Understanding Your Risk and Impacts

 A Comparison of Droughts, Floods, and Hurricanes in the United States

Temporal Aspects
Fatalities
Costs and Losses
Spatial Extent

Drought is one of the most complicated and least understood of all natural hazards. Floods and hurricanes, by contrast, are easily recognized hazards, and their effects are immediately visible. However, a comparison of the three hazards reveals that although drought’s temporal and spatial characteristics set it apart, its impacts take an equally high toll.

Temporal Aspects:

Warning Time
Droughts
Drought warning is directly related to the ability to predict the occurrence of atmospheric conditions that produce the physical aspects of drought, primarily precipitation and temperature. Some areas of the world have relatively consistent precipitation, temperature, and dominant weather patterns (such as weather patterns associated with the El Niño–Southern Oscillation), and models can be developed to predict drought approximately a year in advance. However, even in these areas, there may be large regional variations and unforeseen changes that will modify predicted conditions. More often, so many variables can affect the outcome of climatic interactions that currently it is not possible to predict a drought in advance. In fact, an area may already be in a drought before drought is even recognized.

Scientists have been monitoring climatic conditions, including drought, for hundreds of years in some cases and are also finding ways to extract information from prehistoric sites. This information is being used to calculate the frequency of droughts throughout a region’s history and to identify the normal precipitation for a given period of time. Thus, even if a drought cannot be predicted, frequencies can be calculated to yield the probability of drought occurring in a specific region under certain climatic conditions. In addition, indices such as the Standardized Precipitation Index can be developed to compare current climate data to historical data in order to evaluate current precipitation conditions and identify climatic trends over time.

Drought warning is further complicated by the existence of different types of drought. A meteorological drought may be identified first because of a lack of directly measured precipitation. On the other hand, socioeconomic drought is based on the premise that demand has outpaced the supply of water. Thus, a meteorological drought does not necessarily mean that there is also a socioeconomic drought (if supply is still meeting demand).

Therefore, drought warning is based on a complex interaction of many different variables, water uses, and consumer needs. In some areas with dominant and relatively predictable climatic parameters, such as the tropics, a general drought prediction may be made about a year in advance. In most other areas, predictions cannot be guaranteed in advance, although a probability of occurrence can typically be generated along with descriptions of current precipitation and temperature trends.

Floods
Flash floods can strike in seconds (without warning), while snowmelt floods can be predicted months ahead.

Hurricanes
The range of warning capabilities is dependent on people having access to warning systems (television, radio, etc.). Otherwise, hurricane warning depends on people’s perceptions of hurricane conditions. According to the National Oceanic and Atmospheric Administration’s (NOAA) Miami Regional Library at the National Hurricane Center/Tropical Prediction Center, the existing warning system has been effective, but rapid population growth has posed some challenges to the system. Moreover, 80–90% of the population now living in the drought-prone areas have never experienced a major hurricane; they often do not have an understanding of the damage potential.¹

Duration
Droughts
Drought duration is highly variable by region. The duration also depends on when the precipitation is needed for such activities as planting and irrigation. A more detailed description of how drought length may vary by classification can be found in operational definitions of drought.

Floods
Small floods may crest and recede in hours; some floods, such as the North Dakota floods of 1997, may last for weeks.

Hurricanes
A temporal continuum of effects is associated with hurricanes. The effects can range from short-duration initial storm surges to medium-duration winds, rains, and flash flooding to longer-term flooding events.

Frequency
Droughts
The table below lists the number of years that the United States has had severe or extreme drought in the 100 years from 1896 to 1995, based on the Palmer Drought Severity Index (PDSI). The data is further divided and analyzed based on NOAA “river basins,” which do not always correspond to geographical river basins (particularly in the case of the “California” river basin division). The chart shows that some part of the United States has experienced a severe or extreme drought in each year from 1896 to 1995, and that in 72 years, droughts covered more than 10% of the country. At the river basin level, the chart shows that some part of the Missouri River Basin has experienced a severe or extreme drought in 90 of the 100 years, but the Missouri is a particularly large basin. The figure below is a graphic depiction of the data.

Floods
Bankfull stage (when water begins to overflow the stream banks) is used to define a flood. Bankfull stage typically occurs every 1 to 2 years. In general, the average recurrence interval is 1.5 years. This means that on average, discharge in a river will exceed bankfull stage 2 out of 3 years. However, the recurrence interval varies greatly depending on the shape of the channel, the amount of flow regulation on the stream, and other factors. For example, many of the incised channels in eastern Nebraska may be able to contain water from a much larger storm (a 1 in 25 or 1 in 50 year storm).

Hurricanes
On average, 1.6 hurricanes strike the mainland of the United States every year. This number was averaged from data reported by the National Center for Environmental Prediction/Tropical Prediction Center. Similarly, the Federal Emergency Management Agency (FEMA) (1995) reports that, on average, 5 hurricanes strike the United States every 3 years (1.67 per year). Based on historical data from 1900 to 1994, severe hurricanes (category 4 or 5 on the Saffir–Simpson scale) strike the United States on the average of once every 5.75 years; the maximum interval between severe hurricanes (category 4 or 5 on the Saffir–Simpson scale) for the 1900–94 time period in the United States was 20 years (1969–89) (Insurance Services Office, Inc., 1994).

Fatalities:

Annual Average
Drought
Drought in the United States seldom results directly in the loss of life. Deaths associated with drought are usually related to a heat wave or, in developing countries, a disruption in food supply leading to malnutrition and possibly famine. We have not found any references to annual averages of deaths related to these associated factors. However, information on this topic was reported by Warrick et al. (1975), who note that in the United States, deaths were linked to malnutrition in the late 1800s and early 1900s, but by the 1930s this cause of death was primarily eradicated with federal relief aid. Hurt (1981) notes that many deaths were associated with drought-related dust in the 1930s. Today, the greatest number of deaths related to drought in the United States are from heat stress.²

According to statistics from the Center for Disease Control, exposure to extreme temperatures causes an average of 300 deaths a year; during 1979–98, the United States recorded 7,421 heat-related deaths.³ The National Climatic Data Center (NCDC) estimates heat-related deaths at 5,000–10,000 for 1980 and 1988.⁴

Floods
The Corps of Engineers reported an average of 94 deaths per year due to floods in fiscal years 1986–95 (Myers, 1997; Myers notes that a review of cumulative flood fatalities since the beginning of the century by Stuart Nishenko of the United States Geological Survey indicates a similar average). A graph of yearly fatalities due to flooding in Myers (1997) shows values (from NOAA storm data) that are generally lower than the average of 94 deaths per year, which therefore may be the maximum value of reported attributable deaths.

Hurricanes
Using data from Herbert et al. (1996), the Environmental and Societal Impacts Group (ESIG) estimated hurricane-related deaths at 162 per year. However, if the value of 6,000+ deaths from the Galveston Hurricane of 1900 is omitted, the yearly average falls to about 100 deaths per year. For more recent years, from 1966 to 1995, ESIG also reports an annual casualty toll of 25, while averaging data from Herbert et al. (1996) yields a value closer to 29 deaths.

In addition, Barton and Nishenko (1997) report that, based on historical data, there is a 39% chance that a hurricane in the United States will cause 10 casualties every year. Similarly, there is a 6% yearly chance that a 1,000-casualty hurricane will strike the United States. The probability of hurricanes with these impacts (10 and 1,000 casualties) increases to 99% and 46%, respectively, for a 10-year evaluation period.

Worst Recent Event
Floods
The 1993 floods appear to be the worst recent event, based on NCDC listings of recent disasters.⁵ However, Myers (1997) notes that Puerto Rico also had a flash flood in 1985 that killed 180 people.

Hurricanes
There is some discrepancy in this category. NCDC reports that Hurricane Hugo, which occurred in 1989, caused 86 deaths (57 on the U.S. mainland, 29 in the U.S. Virgin Islands),⁶ while FEMA (1995) reports 49 deaths. In addition, NCDC reports 58 deaths associated with Hurricane Andrew in 1992, while FEMA (1995) reports only 15 deaths. Neither of these storms is listed in the top 30 of USA TODAY’s “20th century hurricanes: The deadliest”.⁷

Worst Recorded
Droughts
Although recent assessments show thousands of drought-related deaths in 1980 and 1988–89, the 1930s droughts may have yielded similar numbers. Although the total number of drought-related deaths in the 1930s apparently was not documented, Hurt (1981) refers to deaths from suffocation in dust storms, poisoning of the body from inhalation of silica dust (which reduced the body’s ability to fight other disease), acute respiratory infections, and dust pneumonia (which is similar to silicosis, contracted by quarry miners). These dust-related problems were so extensive that in the spring of 1935, the Red Cross opened 6 emergency hospitals in Colorado, Texas, and Kansas; issued 17,700 dust masks; and sent nurses to 1,631 homes, mostly for dust-related illnesses. During this same time, a measles outbreak in Kansas affected 40,000 people, with 145 deaths reported in its first 4 months, although it is not known whether this outbreak had anything to do with the conditions created by the drought (i.e., living conditions, nutritional deficiency, lack of medical care).

In terms of nutrition, Warrick et al. (1975) notes that food supply was not a problem in the 1930s and that there were no food-related health effects, although movies and books of the time have shown limitations on food variety and availability (i.e., The Grapes of Wrath). Statistics on nutrition, heat-related deaths, and other health problems were not recorded as comprehensively in the past as they are today. Thus, it is not known how many deaths occurred that could be associated with drought-related stresses before recent droughts, but the 1930s drought had great potential for high morbidity with the combined effects of physical stress, migrations, dust storms, extensive duration, lack of air conditioning, and limited medical capacities.

Floods
The Johnstown Flood of 1889, also known as the Conemaugh Calamity,⁸ claimed more than 2,200 lives.

Hurricanes
The Galveston Hurricane of 1900 is reported to be responsible for more than 6,000 deaths.⁹

Costs and Losses:

Annual Average
Droughts
FEMA has estimated that drought costs the United States $6–8 billion annually (FEMA, 1995). Warrick et al. (1975) noted that drought losses average $200 million to $1.24 billion annually in the Great Plains. This range is based on crop losses and other direct and indirect losses, as well as many self-admitted rough estimates and crude approximations.

Floods
The average annual cost of floods has been calculated at $2.41 billion (Myers, 1997). This figure represents average yearly costs and losses from 1903 to 1993. A listing of yearly losses from 1903 to 1999 can also be found in tables compiled by NOAA/National Weather Service’s (NWS) Hydrologic Information Center.¹⁰

Hurricanes
A value of $1.2 billion/year was calculated from yearly cost totals from 1900 to 1995 (in current and 1992 dollars; adjusted to inflation using a construction consumer price index) presented in a hurricane data table¹¹ by ESIG, which was derived from data by Herbert et al. (1996). However, Pielke and Landsea (1998) note that if U.S. hurricane data are normalized relative to inflation, changes in coastal population, and wealth, average yearly mainland hurricane losses are on the order of $4.8 billion/year (in 1995 dollars).

Pielke and Landsea (1998) also note that if U.S. hurricane data are normalized relative to inflation, changes in coastal population, and wealth, the United States has at least a 1 in 6 chance each year of experiencing hurricane-related losses of at least $10 billion. Barton and Nishenko (1997) also present information regarding hurricane costs throughout the twentieth century, as well as current trends.

Worst Recent Event
Droughts
NCDC estimates the cost of the 1988 drought at $40 billion.¹² In addition, Riebsame et al. (1991) estimate the total cost of the 1988 drought at $39.2 billion.

On the state level, the state of Texas estimated that the drought that struck in 1996 caused more than $5 billion in costs and losses related to agriculture.

Floods
All sources agree that 1993 was the worst recent flooding year in terms of costs. (It is also the most costly flood event on record.) However, estimates of the actual costs differ. FEMA (1995) puts the costs at $15–20 billion (in 1993 dollars). NCDC reported the same figure.¹³ ESIG calculated the cost at $16.3 billion, using NWS data.¹⁴ However, Myers (1997) estimates the costs at $27.6–276 billion, based on a review of NOAA’s storm data. The ten-fold error margin results from disaster evaluators’ practice of grouping disaster costs into categories of ten.

Hurricanes
All sources agree that Hurricane Andrew is the most costly recent hurricane event to affect the United States, with many estimates now topping $30 billion. USA TODAY, using National Hurricane Center data, estimates the cost of Andrew at $30.47 billion, in 1996 dollars.¹⁵ NCDC’s estimate is $32.4 billion (1998 dollars).¹⁶ Pielke and Landsea (1998) note that if U.S. hurricane data are normalized relative to inflation, changes in coastal population, and wealth, damage from Hurricane Andrew totaled $33.094 billion dollars (in 1995 dollars). Pielke (1997) also notes that if hurricane costs are normalized as a percentage of the gross domestic product, the 1938 New England Hurricane was nearly as costly as Hurricane Andrew.

Worst Recorded
Droughts
The Dust Bowl years of the 1930s and the drought of 1988–89 are good contenders for the worst drought on record in the United States. Past and present economic losses are often hard to calculate and compare for a variety of reasons: lack of historical records and economic models, and past and present costs that are often based on different criteria. For example, today, many different types of losses are often included in an economic analysis, such as energy losses, ecosystem losses, and consumer purchasing losses, but they were not typically included in previous analyses and are difficult to assess in retrospect.

Warrick et al. (1975) notes that the 1930s droughts were generally considered to be the most economically damaging droughts to affect the United States. Riebsame et al. (1991) put 1988 total drought costs, including losses in agriculture, energy, water, ecosystems, and other sectors of the economy, at roughly $39 billion, making it the most expensive natural disaster ever to affect the nation. To our knowledge, a definitive dollar amount has not been calculated for the 1930s drought losses.

Spatial Extent:

Annual Average, Worst Recent Event, Worst Recorded
Droughts
The National Drought Mitigation Center (NDMC) has calculated values showing the spatial extent of drought based on historical Palmer Drought Severity Index (PDSI) data. The annual average of 18.1% was calculated by selecting the month of each year from 1895 to 1995 with the greatest spatial extent of severe or extreme drought and averaging the values. Using PDSI data, the NDMC created a series of maps showing the percent of time each climate division in the United States was in severe or extreme drought, 1896–1994. The maps show the spatial extent of drought for various time periods. The worst recent drought event occurred in July 1988, with 36% of the country in severe or extreme drought. The worst drought event ever recorded occurred in July 1934, with 65% of the United States experiencing severe to extreme drought.

Worst Recent Event and Worst Recorded
Floods
The 1993 floods in the Mississippi Valley covered an extensive area—certainly the largest spatial area in recorded history.¹⁷

Notes
¹ “The United States Hurricane Problem”. Additional warning information can be obtained through the warning entry of USA TODAY’s weather section.
² More information on heat stress can be found on the National Weather Service’s Internet Weather Source web page, Heat Wave, and NOAA’s Heat Wave: A National Problem.
³ Statistics taken from data provided by the Center for Disease Control.
⁴ Billion Dollar U.S. Weather Disasters, 1980–2001.
5 Billion Dollar U.S. Weather Disasters, 1980–2001.
⁶ Billion Dollar U.S. Weather Disasters, 1980–2001.
⁷ 20th century hurricanes: The deadliest.
⁸ The Johnstown Flood contains a contemporary newspaper account of the flood and links to related sites.
⁹ This information was obtained from USA TODAY’s 20th century hurricanes: The deadliest (based on the National Hurricane Center’s data), and Barton and Nishenko (1997). For additional information on the history of deadly cyclones, see The Deadliest Atlantic Tropical Cyclones, 1492–Present by Rappaport and Fernandez-Partaga of the National Hurricane Center.
¹⁰ Flood Losses: Compilation of Flood Loss Statistics.
¹¹ Annual death and damages statistics for hurricanes.
¹² Billion Dollar U.S. Weather Disasters, 1980–2001.
¹³ Billion Dollar U.S. Weather Disasters, 1980–2001.
¹⁴ Annual death and damages statistics for floods.
¹⁵ 20th century hurricanes: The costliest.
¹⁶ Billion Dollar Weather Disasters Since 1980.
¹⁷ For an account of the 1993 floods, see Lott, J.N. 1994. The US summer of 1993: A sharp contrast in weather extremes. Weather 49(11):370–83.

References
Barton, C.; and S. Nishenko. 1997. Natural disaster: Forecasting economic and life losses. USGS Fact Sheet. U.S. Geological Survey, Marine and Coastal Geology Program, St. Petersburg, Florida.
FEMA. 1995. National Mitigation Strategy; Partnerships for Building Safer Communities. Mitigation Directorate, p. 2. Federal Emergency Management Agency, Washington, D.C.
Herbert, P.J.; J.D. Jarrell; and M. Mayfield. 1996. The Deadliest, Costliest, and Most Intense Hurricanes of this Century (and Other Frequently Requested Hurricane Facts). NOAA Technical Memorandum NWS TPC-I. NOAA, Washington, D.C.
Hurt, R.D. 1981. The Dust Bowl: An Agricultural and Social History. Nelson-Hall, Chicago.
Insurance Services Office, Inc. 1994. The Impact of Catastrophes on Property Insurance. Insurance Services Office, Inc., Jersey City, New Jersey. The executive summary of this report is online.
Myers, Mary Fran. 1997. “Trends in Floods,” Proceedings of the Workshop on the Social and Economic Impacts of Weather, Roger A. Pielke, Jr., editor. National Center for Atmospheric Research, Boulder, Colorado. pp. 77-86.
Pielke, Jr., R.A.1997. Trends in hurricane impacts in the United States. Crop Insurance Today 30(3):8–10, 18.
Pielke, R.A. Jr.; and C.W. Landsea. 1998. Normalized hurricane damages in the United States: 1925–1995. Weather and Forecasting 13:621–31.
Riebsame, W.E.; S.A. Changnon, Jr.; and T.R. Karl. 1991. Drought and Natural Resources Management in the United States: Impacts and Implications of the 1987-89 Drought. Westview Press, Boulder, Colorado.
Warrick, R.A.; P.B. Trainer; E.J. Baker; and W. Brinkman. 1975. Drought Hazard in the United States: A Research Assessment. Program on Technology, Environment and Man Monograph #NSF-RA-E-75-004, Institute of Behavioral Science, University of Colorado, Boulder.