sign in sign up
<<
Home , Climate of Florida, Great Freeze

3. Dillon, D. F. 1981. Propagating dwarf citrus with hydronic radiant ples and Practices. Prentice-Hall, New Jersey. heated benches. Combined Proc. Int. Plant Propagators' Soc, Four 9. Jauhari, O. S. and S. F. Rahman. 1959. Further investigations on Winds Growers, Fremont, Calif. rooting in cuttings of sweet lime (Citrus limettoides) Tanaka. Sci. Cult. 4. Dore, J. 1953. Seasonal variation in the regeneration of root cuttings. 24:432-434. Nature.172:1189. 10. Johnston, J. C, K. W. Opitz, and E. F. Frolich. 1959. Citrus propaga 5. Ford, H. W. 1957. A method of propagating citrus rootstockclones tion. Calif. Agr. Expt. Sta. Cir. 475. by leaf bud cuttings. Proc. Amer. Soc. Hort. Sci. 69:204-207. 11. Kossuth, S. V., R. H. Biggs, P. G. Webb, and K. M. Porter. 1981. 6. Gates, C. T., D. Bouma, and H. Groenewegen. 1961. The develop Rapid propagation techniques for fruit crops. Proc. Fla. State Hort. ment of cuttings of the navel to the stage of fruit Soc. 94:323-328. set. I. The development of the rooted cutting. Aust. J. Agr. Res. 12. Platt, R. G. and K. W. Opitz. 1973. The propagation of citrus, p. 12:1050-1065. 4-47. In: W. Reuther (ed.). The citrus industry Vol. III. Univ. of 7. Halma, F. F. 1931. The propagation of citrus by cuttings. Hilgardia. Calif., Berkeley. 6:131-157. 13. Umarov, A. Raising transplants from softwood cuttings. 1985. 8. Hartmann, H. T. and D. E. Kester. 1983. Plant Propagation: Princi Hort. Abstr. 55:1513.

Proc Fla. State Hort. Soc. 98:42-46. 1985.

MINIMUM TEMPERATURE CYCLES IN

E. Chen and J. F. Gerber and 7-10 Feb. 1895) occurred. They " . . destroyed more Fruit Crops Department, IFAS property than any other freeze in the history of the state University of Florida . . ." (5, 16). The freeze of 1899 (Jacksonville minimum Gainesville, FL 32611 temperature, -12°C) killed many trees in north and was instrumental in movement of groves Additional index words. , trend, freeze, time southward. Next came the 11 and 13 Dec. 1962 freeze, series. which was recorded as the most severe freeze of this cen tury (10) until another pair of freezes occurred on Dec. Abstract. Time series formed from 88 years (1898-1985) of 1983 and Jan. 1985. Thus, available historical records of minimum temperature observations from Jacksonville, Ocala, citrus culture indicate atleast 3 periods of 60 to 70 years Clermont, Bartow, Arcadia, Fort Myers, and , and 151 in the recent past when the temperature environment was years (1835-1985) of annual minimum temperatures from supportive to citrus cultivation in : Jacksonville were analyzed. Trends, patterns, and possible the St. Augustine area (pre 1835), the Palatka area (1835- cycles and periodicities of the time series were identified. 1899), then movement southward (post 1899). Now there Linear correlations among the seven stations were calculated is doubt about citrus cultivation in Marion County. to determine the change of minimum temperatures with This study analyzes the long term records of absolute latitude and distance to coast. Our present position in the minimum temperatures in Florida for 7 cities in interior time series is compared to similar situations in the past to and coastal locations. The purpose is to identify trends infer possible future minimum temperatures. and periodicities for short term prediction and future planning. Analyses were done on the long term annual data from Jacksonville, and statistical correlations were em Extreme low temperature events have large negative ployed to show the change of minimum temperature in impacts on horticulture. Extreme events which occur back thepeninsula. Seasonal minimum temperatures were to back are even more devastating. An example is the 25 selected for this study because of their critical importance Dec. 1983 freeze (8) which was followed 13 months later to horticulture. Some points in the time series may reach by the 21 Jan. 1985 freeze (9). Four severe freezes occur the observed value for as little as 1 to 2 hours out of the red during the past 5 winters (1980-81 to 1984-85); the entire season (1 Nov. to 15 Mar.) or the annual calendar one winter without a severe freeze was the 1982-83 season. year (1 Jan. to 31 Dec). This recent cluster of freezes raised questions concerning possible short-term changes in the temperature regime in Materials and Methods the Florida peninsula. Historical observations appeared to indicate that these events tend to occur in cycles and with Time series formed from absoluteseasonal minimum some periodicities (5). Citrus products (fresh fruit and temperatures (1898-1985) from Jacksonville, Ocala, Cler juice) were exported from the St. Augustine area as early mont, Bartow, Arcadia, Fort Myers, and Miami were as 1778 (6, 20). Many trees in the area were 100-years-old analyzed. In addition, a longer period (151 years) of an (7) when the freeze of 7-8 Feb. 1835 struck—" . . there nual minimum temperature (1835-1985) from Jacksonville came a frost, a killing frost, which destroyed every orange, were also used. The cities were selected on the basis of the lime, and lemon tree in Florida, a circumstance which length and completeness of their temperature records and could not have been foreseen, as such a thing had never their locations in the peninsula to give spatial and coastal before occurred ..." (7). The point is there was a period representation. The annual data from Jacksonville were of some 50-100 years where minimum temperatures were compiled from 1835-1905 (14) and from Climatological mild enough to support citrus cultivation in the St. Augus Data for Florida (4). Data before about 1897 were deemed tine area. Then 2 back-to-back freezes (27-28 Dec. 1894 reliable but unofficial. Seasonaldata are more suitable for analysis because the annual accounting method separates Florida Agricultural Experiment Stations Journal Services No. 6904. observations in a season into 2 calendar years. A long series

42 Proc. Fla. State Hort. Soc. 98: 1985. Table 1. Linear regression of seasonal observations and correlation of each city with its nearest neighbor (NN) and with Jacksonville (Jax). JflCKSONVILLE RNNURL TEMPERflTURE The correlation between annual and seasonal observations of each city is also shown (Ann).

Linear regress, Linear correlation with

City Slope r NN Jax Ann

Jacksonville -.029 -.255** 1.00 1.00 0.62 Ocala -.004 -.042 0.83 0.83 0.36 Clermont -.021 -.214** 0.82 0.88 0.61 Bartow -.003 -.030 0.80 0.72 0.48 Arcadia -.038 -.385** 0.67 0.67 0.63 Fort Myers -.018 -.207 0.75 0.76 0.51 Miami .000 .000 0.77 0.70 0.60

1635 1855 1875 1695 1915 1935 1955 1975 of seasonal observations is not available but is needed for YEfiRS cycle analyses. The average correlation between seasonal Fig. 1. Annual minimum temperatures from 1835 to 1985. The hori and annual data for the 7 stations is 0.54 (Table 1). The zontal dash line is the mean of the observations, the continuous straight correlation points out the difference between annual and line is the linear trend, the smooth line is the 5-year weighted running seasonal data which are subsets of one set of observations. average. Black triangles indicate periods in the observations where the Comparison of seasonal and annual data for ten stations standard deviation remained less than 1.0 over 4 and 5 consecutive years. in the show an average of 5% in decadal variance (18). Missing data were extrapolated linearly fol tracted from an observed point and the square of the dif lowing the method of Brooks and Carruthers (2). ference summed to obtain a variance. Cycles with the smal During the 88-year period every station had experi lest variance, the "best fit" were kept. Cycles which were enced relocation. Relocations were generally to airports kept were tested for amplitudes of the coefficients until and rural agricultural stations where temperatures are another "best fit" was obtained. These were combined and generally lower. In addition, Jacksonville, Fort Myers, and used to estimate the expected seasonal temperature as a Miami probably also experienced urban effects (3). Urban function of time for each location (Fig. 3). effects produce an average warming trend of 0.0006°C/yr. (11, 15) or a total of 0.05°C from 1900 to 1985. Approxi Results and Discussion mately 30% of the stations also changed observation times from AM to PM, which can produce a cold bias of about Trend. An advective freeze, following our definition, 0.3°C for a 50 year (1931-1985) period (11). generally occurred in a "valley" — an observation lower The annual and seasonal absolute minimum tempera than its 2 neighbors (Fig. 2). Advective freezes exemplify tures are plotted in Figs. 1 and 2. The dashed line is the worst conditions in this series of minimum temperatures; mean of the period. The continuous straight line is the approximately 75% are found below the mean of annual linear least square trend. The thick line is a weighted run minima. Only 2 advective freezes occurred in November; ning average of 5 years; it was applied to smooth the data these were the two mildest of the advective freezes. Each and to remove noise. Each open circle represents a advective freeze is an indirect indication of the penetration minimum temperature (Fig. 2) which occurred in all 7 sta of the polar front through the Florida peninsula and the tions within ± 1 day of one another. This is the criterion lack of energy in the southern region to block it. Fifty- we applied to separate advective freezes from radiative seven percent of the advective freezes occurred in the 30- freezes. An advective freeze cools by active transport of a year period since 1955; 43% occurred in the 60-year cold air mass, a process which is usually accompanied by period before 1955. Thus, the occurrence of worst condi strong wind. The active influx of cold air cools the entire tions approximately doubled during the past 30 years. This peninsula within 1 to 2 days. It would have been more partially accounts for the decreasing trend line in Jackson appropriate to use wind data to identify advective freezes. ville (Figs. 1 and 2). As a group the time series show a However, historical wind data are difficult to obtain and negative trend with an average slope of -0.02°C/yr (Table probably unreliable. The alternative is this indirect 1). This amounted to a drop in temperature of 1.8°C method. The other situation, i.e. when the freeze dates (3.2°F) over 88 years. The trend line is significant at the were well separated, is more indicative of radiative freezes. 0.05 level (F-test) for Jacksonville, Clermont, and Arcadia. Several ways can be used to determine if there are cy Comparing the first 55 years (1835-1890) with the last 55 cles in the temperature record. The most obvious is to plot years (1930-1985) in the series, the average minimum tem the data. If the cycles are simple, a visual inspection will peratures have decreased from -2.7°C (27.1°F) to -5.3°C identify them. But in many cases the time series will have (22.5°F). This decrease in the means brings about a non multiple cycles, phase and length differences, and random linear increase in the number of extreme events in the tail noise. Then the series can be approximated by a sine func end of a frequency distribution (13, 19). The probability tion: of an extreme event having a minimum temperature of f(t) = (A)sin(t) + (B)sin(t) 4- —. [1] -12°C (10°F) may have been a 1 in 100 years occurrence The coefficients (A, B, —) cannot be easily determined. in the 1850s, but because of a decrease in the means, the Eq. [1] was employed to generate points for periods of 1 area under the frequency distribution is now about 8 times to a maximum of 250 years with all possible displacements as great (shaded area, Fig. 4), hence the probability of its for 6 \ocations in Florida. Each calculated point was sub- occurrence is increased 8-fold. Instead of a 1 in 100 year

Proc. Fla. State Hort. Soc. 98: 1985. 43 SERSONflL TEMPERRTURE RRCflDrfi JflCKSONVILLE

•*-

1895 1905 19J5 1925 1935 1945 1955 1965 1975 1985

TORT MYERS 995 1905 1915 1925 1935 1945 1955 1965 1975 1985

OCflLfl

1895 1905 1915 1925 1935 1945 1955 1965 1975 1985

MIRMI jg 1895 1905 1915 19 J5 1935 1945 1955 1*65 19?5 1985

CLERMONT k

1995 1905 1915 1925 1935 1945 1955 1965 1975 1985 YERRS 1895 1905 1915 1925 1975 1945 1955 1%5 1975 1985 Fig. 2. Continued. BflRTOW

JACKSONVILLE

1695 1905 1915 1925 1935 1945 1955 1965 1975 1965

Fig. 2. Seasonal minimum temperatures from 1897 to 1985, the hori zontal dash line is the mean of the observations, the continuous straight line is the linear trend, the smooth line is the 5-year weighted running Fig. 3. The fitted Jacksonville data where the model reduced the vari average. ance from 4739.947 to 3733.689. occurrence, it is now more likely to be a 1 in 12 year occurr Noise in the data is large (average SD = 2.50) and it ence. There is no convincing evidence that the worldwide masks patterns which may be present in the time series. In is changing, though indications of decreasing order to locate periods when the variance is small (less winter temperature trends were found in some parts of than 1.0), a variance was calculated for each consecutive the United States (18). We appear to be in a period of large 4-, 5-, 6-, and 7-year period over the entire time series. On fluctuations (11). There are indications that lengths of each run the calculation for each period was advanced one growing seasons have become longer in Massachusetts (1) observation along the time series. The result shows 5 and in (17). Combined with other climatic periods where the variances were less than 1.0 (triangle, studies, the work of Baron (1) and Thompson (17) imply Fig. 1). These occurred most prominently in the 4- and that winters have become shorter but more severe. Evi 5-year periods; much less distinctly in the 6- and 7-year dence of a decreasing diurnal temperature range is also periods. This suggests that the longest time a minimum present (12). temperature did not deviate by more than one standard

44 Proc. Fla. State Hort. Soc. 98: 1985. years. In all cases the use of Eq. [1] to estimate the expected minimum seasonal temperature decreased the variance from the mean of the observed variance by 25 to 35%. The 6, 11, 21, and 80 years were similar to cycles from visual observation of the plotted data. The 6-year cycle appears to be real though it cannot be explained. The 11- and 22-year cycles are most likely related to sunspot cycles. Whether the 80-year cycle is real is uncertain because the observed series is too short (88 years). The 212-year cycle is close to the cycle needed to reverse the downward trend. Statistical analyses can only be used to infer information from observations but do not explain the cause and effect of physical interactions of the atmosphere. Disturbances such as volcanic eruptions, sun spots, etc, can cause ripples and wave-like motions in the atmosphere which are dif ficult to separate for analysis. But because the atmosphere is a dynamic system with constant motion, wave-like fluctu /// 1835-1890 ations must be present for the longer term climate to re main stable. Station correlation. Since data were analyzed for only six stations they were correlated to obtain spatial distribution

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 of minimum temperatures among the stations and FREQUENCY CLHSSES H Fl generalized to other areas in the state. Nearest neighbor correlations were greater than 0.80 for areas around Bar- Fig. 4. Negatively skewed frequency distribution (thick line) of 151 tow and northward; less than 0.80 south of Bartow (Table years of observed Jacksonville annual minimum temperatures. Each fre quency class is 4°F (2.2°C). Thin lines at the lower tail are projected fre 1). Correlations of each station with Jacksonville also indi quency distribution associated with the decrease in mean temperature cate lower correlations for . Linear regression from -2.7°C (1835-1890) to -5.3°C (1930-1985). indicates an average warming of 1.7°C (3.1°F) per degree change in latitude southward (Fig. 4). From interior to coastal location the warming is likely to be 2 to 3°C (4 to T 5°F).

O Conclusion Y = 43.6 - 1.62X As a group, the time series show a decreasing trend. R = 0.89 LJ The most obvious is from the 151 annual observations from Jacksonville. We are at present, in a trough of the time series; this position is most similar to the series of LJ o: X freezes which occurred between 1894 and 1899 in both intensity and length. If past observations can be used as a QJ \ • yardstick, we are at the bottom of the trough and can ex pect some warming effects. However, because of the de creasing trend line, more extreme events with worst condi *«~ i tions than the Jan. 1985 freeze are more probable now than if there were no trend or if there were an increasing

o - trend. The zigzag pattern of the time series leads us to CD,, expect a higher minimum temperature than the previous 25.0 26.0 27.0 28.0 29.0 30.0 31.0 season. Comparing with the 1890 series of freezes, it will LRTITUDE be a few years before we can expect to emerge from the trough. Finally, if the trend is real, there is strong indica Fig. 5. Regression of mean seasonal minimum temperature on latitude (Jacksonville to Miami). tion that, even after we have emerged from the trough, minimum temperature conditions will not be as " mild" as deviation is 4 to 5 years. This occurred on the average conditions in the 1850s. One feature of the time series of once every 21.6 years and is suggestive of some cyclicity in the past few years has been the persistence of very low the observations. minimum seasonal temperatures. Usually it has been as Periodicity and modeling. The 5-year running averages sumed that each winter is independent and the chances of show that the most prominent cycles appear to be about low temperatures is a random event. The past few winters 5-7 years, 11 years, 22 years, 64 years (1835/1899/1962), called this into serious question because if the minimum and perhaps 84 years, if 1962 can be considered an ano temperatures experienced during recent freezes were even maly. A longer cycle, perhaps in the order of 200 years or a 1 in 20-year event on the average, then 2 events in a row more, is needed so that the downward trend can be re would be a 1 in 400-year event. Clearly this was counter to versed. The present observed series is not long enough to the record for Jacksonville. Our approach clearly indicates detect this longer cycle. The "best-fit" calculated cycles that the minimum winter seasonal temperatures are not were found to be 3, 6, 11, 13, 16, 21, 37-41, 83, and 212 independent, but are clustered due to causes whose inves-

Proc. Fla. State Hort. Soc. 98: 1985. 45 tigation is beyond the scope of this present study. We plan 10. Johnson, W. O. 1963. The big freeze of December 1962. Weather further studies of Florida minimum temperature records Forecasting Mimeo 63-1. Federal-State Frost Warning Service, Lake which may help to ascertain the usefulness of these land, Fla. 11. Karl, T. R., R. E. Livezey, and E. S. Epstein. 1984a. Recent unusual techniques in connection with other techniques to better mean winter temperatures across the contiguous United States. Bui. predict and understand the risk of cold weather damage Amer. Meteor. Soc. 65:1302-1309. to horticultural crops in Florida. 12. Karl, T. R., G. Kukla, and J. Gavin. 1984b. Decreasing diurnal tem perature range in the United States and Canada from 1941 through Literature Cited 1980. J. Climate and Appl. Metero. 23:1489-1504. 13. Mearns, L. O., R. W. Katz, and S. H. Schneider. 1984. Extreme high- 1. Baron, W. R., G. A. Gordon, H. W. Borns, Jr., D. C. Smith. 1984. temperature events: Changes in their probabilities with changes in Frost-free record reconstruction for eastern Massachusetts, 1733- mean temperature. J. Climate and Appl. Metero. 23:1601-1613. 1980. J. Climate and Appl. Meteor. 23:317-319. 14. Mitchell, A. J., and M. R. Ensign. 1928. The . Bui. 2. Brooks, C. E. P., and N. Carruthers. 1953. Handbook of statistical 200. Univ. Fla. Agr. Expt. Sta. methods in meteorology. AMS Press. N.Y. 15. Mitchell, J. M., Jr. 1953. On the causes of instrumentally observed 3. Cayan, D. R., and A. V. Douglas. 1984. Urban influences on surface secular temperature trends. J. Meteor. 10:244-261. temperatures in the southwestern United States during recent dec 16. Sanders, M. L. 1980. The of 1894-95 in Pinellas County. ades. J. Climate and Appl. Meteor. 23:1520-1530. Tampa Bay History. 2:5-14. 4. Climatological Data. 1897-85. Climacological Data for Florida. Na 17. Thompson, R. C, R. A. Miller, and S. H. Crawford. 1983. Climate tional Climatic Data Center, NOAA, Asheville, N.C. at the northeast research station. Louisiana Agr. Expt. Sta. Louisiana 5. Davis, T. F. 1910. The cold waves of the Florida Peninsula. A. B. State Univ. Caldwell, Tallahassee, Fla. 18. Vining, K. C, and J. F. Griffiths. 1985. Climatic variability at ten 6. Davis, T. F. 1937. Early orange culture in Florida and the epochal stations across the United States, J. Climate and Appl. Meteor. cold of 1835, Fla. Hist. Quart. 15:232-241. 24:363-370. 7. Douglas, T., 1856. Autobiography of Thomas Douglas, Calkins and 19. Wigley, T. M. L. 1985. Impact of extreme events. Nature. 316:106- Stiles, N.Y. 107. 8. Florida Citrus Mutual. 1984. Triangle. Vol. 34, No. 31. 20. Wilbur, H. S. 1929. Loyalists in . Fla. State Hist. Soc. 9. Florida Citrus Mutual. 1985. Triangle. Vol. 35, No. 31. 1:68-69.

Proc. Fla. State Hort. Soc. 98:46-48. 1985.

CITRUS TREE LOSSES FROM 1983 AND 1985 FREEZES IN FOURTEEN NORTHERN COUNTIES

Harry M. Whittaker, Mutual; Department of Citrus; Florida Citrus Processors; Agricultural Statistician Supervisor University of Florida; and several large independent grow Florida Crop and Livestock Reporting Service ers, had an even greater desire to know the extent of the 1222 Woodward Street, Orlando, Florida 32803 freeze losses. Total Florida citrus acreage as of Aug. 1985 is no more Abstract. Florida's Citrus Industry suffered the greatest tree than 642,856 acres. This is the smallest number of total losses of the century during the Dec. 1983 and Jan. 1985 acres in Florida since Dec. 1956 when there was 603,060 freezes. These freezes seriously curtailed all acres. By 1965 there were 858,082 acres and in Dec. 1969 in Florida, previously the world's largest producer of citrus there was the all time high acreage for all citrus in Florida products. These freezes also played havoc with the Crop and at 941,471. The current acreage is 24% less than was re Livestock Reporting Service's citrus forecast projections. A reli ported in the Jan. 1982 census. The 1984 and 1985 set able citrus forecast depends upon accurate assessment of tree trees were not considered in this report so as to make the age, type, and numbers. In an attempt to measure tree loss, data comparable to the 1984 Commercial Tree Inventory. this office began a survey in the spring of 1985 of the fourteen Neither time nor aerial photography was available in 1985 Upper West Coast and Upper Interior counties where damage for a complete acreage update. was most severe. Prior to these freezes there were more than The counties most severely affected by the freeze were 275,000 acres of commercial citrus in these 14 counties. Hillsborough, Pasco, Hernando, Citrus, Lake, Orange, Fieldmen visited each separate block and determined the per Sumter, , Volusia, Marion, Alachua, Putnam, centage of live trees of each age group and type of citrus. All Flagler, and St. Johns (Fig. 1). It was clear that each sepa trees set before 1984 were counted and included in this re rate variety or age block would have to be visited in order port. Many growers and caretakers helped by giving tree to provide the citrus industry with sound, reliable informa counts and variety identification. tion. We had all read reports on various surveys, observa tions, and quotes from newspapers and there was a distinct The northern portion of the Florida Citrus Belt has need for a block by block report. suffered severe acreage and tree loss as a result of catas trophic freezes in Dec. 1983 and Jan. 1985. In Mar. 1984, Methodology growers and other citrus related organizations began ask ing the Florida Crop and Livestock Reporting Service for The Florida Crop and Livestock Reporting Service has a report on the acreage lost during the first of these each block of citrus in the State of Florida identified as to freezes. In the aftermath of the 1985 freeze, these organi variety, year planted, spacing between rows, and number zations, Citrus Crop Estimates Committee; Florida Citrus of trees and acres. This information is indexed by section,

46 Proc. Fla. State Hort. Soc. 98: 1985.