ABSTRACT
      Although blessed by generally temperate weather conditions, Florida, according to the official National Weather Service records published in Storm Data. lists a total of 27 temperature-related deaths from 1979 through 1999. How close is this to the actual number? For comparison, data from Florida's Office of Vital Statictics (FOVS) were gathered during this same period, and the number of direct temperature-related deaths was listed as 249! This large discrepancy illustrates the difficulty of adequately indetifying weather-related casualties in Storm Data, especially those associated with the less violent, and therefore less newsworthy, weather-related events. The 249 deaths are more than the number of deaths during this time from more notorious weather-related hazards such as hurricanes, tornadoes and even lightning. Statistics of temperature-related deaths concerning the annual, seasonal and monthly distribution, and various demographic parameters are analyzed and interpreted. Among some of the more interesting results are the differences in age between heat and cold victims and the distribution of deaths with respect to heat and cold waves.

INTRODUCTION
      Before steps can be taken to reduce the effects of weather-related hazards, a vlunerability study needs to be accurately conducted. The National Weather Service (NWS) attempts to keep track of weather-related casualties and damage through the monthly publication, Strom Data. Input to this publication, for all but the most significant events such as hurricanes, tornadoes and floods, usually comes from second-hand sources such as newspapers. These sources have limitations which make the completeness of the casualty records questionable for events such as lightning and excessive temperature. Lopez, et al, 1993, discuss a variety of reasons why lightning deaths may not be reported in newspapers, including limited coverage, priority of news on a particular day, delayed deaths, simple omissions, etc. The same arguments can be applied to temperature-related casualties. Although overall temperature extremes in Florida are less than and of the 48 contiguous states, occasional periods of cold or hot weather, along with sociological factors such as rapidly increasing and aging population, make the state vulnerable to both heat and cold-related casualties.

DATA
Storm Data
      The NOAA publication Storm Data has been compiled at the National Climatic Data Center (NCDC) in Asheville, North Carolina since 1959. Input to the monthly publication has been from reports submitted by local NWS offices. The data includes the number of direct weather-related casualties, including their location and date, along with other factual and anecdotal information. The principal resource used by Florida NWS offices for gathering temperature-related casualty information is newspaper clippings, obtained from a statewide clipping service. Other sources include those available from the broadcast media, emergency management and law enforcement.

Florida Death Certificates
      The Public Health Statistics Section, Office of Vital Statistics, Florida Department of Health and Rehabilitative Services, is responsible for tabulating deaths that result from accidents occurring in the state of Florida. This is done either by collecting death certificate data from the 67 individual Florida counties or obtaining death certificates from other states when the accident causing a death actually occurred in Florida. Deaths are classified as to the underlying cause using the International Classification of Disease, 9th Edition from 1979-1998 and the 10th Edition for 1999. The code number for deaths associated with "excessive heat/cold due to weather conditions" in the 9th Edition is E900.0/901.0 and in the 10th Edition X30/31. Statistics on temperature-related deaths in Florida, from death certificates by county, were available for the 21-year period from 1979 through 1999. The information from FOVS included the date and county of the incident, and the gender, age and residency of the victim.

RESULTS
      Examining the time period from 1979 through 1999, Storm Data listed 27 temperature-related deaths in Florida, 6 from excessive heat and 21 from excessive cold while the FOVS data listed 249 deaths, 125 from excessive heat and 124 from excessive cold Figure 1. Figure 2 shows that the number of temperature-related deaths during this period is greater than those from hurricanes, tornadoes, and even lightning.

      The annual frequency of the FOVS temperature-related deaths is shown in Figure 3. The number of combined heat and cold deaths averaged 12 per year from 1979 through 1999, 6 from heat and 6 from cold, but the annual distribution of heat and cold deaths was quite different. For the most part the distribution of heat deaths from year to year are relatively uniform while the cold deaths are more concentrated in certain years. If the cold deaths are classified by season rather than by year (Figure 4), the distribution of cold deaths is even more concentrated in certain seasons. The winters of 1983-84, 1984-85, and 1989-90 accounted for 65 of the 124 cold deaths. Major freezes occurred in Florida during each of these winters.

      The monthly distribution of FOVS deaths is depicted in Figure 5. Every month but April has recorded a temperature-related death. All but one of the heat deaths occurred from May through September, with the most deaths, 47, occurring in the month of August. Interestingly, May had more heat deaths than did the climatologically warmer September. The cold deaths occurred in the months from November through February with December recording the most deaths, 53. The coldest month of the year in Florida is normally January, but during the period from 1979-1999 the worst cold snaps occurred in December.

      Figure 6 is a chart showing the age of the temperature-related deaths according to FOVS data. The heat and cold-related deaths show opposite age correlations to age, with the majority of heat-related deaths occurring in people 50 years or younger and the majority of cold-related deaths occurring in those older than 60.

      To try to determine any relationship between the deaths, and concentrated periods of heat and cold, the daily distribution of deaths was examined. After examining the data, the definition of a period in which multiple deaths were observed, a so called "heat wave" or "cold wave", was formulated. Heat or cold wave deaths were defined as those in which four or more persons died within a week-long period. Using this definition, according to the FOVS data during the period from 1979 through 1999 there were four heat waves and six cold waves. Out of a total of 125 heat deaths from 1979-1999, 25 occurred during heat waves while out of the 124 cold deaths, 56 occurred during cold waves.

      The gender of temperature-related deaths showed 73% were male and 27% were female with both heat and cold-related deaths nearly identical proportions. This percentage is similar to other weather-related hazards with male victims far outnumbering females. About 94% of the victims were residents of Florida with 6% residing out of state.

CONCLUSIONS AND DISCUSSION
      The 249 temperature-related deaths in Florida from 1979-1999 derived from the Florida Office of Vital Statistics dwarf the 27 deaths listed in Storm Data. The reason for this disparity likely involves the limitations of input to Storm Data for normally less publicized events. A solution to this situation could be accomplished if FOVS data were available within the two month deadline of the Storm Data publication. This might be possible in the future as FOVS data processing becomes more timely (Gary Sammett, personal communication).

      Heat deaths in Florida occur more frequently in younger people than do cold deaths. A likely explanation is that heat deaths are more often due to overexertion in the more active younger persons.

      Temperature-related deaths, both heat and cold, occur three times more often in males than in females. While more heat deaths among the males might be explained by their greater physical exertion, it is difficult to understand the large gender gap among the victims of hypothermia.

      The vast majority of victims are residents of Florida. Whether these residents live in Florida year-round or whether they are "snow birds" cannot be discerned from the available data.

      Deaths from cold more frequently occur during multi-day cold waves while deaths from heat more frequently occur one day at a time. This has important implications when trying to warn people about severe heat and cold conditions

FUTURE WORK
      The FOVS will be scrutinized in detail and compared to temperature records to try and better define specific criteria for heat and wind chill advisories in Florida.

Lopez, R.E., R.L. Holle, T.A. Heitkamp, M. Boyson, M. Cherrington, and K. Langford, 1993; The Underreporting of Lightning Injuries and Deaths in Colorado. Bull. Amer. Meteor. Soc., 74, 2171-2178.
Sammett, Gary J., 2000: Program Administrator, State Office of Vital Statistics, Jacksonville, Florida.