HISTORIC NOTE
The publications in this collection do
not reflect current scientific knowledge
or recommendations. These texts
represent the historic publishing
record of the Institute for Food and
Agricultural Sciences and should be
used only to trace the historic work of
the Institute and its staff. Current IFAS
research may be found on the
Electronic Data Information Source
(EDIS)
site maintained by the Florida
Cooperative Extension Service.
Copyright 2005, Board of Trustees, University
of Florida
Bulletin 777 cal)
Bulletin 777 (tjechical)
Agricultural Experiment Stations
Institute of Food and Agricultural Sciences
University" of Florida Gainesville
June 1981
ILITIE!
.
FREEZE PROBABILITIES IN FLORIDA
James T. Bradley
Former State Climatologist, National Weather Service
and University of Florida Agricultural Research Center,
Lakeland, Florida.
First Printing November 1975
Second Printing June 1981
CONTENTS
Page
Introduction  1
Analysis of Weather Data _........... .............. ............ 1
Computation of Freeze Probability Dates ... .. 2
Sample Computation .... .......... 3
Discussion of Results .......  3
Using the Probability Tables _____ 4
Limitations ..........................................  8
Conclusions ............................................................. 8
Literature Cited __. 22
Acknowledgments ....._... .......... 22
INTRODUCTION
Freeze probabilities are useful to Florida agricultural indus
tries which produce cold sensitive crops as well as to homeowners
and engineers. The time interval between the 10% risk and the
50% risk of the first freezing temperature is 35 to 40 days, which
is much longer than in many states. This time interval is of im
portance to farmers who often make decisions and long range
plans based on freeze probabilities.
The usual methods used in determining climatological probabili
ties of the first and last occurrence of low temperatures are not
particularly applicable to Florida because there is not a well
developed fall and spring season. For example, temperatures be
low 32'F or 280F do not occur each year. Also, in seasons
where freezes do occur, freezing temperatures may not occur be
fore January 1, while in other seasons the last freeze may occur
before December 31. Finally, there are some cold seasons in which
there is only one cold day which is both the first and last freezing
temperature. All three of these climatic anomalies must be prop
erly treated to estimate freeze probabilities for Florida.
For those stations with freezing temperatures every year, the
dates of first and last freezing temperatures have been shown to
be random variables with a normal frequency distribution (6).
For stations where freezes do not occur each year, Thom (5) de
veloped techniques to combine the series of freeze and freezeless
years. Vestal (7) developed a method for analyzing data for the
"cold season," i.e., from July 1 to June 30, and has shown that
data from such a season fit a normal distribution.
ANALYSIS OF WEATHER DATA
The Florida "cold season" was defined as the period from July
1 to June 30. The dates of the first and last occurrences of each
selected temperature were abstracted from the station records of
23 climatological and first order National Weather Service sta
tions, mostly for the 30year period 19411970. Most of the data
retrieval and computation was done by the National Oceanic and
Atmospheric Administration, Environmental Data Service, Na
tional Climatic Center at Asheville, North Carolina. Separate com
putations were made for five stations to compare the methods
used and verify the methods.
Eleven stations tested for randomness with a runs test were
all random at the 0.10 probability level. Selected data from spring
1
INTRODUCTION
Freeze probabilities are useful to Florida agricultural indus
tries which produce cold sensitive crops as well as to homeowners
and engineers. The time interval between the 10% risk and the
50% risk of the first freezing temperature is 35 to 40 days, which
is much longer than in many states. This time interval is of im
portance to farmers who often make decisions and long range
plans based on freeze probabilities.
The usual methods used in determining climatological probabili
ties of the first and last occurrence of low temperatures are not
particularly applicable to Florida because there is not a well
developed fall and spring season. For example, temperatures be
low 32'F or 280F do not occur each year. Also, in seasons
where freezes do occur, freezing temperatures may not occur be
fore January 1, while in other seasons the last freeze may occur
before December 31. Finally, there are some cold seasons in which
there is only one cold day which is both the first and last freezing
temperature. All three of these climatic anomalies must be prop
erly treated to estimate freeze probabilities for Florida.
For those stations with freezing temperatures every year, the
dates of first and last freezing temperatures have been shown to
be random variables with a normal frequency distribution (6).
For stations where freezes do not occur each year, Thom (5) de
veloped techniques to combine the series of freeze and freezeless
years. Vestal (7) developed a method for analyzing data for the
"cold season," i.e., from July 1 to June 30, and has shown that
data from such a season fit a normal distribution.
ANALYSIS OF WEATHER DATA
The Florida "cold season" was defined as the period from July
1 to June 30. The dates of the first and last occurrences of each
selected temperature were abstracted from the station records of
23 climatological and first order National Weather Service sta
tions, mostly for the 30year period 19411970. Most of the data
retrieval and computation was done by the National Oceanic and
Atmospheric Administration, Environmental Data Service, Na
tional Climatic Center at Asheville, North Carolina. Separate com
putations were made for five stations to compare the methods
used and verify the methods.
Eleven stations tested for randomness with a runs test were
all random at the 0.10 probability level. Selected data from spring
1
and fall freezes, (Lakeland WSO, Tampa WSO, St. Leo, Quincy,
and Bushnell) were tested at five different temperature levels for
goodness of fit to the Gaussian probability distribution. Twenty
six of 32 data sets examined met the criteria for normal distri
bution given in Lilliefors' (4) table of critical values for the
KolmogorovSmirnov goodnessoffit test at the 0.1 level. With
these tests and in light of previous studies (1, 5, 7), it was con
cluded that the data for each station could be treated as if they
were random and normally distributed.
COMPUTATION OF FREEZE PROBABILITY DATES
Each data series was analyzed to determine the climatological
probability that a preselected low temperature would occur before
a certain date and to determine the date, for a specified probabili
ty level, when the low temperature would occur. The specified
probability level is the risk the user wants to take.
P(B) =P(B/A) x P(A)
Where: A=the threshold temperature.
B =the date upon which the first (last) event A oc
curs.
P(B) =the probability that the first (last) event A will
occur before (after) the date B in the cold season.
P(A) =the probability that event A will occur sometime
during the cold season.
P(B/A) =the probability that the first (last) event A will
occur before (after) date B, given that event A
actually does occur.
P(A) can be estimated by letting P(A)=m/n where m is the
number of cold seasons during which the preselected temperature
was reached and n is the total number of seasons in the data
sample. The estimate of P(B/A) applies in general only for those
cases in which event B is contained in event A.
Since it was assumed that the preselected temperature thres
holds were normally distributed, the cold season day number can
be defined by:
x =ts+Y.
Where: x =the cold season day number and is derived by as
signing to each calendar date a unique number be
ginning with July 1 as one, July 31 as 31, and so
on to June 30 as 365 (366 in leap years).
2
K= the mean day number of the first (last) occurrence
of the event A, calculated using the data for only
those years in which the event A actually occurs.
s=the standard deviation of the date of the first
(last) occurrence of the event A, calculated by the
formula =_m x2 (x)2
m(ml)
t=the number of standard deviations from the mean
of the normally distributed dates of the event A.
The value of t can be obtained from a table of the
inverse normal probability distribution.
SAMPLE COMPUTATION
The record used for Tampa were the cold seasons from 1939
1940 to 19691970. During this 31year period, there were 23 cold
seasons with temperatures of 320F or lower, yielding a probability
estimate of cold temperatures, P(A), of 23/31 or 0.742. x was
calculated to be 182.696 (December 30th).
s= m x2 (yx)2 1/2
m(ml)
s=27.08 days.
Assuming one wishes to know the latest date before there is a
25% chance of having a temperature as low as 320F, then
P(B/A) P(B) 0.25 =0.337
P(A) 0.742
From a table of the inverse normal distribution one finds that a
value of t=0.42 is associated with the probability of 0.337.
These values are substituted to yield
x=ts+x, giving
x=(0.42) (27.080)+182.696
x= 11.374+ 182.696=171.3 or December 18.
Thus there is a 25% chance of 320F or lower on or before Decem
ber 18 at Tampa.
DISCUSSION OF RESULTS
The mean (50% probability) date of the first 320F in the cold
season ranges from November 20th in the Panhandle to Decem
3
K= the mean day number of the first (last) occurrence
of the event A, calculated using the data for only
those years in which the event A actually occurs.
s=the standard deviation of the date of the first
(last) occurrence of the event A, calculated by the
formula =_m x2 (x)2
m(ml)
t=the number of standard deviations from the mean
of the normally distributed dates of the event A.
The value of t can be obtained from a table of the
inverse normal probability distribution.
SAMPLE COMPUTATION
The record used for Tampa were the cold seasons from 1939
1940 to 19691970. During this 31year period, there were 23 cold
seasons with temperatures of 320F or lower, yielding a probability
estimate of cold temperatures, P(A), of 23/31 or 0.742. x was
calculated to be 182.696 (December 30th).
s= m x2 (yx)2 1/2
m(ml)
s=27.08 days.
Assuming one wishes to know the latest date before there is a
25% chance of having a temperature as low as 320F, then
P(B/A) P(B) 0.25 =0.337
P(A) 0.742
From a table of the inverse normal distribution one finds that a
value of t=0.42 is associated with the probability of 0.337.
These values are substituted to yield
x=ts+x, giving
x=(0.42) (27.080)+182.696
x= 11.374+ 182.696=171.3 or December 18.
Thus there is a 25% chance of 320F or lower on or before Decem
ber 18 at Tampa.
DISCUSSION OF RESULTS
The mean (50% probability) date of the first 320F in the cold
season ranges from November 20th in the Panhandle to Decem
3
ber 30th in Central Florida and January 20th in South Florida.
Coastal sections south of Fort Pierce on the East Coast and
Punta Gorda on the West Coast have less than a 50% chance of
32F or lower in any year (Figures 1 and 2).
Since the standard deviation of the date of first (last) occur
rence of some freezing temperature is 20 to 25 days, the interval
between the 10% risk and the 50% risk is 35 to 40 days. At those
stations where the temperature threshold is not reached every
year, this interval becomes even larger. The standard deviation
of the first 320F temperature averages 18 days in the Panhandle,
20 days in North Florida, and 25 days in South and Central
Florida.
USING THE PROBABILITY TABLES
Results of the statistical analysis of the first and last occur
rence of preselected temperature thresholds are given in Tables
1 through 11. The probability that the indicated temperatures
will be reached sometime in the cold season is shown in Table 1.
Table 2 shows the dates by which there is a 10, 25, 50, 75, and
90% probability that 320F temperature will occur. Tables 36 are
similar tables for 28, 24, 20, and 160F. Tables 711 present the
last dates and probabilities for freezing temperatures. An index
and description of the stations used is this study is given in
Table 12.
The use of these tables can best be illustrated by examples. On
February 3rd a producer wishes to know the risk of cold damage
to sensitive crops such as tomatoes or peppers in the vicinity of
Lake Alfred and Fellsmere. By using Table 7, he can determine
there is a 50% probability of temperatures 320F or lower after
that date at Lake Alfred. Near Fellsmere, there is only a 25%
chance. If he waits until February 22, the risk of a 320F tempera
ture will decrease to 25% at Lake Alfred and 10% chance at
Fellsmere. Rather than wait he may decide to plant hardy or
semihardy crops like cabbage or celeryVat Lake Alfred.
Figures 1 and 2 are maps of the mean date (50% probability)
of the first 320F in the season and the last 320F in the season.
These maps can assist users in determining the freeze risks at
locations other than those given in the tables. Similar maps can
be made for other temperature thresholds.
Figures 3 and 4 graphically portray the dates associated with
the various probabilities of the first or last 320F temperatures in
a cold season for Clermont. The probability for any date or the
12/20
FIGURE 1.
MEAN DATE (50% PROBABILITY) OF FIRST
(320F) FREEZE OCCURRENCE.
3/1
r,j .Jo'
FIGURE 2. MEAN DATE (50%7 PROBABILITY) OF LAST
(320F) FREEZE OCCURRENCE.
Jan 29 Dec 30
Jan 19 Jan 09
Jan 09 Jan 19
Dec 30 Jan 29
Dec 20 Feb 08
Dec 10 Feb I8
Nov 30 Feb 28
Nov 20 Mar 10
Nov 10 Mar 20 
Nov O i Mar 3
0.01 0.05 02 0.5 2 5 10 20 30 40 50 60 70 80 90 aoI 0.05 02 0.5 I 2 5 10 20 30 40 50 60 70 80 90
PERCENT PROBABILITY OF (2*F) FREEZE PERCENT PROBABILITY OF (32F) FREEZE
BEFORE GIVEN DATE IN FALL AFTER GIVEN DATE IN SPRING
FIGURE 4. STATION: CLERMONT.
FIGURE 3. STATION: CLERMONT.
date corresponding to any specific probability can be determined
from these figures.
The days (dates) in the cold season correspond to the horizontal
lines and are plotted at regular spaced intervals along the left
side of the graph. For estimating the probabilities of the first
freeze, the days are plotted so that the earlier dates are at the
bottom of the graph and the later dates towards the top. The
reverse is done for the last freeze. The vertical lines labeled with
numbers along the bottom of the graph correspond to various
probability levels (e.g. 10 indicating 10 % probability, 70 indi
cating 70% probability).
From Table 2, the date corresponding to 10% probability at
Clermont for 320F temperature is December 1. This date is
plotted in Figure 3 at the intersection with line for December 1.
This intersection is marked with a small circle. Using another
probability value for Clermont, the procedure is repeated. For ex
ample, the date corresponding to the 50% probability is Decem
ber 31. This is plotted as a second circle. These points are con
nected with a straight line extending to the bottom and top of
the graph. Any desired probability can now be estimated from
this line. From Figure 3, it is estimated that there is a 5% prob
ability that the first 320F temperature will occur on or before
November 22nd and a 70% probability it will occur on or before
January 14th. In certain cases, when a preselected temperature
does not occur every year, a given probability may not exist.
A similar graph can be made for the probability of a spring
freeze after a given date. The graph for Clermont showing the
probability of receiving the last 320F after a given date (Figure
4) was constructed with the latest date at the bottom of the
graph. The 50% probability is January 17. A circle is placed at
the intersection of the 50% probability at January 17. The 10%
probability is February 26, which is similarly plotted on the
graph. A straight line is drawn through these points from the
top to the bottom of the graph. A vertical line is drawn at the
90% value since Table 1 shows that the probabilities beyond do
not exist.
The paper used for these graph is special probability paper.
On this paper, the socalled "normal probability curve" appears
as a straight line. If one wishes to estimate probabilities for other
locations it can be done by plotting the tabular data on probability
paper.
LIMITATIONS
The minimum temperatures used in this study were observed
in standardtype louvered shelters at a height of 5 feet above
the ground. (Lakeland WSO was an exception.) Although the
sites for weather stations are chosen to be as representative as
possible, local conditions do affect the data. Topography is the
most significant influencing factor (1). Cold air drainage at night
from the top to the bottom of a slope can produce significantly
different minimum temperatures at adjacent sites. This factor
assumes an added dimension in Florida where many of the freezes
are of the radiationtype with little cold advection and near calm
conditions. Minimum temperature differences as much as 18F
have been noted between high and low ground locations less than
onequarter mile apart in the central section of the peninsula (3).
Consideration must be given in relating shelter temperatures to
ground temperatures. Studies at the University of Vermont (2)
show that daily minimum temperatures at the 3inch level aver
age 40F below those at the 5foot level in the spring and fall
seasons. Thus the shelter temperature will have a different mean
ing to the citrus grower than to the truck farmer.
Data from Lakeland WSO illustrate the effect of observation
height on temperatures. The site at Lakeland is located at the
City Hall with thermometers some 56 feet above the ground level.
Table 2 shows that for this station the mean date of the first
320F occurrence is January 24. This date is almost a month later
than the mean date for surrounding stations. This station also
shows an average of two days per year on which the temperature
falls to 320F, or below; the duration of temperatures in this range
averages 11 hours per year. At a station located on the northern
edge of the city, (Lakeland No. 2), where the thermometers are
exposed at a standard height of 5 feet above ground level, similar
values are 7 days and 35 hours respectively.
CONCLUSIONS
The usefulness of these probability tables and graphs is mainly
for planning agricultural and agribusiness operations. Individual
users must make subjective corrections for differences in topog
raphy, for nearness to urban locations or bodies of water, and for
temperatures at other than the standard level. Finally, during
any particular cold season, daytoday operational decisions
should be based on the latest weather forecasts and outlooks.
LIMITATIONS
The minimum temperatures used in this study were observed
in standardtype louvered shelters at a height of 5 feet above
the ground. (Lakeland WSO was an exception.) Although the
sites for weather stations are chosen to be as representative as
possible, local conditions do affect the data. Topography is the
most significant influencing factor (1). Cold air drainage at night
from the top to the bottom of a slope can produce significantly
different minimum temperatures at adjacent sites. This factor
assumes an added dimension in Florida where many of the freezes
are of the radiationtype with little cold advection and near calm
conditions. Minimum temperature differences as much as 18F
have been noted between high and low ground locations less than
onequarter mile apart in the central section of the peninsula (3).
Consideration must be given in relating shelter temperatures to
ground temperatures. Studies at the University of Vermont (2)
show that daily minimum temperatures at the 3inch level aver
age 40F below those at the 5foot level in the spring and fall
seasons. Thus the shelter temperature will have a different mean
ing to the citrus grower than to the truck farmer.
Data from Lakeland WSO illustrate the effect of observation
height on temperatures. The site at Lakeland is located at the
City Hall with thermometers some 56 feet above the ground level.
Table 2 shows that for this station the mean date of the first
320F occurrence is January 24. This date is almost a month later
than the mean date for surrounding stations. This station also
shows an average of two days per year on which the temperature
falls to 320F, or below; the duration of temperatures in this range
averages 11 hours per year. At a station located on the northern
edge of the city, (Lakeland No. 2), where the thermometers are
exposed at a standard height of 5 feet above ground level, similar
values are 7 days and 35 hours respectively.
CONCLUSIONS
The usefulness of these probability tables and graphs is mainly
for planning agricultural and agribusiness operations. Individual
users must make subjective corrections for differences in topog
raphy, for nearness to urban locations or bodies of water, and for
temperatures at other than the standard level. Finally, during
any particular cold season, daytoday operational decisions
should be based on the latest weather forecasts and outlooks.
TABLE 1. PROBABILITY THAT INDICATED TEMPERATURE
THRESHOLD IS REACHED.*
Threshold
Station 320F 280F 240F 200F 160F
Avon Park 0.833 0.400 0.067 0.033 0
Bushnell 1.000 0.933 0.733 0.200 0.067
Carrabelle 1.000 0.900 0.767 0.467 0.133
Chipley 1.000 1.000 0.900 0.733 0.233
Clermont 0.900 0.467 0.100 0.033 0
Cross City 1.000 1.000 0.900 0.600 0.167
Everglades 0.333 0.033 0 0 0
Fellsmere 0.767 0.533 0.067 0 0
Fernandina Beach 0.967 0.733 0.433 0.133 0.033
Homestead 0.700 0.133 0 0 0
Jacksonville Beach 0.967 0.833 0.367 0.067 0
Lake Alfred 0.967 0.700 0.200 0.033 0.033
Lake City 1.000 0.967 0.833 0.433 0.067
Lakeland WSO 0.679 0.250 0.033 0.033 0
Madison 1.000 0.933 0.833 0.433 0.133
Milton 1.000 1.000 0.900 0.633 0.267
Monticello 1.000 0.933 0.900 0.667 0.233
Palatka 0.967 0.700 0.333 0.033 0.033
Quincy 1.000 0.968 0.903 0.548 0.226
St. Leo 0.833 0.567 0.167 0.033 0
Tallahassee WSO 1.000 0.900 0.633 0.400 0.226
Tampa WSO 0.742 0.290 0.161 0.033 0
Titusville 0.967 0.667 0.233 0 0
* P [A]= m/n
m = number of cold seasons threshold reached.
n = total number of seasons of record.
TABLE 2. FALL DATES BEFORE WHICH THE FIRST TEMPERA
TURE OF 320F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.10 0.25 0.50 0.75 0.90
Avon Park Dec. 3 Dec. 20 Jan. 9 Feb. 4 *
Bushnell Nov. 7 Nov. 21 Dec. 6 Dec. 22 Jan. 5
Carrabelle Oct. 30 Nov. 15 Dec. 3 Dec. 20 Jan. 5
Chipley Oct. 26 Nov. 4 Nov. 15 Nov. 25 Dec. 5
Clermont Dec. 1 Dec. 15 Dec. 31 Jan. 18
Cross City Oct. 23 Nov. 7 Nov. 24 Dec. 10 Dec. 25
Everglades Dec. 24 Jan. 25 *
Fellsmere Dec. 7 Dec. 24 Jan. 14 Feb. 24
Fernandina Beach Nov. 14 Nov. 26 Dec. 10 Dec. 24 Jan. 7
Homestead Dec. 15 Jan. 1 Jan. 22 *
Jacksonville Beach Nov. 16 Nov. 30 Dec. 15 Dec. 31 Jan. 16
Lake Alfred Nov. 24 Dec. 10 Dec. 27 Jan. 14 Feb. 1
Lake City Nov. 4 Nov. 15 Nov. 28 Dec. 11 Dec. 22
Lakeland WSO Dec. 13 Dec. 31 Jan. 24 *
Madison Nov. 5 Nov. 14 Nov. 24 Dec. 4 Dec. 13
Milton Oct. 26 Nov. 6 Nov. 18 Nov. 30 Dec. 10
Monticello Oct. 31 Nov. 11 Nov. 23 Dec. 5 Dec. 16
Palatka Nov. 17 Nov. 30 Dec. 14 Dec. 29 Jan. 13
Quincy Oct. 30 Nov. 9 Nov. 20 Dec. 1 Dec. 11
St. Leo Nov. 18 Dec. 6 Dec. 27 Jan. 17
Tallahassee WSO Oct. 28 Nov. 8 Nov. 21 Dec. 3 Dec. 14
Tampa WSO Nov. 30 Dec. 18 Jan. 11 *
Titusville Nov. 29 Dec. 15 Jan. 2 Jan. 20 Feb. 8
* The chances are less than the indicated probability that a temperature of
32F or lower will occur.
** Fewer than five occurrences.
TABLE 3. FALL DATES BEFORE WHICH THE FIRST TEMPER
ATURE OF 280F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.10 0.25 0.50 0.75 0.90
Avon Park Dec. 26 Jan. 14 *
Bushnell Nov. 20 Dec. 6 Dec. 24 Jan. 13 Feb. 6
Carrabelle Nov. 7 Nov. 21 Dec. 7 Dec. 26
Chipley Oct. 31 Nov. 13 Nov. 29 Dec. 14 Dec. 28
Clermont Dec. 15 Jan. 7 *
Cross City Nov. 4 Nov. 22 Dec. 12 Jan. 1 Jan. 19
Everglades ** ** ** ** **
Fellsmere Dec. 19 Jan. 6 Feb. 12 *
Fernandina Beach Dec. 3 Dec. 21 Jan. 12 *
Homestead ** ** ** **
Jacksonville Beach Dec. 4 Dec. 19 Jan. 6 Jan. 30 *
Lake Alfred Dec. 10 Dec. 26 Jan. 20 *
Lake City Nov. 10 Nov. 24 Dec. 10 Dec. 27 Jan. 13
Lakeland WSO Jan. 2 *
Madison Nov. 14 Nov. 28 Dec. 14 Jan. 2 Jan. 24
Milton Oct. 31 Nov. 13 Nov. 28 Dec. 13 Dec. 26
Monticello Nov. 5 Nov. 18 Dec. 3 Dec. 19 Jan. 8
Palatka Dec. 4 Dec. 23 Jan. 19 *
Quincy Nov. 12 Nov. 24 Dec. 7 Dec. 21 Jan. 4
St. Leo Dec. 16 Jan. 6 Feb. 11 *
Tallahassee WSO Nov. 2 Nov. 22 Dec. 14 Jan. 8
Tampa WSO Jan. 1 Feb. 8 *
Titusville Dec. 16 Jan. 2 Jan. 26 *
* The chances are less than the indicated probability that a temperature of
28F or lower will occur.
** Fewer than five occurrences.
TABLE 4. FALL DATES BEFORE WHICH THE FIRST TEMPERA
TURE OF 240F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.10 0.25 0.50 0.75 0.90
Avon Park ** ** ** ** **
Bushnell Dec. 6 Dec. 24 Jan. 16 *
Carrabelle Nov. 27 Dec. 14 Jan. 3 Feb. 12
Chipley Nov. 17 Dec. 2 Dec. 20 Jan.8 *
Clermont ** ** ** ** **
Cross City Nov. 27 Dec. 13 Jan. 1 Jan. 21
Everglades ** ** ** ** **
Fellsmere ** ** ** ** **
Fernandina Beach Dec. 17 Jan.13 *
Homestead ** ** ** ** **
Jacksonville Beach Dec. 26 Jan.25 *
Lake Alfred Jan. 13 *
Lake City Nov. 29 Dec. 16 Jan. 5 Jan. 31
Lakeland WSO ** ** ** ** **
Madison Nov. 26 Dec. 14 Jan.5 Feb. 3 *
Milton Nov. 12 Nov. 28 Dec. 16 Jan.6 *
Monticello Nov. 19 Dec. 4 Dec. 22 Jan. 10 *
Palatka Dec. 23 Jan. 29 *
Quincy Nov. 21 Dec. 8 Dec. 27 Jan.18 Mar. 6
St. Leo Jan. 16 *
Tallahassee WSO Nov. 15 Dec. 9 Jan. 14 *
Tampa WSO ** ** ** ** **
Titusville Dec. 31 *
* The chances are less than the indicated probability that a temperature of
24F or lower will occur.
** Fewer than five occurrences.
TABLE 5. FALL DATES BEFORE WHICH THE FIRST TEMPERA
TURE OF 200F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.10 0.25 0.50 0.75 0.90
Avon Park ** ** ** **
Bushnell Jan. 10 *
Carrabelle Dec. 26 Jan. 18 *
Chipley Dec. 12 Dec. 28 Jan. 19 *
Clermont ** ** ** ** *E
Cross City Dec. 12 Dec. 31 Jan. 29 *
Everglades ** ** ** **
Fellsmere ** ** ** ** *
Fernandina Beach ** ** ** ** *
Homestead ** ** ** **
Jacksonville Beach ** ** ** **
Lake Alfred ** ** ** ** **
Lake City Dec. 15 Jan. 11 *
Lakeland WSO ** ** ** *
Madison Dec. 19 Jan. 13 *
Milton Dec. 9 Dec. 27 Jan. 23 *
Monticello Dec. 9 Dec. 28 Jan. 23 *
Palatka *' ** ** ** 
Quincy Dec. 6 Dec. 27 Feb. 2 *
St. Leo ** ** ** ** **
Tallahassee WSO Dec. 18 Jan. 7 *
Tampa WSO ** ** ** ** **
Titusville :* ** ** ** **
* The chances are less than the indicated probability that a temperature of
20 F or lower will occur.
** Fewer than five occurrences.
TABLE 6. FALL DATES BEFORE WHICH THE FIRST TEMPERA
TURE OF 160F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.10 0.25 0.50 0.75 0.90
Avon Park ** ** ** **
Bushnell ** ** ** ** **
Carrabelle ** ** ** ** **
Chipley Jan.7 *
Clermont ** ** ** ** **
Cross City Dec.24 *
Everglades ** ** ** ** **
Fellsmere ** ** ** ** **
Fernandina Beach ** ** ** ** **
Homestead ** ** ** ** **
Jacksonville Beach ** ** ** ** **
Lake Alfred ** ** ** ** **
Lake City ** ** ** ** **
Lakeland WSO ** ** ** ** **
Madison ** ** ** ** **
Milton Jan. 4 Feb. 9 *
Monticello Dec. 29 *
Palatka ** ** ** ** **
Quincy Jan. 7 ** *
St. Leo ** ** ** ** **
Tallahassee WSO N/A *
Tampa WSO ** ** ** ** **
Titusville ** ** ** ** **
* The chances are less than the indicated probability that a temperature of
16F or lower will occur.
** Fewer than five occurrences.
TABLE 7. SPRING DATES AFTER WHICH THE LAST TEMPERA
TURE OF 320F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.90 0.75 0.50 0.25 0.10
Avon Park Dec. 16 Jan.14 Feb.5 Feb.24
Bushnell Jan. 18 Feb. 4 Feb. 23 Mar. 14 Mar. 31
Carrabelle Jan. 29 Feb. 13 Mar. 1 Mar. 17 Apr. 1
Chipley Feb. 5 Feb. 17 Mar. 3 Mar. 17 Mar. 30
Clermont Dec. 25 Jan. 17 Feb. 8 Feb.26
Cross City Feb. 11 Feb. 25 Mar. 12 Mar. 27 Apr. 9
Everglades Dec. 25 Jan. 26
Fellsmere Nov. 28 Jan. 12 Feb. 4 Feb. 23
Fernandina Beach Dec. 30 Jan. 21 Feb. 12 Mar. 5 Mar. 24
Homestead Jan. 2 Jan. 24 Feb. 10
Jacksonville Beach Dec. 31 Jan. 19 Feb. 8 Feb. 26 Mar. 15
Lake Alfred Dec. 28 Jan. 16 Feb. 3 Feb. 21 Mar. 8
Lake City Feb. 7 Feb. 19 Mar. 5 Mar. 18 Mar. 30
Lakeland WSO Jan. 7 Jan. 21 Feb. 15
Madison Jan. 28 Feb. 11 Feb. 26 Mar. 13 Mar. 26
Milton Feb. 9 Feb. 20 Mar. 5 Mar. 18 Mar. 30
Monticello Jan. 31 Feb. 14 Mar. 3 Mar. 20 Apr. 4
Palatka Dec. 27 Jan. 14 Feb. 1 Feb. 19 Mar. 7
Quincy Feb. 8 Feb. 21 Mar. 7 Mar. 21 Apr. 3
St. Leo Jan. 2 Jan. 23 Feb. 8 Feb. 21
Tallahassee WSO Feb. 4 Feb. 18 Mar. 7 Mar. 23 Apr. 6
Tampa WSO Jan. 12 Feb. 3 Feb. 19
Titusville Dec. 20 Jan. 10 Feb. 1 Feb. 21 Mar. 11
* The chances are less than the indicated probability that a temperature of
32 F or lower will occur.
** Fewer than five occurrences.
TABLE 8. SPRING DATES AFTER WHICH THE LAST TEMPERA
TURE OF 280F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.90 0.75 0.50 0.25 0.10
Avon Park Jan. 10 Jan. 29
Bushnell Dec. 12 Jan. 11 Feb. 3 Feb. 25 Mar. 16
Carrabelle Jan. 29 Feb. 15 Mar. 1 Mar. 14
Chipley Jan. 17 Feb. 1 Feb. 18 Mar. 7 Mar. 22
Clermont Jan. 14 Feb. 5
Cross City Jan. 13 Jan. 29 Feb. 15 Mar. 5 Mar. 21
Everglades *
Fellsmere Dec. 18 Jan. 19 Feb. 4
Fernandina Beach Jan. 19 Feb. 8 Feb. 23
Homestead ** ** ** *
Jacksonville Beach Dec. 27 Jan. 19 Feb. 5 Feb. 19
Lake Alfred Jan. 12 Feb. 1 Feb. 16
Lake City Jan. 16 Feb. 1 Feb. 17 Mar. 5 Mar. 18
Lakeland WSO Jan. 26
Madison Dec. 24 Jan. 16 Feb. 4 Feb. 21 Mar. 9
Milton Jan. 31 Feb. 11 Feb. 23 Mar. 7 Mar. 18
Monticello Jan. 4 Jan. 27 Feb. 14 Mar. 3 Mar. 18
Palatka Jan. 6 Jan. 30 Feb. 18
Quincy Jan. 4 Jan.23 Feb.11 Feb.28 Mar. 16
St. Leo Dec. 21 Jan. 24 Feb. 13
Tallahassee WSO Dec. 30 Jan. 29 Feb. 26 Mar. 21
Tampa WSO Dec. 23 Jan. 28
Titusville Jan. 1 Jan. 26 Feb. 12
* The chances are less than the indicated probability that a temperature of
28F or lower will occur.
** Fewer than five occurrences.
TABLE 9. SPRING DATES AFTER WHICH THE LAST TEMPERA
TURE OF 240F OR LOWER WILL OCCUR FOR SE
LECTED PROBABILITIES.
Probabilities
Station 0.90 0.75 0.50 0.25 0.10
Avon Park ** ** ** ** **
Bushnell Jan.11 Feb. 2 Feb. 20
Carrabelle Nov. 30 Jan. 12 Feb. 4 Feb. 22
Chipley Dec. 30 Jan. 18 Feb. 5 Feb. 20
Clermont ** ** ** :* **
Cross City Jan. 6 Jan. 22 Feb. 6 Feb. 19
Everglades ** ** ** ** **
Fellsmere ** ** ** ** **
Fernandina Beach Jan. 20 Feb. 10
Homestead ** ** ** ** **
Jacksonville Beach Jan. 15 Feb. 11
Lake Alfred Jan. 14
Lake City Dec. 25 Jan. 17 Feb. 3 Feb. 18
Lakeland WSO ** ** ** ** **
Madison Dec. 19 Jan.14 Feb. 3 Feb. 19
Milton Jan. 8 Jan. 28 Feb. 15 Mar. 2
Monticello Dec. 30 Jan. 22 Feb. 11 Mar. 1
Palatka Jan. 4 Feb. 3
Quincy Oct. 28 Dec. 21 Jan. 15 Feb. 6 Feb. 25
St. Leo Jan. 7
Tallahassee WSO Jan. 15 Feb. 10 Feb. 28
Tampa WSO ** ** ** ** **
Titusville ** ** ** ** **
* The chances are less than the indicated probability that a temperature of
24F or lower will occur.
** Fewer than five occurrences.
TABLE 10.
SPRING DATES AFTER WHICH THE LAST TEM
PERATURE OF 200F OR LOWER WILL OCCUR FOR
SELECTED PROBABILITIES.
Probabilities
Station 0.90 0.75 0.50 0.25 0.10
Avon Park ** ** ** ** **
Bushnell Jan. 11
Carrabelle Jan.24 Feb.9
Chipley Jan. 10 Jan.29 Feb. 13
Clermont ** ** ** ** **
Cross City Dec. 31 Jan. 23 Feb. 7
Everglades ** ** ** ** **
Fellsmere ** ** ** ** **
Fernandina Beach ** ** **
Homestead ** ** ** ** **
Jacksonville Beach ** ** ** ** **
Lake Alfred ** ** ** ** **
Lake City Jan. 15 Feb. 6
Lakeland WSO ** ** ** ** **
Madison Jan. 20 Feb. 5
Milton Jan. 7 Jan. 26 Feb. 8
Monticello Jan. 11 Jan. 29 Feb. 12
Palatka ** ** ** ** **
Quincy Dec. 7 Jan.15 Feb.4
St. Leo ** ** ** ** **
Tallahassee WSO Jan. 9 Feb. 2
Tampa WSO ** ** ** ** **
Titusville ** ** ** **
* The chances are less than the indicated probability that a temperature of
20 F or lower will occur.
** Fewer than five occurrences.
TABLE 11.
SPRING DATES AFTER WHICH THE LAST TEM
PERATURE OF 160F OR LOWER WILL OCCUR FOR
SELECTED PROBABILITIES.
Probabilities
Station 0.90 0.75 0.50 0.25 0.10
Avon Park ** ** ** ** **
Bushnell ** ** ** ** **
Carrabelle ** ** ** ** **
Chipley Jan. 20
Clermont ** ** ** ** **
Cross City Dec. 26
Everglades ** ** ** ** **
Fellsmere ** ** ** ** **
Fernandina Beach ** ** ** **
Homestead ** ** ** ** **
Jacksonville Beach ** ** ** ** **
Lake Alfred ** ** ** ** **
Lake City ** ** ** ** **
Lakeland WSO ** ** ** ** **
Madison ** ** ** **
Milton Dec. 23 Jan. 25
Monticello Jan. 18
Palatka ** ** ** ** **
Quincy Jan.20
St. Leo ** ** ** ** **
Tallahassee WSO NA
Tampa WSO ** ** ** **
Titusville ** ** ** ** **
* The chances are less than the
16F or lower will occur.
** Fewer than five occurrences.
indicated probability that a temperature of
Station
NORTH
Cross City 2WNW
Fernandina Beach
Jacksonville Beach
Lake City 2E
Madison
Palatka
Tallahassee WSO
NORTHWEST
Carrabelle 1NNW
Chipley 3E
Milton Exp. Stn.
Monticello 3W
Quincy 3SSW
County Elevation (ft.)
Dixie
Nassau
Duval
Columbia
Madison
Putnam
Leon
Franklin
Washington
Santa Rosa
Jefferson
Gadsden
42 Over sod on grounds of Florida Forest Service district
office. Good exposure.
25 Over sod at city water plant. Surrounded by low to
medium height shrubs. Fair exposure.
10 Over sod at Fire Dept. Good exposure.
195 Over sod in fenced enclosure. Excellent exposure.
190 Shelter over sod near observer's residence. Good ex
posure. SRG over dirt. Excellent exposure.
20 Over sod at Fire Dept. Only a few feet away from as
phalt parking lot. Fair exposure.
55 Over sod in low depression. Poor exposure. Much too
cold for area.
10 Over sod near observer's residence. Fair exposure.
130 Over sod at U of F Exp. Stn. Heavily wooded in gen
eral area of shelter. Fair exposure.
217 Over sod in fenced enclosure. Excellent exposure.
148 Over sod at U of F Exp. Stn. Excellent exposure.
Ground elevation drops 15 feet 600' north.
245 Over sod in fenced enclosure. Excellent exposure. Lo
cated at U of F Exp. Stn.
Remarks
NORTHCENTRAL
Bushnell 2E
Clermont 6SSE
St. Leo
Titusville 3NW
SOUTHCENTRAL
Avon Park
Fellsmere 7SSW
Lake Alfred Exp. Stn.
Lakeland WSO
Tampa WSO
EVERGLADES &
COAST
Everglades
LOWER EAST
COAST
T~inestead Exp. Stn.
Sumter
Lake
Pasco
Brevard
Highlands
Indian River
Polk
Polk
Hillsborough
Collier
Dade
75
125
190
30
149
20
145
214
19
Over sod in observer's back yard. Excellent exposure.
Over thin sod very near observer's large citrus grove.
Good exposure.
Over sod at St. Leo College. Surrounding area very
hilly.
Over very thin sod near observer's residence. Heavily
wooded with fair exposure.
Over sod across street from Fire Dept. Excellent ex
posure.
Over sod at Diamond G Farm. Completely open with
excellent exposure.
Over sod in fenced enclosure. Excellent exposure.
On roof of City Hall, in immediate downtown area.
Very poor exposure.
Over sod on airport property. Excellent exposure.
5 Over hardpan and coral rock. Chokoloskee Bay 20 feet
to the south. Good exposure for this area.
11 Over sod in fenced enclosure. Excellent exposure.
LITERATURE CITED
1. Butson, K. D. and J. F. Gerber. 1964. Temperature hazards to peaches
in Florida. Proc. Fla. State Hort. Soc. 77:395401.
2. Hopp, R. J., K. E. Varney, and R. E. Lautzenheiser. 1964. Late spring
and early fall low temperatures in Vermont. Univ. Vt. Agr. Exp.
Stat. Bull. 639, 23 pp.
3. Johnson, W. O. 1970. Minimum temperatures in the agricultural areas
of peninsular Florida, Summary of seasons 193767. Univ. Fla. Inst.
Food Agr. Sci. Public. 9, 154 pp.
4. Lilliefors, H. W. 1967. On the KolmogorovSmirnov test for normality
with mean and variance unknown. Amer. Stat. Assoc. 62(318):
399402.
5. Thom, H. C. S. 1959. The distribution of freezedate and freezefree
period for climatological series with freezeless years. Monthly
Weather Rev. 87(4):136144.
6. Thom, H. C. S., and R. H. Shaw. 1958. Climatological analysis of freeze
data for Iowa. Monthly Weather Rev. 86(7):251257.
7. Vestal, C. K. 1971. First and last occurrence of low temperatures during
cold season. Monthly Weather Rev. 99(8):650652.
ACKNOWLEDGMENTS
The author thanks W. O. Johnson and J. Georg of the Lakeland
Weather Service Office for many fruitful discussions and T.
Clarke, the National Weather Service Substation Network Spe
cialist, for compiling Table 12. Finally, we acknowledge our debt
to the many dedicated and publicspirited Cooperative Weather
Observers, whose daily observations form the basis for this work.
LITERATURE CITED
1. Butson, K. D. and J. F. Gerber. 1964. Temperature hazards to peaches
in Florida. Proc. Fla. State Hort. Soc. 77:395401.
2. Hopp, R. J., K. E. Varney, and R. E. Lautzenheiser. 1964. Late spring
and early fall low temperatures in Vermont. Univ. Vt. Agr. Exp.
Stat. Bull. 639, 23 pp.
3. Johnson, W. O. 1970. Minimum temperatures in the agricultural areas
of peninsular Florida, Summary of seasons 193767. Univ. Fla. Inst.
Food Agr. Sci. Public. 9, 154 pp.
4. Lilliefors, H. W. 1967. On the KolmogorovSmirnov test for normality
with mean and variance unknown. Amer. Stat. Assoc. 62(318):
399402.
5. Thom, H. C. S. 1959. The distribution of freezedate and freezefree
period for climatological series with freezeless years. Monthly
Weather Rev. 87(4):136144.
6. Thom, H. C. S., and R. H. Shaw. 1958. Climatological analysis of freeze
data for Iowa. Monthly Weather Rev. 86(7):251257.
7. Vestal, C. K. 1971. First and last occurrence of low temperatures during
cold season. Monthly Weather Rev. 99(8):650652.
ACKNOWLEDGMENTS
The author thanks W. O. Johnson and J. Georg of the Lakeland
Weather Service Office for many fruitful discussions and T.
Clarke, the National Weather Service Substation Network Spe
cialist, for compiling Table 12. Finally, we acknowledge our debt
to the many dedicated and publicspirited Cooperative Weather
Observers, whose daily observations form the basis for this work.
A JUW 7W
This public document was promulgated at a cost of $688.20, or
34 cents per copy to provide information on climatological
probabilities of occurrence of freezing temperatures in Florida.
All programs and related activities sponsored or assisted by the Florida
Agricultural Experiment Stations are open to all persons regardless of race,
:olor, national origin, age, sex, or handicap.
I TEACHING RESEARCH EXTENSI
