terms, but leaves it to the user of the report
to make final judgments that are suited to his particular problem.

The discussion is restricted to interrelation of the basic hurricane param-
eters with each other and with latitude. In storm tide frequency analysis,
the joint probability of the hurricane surge and the astronomical tide must
also be taken into account, but this. particular application is outside the
scope of the present report.

6.1 Central Pressure, po, vs. Radius of Maximum Wind, R

6.1.1 Overall Relation

A significant joint probability question is whether hurricane size (R) and
intensity (po) are independent. A storm that is both large and intense would
have enormous destructive power. Myers, in Hydrometeorological Report No. 32
(1954, chap. 7), applied a kinetic energy index to coastal hurricanes and
found a suggestion of an inverse relation between size and intensity, at
least at the extreme. Our analysis indicates the same conclusion. Figures
29 and 30 are plots of R vs. p0 from tables 1 and 2 for gulf coast and
Atlantic hurricanes, respectively.

Probability distribution curves (expressed in percent) have been sketched
on the two graphs. These depict the R distribution for a given po band; they

I I I I I IiII

LIMIT OF DATA
980 - 0 o 0 0

~~~~~Oo� o

970 - 0
O~~~
O ~O
~~~0 00
O~~~0

960 - 0 0
0
O 00O O

960 - 8

0 0
0 l
95 ~ ~ ~o-

wO 0
_930- 0

-J 0
In
~J3O O

'- 930 - /

z
920-
/~~~

910 -

900 -

890 -

880 -

870 I I I I I I
0 5 10 15 20 25 30 35 40 45 50
RADIUS OF MAXIMUM WINDS (n mi)

Figure 29.--R vs. Po, Gulf of Mexico hurricanes. Data from table 1.
Numbered lines denote the percentile of storm occurrences.

LIMIT OF DATA
980- 0 0 0

O
970- 0
0 � 00
880
0 0

960 - 0 0 0

90C

00
950 o o

940

930

890 -

880 --

870
0 5 10 15 20 25 30 35 40 45 50
RADIUSOF MAXIMUM WINDS (n mi)

Figurd 30.--Same as figure 29 except for east coast hurricanes: data
from table 2.

do not show the p0 distribution for a given R band -- a different set of
curves. The curves on the two graphs have been forced to agree at the low-
pressure end of the diagrams. The lowest p0 of 892 mb in both diagrams is
from the September 1935 Florida Keys hurricane.

6.1.1.1 Conclusions and Judgments from Figures 29 and 30.

a. Both hurricanes with central pressures below 920 mb have small Rs. It
is inferred that these are intense, well-formed vortices. It is a matter of
judgment as to what the largest R is that can be expected in a hurricane with
a very low central pressure.

b. Hurricanes with very large Rs (in excess of 45 n.mi.) are,as expected,
of moderate or weak intensity. In hurricanes moving northward in the
Atlantic and becoming extratropical, the radius of maximum winds becomes
larger and more diffuse and the central pressure rises. The one exception to
this pressure rise tendency (R = 50 n.mi., P0 = 940 mb) is the New England
hurricane of 1938. Not all the weaker storms have large Rs; some are small
(upper left-hand corner of figs. 29 and 30).

c. If the extremes are excluded--for example, on figure 30 if the pressure
range is restricted to 920 to 970 mb--there is no detectable relation between
R and po.

6.1.2 Local Relation

The relation shown in figure 30 is accounted for, at least in part, by the
trend toward larger, weaker hurricanes off the northern part of the Atlantic
seaboard as compared with the southern part. This trend and synoptic mete-
orological factors related to it were discussed in chapter 4. Locally, is
there any relation between p0 and R? Any such relation is obscured by the
latitudinal trend present in both p0 and R. This was removed for comparison
purposes for the hurricanes in table 2 by converting the p0 to a probability
level by entering figure 11 at the appropriate coastal point and by convert-
ing the Rs to local probability levels in the same way from figure 19. A
plot (not shown) of the resulting p probability level in individual storms
vs. the R probability level revealed a random scatter. This implies that, in
general, there is little local interrelation between p0 and R. It is left to
the user of this report to look at this more closely at latitudes where low
Po Is may be expected.

A similar analysis (in a different sequence) was made for gulf hurricanes
in NHRP No. 33 (Graham and Nunn 1959). It was found that when the R vs. p0
interrelation is removed, no latitudinal trend in R remains (fig. 18 of NHRP
No. 33).

6.2 Forward Speed, T, vs. Direction of Motion, a

Another joint probability question is whether there is an interrelation
between hurricane forward speed and direction of motion. On the east coast,
hurricanes that have recurved from their low latitude east-to-west motion to
a track toward the north-northeast generally move faster than the storms at

the same latitude still having a westward component of motion. With this in
mind, separate probability distribution analyses were made of forward speed
for alongshore hurricanes, all of which have recurved, and landfalling-
hurricanes, most of which are still drifting westward (figs. 23 and 24).

In the central Gulf of M-exico, the situation is reversed, as discussed in
chapter 5. Hurricanes that have recurved are typically gaining speed as they
move across the coast. Storms moving parallel 1~o the mid-gulf coast, either
eastward or westward, are more apt to be drifting slowly in an indifferent
steering current. This behavior difference, too, is taken care of, at least
in part, by the separate analysis of forward speeds for alongshore and land-
falling hurricanes. Along the north part of the west coast of the Florida
peninsula alongshore and landfalling hurricanes would be expected to have the
same forward speed characteristics as on the Atlantic east coast; the
respective forward speed diagrams (figs. 23 and 24) reflect this.

In summary, providing two forward speed probability graphs is a partial
recognition of T and 0 interdependence, and may be a sufficient recognition
for practical applications.

6.3 Other Joint Probability Questions

Is there any dependence of p0 and R on track direction or forward speed?
There may well be some subtle relation. For example, we suspect that storms
landfalling from the east-northeast on the Atlantic coast between 350 and
39'N (see fig. 28b) are mostly weak. It is left to the user of the report to
discover other relations from the data in tables I and 2 or deduce them from
synoptic experience.

7. SUMMARY

This report presents a probability analysis of the geographical distribu-
tion of major hurricane and tropical storm factors. The charts depicting
our solutions are shown in figures 6, 7, 8, 11, 19, 23, 24, 27, and 28.
Judicious smoothing was employed along the gulf and east coasts and across
the frequency spectra.

Table 5 shows the source of data and the classes of tropical cyclones rep-
resented. These are not the same for the several factors, for the reasons
stated in the report.

7.1 Highlights

7.1.1 Frequency of Hurricane and Tropical Storm Occurrences

The frequency of landfalling, exiting, and alongshore (within 150 n.mi. of
the coast) tropical storms and hurricanes (figs. 6, 7, and 8) were
summarized.

The fraction of storms that are hurricanes is presented as a continuous
smooth curve along both coasts. A striking result applies to the northwest

Table 5.--Data for probability analysis

Climatological Data Range of resulting Figures
characteristic source Adjustments probability distribution for results

Storm frequency Cry's tracks 1871-1963; Hurricanes plus Fig.6, 7, 8
(landfalling, Monthly Weather Review tropical storms
alongshore, since 1964, (hurricanes
exiting) plus tropical storms)

Central Tables 1 and 2 Statistically adjusted Hurricanes plus Fig. 11
pressure (hurricanes with to include tropical tropical storms
p9 < 982 mb storms
since 1900)

Radius of Tables 1 and 2 Rs > 45 n.mi. Hurricanes Fig. 19
maximum (hurricanes with excluded
winds Po < 982 mb
since 1900)

Forward Cry's tracks 1886-1963; Hurricanes Fig. 23, 24
speed Monthly Weather Review
since 1964, (hurricanes
only)

Direction Cry's tracks 1871-1963; Hurricanes or hurricanes Fig. 27,
of Monthly Weather Review plus tropical storms 28a, 28b
motion since 1964, (hurricanes (assumed the same)
plus tropical storms)

Florida coast. Here, frequency of landfalling storms has its highest value,
but only 42% of these storms have attained hurricane intensity. Elsewhere,
areas most vulnerable to tropical cyclones experience a high percentage of
storms of hurricane intensity.

7.1.2 Probability Distributions of Central Pressure, Radius of Maximum Winds,
Forward Speed and Direction of Storm Motion

Analysis of the data led to a set of graphs depicting the probability dis-
tribution of central pressure, radius of maximum winds, forward speed, and
direction of storm motion. The central pressure distribution (fig. 11) is
for hurricanes and tropical storms and is broken down for illustrative pur-
poses into 7 probability levels (percentiles) ranging from 1 to 90%. The
radius of maximum winds distribution (fig. 19) is for hurricanes only and is
separated into three probability levels--16-2/3, 50, and 83-1/3%. The forward
speed distribution consists of two charts--one for landfalling hurricanes
(fig. 23) and one for alongshore hurricanes (fig. 24). Both charts consist
of 6 selected probability level curves, ranging from 5 to 95%. The direction
of storm motion distribution for landfalling hurricanes and tropical storms
(figs.27, 28a and 28b) is illustrated by three probability levels--16-2/3,
50, and 83-1/3%.

A detailed analysis was made of the radius of maximum winds of Hurricane
Camille, the most intense hurricane of record on the middle gulf coast.

7.1.3 Joint Probability

Establishing the joint probability of two factors, such as central pressure
(po) and radius of maximum winds (R), with a given degree of reliability re-
quires a much larger sample of data than that present in tables 1 and 2.
Knowing this to be true, we can say that the interrelation between po and R
indicates that for the gulf and east coasts, 1) hurricanes with central
pressures below 920 mb have small Rs; 2) hurricanes with large Rs are nearly
always of moderate or weak intensity; 3) not all the weaker storms have large
Rs; 4) if the pressure range is restricted to the range 920 to 970 mb there
is no detectable interrelation between R and po; 5) if the latitudinal trend
is removed from po and R little local interrelation between p0 and R remains.

There is an interrelation between hurricane forward speed and direction of
motion along both coasts. Hurricanes that have recurved toward the north-
northeast generally move faster than the storms at the same latitude still
moving westward.

Other joint probability questions involving the various hurricane factors
(i.e., a dependenc& of p0 and R on track direction or forward speed) also
exist, although these relations are often not as obvious.

REFERENCES

Alaka, Mikhail A., 1968: "ZClimatology of Atlantic Tropical Storms and
Hurricanes," ESSATechnicalReport WB-6, Environmental Science Services
Administration, U.S. Department of Commerce, Washington, D.C., 18 pp.

Bradbury, Dorothy L., 1971: "The Filling Over Land of Hurricane Camille,"
Satellite and MesometeorOlogy Research Project Paper No. 96, NOAA Grant
E-198-68(B) Mod No. 2, Department of the Geophysical Sciences, The University
of Chicago, Chicago, Ill., 25 pp.

Cline, Issac M., 1926: Tropical Cyclones, McMillan Company, New York, N.Y.,
301 pp.

Colon, Jose A., 1953: "A Study of Hurricane Tracks for Forecasting Purposes,'
Monthly Weather Review, Vol. 81, No. 3, March, pp. 53-66.

Craddock, J. M., 1969: Statistics in the Computer Age, American Elsevier
Pub. Co., New York, N.Y., 214 pp.

Crutcher, Harold L., 1971: "Atlantic Tropical Cyclone Statistics," National
Oceanic and Atmospheric Administration, U.S. Department of Commerce,
Washington, D.C., 83 pp.

Cry, George W., 1965: "Tropical Cyclones of the North Atlantic Ocean. Tracks
and Frequencies of Hurricanes and Tropical Storms, 1871-1963," Technical
Paper No. 55, Weather Bureau, U.S. Department of Commerce, Washington, D.C.,
148 pp.

DeAngelis, Richard M. and Nelson, Elmer R., 1969: "Hurricane Camille,
August 5-22," Climatological Data National Summary, Vol. 20, No. 8, August,
Environmental Science Services Administration, U.S. Department of Commerce,
pp. 451-474.

Dunn, Gordon E., Davis, W.R., and Moore, P.L., 1955: "Hurricanes of 1955,"
Monthly Weather Review, Vol. 83, No. 12, December, pp. 315-326.

Goodyear, Hugo V., 1968: "Frequency and Areal Distributions of Tropical
Storm Rainfall in the United States Coastal Region on the Gulf of Mexico,"
ESSA Technical Report WB-7, Environmental Science Services Administration,
U.S. Department of Commerce, Silver Spring, Md., 33 pp.

Graham, Howard E. and Hudson, Georgina N., 1960: "Surface Winds Near the
Center of Hurricanes (and other Cyclones)," National Hurricane Research
Project Report No. 39, Weather Bureau, U.S. Department of Commerce,
Washington, D.C., 200 pp.

Graham, Howard E. and Nunn, Dwight E., 1959: "Meteorological Considerations
Pertinent to Standard Project Hurricane, Atlantic and Gulf Coasts of the
United States," National Hurricane Research Project Report No. 33, Weather
Bureau, U.S. Department of Commerce, Washington, D.C., 76 pp.

Harris,D. Lee, 1959: "An Interim Hurricane Storm Surge Guide," National
Hurricane Research Project Report No. 32, Weather Bureau, U.S. Department of
Commerce, Wasbington, D.C., 24 pp.

Jelesnianski, Chester P., 1972: "SPLASH (Special Program to List Amplitudes
of Surges from Hurricanes) I. Landfall Storms," NOAA Technical Memorandum
NWS TDL-46, National Oceanic and Atmospheric Administration, U.S. Department
of Commerce, Silver Spring, Md., 52 pp.

Miller, Banner I., 1963: "On the Filling of Tropical Cyclones Over Land,
with Particular Reference to Hurricane Donna of 1960," National Hurricane
Research Project Report No. 66, Weather Bureau, U.S. Department of Commerce,
Washington, D.C., 82 pp.

Mitchell, Charles L., 1928: "The West Indian Hurricane of September 10-20,
1928," Monthly Weather Review, Vol. 56, No. 9, September, pp. 347-350.

Myers, Vance A., 1954: "Characteristics of United States Hurricanes Pertinent
to Levee Design for Lake Okeechobee, Florida," Hydrometeorological Report
No. 32, Weather Bureau, U.S. Department of Commerce, and Corps of Engineers,
U.S. Department of the Army, Washington, D.C., 106 pp.

Myers, Vance A., 1970: "Joint Probability Method of Tide Frequency Analysis
Applied to Atlantic City and Long Beach Island, N.J.," ESSA Technical
Memorandum WBTM Hydro-11, Environmental Science Services Administration,
U.S. Department of Commerce, Silver Spring, Md., 109 pp.

Myers, Vance A., and Jordan, Elizabeth S., 1956: "Winds and Pressures Over
the Sea in the Hurricane of September 1938," Monthly Weather Review, Vol. 84,
No. 7, July, pp. 261-270.

National Oceanic and Atmospheric Administration, 1973: "Flood Insurance
Study, Puerto Rico," study for Federal Insurance Administration, Department
of Housing and Urban Development, U.S. Department of Commerce, Washiaigton,
D.C., 42 pp.

Riehl, Herbert, 1954: Tropical Meteorology, McGraw-Hill Book Company,
New York, N.Y., 392 pp.

Schloemer, Robert W., 1954: "Analysis and Synthesis of Hurricane Wind
Patterns Over Lake Okeechobee, Florida," Hydrometeorological Report No. 31,
Weather Bureau, U.S. Department of Commerce, and Corps of Engineers,
U.S. Department of the Army, Washington, D.C., 49 pp.

Shea, Dennis J., 1972: "'The Structure and Dynamics of the Hurricane's Inner
Core Region," Atmospheric Science Paper No. 182, NOAA Grant N22-65-72(G),
NSF Grant GA 19937, Department of Atmospheric Science, Colorado State
University, Fort Collins, Colo., 134 pp.

Sugg, Arnold L., Pardue, Leonard G., and Carrodus, Robert L., 1971:
"Memorable Hurricanes of the United States Since 1873," NOAA'TeChnical
Memorandum NWS SR-56, National Oceanic and Atmospheric Administration,
U.S. Department of Commerce, Fort Worth, Tex., 52 pp.

Thom, Herbert C. S., 1968: "New Distributions of Extreme Winds in the United
States," Journalof tb-e Structural Division, Proceedings of the American
Society of Civil Engineers, Vol. 94, No. ST7, July, pp. 1787-1801.

Weather Bureau, 1968: "Interim Report--Meteorological Characteristics of the
Probable Maximum Hurricane, Atlantic and Gulf Coasts of United States,"
unpublished memorandum HUR 7-97, U.S. Department of Commerce, Silver Spring,
Md.. 45 pp.

Weather Bureau, 1957: "Survey' of Meteorological Factors Pertinent to
Reduction of Loss of Life and Property in Hurricane Situations,"
Kutschenreuter,.Paul H., Compiler, National Hurricane Research Project Report
No. 5, U.S. Department of Commerce, Washington, D.C., 87 pp.

ACKNOWLEDGMENTS

The authors are grateful for the help of numerous members of the Water
Management Information Division, Office of Hydrology, National Weather
Service during the course of this study. The guidance and critical review of
the manuscript given by Vance A. Myers, Chief of the Special Studies Branch;
John T. Riedel, Chief of the Hydrometeorological Branch; and John F. Miller,
Chief of the Water Management Information Division was especially helpful.
Roger Watkins is credited with providing valuable information concerning
previous hurricane studies carried out by the Hydrometeorological Branch.
Normalee Foat, Miriam McCarty, Marion Choate, Owen Gourley, Samuel Otlin,
Keith Bell, Raymond Evans, and Samuel Srbljan provided valuable research
support. Typing through various stages of the report was done by Clara Brown,
Laura Nelson, Virginia Hostler, Cora Ludwig, Kathryn Carey, and the late
Edna Grooms.

Dr. Gerald F. Cotton, National Weather Service, advisedus on some
statistical procedures. Dr.-Harry F. Hawkins, Assistant Director of the
National Hurricane Research Laboratory was kind enough to provide us with
radius of maximum winds data for some recent hurricanes.

APPENDIX

TRACK DENSITY METHOD FOR STORM FREQUENCY

Assessing the frequency with which tropical cyclones landfall on the coast
of the Gulf of Mexico is more complicated than for the east coast because of
the small angle between prevailing track. directions and the coast, on the one
hand, and varying coastal directions (e.g., the Mississippi Delta) on the
other. In order to handle this in a straightforward manner we resort to the
track density method in which storm track behavior is assessed independently
of coastal direction and of whether the track is over water or over land.
The method described here is applied in chapter 2 of the report.

Track Frequency

Definitions

a. Storm track density. The frequency of occurrence of tropical cyclones
may be expressed as storm track density at a point. This is defined as the
number of storm tracks which cross that point, from any direction, per unit
length normal to track per unit time. In concept~, one obtains this number by
a limit process which may be expressed as:

Storm track density at a point = lim( t)
Dt (A-i)
as D4O, t-)-o

where N is the count of tracks passing through a circle of diameter D in time
t. Practically, it is necessary to count storm tracks passing through a
large enough circle over a long enough period of time to smooth out random
fluctuations.

Crutcher has extensively analyzed oceanic hurricane track frequencies by 50
squares (e.g., Crutcher 1971) and has developed a concise method of bivariate
presentation of direction and speed. We made a new track count in our
limited area to obtain information by 2.50 areas. We used a more empirical
method of analysis rather than Crutcher's method because a bimodal direction
distribution is especially prominent in some parts of the gulf area (e.g.,
fig. A-3) and would have been smoothed out. Our method retains more detail
of the direction distribution, which we needed, but yields no information on
the association of direction and speed, which his method would give. We had
to judge this separately and subjectively (chap. 6).

b. Storm track count. In this study the track density of tropical cyclones
(hurricanes plus tropical storms) was approximated by counting tracks from
Cry's maps (1965) for the period 1871-1963 (and from the annual Monthly
Weather Review articles starting in 1964) that passed through 2.50 longitude-
latitude squares with the corners cut off to form octagons. This was done for
each 2'5� square in the Gulf of Mexico and adjacent land areas. Such an
octagon is shown schematically in figure A-l. Octagons are used to approximate
circles, the theoretically correct form for the later track direction analysis
(par. 8.3.4). The resulting count for the Gulf of Mexico and adjacent areas

is shown in figure A-2 together with an isopleth analysis made by eye. The
average "diameter" of the octagon is 142 n.mi. The analysis of figure A-2 is
construed as depicting the point track density per 103 years per 142 n.mi.
T?0,onvert to track density per 10 n.mi. per 100 years, multiply by
('~--x __l0l) or 0.068.
103 142
Direction Probability

The second part of the track density analysis for each 2.50 octagon is the
track direction. The prevailing direction of each track within each octagon
was tabulated by 150 class intervals and then histograms like figure A-3 were
constructed for each octagon. The ordinate is the normalized frequency of
occurrences (f/Ni), where f is the count in a direction class interval, N is
the total count for all directions for the octagon, and (i) the class interval
(150).

Smoothing of Track Direction Histograms

There are several procedures for smoothing the direction histograms into
continuous probability distributions. In another study (NOAA, Flood Insurance
Study, Puerto Rico, 1973) in which the distributions exhibited a single mode,
"beta" distributions were fitted using a library computer program. With the
distributions like figure A-3 the distributions were sketched and successively
revised by eye until the area equaled 1.0 under each curve, and smooth tran-
sitions between adjacent octagons, were obtained. For some octagons a normal
(Gaussian) curve was fitted as an aid to the' eye in obtaining the proper area
under the curve. The resulting histograms and eye-fitted probability distri-
butions for gulf coast octagons are shown in figures A-3 to A-16.

Land falling Frequency

The track frequency (fig. A-2) and the track direction distribution (figs.
A-3 to A-16) are presented as continuous spatial variables, the latter by
interpolation between the respective octagon diagrams. The storm landfalling
frequency at a coastal point, Fn, is given by the integral

180 f
F f= (-- ) sin a de (A-2)
n 0 Nic a

where F = track density (all directions) from figure A-2,

f/Ni = normalized direction probability (in degrees -1), and
= angle between track direction and coast direction (degrees).
This is evaluated numerically from
12 f
Fn- F ( k sin -k i (A-3)
n k=l N
Where the k's denote 15� class intervals and Y is the mean angle for the class
interval between track and coast.

Exiting and alongshore track frequencies may be counted by the same proce-

dure by the appropriate choice of a. This application was not made in the
present study.

Necessity for Octagons

Counting tracks in complete latitude-longitude squares and then applying
the type of direction analysis used here biases the direction probability
toward higher values normal to the main diagonals of the square. A test
showed that this bias is significant for the purposes of this report and
needs to be avoided.

30� N

27.50 N

87.50 W 85 W,

Figure A-l.--Octagon for counting hurricane tracks.

~~~~~~~~~~~~~20~

~~~~~~~~~~~~~~~~~0

~~~~~~~~4 46

2030 40 52 38 65~h3

Figure A-2.---Hurricane and tropical storm track -'ount for 2.50 octagons in the Gulf of Mec.
1871-1973.

3.4 - I.
HURRICANE DIRECTIONS 1871-1973

3.2- N.48 LAT. 25 27.50N
LONG. 80.82.50W I N
3.0-

2.8- 25

2.6- 2

2.4-

2.2-

, 2.0-

1.8-

1. -

1.4 -

1.0 -

o.- 1 - _

:.6- I,. I

300 270 240 210 180 150 120 090 060 030 0
DIRECTION {DEGREESI

3.4 -
HURRICANE DIRECTIONS 1871.1973 --' | i -

3.2- N= 55 LAT. 25 - 27.5N' .-...
LONG. 82.5 850W ..I
3.0-

2.8

2.62
i00 550 250 85�
2.4-

2.2 -

x 1.8-
z

1.4 - _

1.2 -

1.0 -

/
e.8-

300 270 21 150 120 090 60 030
300 278 240 210 180 150 120 090 060 030 0
DIRECTION IDEGREESI

Figure A-3(upper), Figure A-4(lower).--Hurricane and tropical storm

track direction histograms and probability distributions for gulf

coast octagons. 1871-1973.

34 HURRICANE DIRECTIONS 1871.1973.

3.2 - N 54 LAT. 27.5 - 300 N .
LONG. 82.5-850�W
3.0-

2.8 -

2.6-

2.4-

2.2-

2.0-
g
1.8

1.6 -

1.4-

1.2 -

0.8-

0.-

0.4 --

0.2

.11 . I i . I II fI1 ,]iL
300 270 240 210 180 150 120 090 060 030 0
DIRECTION (DEGREESI

.4 - HURRICANE DIRECTIONS 1871-1973 .

3.2- N 43 LAT. 275 - 30 N
LONG. 85 87.5� N
3.0-

2.820-100

2.4-

2.2 -

2.0-

1.8

1. -

1.4

1.2 --

1.0 -

0.6 -

0.4-

300 270 240 210 180 150 120 090 060 030 0
DIRECTION IDEGREESI

Figure A-5(upper), Figure A-6(lower).--Hurricane and tropical storm

track direction histograms and probability distributions for gulf

coast octagons. 1871-1973.

HURRICANE DIRECTIONS 1871-1973 ., . ... -

3.2- N= 53 LAT. 27.5 - 300N .- .. ..-
LONG. 87.5 - 900W ,

2.6-

2.4 -

2.2 -

2.0-
3 -
1.8-

1.6-

1.4-

1.2 -

1.0
1.0 -

0*.4-

300 270 240 210 . 180 150 120 090 060 030 0
DIRECTION (DEGREESI

HURRICANE DIRECTIONS 1871-1973 J '' .. ' .

.2- \
3.2- N 50 LAT. 27.5-30�N -. .._ ' ,
LONG. 90 - 92.50 W

2.0-
2.8 -

2.4-

2.2 -

-2.0

1.8-

1.6

1.4-

1.2 -

1.0

300 270 240 210 180 150 120 090 060 030 0
DIRECTION (DEGREES)

Figure A-7(upper), Figure A-8(lower).--Hurricane and tropical storm

track direction histograms and probability distributions for gulf

coast octagons. 1871-1973.

3.4
HURRICANE DIRECTIONS 1871.1973 .-- ::... J :
3.2- N. 36 LAT. 27.5-30�N . -.
LONG. 92.5 -950W V
3.0-

2.8 -

2.6 -
100� 95� 90
2.4

2.2

2.0-
.
1.8 -

1.6 -

1.4

1.2 -

1.0 -

0.6 '

0.2

300 270 240 210 180 150 12 00 090 060 030
DIRECTION (DEGREES)

HURRICANE DIRECTIONS 1871-1973

3.2- N 30 LAT. 27.5 .30 N
LONG. 95.9750W , -
3.0-

2.8 -

2.8 -25

2.4-

:.2-

2.0- -

1.8-

z 1.6 -

1.4-

1.2-

o- _,

0.28-

0 4 O I ;I II
300 270 240 210 180 150 120 090 060 030 0
DIRECTION IDEGREES)

Figure A-9(upper), Figure A-lO(lower).--Hurricane and tropical storm

track direction histograms and probability distributions for gulf

coast octagons. 1871-1973.

HURRICANE DIRECTIONS 1871-1973 ...

3.2- N= 3 LAT. 25-27.5�N _ j .-;
LONG. 95-97.5�W , i L '--

2.2 -

2.0

1.8-

1.6 -

1.4 -

1.2-

1.0 -

0.8- /
0.6-

I I ,, I , I , I
300 270 240 210 180 150 120 090 060 030 0
DIRECTION IDEGREESI

I I I I I I ' I ' I [ ' .

HURRICANE DIRECTIONS 1871-1973 .....

2.6-
2.83.0- =4 LAT. 30.2.5�N j "i :.'1 ' 5' 0 -'
LONG 82,5- 85� W i-

2.4 -

2.2-

2.0 -

1.8

1.6-

1.4 -

1.2 -

1.0 -

0.6 -

300 270 240 210 180 150 120 090 060 030 0
DIRECTION (DEGREES)

Figure A-ll(upper), Figure A-12(lower).--Hurricane and tropical storm

track direction histograms and probability distributions for gulf
coast octagons. 1871-1973.

3.4- ---
HURRICANE DIRECTIONS 1871-1973 - ...., ..

3.2 - N 29 LAT. 30 - 32.6 0N
LONG. 85 - 87.50W .
3.0-

2.8-

2.6
00�' 50 go900 850
2.4-

2.2-

m 2.0

1.6-

1.4 - T

1.2 -

1.0 -

0.68- 8 t .

�o , . I I
300 270 240 210 180 150 120 090 060 030 0
DIRECTION (DEGREES)

I ' I ' I ' I ' I ' I ' I ' I I ' .

HURRICANE DIRECTIONS 1871-1973 . r

3.2- N- 39 LAT. 30.32.'0N.. r
LONG. 87.5 - 90W -...
.0 -

2.8-

2.2- -
1000 550 950 55
2.4 -

2.2

'o 2.0- -
1.8-

1.8 -

1.2 -

1.0 -

0.6 /

0.2 -

270 I I 1 120 ao 060 030 0
300 270 240 210 180 150 120 090 060 030 0
DIRECTION IDEGREES)

Figure A-13(upper), Figure A-14 (lower) .--Hurricane and tropical storm
track direction histograms and probability distributions for gulf
coast octagons. 1871-1973.

34 HURRICANE DIRECTIONS 171-1973 . 3 v.

3.2 N 33 LAT. 30 - 32.5� N
LONG. 90- 92.5 W
3.0-

2.8-

2,6-

2.4-

2.2-

O 2.0-

'~ 1.8-

1.4 -

1.2 -

1.0 -

0.4-

0.2-

0
I I J , 1 , ,. I I
300 270 240 210 180 150 ' 120 090 060 030 0
DIRECTION (DEGREESI

3,
HURRICANE DIRECTIONS 1871-1973 ;

3.2 - N= 16 LAT. 30 - 32.5� N . ;.
LONG. 925 .950W W
3.0-

2.6 -
1000 950 �00 05
2.4-

2.2-

C 2.0-
2,0 --

1.8

1.6 -

1.4 -

1.2 -_

0.6 - -

0.4 -

0.2-- /

o I , , , , \
300 270 240 210 180 150 120 090 060 030 0
DIRECTION (DEGREES)

Figure A-15(upper), Figure A-16(lower).--Hurricane and tropical storm

track direction histograms and probability distributions for gulf

coast octagons. 1871-1973.