Henry E. Fuelberg
Florida State University
National Weather Service
The area of warning responsibility for the Tallahassee National Weather Service office (TLH) expanded from 15 to 47 counties during the summer of 1995. This added responsibility presents numerous operational challenges. It is important for forecasters to have a working knowledge of the severe weather threat in this new CWA (County Warning Area). A climatology of severe weather and an analysis of reporting tendencies in this area are essential to understanding such a threat.
This paper discusses the seasonal and diurnal variations of tornado, severe hail, and severe wind reports in the TLH CWA, as well as the changes in diurnal variation that occur throughout the year. We then present a size/magnitude distribution of the thr ee types of reports to determine the overall strength of severe weather in this area. Finally, we consider the annual number of severe weather reports and contrast this to the annual number of severe weather days, which are defined simply as days on which at least one severe weather event was reported.
Severe weather reports from the new CWA were examined using the NSSFC (National Severe Storms Forecast Center) database. Wind and hail reports extend back to 1955 (excluding 1972), while tornado reports go back to 1950. The database was filtered by the authors so that only those reports from counties in the new warning area were included in the study.
The exclusion of only one year of wind and hail data (1972) is not a significant limitation to our long-term climatology. There are, however, several important limitations to the NSSFC database that must be considered. These limitations, mentioned in s everal previous papers (e.g., Anthony, 1994; Gaffin and Smith, 1995; Kelly et al., 1985), result from the rather subjective nature of the severe weather reports. Since this subjectivity affects the data in different ways, the various limiting factors wil l be discussed at the appropriate parts of the paper.
3. TEMPORAL DISTRIBUTION
The temporal distribution of the combined three severe weather types (Fig. 1) shows an afternoon maximum during the spring and summer that is not present during the fall and winter. In addition, when all three types are plotted together, the warm season has the greatest number of reports. This can be seen qualitatively in Fig. 1. In fact, the number of reports from April to July exceeds the total for the remaining eight months of the year (959 vs. 871).
Tornado reports show a uniform hourly distribution throughout the year, with no obvious afternoon maximum during any season (Fig. 2). The months of April and May have the most reports, though tornadoes are not uncommon in the cool season. There is also less annual variation in the overall distribution of tornado reports than in the distribution of all three types of events together (Fig. 1). This suggests that buoyancy effects associated with diurnal heating are less important to tornado-producing sto rms than to severe hail or wind storms.
Severe hail, in sharp contrast to tornadoes, shows a pronounced afternoon maximum (Fig. 3). In addition, hail is reported almost exclusively during the spring and early summer. From March through July there were 211 reports of hail, compared to only 34 f or the rest of the year.
While hail is found almost exclusively during the spring and early summer, the temporal distribution of severe wind reports (Fig. 4) appears to be a composite of the tornado and hail distributions (Figs. 2 and 3). Like hail reports, wind reports show an afternoon maximum during the warm season, with most occurring in the warm season. However, their relatively uniform hourly distribution and high frequency during the rest of the year are similar to that of the tornado distribution. This suggests that dy namical processes are necessary for both tornado and severe wind events during the cool season, when diurnal heating is less of a factor.
4. SIZE DISTRIBUTION
It is important to analyze reported magnitudes in each severe category to determine the strength of severe weather in the TLH CWA. Moreover, such an analysis reveals the subjectivity of severe reports, which is different for each category.
The highest percentage of tornadoes in the TLH CWA are in the F1 category (Fig. 5); F0 is the second most-commonly reported magnitude. This second category likely includes a number of straight-line wind reports that were mistakenly classified as tornadoes, especially during the earlier years of the period when less was known about distinguishing between tornado and wind damage. Even so, tornadoes in the TLH CWA are rather weak, with only four F4 and no F5 tornadoes reported. The largest percentage of tornadoes likely are the result of short-lived, boundary-layer (spin-up) tornadogenesis mechanisms. The extremely small number of strong/ violent tornadoes suggests rotational storms (i.e., supercells) are extremely rare in the TLH CWA.
The hail size distribution (Fig. 6), more so than the tornado distribution, suggests subjectivity in severe storm reporting. The most-commonly reported hail diameter is 4.4 cm (1.75 inches), followed by 1.9 and 2.5 cm (0.75 and 1.00 inches), respectivel y. These three distinct peaks in the distribution correspond to the most familiar objects used when reporting hail size, and agree with the distribution seen at the national level (Sammler, 1993). Dime-sized hail, i.e., hail with a diameter of 1.9 cm (3 /4 in), is the smallest size that is considered severe. The other common sizes correspond to quarter- and golfball-sized diameters.
The reported wind speed distribution (Fig. 7) also suggests subjectivity. Peak frequencies occur near 25.8, 28.4, 31.0, and 33.5 m/s, corresponding to 50, 55, 60, and 65 knots, respectively. One would expect a more even distribution if these reports we re always exact measurements from an anemometer. In reality, many reported wind speeds are rounded to the nearest 5 knots. One should note that no reported gust has exceeded 52 m/s (100 kts) and that most winds range from 25.8 to 31.0 m/s (50 to 60 knot s). These certainly are on the low end of the distribution.
5. ANNUAL FREQUENCY
A recent, dramatic increase in the annual number of reports is a notable characteristic of the TLH CWA data (Fig. 8). In contrast, Fig. 9 shows that the annual number of severe weather days has increased more gradually. For example, the six most recent years have had the greatest number of severe weather reports. This is not, however, the case for severe weather days.
The largest percentage of this recent increase has been in the number of severe wind reports. Figure 8 demonstrates that the annual number of tornado and hail reports was relatively constant throughout the period of study. Conversely, reports of wind dam age or of measured gusts greater than 26 m/s (50 kts) have increased dramatically, especially since 1988.
One possible explanation for the increase in wind reports is the attempt in recent years to distinguish between tornado and straight-line wind damage. We noted above that straight-line wind damage often is incorrectly reported as tornado damage, especia lly for weak tornadoes. The more recent data (Figs. 8-9) better reflect the far more frequent occurrence of straight-line wind damage and the rather rare occurrence of tornadoes. Such a large increase in severe wind reports certainly is not entirely lin ked to a better distinction of non-tornadic wind damage, however. The implications of this are discussed further in the next section.
Another noticeable aspect of Fig. 9 is the scarcity of days with more than one category of severe report. This again suggests that supercells, in which all three events frequently occur together, are very rare in the TLH CWA. While it is likely that the less significant events often are overlooked (i.e., hail with a tornado), the majority of severe events in the Tallahassee CWA are produced by either non-supercell organized or pulse-type severe storms.
The Tallahassee CWA is not unique in the trend toward recent increases in the annual number of severe reports. Also, since eight offices previously had warning and verification responsibility for counties in the new TLH CWA, this recent trend is not uni que to a particular office.
Hales and Kelly (1985) noted this tendency at the national level, attributing it to the initiation of warning verification by the National Weather Service in 1979. They stated that to improve their verification scores, offices either have to decrease th e number of warnings being issued, or develop new post-storm procedures for finding severe reports. Since little effort is made to look for severe events on unwarned storms (Hales, 1993), the increase in reports also reflects an increase in the number of warnings being issued. This leaves offices with only one option when attempting to improve their verification scores: to develop new post-storm verification procedures.
We have mentioned that the largest percentage of the recent increase has been in the number of severe wind reports. This reveals important changes in post-storm verification efforts. To understand these changes, it is helpful to examine the recent seve re wind reports in more detail.
There are essentially two types of severe wind reports. Type I contains easily validated data, such as the number of casualties from an event or a measured wind speed, while Type II uses either a dollar estimate or a qualitative account of storm damage. Quite often Type II reports are something equivalent to “downed trees and power lines,” though at times they contain detailed damage surveys. In either case, an increase in the number of Type II reports, which generally require more work to obtain, wou ld reflect a corresponding increase in verification efforts.
Since 1980, Type II reports have indeed been on the rise, accounting for 89% of all wind reports in the TLH CWA, compared to 77% prior to that time. Even though some of these Type II reports reflect marginal, perhaps questionable, severe events, this sti ll indicates that the National Weather Service has increased their post-storm verification efforts. Refinements in such efforts will improve public service, and more importantly, provide the ground-truth data vital for a more complete understanding of sev ere weather situations.
A climatology of severe weather and an analysis of reporting tendencies are the first steps to a better understanding of the severe weather threat in this new CWA. The results of this study also indicate several areas for future work.
One area of future research follows from what we have learned about the temporal distribution of severe weather reports for the TLH CWA. We have shown that the temporal distribution of severe wind reports (Fig. 4) appears to be a composite of the tornad o and hail distributions (Figs. 2 and 3). This suggests a forecast problem in distinguishing severe wind storms from storms that produce other types of severe events. Efforts to make this distinction more apparent would be of value to forecasters in this office.
We also determined that wind, hail, and tornado events frequently are not reported together, and that there have been very few strong/violent tornadoes in this area. We believe this indicates severe storms in our area very rarely are supercells. Theref ore, future research into the spectrum of severe storms that are not supercells would benefit forecasters in this office most directly.
Finally, we discovered that while the annual number of reports has grown very rapidly in recent years, the number of severe days has remained much more constant. This information, along with the increase in Type II wind reports, suggests the Weather Ser vice has been successful in its attempts to increase verification efforts. The focus should now be on refining these efforts to gain a more complete, more accurate database.
The authors thank John Hart and Mike Vescio (NSSFC) for their explanation of the NSSFC database and Paul Close (NWS-MEM) for assistance with Quattro Pro while he was still at the Tallahassee office. We also thank Josh Korotky, Science and Operations Off icer at NWS-TLH for valuable suggestions regarding the final form and content of this paper.
This research was funded by the Cooperative Program for Operational Meteorology, Education and Training (COMET) under subaward S94-49193 to Florida State University, and by the National Weather Service.
Anthony, R., 1994: Severe thunderstorm climatological data for the new Jacksonville, Florida county warning area. NOAA Tech. Memo. NWS SR-155. 19 pp.
Gaffin, D., and R. Smith, 1995: Severe weather climatology for the NWSFO Memphis county warning area. NOAA Tech. Memo. NWS SR-169. 15 pp.
Hales, J.E., 1993: Biases in the severe thunderstorm data base: ramifications and solutions. Preprints, 13th Conf. Wea. Forecasting and Analysis, Vienna, VA, Amer. Meteor. Soc., 504-507.
_____, J.E., and D.L. Kelly, 1985: The relationship between the collection of severe thunderstorm reports and warning verification. Preprints, 9th Conf. On Severe Local Storms, Indianapolis, Amer. Meteor. Soc., 13-16.
Kelly, D.L., J.T. Schaefer, and C.A. Doswell III, 1985: Climatology of non-tornadic severe thunderstorm events in the United States. Mon. Wea. Rev., 113, 1997-2014.
Sammler, W.R., 1993: An updated climatology of large hail based on 1970-1990 data. Preprints, 17th Conf. On Severe Local Storms, St. Louis, Amer. Meteor. Soc., 32-35.