Introduction

The long-term trends in hail incidence and hail losses across the United States, and within states and discrete regions within the nation, were examined using readily available data and results. Data on hail sufficient to be used in multi-decadal analyses comes from two sources: the records of hail from the National Weather Service (NWS), and insurance records of hail loss. In preparing this analysis, I have drawn on the available data bases.

Data. Since the 1890s, the U.S. Weather Bureau (now NWS) has recorded days when thunder was heard (thunder days) and when hail fell, identified as "hail days." This was done at the 200+ first-order stations (FOS) across the nation, and these are manned by staff who make observations around the clock. At the 10,000+ cooperative substations of the NWS manned by volunteer weather observers, the incidence of hail could be recorded, and considerable research of these data has shown that weather observers at some stations have, over the years, accurately recorded the incidence of hail. Data on hail presented in the NWS's Storm Data, since its publication began in 1955, are unfortunately badly biased and are unsuitable for temporal analyses of hail or any form of severe local storms. Thus, the hail-day data of the NWS stations, all FOS and some cooperative stations, are one major data set suitable for temporal analysis.

The other long-term records of hail are from insurance records of hail-produced loss. These come from different sources. The crop-hail insurance data have been systematically collected since 1948 by companies acting through a central association which has compiled the data and archived it. These data have limitations including the fact that not all farmers have taken insurance coverage (hail insurance is estimated to cover 25 to 30 percent of all crop losses caused by hail). Second, records of loss occur only when crops are growing and susceptible to loss, and this susceptibility to hail damage changes during the growing season and varies between crops (e.g., tobacco and tea are more susceptible to hail damage than corn). Third, crop-hail losses for a state or the nation shift with time due to the amount of coverage (liability) and the crop value, as well as the temporal variations in hail occurrences (which are large). Fortunately, the industry devised an adjustment factor named "loss cost," which is the amount of loss per year ($) divided by the annual amount of liability ($) multiplied by $100. The loss cost values for 1948 to 1996 provide a useful measure of the temporal fluctuations of insured loss.

The property insurance industry has not kept loss records for hail alone (or for any other individual form of severe damaging weather). However, since 1949, the industry, through its centralized Property Claim Services of the American Insurance Services Group, has recorded each catastrophe, defined as storm situation producing $5 million or more loss to property (not crops). For each catastrophe, an estimated amount of dollar loss is available along with the weather factors causing the damage and states where the losses occurred. The catastrophe data available since 1949 have several biases limiting their use in temporal analyses. Fortunately, one major insurance company systematically over time made major adjustments to the catastrophe data base (done each year) to adjust for the ever changing dollar value, for changes in property density-location, and changing of costs of construction. This "adjusted" catastrophe data base offers an opportunity to meaningfully examine the temporal trends in catastrophes related to hail.

Impacts of Hail

Crop-Hail Losses. The national annual values of insured crop-hail losses appear in figure 1, along with amount of liability. This shows ever increasing losses ranging from $15 million in 1948 to $129 million in 1974, then jumping to $265 million in 1980, and approaching $400 million in the early 1990s. Liability also climbed steadily in this 48-year period.

Figure 1 also presents the adjusted loss cost values, seen as the best way to examine the climatological trends in crop-hail losses. This shows relatively high values in the 1950s, early 1960s, and again in the early 1990s. After the peak centered at 1962-63, values declined slowly until the recent 3-year high in 1992-1994. The long-term average loss cost for the U.S. is $2.55, and the highest 3-year loss costs since 1947 are $3.38 in 1961-63, $3.27 in 1954-56, and $3.25 in 1992-94. From a risk standpoint, the recent peak was preceded by an 11-year period with relatively low-loss cost values, as shown in table 1. This circumstance only acted to emphasize the recent losses, but when put in a 48-year time frame, the highs in the early 1990s rate third. No statistically significant long-term trend of decrease or increase in crop-hail losses is evident. This distribution also illustrates another key aspect of hail loss found at all scales -- the county, state, region, and national level -- the losses are skewed over time with 1 to 3 years of high losses often separated by many years (5 to 15) with low losses.

Regionally, one finds startling differences in crop-hail loss trends, a not unexpected outcome since hail is so notoriously variable in both space and time. As shown in figure 2, upward trends exist in recent years in the Northern High Plains and since about 1970 in certain East Coast states (VA, NC, SC, and GA). Conversely, trends in loss costs in the Midwest and Tennessee Valley states show continuing decreases over the past 30 years.

Property Catastrophes. The adjusted catastrophe data for 1949-1995 were examined to identify only those events when hail was part of the cause of loss. Hail, as one cause of damage, was further assessed for those catastrophes when the loss was due only to hail with wind. Figure 3 presented two curves based on the pentad values for these "hail-only" catastrophes during 1950-94. One curve shows the frequency of the hail-only catastrophes, revealing a peak in 1965-69 (30 storms) and a minimum of 5 storms in 1955-59. The frequency distribution does not indicate a long-term trend upwards or downwards, particularly since 1960. The other curve on figure 3 is based on the average hail-only catastrophe loss values per pentad. This shows peaks early, in 1950-54 and 1960-64, followed by low values until higher averages re-appeared in 1980-84 and 1990-94. The values of dollar loss per storm for these 177 hail-only catastrophes suggest a slight downward trend with time.

The top 20 most damaging hail-only catastrophes during 1949-1994 are listed in table 2. This reveals three important findings. First, the distribution over time is bi-modal with 13 of the 20 events in two pentads, 1960-64 with 7 top storms, and 1990-94 with 6 events. The 1949-54 period had 3 top storms and 1965-69 had 2 events. Thus, the distribution forms an early peak and a recent one. Second, the storm losses are confined to one or two states with 17 events causing damage in 1 or 2 states, a much smaller areal extent than found with most weather catastrophes. Third, the states where the top 20 events most frequently produced property-hail losses formed a SW-NE oriented area including Texas (8 occurrences), Oklahoma (5), Kansas (4), Missouri (4), and Illinois (5 occurrences). This distribution likely reflects a combination of large hailstorm incidence with the target at risk.

Many other weather catastrophes included hail damage along with damages due to two or more conditions like tornadoes, heavy rains-flooding, high winds, lightning, etc. These cases also were analyzed and their frequencies per pentad appear in figure 4 is the U.S. population distribution with time. This and other studies of the nation's weather catastrophe data reveal that the time-related increase in catastrophic events and their losses is largely a function of the increased target-at-risk, as indicated by population as a surrogate measure of the property at risk. The insurance company's adjustment is the catastrophe data for shifting property-at-risk obviously did not capture all the societal changes affecting at risk, such as to growth of property density by location and the changing value of property.

Furthermore, study of the catastrophe data for events causing >$100 million in losses revealed (figure 5) that the greatest relative increases in catastrophes have occurred in the southeast and south where population growth has been greatest since 1950. The annual losses produced by all weather catastrophes causing >$100 million in losses were divided by U.S. population (figure 6) to obtain a population normalized time distribution. This reveals an oscillating but generally unchanging distribution with time. The five peaks are a result of major hurricanes like Hugo and Andrew.

In summary, the major insurance-based expressions of damaging hailstorms (crops and property catastrophes for hail-only events) do not suggest long-term trends up or down. They do show periods lasting from 1 to 5 years with extremely high losses, followed by longer periods of relatively low loss. upward trends in losses due to hail exist and these are largely a result of the changing dollar values, questionable construction practices and materials, and growth in the property at risk.

Hail-Day Data

A key finding revealed in the hail-day values for these three states, and in the state values of loss cost and in the regional shifts of catastrophe frequency, is that trends in hail incidence and damages vary considerably across the nation, and even within large states like Texas. Since hail is a product of thunderstorms, available temporal results on thunder-day incidences since 1900, being studied in an on-going project, were examined. The 1901-95 data from 200 FOS across the U.S., when analyzed statistically to define regions of similar temporal behavior, defined five discrete regions, each with a different time distribution. Figure 10 presents the results for these five regions that comprise the 48 contiguous states. Basically, the distribution in the western mountains and the southeastern U.S. show marked downward trends in thunderstorms since the 1920s. The incidences of thunderstorms in High Plains and Midwest show an up-then-down distribution centered on a peak in the 1940s. This is similar to the hail-day distribution in Illinois and Nebraska. The West Coast stations suggest a slight upward trend with time, whereas the stations in and adjacent to Texas show a major increase in thunderstorm activity with time, also similar to the hail-day distributions. These thunder results reflect the findings from the hail data -- different long-term trends occur in different regions of the nation.

Limitations of the Data

The property losses due to hail are not well defined and conflicting information exists. For example, in 1992 the Property Claims Service declared that "hailstorms across the country (in 1992) ran up a bill of $1.57 billion." yet, their data on all weather catastrophes shows that hail plus other conditions caused $3.9 billion in insured losses in 1992, and only one storm was a hail-only event, and it caused losses listed at $275 million. So, where did the $1.57 billion value come from? Other recent insurance publications have claimed that all losses from catastrophes listed as caused by 3, 4 or 5 weather conditions were solely due to hail, an amazing overstatement.

This points to the lack of good data on the property losses due to hail. In an economic study of hail losses done in 1975, it was shown that crop-hail losses over a 20-year period were about ten times greater than the property-hail losses. In recent years several major hail-caused property losses occurred in cities like Denver with $650 million in hail damage in July 1990, Orlando with $85 million in April 1992, Wichita with $215 million in June 1992, Oklahoma City with $200 million in April 1992, Dallas with $227 million in April 1995, and $300 million in Ft. Worth in May 1995. Recall that the insured crop-hail losses during these years were less than $400 million, and thus, these huge big-city property losses provide annual values in excess of the crop-hail losses in 1990, 1992, and 1995. This suggests that the ratio of crop to property losses has drastically shifted, at least in recent years. The results further suggest that property losses have been increasing with time due largely to the ever increasing property target. The good news is that the property insurance industry, through its Insurance Institute for Property Loss Reduction, has begun keeping records of weather-induced property losses.

Unfortunately the NWS hail data since 1900 is only for hail incidence with no other information. However, careful studies of the cooperative station data have found that some observers also reported hail sizes when storms occurred. A project will soon start to get the hail-day data for most of the nation complete for the 1901-196 period. Field studies of hail in Illinois and Colorado both agree -- most hailstones are windblown, but there is no data to analyze the temporal aspects of this condition. Inexpensive hail sensors, developed 35 years ago, could be used at many weather station locations to begin collecting data that would allow measurement of hailstone sizes, number of hailstones, and the windblown incidence of hail.

Causes of Trends

A recent study of very costly weather catastrophes, those causing >$100 million per event, revealed the time trends of these 189 storms (1949-94) were more closely associated with weather conditions, such as extra-tropical cyclone activity in the U.S. For example, figure 11 illustrates the time distribution of these costly catastrophes (many due to conditions other than hail) for 1949-94 along with the frequency of cyclones. A moderately good relationship exists and the constantly increasing population was not found to be an important factor affecting the fluctuations of these more damaging weather events. Recall, however, that only 19 of the 177 hail-only catastrophes since 1949 caused losses >$100,000. The more costly weather catastrophes are a result of hurricanes, massive outbreaks of tornadoes and associated thunderstorm conditions (including hail), major flooding events, or severe winter storms.

Societal Aspects of Weather

Workshop's Main Page

Table of Contents