Subpart A 
ATMOSPHERIC
HAZARDS

6 NATURAL HAZARDS

ATMOSPHERIC HAZARDS 7
Introduction
F
or purposes of this report, phenomena 
associated with certain weather-generated 
events are grouped as atmospheric 
hazards. The individual hazards 
included are: 
Tropical Cyclones
Thunderstorms and Lightning
Tornados
Windstorms
Hailstorms
Snow Avalanches
Severe Winterstorms
Extreme Summer Weather
Each of the atmospheric hazards may have 
its own natural characteristics, geographic 
area where it occurs (areal extent), time of 
year it is most likely to occur, severity, and 
associated risk. While these characteristics 
allow identification of each hazard, many 
atmospheric hazards are interrelated. In 
most cases, a natural disaster or event 
involves multiple hazards: severe thunder-
storms spawn tornados; wind is a factor in 
thunderstorms, severe winterstorms, tropical 
cyclones, and hailstorms; snowfall from a 
severe winterstorm can prompt avalanches. 
Because several atmospheric hazards may 
occur concurrently, it may be difficult to 
attribute damage to any one hazard or to 
assess the risk a particular hazard. On the 
other hand, mitigation efforts directed at a 
specific hazard, windstorms for example, 
often have beneficial effects on related 
atmospheric hazards. 
Although atmospheric hazards are presented 
separately from geologic, hydrologic, seismic, 
and other natural hazards, atmospheric 
hazards may be related to these natural 
events and often to technological hazards, as 
well: extreme summer weather contributes to 
drought; earthquakes cause snow avalanches, 
landslides, subsidence, and dam failures; 
and tropical cyclones can exacerbate coastal 
erosion and flooding. 
Photo of lightningPhoto: Red Cross 


Photo of a Tropical Cyclone
10 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
H
urricanes, tropical storms, and typhoons, collectively 
known as tropical cyclones, are among the 
most devastating naturally occurring hazards in 
the United States and its territories. Recent events reveal 
the magnitude of damage that is possible. In 1992, 
Hurricane Andrew resulted in the highest total damage of 
any natural disaster in U.S. history, an estimated $25 billion. 
Tropical Storm Alberto in 1994 caused 30 deaths and 
$500 million in damage. Puerto Rico and the U.S. Virgin 
Islands suffered more than $1 billion in damage from 
Hurricane Hugo in 1989. Hurricane Iniki in 1992 inflicted 
almost $2 billion in damage to Hawaii. During the past 20 
years, more than 75 Federal disaster declarations have 
involved tropical cyclone activity. 
More than 36 million people live in the counties along the 
Gulf of Mexico and Atlantic Ocean coast, the area of the 
conterminous United States most susceptible to tropical 
cyclones. These are also the regions with the highest 
growth rates and rising property values. The trend of 
increasing development in coastal zones magnifies the 
exposure of those areas to catastrophic losses from tropical 
cyclones. Although the Western States are less prone to 
landfall by tropical cyclones than the Atlantic and Gulf 
regions, storms in the eastern North Pacific Ocean have 
produced damaging rainfalls in California, Nevada, 
Arizona, and New Mexico. Hawaii and the Pacific territories 
are at risk from hurricane and typhoon activity year 
round. 
Computer models, improved radar technology, and historical 
data have enhanced the ability to predict the paths and 
impacts of tropical cyclones. However, predicting landfall 
locations with sufficient advance warning remains uncertain. 
A low-level tropical storm can develop into a hurricane 
within 6 to 12 hours, yet heavily developed, hurricane-
prone areas often require 24 to 36 hours to complete 
evacuations. 
Current mitigation and response efforts rely heavily on 
public awareness and development of evacuation plans. 
Adoption and enforcement of strict building codes is one 
of the more successful long-term mitigation measures for 
minimizing structural damage. Other efforts include buy-
out programs, relocation and/or elevation of structures, 
improved open-space preservation, and land-use planning 
in high-risk areas. 
PhotoPhoto: Red Cross 

TROPICAL CYCLONES 11
Photo of a damaged housePhoto: Red Cross 

12 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
A tropical cyclone is defined as a low pressure area of 
closed circulation winds that originates over tropical 
waters. Winds rotate counterclockwise in the Northern 
Hemisphere and clockwise in the Southern 
Hemisphere. 
A tropical cyclone begins as a tropical depression with 
wind speeds below 39 mph (18 m/s). As it intensifies, 
it may develop into a tropical storm, with further development 
producing a hurricane or typhoon. As a storm 
travels over land or colder waters, it eventually weak-
ens. Table 1-1 defines the classification criteria for 
tropical cyclones based on the stage of development, 
wind speed, and tropical or subtropical environment. 
In the North Atlantic and Central and South Pacific 
basins east of the International Date Line, tropical 
cyclones with wind speeds between 39 mph (19 m/s) 
and 74 mph (33 m/s) are commonly known as tropical 
storms. When wind speeds exceed 74 mph (33 m/s), 
they are commonly known as hurricanes. In the west-
ern North Pacific basin, tropical cyclones are called 
typhoons. Typhoons that attain maximum sustained 
wind speeds of 150 mph (67 m/s) or greater are classified 
as super typhoons. 
The distinguishing feature of tropical cyclones is the 
eye around which winds rotate. The eye, the storm's 
core, is an area of low barometric pressure that is generally 
10 to 30 nautical miles in diameter. The surrounding 
storm may be 100 to 500 nautical miles in 
diameter, with intense windfields in the eastern and 
northern quadrants. The eye can be seen distinctly in 
satellite and radar imagery. 
The Saffir/Simpson Hurricane Scale is used to classify 
tropical cyclones by numbered categories (Table 1-2) in 
the North Atlantic Basin, eastern and central North 
Pacific Basin, and the South Pacific basin. Hurricanes 
are classified as Categories 1 through 5 based on central 
pressure, wind speed, storm surge height, and damage 
potential. 
Tropical cyclones involve both atmospheric and hydro-
logic characteristics. Those commonly associated with 
tropical cyclones include severe winds, storm surge 
flooding, high waves, coastal erosion, extreme rainfall, 
thunderstorms, lightning, and, in some cases, tornadoes. 
These individual phenomena are addressed separately 
in separate chapters in this report. 
TABLE 1-1.—Classification criteria for tropical, subtropical, and extratropical cyclones 
Source: Modified from Neumann and others, 1993 
Stage of Development Criteria 
Tropical depression 
(development) 
The formative stages of a tropical cyclone in which the 
maximum sustained (1-min mean) surface wind speed is <39 
mph (<18m/s). 
Tropical storm A warm core tropical cyclone in which the maximum sustained 
surface wind speed (1-min mean) ranges from 39 to <74 mph 
(18 to <33m/s). 
Hurricane A warm core tropical cyclone in which the maximum sustained 
surface wind speed (1-min mean) is at least 74 mph (33 m/s). 
Tropical depression 
(dissipation) 
The decaying stages of a tropical cyclone in which the 
maximum sustained surface wind speed (1-min mean) has 
dropped below 39 mph (18 m/s). 
Extratropical cyclone Tropical cyclones modified by interaction with nontropical 
environment. There are no wind speed criteria, and maximum 
winds may exceed hurricane force. 
Subtropical depression A subtropical cyclone in which the maximum sustained surface 
wind speed (1-min mean) is below 39 mph (18 m/s). 
Subtropical storm A subtropical cyclone in which the maximum sustained surface 
wind speed (1-min mean) is at least 39 mph (18 m/s). 

TROPICAL CYCLONES 13
TABLE 1-2.—Saffir/Simpson hurricane scale ranges 
Scale Number 
(Category) 
Central Pressure 
(mbar) (in) 
Wind Speed 
(mph) 
Storm 
Surge 
(ft) 
Potential 
Damage 
1 > 980 > 28.94 74 - 95 4 - 5 Minimal 
2 965 - 979 28.50 - 28.91 96 - 110 6 - 8 Moderate 
3 945 - 964 27.91 - 28.47 111 - 130 9 - 12 Extensive 
4 920 - 944 27.17 - 27.88 131 - 155 13 - 18 Extreme 
5 < 920 < 27.17 > 155 > 18 Catastrophic 
Source: Hebert and others, 1995 
Atlantic Region. For the coastline from Texas to 
Maine, tropical cyclones develop over the warm waters 
of the Gulf of Mexico, Caribbean Sea, and Atlantic 
Ocean south of latitude 35 degrees. The cooler waters 
of the North Atlantic Ocean off the central and north-
eastern U.S. shorelines usually reduce the strength of 
storms approaching the shore. The strong steering cur-
rent of the Gulf Stream, flowing northeasterly from the 
coast of Florida past the Outer Banks of North Carolina 
and up to the maritime areas off Nova Scotia, often 
directs storms on a track away from the coast. There, 
storms transition into the extratropical phase and dissipate. 
The primary impact of a tropical cyclone during 
its rapidly moving extratropical phase is on shipping 
traffic. 
The North Atlantic and northern Pacific Ocean season 
for hurricanes and typhoons lasts from June through 
November, when sea and surface temperatures peak. 
The majority of hurricane activity in the North Atlantic 
basin occurs during August and September. For the 
Northern Hemisphere and the North Atlantic basin, 
Figure 1-1 shows the monthly distribution of land-
falling hurricanes striking the U.S. coastline from 1900 
to 1994 (Hebert and others, 1995). Table 1-3 lists the 
most intense hurricanes, categories 4 and 5, to strike 
the United States between 1900 and 1994. 
The peak typhoon season for the Southern Hemisphere 
and the South Pacific Ocean occurs from October to 
April. However, the meteorological history of the 
American Samoa region indicates that major catastrophic 
tropical cyclones can strike at any time of year. 
Figure 1-1FIGURE 1-1.—Total number of U.S. hurricanes per month: 1900-1994. 
Source: Based on data from Hebert and others 1995 

14 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
TABLE 1-3.—Most intense U.S. hurricanes at time of landfall: 1900 - 1994 
Hurricane Year Category 
Central Pressure 
(mbar) (in) 
Florida (Keys) 1935 5 892 26.35 
Camille (MS/SE LA/VA) 1969 5 909 26.84 
Andrew (SE FL/SE LA) 1992 4 922 27.23 
Florida (Keys) /S TX 1919 4 927 27.37 
Florida (SE/Lake 
Okeechobee) 
1928 4 929 27.43 
Donna (FL/Eastern U.S.) 1960 4 930 27.46 
Texas (Galveston) 1900 4 931 27.49 
Louisiana (Grand Isle) 1909 4 931 27.49 
Louisiana (New Orleans) 1915 4 931 27.49 
Carla (N & Central TX) 1961 4 931 27.49 
Hugo (SC) 1989 4 934 27.58 
Florida 
(Miami/Pensacola)/MS/AL 
1926 4 935 27.61 
Hazel (SC/NC) 1954 4 938 27.70 
SE FL/SE LA/MS 1947 4 940 27.76 
N TX 1932 4 941 27.79 
Source: Hebert and others, 1995 
For the period 1886-1994, an average of five hurricanes 
a year have occurred in the North Atlantic basin (Hebert 
and others, 1995). This region is particularly vulnerable 
because hurricanes occur frequently, the areas are 
prone to storm surge and coastal riverine flooding, and 
the population has climbed to an estimated 45 million 
people. Puerto Rico and the U.S. Virgin Islands are 
affected by both Atlantic Ocean and Caribbean Sea hurricanes. 
Pacific Region. California, Oregon, and 
Washington are less prone to landfall by tropical 
cyclones originating in the eastern North Pacific Ocean. 
From 1952 through 1971, an average of 6 tropical 
cyclones occurred each year in this area (Bryant, 1991). 
Storms tend to move away from the coast, heading 
toward Hawaii and the open ocean of the central North 
Pacific. The only effects of tropical cyclones felt in the 
eastern North Pacific Ocean have been in the form of 
extreme rainfall in California, Nevada, Arizona, and 
New Mexico. 
Tropical systems approaching the western coastal 
States tend to lose cyclonic characteristics but retain 
large areas of convection. As remnants of tropical systems 
move into the southwestern region, voluminous 
amounts of rainfall may result. Table 1-4 shows the 
two-day precipitation totals associated with tropical 
cyclones affecting the Southwestern States from 1900 
to 1984 (Smith, 1986). 
Hurricanes impact Hawaii more often than the western 
States. The island of Kauai has been struck recently by 
two major storms, Iwa (1982) and Iniki (1992). 
Although the Islands have experienced hurricanes 
throughout recorded history, significant coastal flood 
events had previously been caused by tsunami waves, 
not hurricanes. 
Historical information for hurricanes in the Central 
Pacific region was compiled by U.S. Army Corps of 
Engineers (USACE), Pacific Ocean Division, for a 
1985 hurricane vulnerability study for Honolulu, HI 
(USACE, 1985). The report cites the activity of 
Central Pacific hurricanes based on data from 1832-
1979 and updated information for the period of 1980-
1983. Before Hurricanes Iwa and Iniki struck, the 
Kohala Cyclone of August 1871 and the Mokapu 
Cyclone of August 1938 were considered the most significant 
tropical cyclones in the Central Pacific. 
South Pacific hurricane activity has always been significant 
in the vicinity of the seven islands of American 
Samoa. The likelihood of impact from a tropical storm 

TROPICAL CYCLONES 15
or hurricane places these islands at risk each year. The 
risk increases because of the ancillary hazard effects 
and damage from stream and river flooding, winds in 
excess of 125 mph (56 m/s), and landslides. 
Annual typhoon activity also is very high in the western 
North Pacific Ocean in the vicinity of Guam and the 14 
islands of the Northern Mariana Islands. The lack of a 
wide continental shelf typical in the North Atlantic 
basin and the fringing reefs around islands reduce storm 
surge elevations and wave impacts. Consequently, the 
severe winds and wind-driven rainfall have much 
greater effects on island structures and agricultural 
industries than storm surges. 
RISK ASSESSMENT 
The various hazard components and risks associated 
with tropical cyclones come from storm surge, rainfall, 
and wind. Associated damage includes: 
• Storm surge causes coastal flooding, salinization of 
land and groundwater, water supply contamination, 
agricultural losses, coastal erosion, loss of life due 
to drowning, and structural and infrastructure dam-
age; 
• Rainfall causes riverine and flash flooding, land-
slides, loss of life, and damages including structures, 
infrastructure, and agriculture; and 
• Wind impacts utilities and transportation, results in 
loss of life due to downbursts and tornadoes, creates 
tremendous amounts of debris, and causes 
agricultural losses and building damage (Bryant, 
1991). 
Probability and Frequency 
The measure of probability of occurrence for tropical 
cyclones is generally derived from the coastal flooding 
caused by storm surge. The probability or return period 
used to classify the significance of an event is deter-
mined as the percent chance of a flood elevation being 
equaled or exceeded in any given year. The coastal 
flood elevation caused by a 1-percent-annual-chance 
tropical cyclone event is commonly referred to as the 
100-year frequency flood. The 1-percent-annual-
chance flood event is established through detailed 
analyses of stage-frequency relationships of measured 
tide level data or analysis and modeling of specific historical 
tropical cyclone parameters such as direction, 
minimum central pressure, forward speed, and radius of 
maximum winds. 
TABLE 1-4.—Two-day precipitation totals, 
eastern North Pacific tropical cyclones: 1900-1984 
Station Date 
Precipitation 
(in) 
Carlsbad, NM (unofficial) September 20-21, 1941 17.00 
Workman Creek, AZ September 4-5, 1970 11.40 
Mt. Wilson, CA September 10-11, 1976 10.74 
Mt. Wilson, CA September 25-26, 1939 10.62 
Castle Hot Springs, AZ August 28-29, 1951 10.46 
Crown King, AZ August 28-29, 1951 10.44 
Greenville, NM September 21-22, 1941 8.79 
Lake Arrowhead, CA September 10-11, 1976 8.71 
Sunflower, AZ September 4-5, 1970 8.30 
Camp Hi Hill Opids, CA September 10-11, 1976 8.22 
Gladstone, CO October 4-5, 1911 8.16 
Hereford, AZ September 26-27, 1926 8.15 
Newkirk, NM September 21-22, 1941 8.15 
Alamogordo Dam, NM September 21-22, 1941 8.05 
Yates, NM September 21-22, 1941 8.00 
Source: Smith, 1986 

16 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Detailed hydraulic analyses include the establishment 
of the relationship of tide levels, wave heights, and synthetic 
generation of a storm population data base from 
hydrodynamic models such as TTSURGE and SURGE. 
The coastal flood elevation for the 1-percent-chance 
tropical cyclone will be a function of the combined 
influences of tidal rise and wave setup, height, and run-
up along the coastline. A discussion of storm surge elevations 
and probability of occurrence due to tropical 
cyclones and other storms is presented in Chapter 13. 
The frequency of occurrence of tropical cyclones can be 
determined by the number of landfall events over a 
given time period. The frequencies of landfall events 
are measured from historical data for specific geographic 
areas of the United States which have experienced 
direct or indirect hits by hurricanes or typhoons. 
In an analysis of hurricanes experienced by coastal 
county populations from Texas to Maine (Hebert and 
others, 1984), the direct and indirect hurricane landfalls 
in each county were tabulated. Storms in which the eye 
passed directly over a county were considered direct 
hits. Indirect or fringe hits were defined as the areas on 
either side of the direct landfall zone and were accounted 
for by assessing the occurrence of hurricane force 
winds and/or storm surge tides of 4 to 5 ft (1.2 to 1.5 m) 
in adjacent counties. 
In an update of the 1984 study, the assessment was 
expanded to include the number of direct hits by land-
falling hurricanes in coastal States from Texas to Maine 
from 1900 to 1994 (Hebert and others, 1995). The 
assessment was modified further using data from the 
U.S. Department of Commerce to include indirect hits 
from 1984 to 1994 (Neumann and others, 1993). As 
shown in Figure 1-2, Florida had the greatest number of 
direct hits by hurricanes since 1900, with Texas, 
Louisiana, North Carolina, South Carolina, and Rhode 
Island ranked in order behind Florida. Florida also has 
the highest incidence rate of category 3 or greater land-
falls. 
Map 1-1 shows the geographic distribution of direct 
and indirect impacts of landfalling hurricanes affecting 
coastal counties from Texas to Maine from 1900 to 
1994. Map 1-2 depicts the probability of each hurricane 
category based on events having at least a 5-percent 
chance of occurring in any given year. 
In Hawaii, coastal flood elevations from historic hurricanes 
have been combined with statistical information 
on tsunami flood inundation limits and used to establish 
1-percent-annual-chance flood elevations and associated 
wave runup. The combination includes recent significant 
hurricane events that exhibited tsunami-like 
wave bore flooding effects. 
The most recent information available on Central 
Pacific hurricanes is the USACE 1985 report. Reliable 
information about the Hawaiian Islands used was deter-
mined to include only the period 1950-1983. For this 
34-year period, there was an annual average of two to 
three tropical cyclones, either tropical storms or hurricanes. 
The highest number recorded during that period 
was 10, during the 1982 season. Records show that the 
month with the highest frequency of occurrence is 
August, although the devastation of Hurricane Iwa 
occurred in November, 1982. 
USACE found the number of hurricanes peaked during 
years in which strong El Niño conditions occurred. 
However, the El Niño, an area of warm surface water in 
the equatorial region of the eastern Pacific Ocean, is 
only one of many factors that influence hurricane activity. 
For 1950 to 1983, USACE identified 20 hurricanes 
that affected the Hawaiian Islands, but only Hurricane 
Dot in 1959 made landfall. The next major hurricane to 
landfall was Iniki in 1992, which also hit Kauai. 
According to research in the draft Survivable Crisis 
Management Plan for American Samoa (Department of 
Public Safety, 1995), the meteorological history of the 
islands indicates major hurricanes can strike throughout 
the year. The Joint Typhoon Warning Center (JTWC) 
reports that from 1981 through 1993, an average of just 
over 5 tropical cyclones developed each year in the 
South Pacific basin. However, more than 13 tropical 
cyclones were recorded in 1992 (JTWC, 1993). The 
frequency of tropicsl cyclones has not been established 
in American Samoa. 
From 1945 to 1990, 162 tropical cyclones made landfall 
or came within 180 nautical miles of the island of Guam 
(JTWC, 1991). In 1994, the JTWC reported that an 
average of 28 tropical storms developed annually in the 
western North Pacific from 1960 to 1993. An average 
of just under 18 of the 28 storms reached typhoon magnitude. 
Because of the warm tropical waters in the 
region, typhoons formed during every month of the 
year during this 33-year period. 
In 1993, JTWC reported that the western North Pacific 
Ocean experienced an above-normal tropical cyclone 
season, with 30 named storms. In addition, three super 
typhoons developed, although they did not directly 
affect the islands. In 1992, five typhoons affected 
Guam between August 28 and November 24, including 
Typhoon Omar which passed right over the island. 

TROPICAL CYCLONES 17
Figure 1-2
18 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Map 1-1Map 1-1. Total number of direct and indirect impacts from landfalling hurricanes for coastal counties 
from Texas to Maine: 1900-1994.
Source: Data from NOAA, National Weather Service, 1994

TROPICAL CYCLONES 19
Map 1-2Map 1-2. Coastal counties from Texas to Maine and the 5% chance associated with the occurrence of 
landfalling hurricane magnitude (by category) being equaled or exceeded in any given year.
Source: Data from NOAA, National Weather Service, 1994

20 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Exposure 
Hurricanes present one of the greatest potentials for 
substantial loss of life, property damage, and economic 
impact because more than 36 million U.S. residents live 
in the coastal counties from Texas to Maine that have 
the greatest exposure to hurricanes. 
More than 85 percent of coastal residents have never 
experienced the effects of a direct-hit hurricane (Hebert 
and others, 1995). The period from 1970 to 1995 experienced 
low hurricane activity and few direct hits 
occurred. Only about one-fifth (12) of the total number 
of intense hurricanes (Category 3 or higher) since 1900 
occurred during the last 25 years. Of those, only 
Hurricane Agnes in 1972 caused more than 25 deaths. 
The highest population growth rates in the United 
States have been in Gulf and Atlantic coastal counties. 
These areas have experienced an estimated 15 percent 
increase in population, more than 5 million people, 
from 1980 to 1993. For the period from 1988 to 1990, 
the value of insured residential and commercial property 
has increased an estimated 65 percent (Insurance 
Research Council, 1995). Figure 1-3 shows the 1993 
value of insured coastal property exposures by State 
(IRC, 1995). 
The nature of the Hawaiian Islands make the entire 
island chain vulnerable to damage from tropical 
cyclones and related hazards, but major tropical 
cyclones appear to be infrequent. On the whole, more 
significant damage results from tsunami waves, winter 
coastal storm waves, riverine flooding, and volcanic 
activity. Development along the coastline is highly valued, 
and damage from future tropical storms is likely to 
have significant social and economic consequences. 
American Samoa's draft Survivable Crisis Management 
Plan recognizes that devastation caused by hurricanes 
can affect all aspects of island life. Coastal flood inundation, 
severe winds, and wind-driven rain impacts residences, 
transportation, utilities, and agricultural industries. 
Hurricanes also trigger other natural hazards, 
such as riverine flooding and landslides. 
Consequences 
Statistics on the 10 deadliest U.S. hurricanes are presented 
in Table 1-5. Half of the most costly hurricanes 
(more than $500 million in damages) occurred in the 
past 25 years, with Hurricane Andrew in 1992 the most 
expensive. The 10 costliest U.S. hurricanes of the 20th 
century are summarized in Table 1-6. 
Two recent tropical cyclonic events reveal consequences 
in densely populated areas: Hurricane Andrew 
(1992) and Tropical Storm Alberto (1994). Hurricane 
Andrew resulted in the highest total damage of any natural 
disaster in U.S. history, estimated at $25 billion for 
southeastern Florida and southeastern Louisiana. 
Tropical Storm Alberto was significant because of the 
extensive inland flooding that occurred as rain fell over 
the southeastern United States, primarily in Georgia. 
Although only a tropical storm, Alberto caused 30 
deaths and $500 million in property damage. Deaths 
from Alberto are the highest total recorded for a tropical 
cyclone since 1975 (Hebert and others, 1995). 
Figure 1-3FIGURE 1-3.—Value of insured coastal property exposures by mainland State: 1993. 
Source: Insurance Research Council, 1995, from data provided by Applied Insurance Research. 

TROPICAL CYCLONES 21
TABLE 1-5.—Deadliest U.S. hurricanes: 1900-1994 
Hurricane Year Category Deaths 
Texas (Galveston) 1900 4 8,000+ 
Florida (SE/Lake Okeechobee) 1928 4 1,836 
Florida (Keys/S TX) 1919 4 6001 
New England 1938 3 600 
Florida (Keys) 1935 5 408 
Audrey (SW LA/N TX) 1957 4 390 
Northeastern U.S. 1944 3 3902 
Louisiana (Grand Isle) 1909 4 350 
Louisiana (New Orleans) 1915 4 275 
Texas (Galveston) 1915 4 275 
1 600-900 estimated deaths, including 500 lost at sea. 
2 Including 344 lost at sea. 
Source: Hebert and others, 1995 
TABLE 1-6.—Costliest U.S. hurricanes: 1900-1994 
Hurricane Year Category 
Damage 
(1990 dollars) 
Andrew (SE FL/SE LA) 1992 4 $25,000,000,000 
Hugo (SC) 1989 4 7,155,120,000 
Betsy (SE FL/SE LA) 1965 3 6,461,303,000 
Agnes (FL/NE U.S.) 1972 1 6,418,143,000 
Camille (MS/SE LA/VA) 1969 5 5,242,380,000 
Diane (NE U.S.) 1955 1 4,199,645,000 
New England 1938 3 3,593,853,000 
Frederic (AL/MS) 1979 3 3,502,942,000 
Alicia (N TX) 1983 3 2,391,854,000 
Carol (NE U.S.) 1954 3 2,370,215,000 
Source: Based on Hebert and others, 1995 
In Puerto Rico, about a dozen hurricanes have made 
landfall in the past 100 years, causing damage to buildings 
and resulting in significant coastal and riverine 
flooding. Category 5 hurricanes are expected to hit, on 
average, every 15 years. In the U.S. Virgin Islands, the 
primary impacts of hurricanes are severe winds and 
coastal flooding. 
The last major hurricane to affect Puerto Rico and the 
U.S. Virgin Islands was Hurricane Hugo, a category 4 
storm when it passed through the islands on September 
17-18, 1989. The majority of the more than $1 billion 
in damage was a result of severe winds, and nearly 
5,000 homes were destroyed. 
Previous loss of life and damage from tropical cyclones 
affecting Puerto Rico include: Tropical Storm Eloise in 
1975 with 34 fatalities and over $125 million damage; 
September 1932 San Ciprian hurricane with 300 fatalities 
and $30–$50 million damage; the September 1928 
San Felipe II hurricane with 300 fatalities and $50–$85 
million damage; and the August 1899 San Ciriaco hurricane 
and associated Arecibo River flood event with 
2,184 fatalities and $35 million in direct damage (Palm 
and Hodgson, 1993). 

22 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
TABLE 1-7.—Significant Hawaiian hurricanes of the 20th century 
Name Date 
Damage 
(1990 dollars) Deaths 
Mokapu Cyclone Aug. 19, 1938 Unknown Unknown 
Hiki Aug. 15, 1950 Unknown Unknown 
Nina Dec. 2, 1957 $900,000 4 
Dot Aug. 6, 1959 $28,000,000 0 
Iwa Nov. 23, 1982 $394,000,000 1 
Iniki Sept. 11, 1992 $1,800,000,000 4 
Source: Based on Hebert and others, 1995 
Tropical cyclone activity in the Pacific Ocean near the 
Hawaiian Islands is less severe than in the North 
Atlantic Ocean or western North Pacific Ocean. 
Nonetheless, as indicated in Table 1-7, many storms 
have affected the Hawaiian Islands since 1900 (Hebert 
and others, 1995). In 1992, Hurricane Iniki's landfall 
on the south shore of Kauai was by far the most destructive 
hurricane to hit the islands, causing an estimated 
$1.8 billion in damage from both coastal flooding and 
severe winds. The cost of damage along the coast was 
very high due to high property values and the cost of 
construction. 
Damage from recent devastating hurricanes affecting 
American Samoa includes: 
• Hurricane Esau in 1981 caused $1.5 million in 
damage to public facilities, $3.2 million in agricultural 
losses and $0.68 million to private structures; 
• Hurricane Tusi in 1987 damaged more than 300 
structures; 
• Hurricane Ofa in 1990, with waves up to 50 ft (15 
m) in the open ocean, caused $7.7 million damage 
to government facilities, $15 million in damage to 
Ofu Harbor, and total estimated damage of $32 mil-
lion, including coastal roads and utilities on Tutuila 
and Ofu; and 
• Hurricane Val in 1991 caused one fatality, 200 
injuries, and damage in excess of $50 million to 
structures, utilities, and agricultural crops. 
According to the Joint Typhoon Warning Center, 
Typhoon Omar in the vicinity of Guam in 1992 had sustained 
winds of up to 121 mph (54 m/s) and inflicted 
$457 million in damage (JTWC, 1993). Storm surge 
levels were estimated to be 10 ft (3 m) above normal 
high tide, and rainfall of 12 to 19 in (30 to 48 cm) fell 
over 3 days (Coch, 1995). Typhoon Omar's damages 
included the destruction of 2,158 structures. 
From June 1975 to May 1995, more than 76 Federal 
disaster declarations resulted from coastal storm surge 
and severe winds associated with tropical storms, hurricanes, 
and typhoons. Four declarations were issued for 
Florida. New York, Texas, and American Samoa each 
had three. Disaster declarations usually include all 
associated secondary atmospheric and hydrologic hazard 
impacts of inland flooding and high winds, tornadoes, 
heavy rain, thunderstorms, lightning, and coastal 
erosion. 
Secondary economic impacts have been noted in recent 
hurricane disasters in Florida and Georgia, far beyond 
the immediately apparent damage. Agriculture and 
tourism in south Florida were disrupted long after 
Hurricane Andrew's landfall. It is difficult to estimate 
the magnitude of losses to these industries. Some businesses 
relocated from the impacted area to relatively 
safer locations in the central and northern parts of the 
State. This left many skilled workers either without 
employment or faced with relocation, increasing the 
extent of secondary economic losses. 
RESEARCH, DATA COLLECTION, 
AND MONITORING ACTIVITIES 
Key research efforts on tropical cyclone activity, characterization, 
and evolution have been undertaken by the 
National Oceanic and Atmospheric Administration 
(NOAA) and its divisions: National Hurricane Center 
(NHC), Atlantic Meteorological and Oceanographic 
Laboratory (AMOL), and National Weather Service 
(NWS). University researchers have conducted extensive 
work on, and monitoring of, significant storm 
events. 
NOAA has documented the history and tracks of tropical 
cyclones occurring since 1871. Cyclone prediction 
based on available climactic data, atmospheric conditions, 
remote-sensing data collection, and other monitoring 
programs and computer models has progressed 

TROPICAL CYCLONES 23
tremendously since the 1950s. NOAA's ability to predict 
landfall locations to provide adequate advance 
warning for evacuation is still maturing. Tropical 
depressions can develop into hurricanes within 6 to 12 
hours, and changes in track may occur quickly. 
However, some heavily developed hurricane-prone 
areas require 24 to 36 hours to complete evacuations. 
Dr. William Gray, Colorado State University, studies 
tropical cyclones and accompanying atmospheric and 
other weather patterns. He has found direct correlations 
between the level of hurricane activity in the North 
Atlantic basin and influencing factors, including equatorial 
upper level wind patterns, El Niño, sea surface 
temperatures and pressures, and rainfall patterns in the 
Saheel desert region of Northwest Africa. 
Dr. Gray's historical research indicates distinct, 10-year 
periods of weather and hurricane activity. The number 
and locations of hurricanes that are category 3 and higher 
can be attributed to unique combinations of the influencing 
factors. However, this analysis only serves as a 
tool to understand a pattern of past weather behavior, it 
is not a means to make accurate predictions of the 
severity or location of hurricanes. 
The U.S. Army Corps of Engineers and the National 
Hurricane Center use the Sea Lake Overland Surge 
from Hurricanes (SLOSH) model to help FEMA and 
affected States develop specific evacuation plans for 
urban centers along the Atlantic and Gulf Coast shore-
lines. SLOSH accounts for different hurricane categories 
and key combinations of hurricane parameters, 
including central pressure, forward speed, radius of 
maximum winds, and angle of approach. The storm 
surge heights from the models are geographically 
dependent due to the influence of offshore bathymetry 
and nearshore topography. 
The National Hurricane Center uses other models to 
predict and forecast the movement and intensity of hurricanes 
in the North Atlantic basin. The computer models 
include those prepared by AMOL in Key Biscayne, 
FL, and a new model developed by the Geophysical 
Fluid Dynamics Laboratory (GFDL) in Princeton, NJ. 
The GFDL model incorporates new mathematical equations 
of known physical properties of the atmosphere 
and sea surface and incorporates storm variable para-
meters for upper level steering currents, windspeed, air 
and water temperature, and barometric pressure. The 
GFDL model has improved the forecasting capability of 
the National Hurricane Center. 
For the Pacific Ocean, the Joint Typhoon Warning 
Center (JTWC) reports that Next Generation Radar 
(NEXRAD) has been implemented to provide Doppler 
weather radar capabilities (JTWC, 1993). The JTWC 
and its U.S. Air Force satellite reconnaissance component 
are working to improve capabilities in new technology 
such as the Meteorological Imagery Data 
Display and Analysis System. The Naval Research 
Laboratory has begun work on an addition to the 
Automated Tropical Cyclone Forecast System to 
improve forecasting capabilities. 
MITIGATION APPROACHES 
In recent years, loss of life from tropical cyclones has 
been reduced due to two activities: public awareness 
campaigns to educate residents about storm preparedness, 
and development of evacuation plans and actual 
evacuations of high-risk areas during emergencies. 
Regional evacuation planning efforts proved successful 
in moving residents from the paths of hurricanes in 
south Florida during Hurricane Andrew (1992) and in 
South Carolina during Hurricane Hugo (1989). 
However, continued awareness and public education 
programs are required to educate new residents who 
may be unaware of the hurricane threat in high hazard 
coastal areas. 
Mitigation of building damage has been most successful 
where strict building codes for high-wind influence 
areas and designated special flood hazard areas have 
been adopted and enforced by local governments, and 
complied with by builders. Coastal setback and regulatory 
programs have helped limit encroachment by some 
developments near high-risk areas, especially where 
erosion and wave impacts are anticipated. 
During the past 25 years, intense development occurred 
so rapidly along the many reaches of the coast that 
building codes and regulatory programs may not have 
been in place. Without codes, communities did not 
have adequate mechanism to control the type and nature 
of structures or construction techniques needed to resist 
wind and water forces associated with tropical 
cyclones. 
For the most part, buildings constructed prior to adoption 
of building codes remain more susceptible to dam-
age. Some retrofit projects, for example specially 
designed shutters and windows for public schools, are 
expected to reduce future damage. Modification of 
existing buildings to incorporate hurricane-resistant 
measures may come about slowly as buildings are substantially 
improved. 
Post-disaster mitigation efforts include buyout pro-
grams, relocation, elevation of structures, improved 
open-space preservation, and land-use planning within 

24 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
high-risk areas. In many areas of the coast, utility lines 
and critical transportation routes may have to be relocated 
to protect against damage resulting from tropical 
cyclones. 
RECOMMENDATIONS 
In April 1995, the Insurance Research Council (IRC) 
and Insurance Institute for Property Loss Reduction 
(IIPLR) published a report entitled Coastal Exposure 
and Community Protection: Hurricane Andrew's 
Legacy (IRC, 1995). This report summarized a number 
of global and specific recommendations from the 
Southern Building Code Congress International 
(SBCCI) field team investigations in south Florida following 
Hurricane Andrew's landfall in 1992. An inter-
disciplinary group agreed on 15 objectives and strategies 
in a series of symposia in 1993 that would "facilitate 
mitigation of natural hazards and foster a cooperative 
approach among groups that can affect change, 
including government regulators, code officials and 
code-making bodies, planners, civil engineers, and 
insurers" (IRC, 1995). The recommendations for 
improvements were primarily for the south Florida 
region, but are applicable to all hurricane-prone areas. 
BIBLIOGRAPHY AND REFERENCES 
Anders, F., S. Kimball, and R. Dolan. 1989. Map. 
Coastal Hazards: National Atlas of the United States. 
Map. U.S. Geological Survey. 
Byrant, E. 1991. Natural Disasters. New York, NY: 
Cambridge University Press. 
Coch, N.K. 1995. GEOHAZARDS: Natural and 
Human. Upper Saddle River, NJ: Prentice Hall. 
Culliton, T. J., M.A. Warren, T.R. Goodspeed, D.G. 
Remer, C.M. Blackwell, and J.J. McDonough, III. 
1990. The Second Report of a Coastal Trends Series: 
50 Years of Population Change Along the Nation's 
Coasts, 1960-2010. Rockville, MD: National Oceanic 
and Atmospheric Administration. 
Daniels, R.C., V.N. Gornitz, A.J. Mehta, S.C. Lee, and 
R.M. Cushman. 1992. Adapting to Sea-Level Rise in 
the U.S. Southeast: The Influence of Built 
Infrastructure and Biophysical. Oak Ridge National 
Laboratory, Environmental Sciences Division, 
Publication No. 3915. 
Department of Public Safety, TEMCO. 1995. 
Survivable Crisis Management (SCM) Plan for 
American Samoa (draft). Pago Pago, American 
Samoa. 
Doehring, F., I.W. Duedall, and J.M. Williams. 1994. 
Florida Hurricanes and Tropical Storms 1871—1993: 
An Historical Survey. Technical Paper 71, Florida Sea 
Grant College Program, University of Florida. 
Federal Emergency Management Agency, Federal 
Insurance Administration. 1994. Mitigation of Flood 
and Erosion Damage to Residential Buildings in 
Coastal Areas. Washington DC. 
Federal Emergency Management Agency, Federal 
Insurance Administration. 1991. Projected Impact of 
Relative Sea Level Rise on the National Flood 
Insurance Program. 
Finkl, C. W., Jr. (ed.). 1994. Journal of Coastal 
Research, Special Issue No. 12. Coastal Hazards: 
Perception, Susceptibility and Mitigation. The Coastal 
Education and Research Foundation, Inc. 
Golden, J.H., and J.T. Snow. 1991. "Mitigation Against 
Extreme Windstorms." Reviews of Geophysics. 
American Geophysical Union. Vol. 29, No. 4, pp. 477-
504. 
Good, J.W., and S.S. Ridlington. 1992. Coastal 
Natural Hazards: Science Engineering and Public 
Policy. Oregon Sea Grant. Corvallis, OR: Oregon State 
University. 
Gornitz, V.N., R.C. Daniels, T.W. White and K.R. 
Birdwell. 1994. "The Development of a Coastal Risk 
Assessment Database: Vulnerability to Sea-Level Rise 
in the U.S. Southeast." Journal of Coastal Research, 
Special Issue No. 12. Coastal Hazards: Perception, 
Susceptibility and Mitigation. The Coastal Education 
and Research Foundation, Inc. 
Gornitz, V.M., T.W. White, and R.C. Daniels (ed.). 
1994. A Coastal Hazards Data Base for the U.S. Gulf 
Coast. Oak Ridge Laboratory, Environmental Sciences 
Division, Publication No. 4101. 
Gornitz, V.M., and T.W. White. 1992. A Coastal 
Hazards Data Base for the U.S. East Coast. Oak 
Ridge Laboratory, Environmental Sciences Division, 
Publication No. 3913. 
Hebert, P.J., J.D. Jerrel, and M. Mayfield. 1995. "The 
Deadliest, Costliest, and Most Intense United States 
Hurricanes of This Century (And Other Frequently 
Requested Hurricane Facts)." Proceedings for 17th 
Annual National Hurricane Conference, April 11-14, 
1995, Atlantic City, New Jersey. 

TROPICAL CYCLONES 25
Hebert, P.J., G. Taylor, and R.A. Case. 1984. 
Hurricane Experience Levels of Coastal County 
Populations—Texas to Maine. Technical 
Memorandum NWS NHC 24. Miami, FL: National 
Hurricane Center. 
Insurance Research Council and Insurance Institute for 
Property Loss Reduction. 1995. Coastal Exposure and 
Community Protection: Hurricane Andrew's Legacy. 
Wheaton, IL. 
Joint Typhoon Warning Center. 1993. Annual Tropical 
Cyclone Report. U.S. Naval Oceanography Command 
Center, Joint Typhoon Warning Center, Guam, Mariana 
Islands. 
Joint Typhoon Warning Center. 1991. Tropical 
Cyclones Affecting Guam (1671-1990). U.S. Naval 
Oceanography Command Center, Joint Typhoon 
Warning Center, Guam, Mariana Islands. 
NOAA/JTWC Tech Note 91-2. 
National Academy of Sciences. 1994. Natural 
Disaster Studies Volume Six, Hurricane Hugo: Puerto 
Rico, the U.S. Virgin Islands, and South Carolina— 
September 17-22, 1989. Washington, DC: National 
Academy of Sciences/National Research Council, 
Committee on Natural Disasters, Board on Natural 
Disasters, Commission on Engineering and Technical 
Systems, Commission on Geosciences, Environment, 
and Resources. 
National Committee on Property Insurance. 1989. 
Surviving the Storm: Building Codes, Compliance, 
and the Mitigation of Hurricane Damage. Oak Brook, 
IL: All-Industry Research Advisory Council. 
NOAA, National Weather Service. 1993. Natural 
Disaster Survey Report: Hurricane Andrew—South 
Florida and Louisiana—August 23-26, 1992. 
NOAA, National Weather Service. 1993. Natural 
Disaster Survey Report: Hurricane Iniki—September 
6-13, 1992. 
NOAA, National Weather Service. 1994. 
Hurricanes...The Greatest Storms on Earth. 
Neumann, C.J., B.R. Jarvinen, C.J. McAdie, and J.D. 
Elms. 1993. Supplements (1994) to Tropical Cyclones 
of the North Atlantic Ocean—1871 to 1992. National 
Oceanographic and Atmospheric Administration, 
National Climactic Data Center, Ashville, NC, and 
National Hurricane Center, Coral Gables, FL. 
Palm, R.I., and M.E. Hodgson. 1993. Natural Hazards 
in Puerto Rico: Attitudes, Experience, and Behavior 
of Homeowners. Monograph No. 55. Boulder, CO: 
Institute of Behavioral Science, University of Colorado. 
Smith, W. 1986. The Effects of Eastern North Pacific 
Tropical Cyclones on the Southwestern United States. 
NOAA Technical Memorandum NWS WR-197. 
Tucson, AZ: Department of Atmospheric Sciences, 
University of Arizona. 
U.S. Army Corps of Engineers. 1985. Hurricane 
Vulnerability Study for Honolulu, Hawaii, and 
Vicinity, Volume 1—Hazard Analysis. USACE, 
Pacific Ocean Division, Floodplain Management 
Section, Planning Branch. 
Photo

Photo of Lightning
28 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
T
he National Weather Service (NWS) estimates that 
over 100,000 thunderstorms occur each year on the 
U.S. mainland. Approximately 10 percent are classified 
as "severe." Thunderstorms can produce deadly and 
damaging tornadoes, hailstorms, intense downburst and 
microburst winds, lightning, and flash floods. 
Thunderstorms spawn as many as 1,000 tornadoes each 
year. Since 1975, severe thunderstorms were involved in 
327 Federal disaster declarations. 
During the past decade, more than 15,000 lightning-
induced fires resulted in widespread property damage and 
the loss of 2 million acres of forest. On average, 89 people 
are killed by lightning each year, and another 300 are 
injured. Flashloods from thunderstorms, the number one 
cause of deaths associated with thunderstorms, claim more 
than 140 lives in the United States each year. In 1993 
alone, thunderstorm winds caused 23 fatalities and $348.7 
million in property damage, while lightning caused 43 
deaths and damage estimated at $32.5 million. 
Florida has the greatest number of thunderstorms and the 
central region of the State has the highest density of lightning 
strikes in the mainland United States. The Western 
States, around the junction of Arizona, Utah, and Nevada, 
experience the longest duration thunderstorms. The 
lengthier, less frequent thunderstorms in this region can 
create as great a hazard as the more frequent, but shorter 
duration storms in Florida. 
Mitigation efforts are directed at the components of thunderstorms: 
tornadoes, hailstorms, windstorms, lightning, 
and flash floods. The modernization of the NWS weather 
monitoring systems and computer models should increase 
the warning time available to alert local emergency officials 
and citizens. 
Thunderstorms and lightning are underrated killer events 
experienced in nearly every region of the mainland United 
States. Thunderstorms do not pose a hazard in the U.S. territories 
in the Pacific Ocean. Data are not available for 
Alaska, Hawaii, Puerto Rico, or the U.S. Virgin Islands. 
Although individual storms have only a relatively small 
impact area, throughout the world as many as 1,800 thunderstorms 
can occur at a time. 

THUNDERSTORMS AND LIGHTNING 29
Photo of LightningPhoto: Red Cross 

30 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
Thunderstorm and lightning events are generated by 
atmospheric imbalance and turbulence due to the combination 
of conditions: 
• Unstable warm air rising rapidly into the atmosphere; 
• Sufficient moisture to form clouds and rain; and 
• Upward lift of air currents caused by colliding weather 
fronts (cold and warm), sea breezes, or mountains. 
Thunderstorms, sometimes referred to as "thunder 
events," are recorded and observed as soon as a peal of 
thunder is heard by an observer at a NWS first-order 
weather station. A thunder event is composed of lightning 
and rainfall, and can intensify into a severe thunderstorm 
with damaging hail, high winds, tornadoes, 
and flash flooding. 
The duration of a thunder event is determined by measuring 
the time between the first peal of thunder and the 
last. The last peal of thunder is defined to be that which 
is followed by a period of at least 15 minutes without an 
additional peal. A "thunder day" is defined as any day 
in which at least one thunder peal is heard. 
Downburst winds are strong, concentrated, straight-line 
winds created by falling rain and sinking air that can 
reach speeds of 125 mph (200 km/h). The combination 
induces a strong downdraft of wind due to aerodynamic 
drag forces or evaporation processes (Golden and 
Snow, 1991). Microburst winds are more concentrated 
than downbursts, with speeds up to 150 mph (240 
km/h). Severe damage can result from the spreading 
out of downbursts and microbursts, which generally last 
5 to 7 minutes. Due to wind shear and detection difficulties, 
they pose the biggest threat to aircraft departures 
and landings. 
Lightning, which occurs during all thunderstorms, can 
strike anywhere. Generated by the buildup of charged 
ions in a thundercloud, the discharge of a lightning bolt 
interacts with the best conducting object or surface on 
the ground. The air in the channel of a lightning strike 
reaches temperatures higher than 50,000ºF. The rapid 
heating and cooling of the air near the channel causes a 
shock wave which produces thunder (NOAA, 1994). 
The NWS classifies a thunderstorm as severe if its 
winds reach or exceed 58 mph (km/h), produces a tornado, 
or drops surface hail at least 0.75 in (1.91 cm) in 
diameter (Golden and Snow, 1991). 
Compared with other atmospheric hazards such as tropical 
cyclones and winter low pressure systems, individual 
thunderstorms affect relatively small geographic 
areas. The average thunderstorm system is approximately 
15 mi (24 km) in diameter and typically lasts 
less than 30 minutes at a single location. However, 
weather monitoring reports indicate that coherent thunderstorm 
systems can travel intact for distances in 
excess of 600 mi (1,000 km). 
RISK ASSESSMENT 
Dangerous and damaging aspects of a severe thunder-
storm, other than tornadoes and hail, are lightning 
strikes, flash flooding, and the winds associated with 
downbursts and microbursts. Detailed information is 
presented separately on tornadoes (Chapter 3), hail-
storms (Chapter 5), and flash flooding (Chapter 12). 
Probability and Frequency 
The probability of a severe thunderstorm occurring in a 
specific region depends on certain atmospheric and climatic 
conditions. Duration and frequency can be used 
as indicators of potential severity. Damage from lightning 
strikes will likely increase with longer duration 
and more frequent thunderstorm occurrence. 
Therefore, the geographic areas with a high density of 
lightning strikes, measured in units of flashes per 
square kilometer, are at a greater risk for damage or 
potential loss of life during a thunder event. 
The likelihood of a severe thunderstorm occurring 
increases as the average duration and number of thunder 
events increase. Therefore, data collection and 
combined review of these aspects provide information 
to assess the areal extent and frequency of the hazard. 
NWS collected data for thunder days, number and duration 
of thunder events, and lightning strike density for 
the 30-year period from 1948 to 1977. A series of maps 
was generated showing the annual average thunder 
event duration (Map 2-1), the annual average number of 
thunder events (Map 2-2), and the mean annual density 
of lightning strikes (Map 2-3). Together, these maps 
indicate the areal extent and frequency of occurrence of 
thunderstorm and lightning hazards across the mainland 
United States (Changnon, 1988). 
Thunderstorm Frequency. Maps 2-1 and 2-2 
indicate that two areas of the United States are subject 
to the most damaging thunderstorms, but have different 
influencing characteristics. South Florida has the greatest 
number of thunderstorms, with an annual average of 
100 to 130, and an average duration of 80 to 100 min-

THUNDERSTORMS AND LIGHTNING 31
U.S. Map 2-1Map 2-1. Thunderstorm hazard severity based on the annual average duration of thunder events from 
1949 - 1977. Data not available for Alaska, Hawaii, Puerto Rico, U.S. Virgin Islands, and 
Pacific Territories. 
Source: Data from Changnon, 1988. 
U.S. Map 2-2Map 2-2. Thunderstorm hazard severity based on the annual average number of thunder events from 
1948 - 1977. Data not available for Alaska, Hawaii, Puerto Rico, U.S. Virgin Islands, and 
Pacific Territories. 
Source: Data from Changnon, 1988. 

32 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
U.S. Map 2.3Map 2-3. Areal extent and severity of lightning hazard based on mean annual lightning strike 
density: 1948 - 1977. Data not available for Alaska, Hawaii, Puerto Rico, U.S. Virgin 
Islands, and Pacific Territories. 
Source: Data from MacGorman and others, 1984. 
utes. The area around the junction of Arizona, Utah, 
and Nevada has an annual average of 30 to 50 thunder 
events, and the average duration is 110 to 130 minutes. 
The longer duration, less frequent thunderstorms in this 
region can create hazards comparable to those experienced 
in Florida. 
The severity of thunderstorm activity in the desert 
region is influenced by the fact that the annual peak 
occurrence takes place during a shorter period of time, 
3 months during the late summer. In Florida, the season 
lasts from early summer to late fall, due to the warmer 
tropical climate and resulting unstable atmospheric conditions 
favorable for thunderstorm development. 
Although severe thunderstorms occur less frequently in 
the Midwest, the period of activity is not well-defined. 
Significant activity occurs during different months, but 
mostly from spring until early winter (Changnon, 
1988). The unstable air masses and collision of developing 
cold and warm fronts during the summer and fall 
appear to be the cause of activity throughout the 
Midwestern States. 
Lightning Frequency. The lightning hazard component 
of thunderstorms has been documented by 
researchers at the NOAA National Climatic Data 
Center (NCDC), who record the mean annual ground 
flash density (flashes per square kilometer). Review of 
these data shows that the central Florida region has over 
18 flashes/km², the highest density in the U.S. mainland 
(Map 2-3). Southern Alabama has the next highest 
strike density with over 16 flashes/km². Northeastern 
New Mexico and northern Arizona have isolated high 
density areas with over 14 flashes/km². 
Florida has a higher lightning strike density and more 
frequent occurrence of thunderstorms, therefore the risk 
of damaging impacts and loss of life are expected to be 
greatest. Data were not available for Alaska, Hawaii, 
Puerto Rico, or the U.S. Virgin Islands, and the Pacific 
Territories. 
Exposure 
People and property in virtually the entire United States 
are exposed to damage, injury, and loss of life from 
thunderstorms and related hazards such as lightning, 
severe windstorms, hail, tornadoes, and flash floods. 
Everywhere they occur, thunderstorms are responsible 
for significant structural damage to buildings, forest 
and wildfires, downed power lines and trees, and loss of 
life. Damage similar to that caused by tornadoes and 
other cyclonic windstorms can result from severe thunderstorm 
downbursts and microburst winds. As many 
as 1,000 tornadoes each year grow out of thunderstorms 
(Golden and Snow, 1991). 

THUNDERSTORMS AND LIGHTNING 33
Consequences 
NOAA reports that thunderstorm winds were responsible 
for 23 fatalities in 1993, and associated lightning 
strikes caused 43 deaths (NOAA, 1994). For the same 
year, damage from thunderstorm winds amounted to 
$348.7 million, while lightning caused $32.5 million in 
damage. 
According to NOAA, from 1963 to 1993, the average 
loss of life due to lightning was 89 per year, with an 
additional 300 persons injured each year. Most lightning-
related deaths and injuries occurred when people 
were outdoors during summer afternoons and evenings. 
The total number of deaths by State from 1959 to 1993 
is shown on Map 2-4. 
Significant airplane disasters often are associated with 
thunderstorms and lightning. Crashes in 1982 in 
Kenner, LA, and in 1985 in Dallas/Ft. Worth, TX, were 
attributed to thunderstorm downbursts. In 1963, a plane 
struck by lightning near Elkton, MD, killed 38 people. 
Flash flooding from thunderstorms cause more than 140 
fatalities each year and is the primary cause of death 
from thunderstorm events. Most fatalities occur when 
people become trapped in automobiles (NOAA, 1994). 
During the past decade, more than 15,000 lightning-
induced fires nationwide resulted in widespread property 
damage and the loss of 2 million acres of forest. 
Severe thunderstorms were involved in 327 Federal disaster 
declarations from 1975 to 1995. Tornadoes 
spawned from severe thunderstorms accounted for 106, 
while several declarations cited thunderstorms and 
associated phenomena: rain (26), high winds (22), and 
flash flooding (17). 
RESEARCH, DATA COLLECTION, 
AND MONITORING ACTIVITIES 
Thunderstorm wind speeds were included in research 
conducted by Twisdale and Vickery (1993). The 100-
year return period thunderstorm wind speeds were predicted 
for nine locations in the Central and Southern 
U.S. Map 2-4Map 2-4. Total deaths caused by lightning: 1959 - 1993. 
Data not available for Puerto Rico, U.S. Virgin Islands, and Pacific Territories. 
Source: Data from U.S. Department of Commerce, NOAA, 1993. 

34 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
United States, based on daily peak gust data for each 
site obtained from the National Climatic Data Center in 
Asheville, NC. The 100-year return periods for highest 
wind speeds were derived from stochastic models, 
which predict thunderstorm wind velocities of 66 to 84 
mph (106 km/h to 135 km/h) in the Central United 
States. Each station analyzed showed the highest wind 
speed on record to be associated with a thunderstorm 
event. 
The National Weather Service has undertaken modernization 
of weather observation through implementation 
of NEXRAD systems. The various climatology monitoring 
programs that allow forecasting of thunderstorms 
are fairly well-established in each weather-related 
agency. The forecast centers have been integrated into 
warning systems for their respective areas. In conjunction 
with existing Doppler radar weather stations, 
NEXRAD will improve forecaster capability to predict 
the development of, and to detect, severe thunderstorms 
and associated strong winds, hail, lightning, and tornadoes. 
MITIGATION APPROACHES 
There are no clearly defined mitigation approaches 
designed specifically for thunderstorms that are separate 
from the associated hazard phenomena. Mitigation 
measures for tornadoes, hailstorms, windstorms, and 
flash flooding can be expected to achieve a reduction in 
damage caused by or associated with thunderstorms. 
Proven techniques are available to reduce lightning 
damage by grounding techniques for buildings. 
RECOMMENDATIONS 
The National Research Council (NRC) report on wind 
hazards, Wind and The Built Environment: U.S. Needs 
in Wind Engineering and Hazard Mitigation, includes 
recommendations applicable to mitigating damage 
caused by or associated with thunderstorms and lightning 
(NRC, 1993). 
BIBLIOGRAPHY AND REFERENCES 
Byrant, E. 1991. Natural Disasters. New York, NY: 
Cambridge University Press. 
Changnon, Jr, S.A. 1988. "Climatology of Thunder 
Events in the Conterminous U.S., Part I: Temporal 
Aspects" and "Part II: Spatial Aspects." Journal of 
Climate. Vol. 1, No. 4, pp. 389-405. 
Golden, J.H., and J.T. Snow. 1991. "Mitigation 
Against Extreme Windstorms." Reviews of Geophysics. 
American Geophysical Union. Vol. 29, No. 4, pp. 477-
504. 
Kitsmiller, D.H., W.E. McGovern, and R.E. Saffle. 
1992. The NEXRAD Severe Weather Potential 
Algorithm. NOAA Technical Memorandum, NWS 
TDL 81. Silver Spring, MD: Techniques Development 
Laboratory. 
MacGorman, D.R., M.W. Maier, and W.D. Rust. 1984. 
Lightning Strike Density for the Contiguous United 
States from Thunderstorm Duration Records. NOAA, 
Environmental Research Laboratories, National Severe 
Storms Laboratory. 
National Research Council. 1993. Wind and the Built 
Environment: U.S. Needs in Wind Engineering and 
Hazard Mitigation. Washington, D.C.: National 
Academy Press. 
NOAA, National Weather Service. 1994. 
Thunderstorms and Lightning: The Underrated 
Killers. U.S. Department of Commerce, NOAA, NWS. 
Twisdale, L.A., and P.J. Vickery. 1993. "Analysis of 
Thunderstorm Occurrences and Windspeed Statistics." 
In: Proceedings of the 7th U.S. National Conference 
on Wind Engineering. Los Angeles, CA: University 
of California. 

THUNDERSTORMS AND LIGHTNING 35


Photo of a tornado
38 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
A
pproximately 1,000 tornadoes each year are 
spawned by severe thunderstorms. Although 
most tornadoes remain aloft, those that touch 
ground are forces of destruction. Tornadoes have been 
known to lift and move huge objects, destroy or move 
whole buildings long distances, and siphon large volumes 
from bodies of water. Over the past 20 years, 106 Federal 
disaster declarations included damage associated with tornadoes. 
Tornado Alley, portions of Texas, Oklahoma, Arkansas, 
Missouri, and Kansas, is the most susceptible area of the 
United States. Texas alone averaged 128 tornadoes and 11 
tornado-related deaths a year over the 40-year period ending 
in 1993. 
Tornadoes follow the path of least resistance. People living 
in valleys, which normally are the most highly developed 
areas, have the greatest exposure. When a tornado 
warning is issued, local officials typically notify residents 
with radio and television announcements and alarm systems. 
Many tornado-prone areas have public shelters, and 
residents often have specially constructed shelter areas in 
their homes. 
Other hazards that accompany weather systems that produce 
tornadoes include rainstorms, windstorms, large hail, 
and lightning. 

TORNADOES 39
Photo of a tornadoPhoto: Severe Storm Lab 

40 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
A tornado is a rapidly rotating vortex or funnel of air 
extending groundward from a cumulonimbus cloud. 
Most of the time, vortices remain suspended in the 
atmosphere (Golden and Snow, 1991). When the lower 
tip of a vortex touches earth, the tornado becomes a 
force of destruction. Approximately 1,000 tornadoes 
are spawned by severe thunderstorms each year. 
Tornado damage severity is measured by the Fujita 
Tornado Scale. The Fujita Scale assigns numerical values 
based on wind speeds and categorizes tornadoes 
from 0 to 5. The letter "F" often precedes the numerical 
value. Scale values above F5 are not used because 
wind speeds above 318 mph (513 km/h) are unlikely. 
Table 3-1 shows the Fujita Scale values, wind speeds, 
descriptions of damage, and average annual number of 
tornadoes for the period 1953-1989. 
Tornadoes are related to larger vortex formations, and 
therefore often form in convective cells such as thunderstorms 
or in the right forward quadrant of a hurricane, 
far from the hurricane eye. The strength and 
number of tornadoes are not related to the strength of 
the hurricane that generates them. Often, the weakest 
hurricanes produce the most tornadoes (Bryant, 1991). 
In addition to hurricanes, events such as earthquake-
induced fires and fires from atomic bombs or wildfires 
may produce tornadoes. 
TABLE 3-1.—Fujita tornado scale 
Scale 
Value 
Wind Speed* Range and Description of Damage 
F0 40-72 mph (17.8-32.6 m/s): Light damage. Some damage to chimneys; tree 
branches broken off; shallow-rooted trees pushed over; sign boards damaged. 
Average number per year, 1953-1989: 218 (29 percent). 
F1 73-112 mph (32.7-50.3 m/s): Moderate damage. The lower limit is the 
beginning of hurricane wind speed. Roof surfaces peeled off; mobile homes 
pushed off foundations or overturned; moving autos pushed off roads. Average 
number per year, 1953-1989: 301 (40 percent). 
F2 113-157 mph (50.4-70.3 m/s): Considerable damage. Roofs torn off from 
houses; mobile homes demolished; boxcars pushed over; large trees snapped or 
uprooted; light-object missiles generated. Average number per year, 1953-
1989: 175 (23 percent). 
F3 158-206 mph (70.4-91.9 m/s): Severe damage. Roofs and some walls torn off 
well-constructed houses; trains overturned; most trees in forest uprooted; heavy 
cars lifted off ground and thrown. Average number per year, 1953-1989: 43 
(6 percent). 
F4 207-260 mph (92.0-116.6 m/s): Devastating damage. Well-constructed houses 
leveled; structures with weak foundations blown off some distance; cars 
thrown; large missiles generated. Average number per year, 1953-1989: 10 
(1 percent). 
F5 261-318 mph (116.7-142.5 m/s): Incredible damage. Strong frame houses 
lifted off foundations and carried considerable distances to disintegrate; 
automobile-sized missiles fly through the air in excess of 100 yards; trees 
debarked. Average number per year, 1953-1989: 1 (0.002 percent). 
* Wind speeds in the range are defined by Fujita to be 
“the fastest 1/4-mile wind.” 
Sources: From Golden and Snow, 1991: NOAA, NWS Natural Disaster Survey Report, 1991. 

TORNADOES 41
The path width of a single tornado generally is less than 
0.6 mi (l km). The path length of a single tornado can 
range from a few hundred meters to dozens of kilometers. 
A tornado typically moves at speeds between 30 
and 125 mph (50 and 200 km/h) and can generate internal 
winds exceeding 300 mph (500 km/h). However, 
the lifespan of a tornado rarely is longer than 30 minutes. 
A tornado event occurs when a single atmospheric 
condition such as a thunderstorm or hurricane generates 
more than one tornado. Multiple tornadoes generally 
are the result of many thunderstorms embedded in 
one large extratropical cyclone or mesoscale convective 
complex (Golden and Snow, 1991). 
RISK ASSESSMENT 
The NWS National Severe Storms Forecast Center in 
Kansas City, MO, provides information on severe thunderstorms 
and tornadoes to the general public, news 
media, emergency managers, and law enforcement personnel. 
The Center uses the latest Doppler radar, wind 
profilers, and the networks of automated surface 
observing systems (ASOS) across the United States to 
assist in the prediction and identification process for 
severe thunderstorm and tornado watches and warnings. 
A tornado watch is issued for a specific location when 
thunderstorms capable of producing tornadoes are recognized 
and arrival is expected in a few hours. A tornado 
warning is issued when tornadoes are spotted or 
when Doppler radar identifies a distinctive "hook-
shaped" area within a local partition of a thunderstorm 
line that is likely to form a tornado. 
When a tornado watch or warning is issued, local tornado 
spotters, emergency response organizations, and 
ham radio operators are placed on alert to assist in identifying 
and locating possible tornadoes. When a tornado 
is detected, emergency operations personnel and law 
enforcement agencies are alerted immediately. 
Warnings are broadcast to the public on radio, television, 
and alarm systems. Emergency managers and 
local law enforcement officials sound sirens to notify 
those who have not already received the information by 
television, radio, or visual sighting. 
Education about tornado hazards continues to be 
emphasized for schoolchildren in all grades. Residents 
in tornado-prone areas such as the Southern and 
U.S. Map 3-1Map 3-1. Average annual number of tornadoes per State from 1953 - 1993. 
Data not available for Puerto Rico, U.S. Virgin Islands, and Pacific Territories. 
Source: Data from National Oceanographic and Atmospheric Administration, 1993. 

42 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Photo of a tornadoMidwestern States are aware of the dangers and many 
have underground or specially constructed shelters in 
their homes. Employees and occupants of unsafe or 
unprotected facilities are routinely instructed on procedures 
for safe evacuation. 
Probability and Frequency 
Map 3-1 shows the average number of tornadoes that 
occurred each year from 1953 to 1993. Texas experienced 
the highest average annual number of tornadoes 
with 128, followed by Oklahoma (52), Kansas (47), 
Florida (46), and Nebraska (38). Hurricanes Carla 
(1960), Beulah (1967), and Gilbert (1980) produced 26, 
115, and 40 tornadoes, respectively. 
The occurrence of F4 and F5 tornadoes may at times be 
underestimated by as much as a factor of five in some 
areas of the United States: F5 tornadoes may be listed 
as F4 tornadoes, F4 tornadoes may be listed as F3 tornadoes, 
and so on (Twisdale, 1978). This can occur 
because F3 tornadoes have an F3 intensity for approximately 
35 percent of their duration, and an intensity of 
less than F3 for the remainder. Similarly, F4 and F5 tornadoes 
endure at those intensities for 24 percent and 19 
percent of their lifespan, respectively (Ramsdell and 
Andrews, 1986). 
There has been considerable research on the relation-
ship between tornado dimensions and tornado intensity: 
the length, width, and area of the tornado track compared 
to the probability of being exceeded and strike 
probability. The arithmetic averages of length, width, 
area, and strike probability of all tornadoes during the 
period 1954-1983 have been developed (Ramsdell and 
Andrews, 1986). These data are available and can been 
used to assess the risk of tornado hazards. 
Exposure 
An area covering portions of Texas, Oklahoma, 
Arkansas, Missouri, and Kansas is known as Tornado 
Alley, where the average annual number of tornadoes is 
the highest in the United States. Tornadoes occur in this 
area for several reasons. Cold air from the north collides 
with warm air from the Gulf of Mexico, creating a 
temperature differential on the order of 20–30ºC. The 
flat terrain enhances rapid movement of air, while the 
high humidity of the Gulf Stream further induces instability 
in the atmosphere. Most tornadoes in this area 
occur during the spring (Bryant 1991). People, buildings, 
and infrastructure located in Tornado Alley have 
the highest exposure to this hazard. 
People living in manufactured or mobile homes are 
most exposed to damage from tornadoes. Even if 
anchored, mobile homes do not withstand high wind 
speeds as well as some permanent, site-built structures. 
Ramsdell and Andrews (NOAA, 1986) placed tornado 
data for the period of 1954–1983 into 1-degree (longitude 
and latitude) squares on a map of the United States 
(Map 3-2). Combining this information with data on 
the estimated population living within each 1-degree 
area allows assessment of the relative degree of expo-
sure. 
Consequences 
Tornadoes have been known to lift and move objects 
weighing more than 300 tons a distance of 30 ft (10 m), 
toss homes more than 300 ft (100 m) from their foundations, 
and siphon millions of tons of water from water 
bodies. Tornadoes generate a tremendous amount of 
debris, which often becomes airborne shrapnel that 
causes additional damage. Tornadoes are almost 
always accompanied by heavy precipitation (Bryant, 
1991). 

TORNADOES 43
U.S. Map 3-2
44 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
TABLE 3-2.—Ten most deadly tornado events: 1870-1979 
Date 
Name or Location 
of Event 
Number of 
Tornadoes Deaths 
Number of 
States 
Affected 
Estimated 
Damage 
(in millions) 
March 18, 1925 Tri-State 7 740 6 $18 
April 5-6, 1936 Tupelo-Gainesville 17 446 5 $18 
February 19, 1884 Enigma 60 420 8 $3 
March 21-22, 1932 Northern Alabama 33 334 7 $5 
April 3-4, 1974 Super 148 315 13 N/A 
April 24-25, 1908 Louisiana-Georgia 18 310 5 $1 
May 27, 1896 St. Louis, MO 18 306 3 $15 
April 11-12, 1965 Palm Sunday 51 256 6 $200 
March 21-22, 1952 Dierks, AR 28 204 4 $15 
March 23, 1913 Easter Sunday 8 181 3 $4 
Source: J. Galway, Weatherwise Magazine, 1981 
The largest recorded tornado event occurred on April 
3–4, 1974, during which 148 tornadoes across 13 States 
resulted in 315 deaths. Table 3-2 summarizes damage 
caused by the 10 deadliest tornadoes occurring between 
1870 and 1979. Table 3-3 provides similar information 
for tornado events since 1979. 
From 1916 to 1950, 5,204 tornadoes occurred in the 
United States, resulting in the deaths of 7,961 people, 
an average of 234 deaths per year (Bryant, 1991). From 
1953-1989, tornadoes claimed 3,550 lives, an average 
of 96 deaths per year. From 1961-1990, the highest 
number of tornado-related deaths occurred in 
Mississippi, Indiana, Ohio, Alabama, and Arkansas 
(Golden and Snow, 1991). According to NWS, 39 tornado-
related deaths occurred in 1992, and 33 occurred 
in 1993. 
Map 3-3 shows the average annual number of deaths by 
State from 1953 to 1993. During that period, Texas had 
TABLE 3-3.—Six most deadly recent tornado events: 1980-1994 
Date 
Name or Location 
of Event 
Number of 
Tornadoes Deaths 
Number 
of States 
Affected 
Estimated 
Damage 
(in millions) 
May 31, 1985 Pennsylvania-Ohio 41 75 3 $450 
March 28, 1984 Carolinas 22 57 2 $200 
March 27, 1994 Southeastern U.S. 2 42 2 $107 
November 21-23, 1992 Houston to Raleigh 
and Gulf Coast to 
Ohio Valley 
94 26 13 $291 
April 26-27, 1991 Wichita/Andover, KS 54 21 6 $277 
October 3, 1992 Tampa Bay Area 3 4 1 $100 
Source: From NOAA, NWS Natural Disaster Survey Reports, 1985-1994. 

TORNADOES 45
U.S. Map 3-3Map 3-3. Average annual deaths by State caused by tornadoes: 1953-1993. 
Data not available for Puerto Rico, U.S. Virgin Islands, and Pacific Territories. 
Source: Data from National Oceanic Atmospheric Administration, 1993. 
the highest annual average, 11 deaths, followed by 
Mississippi with 8 deaths. Four States had an annual 
average of five deaths during that period, and four others 
had an annual average of four deaths. 
Between 1975 and 1995, 106 major Federal disaster 
declarations included impacts caused by tornadoes. 
The States with the greatest number of tornado-related 
disasters were: Mississippi (14); Alabama and Illinois 
(9 each); Oklahoma (8); Wisconsin (7); Ohio (6); and 
Missouri, Minnesota, Louisiana, Georgia, and Arkansas 
(5 each). 
RESEARCH, DATA COLLECTION, AND 
MONITORING ACTIVITIES 
The NWS evaluates each major tornado to determine 
the accuracy of its predictions and identifications based 
on weather data obtained from radar and other sources, 
local tornado spotters, emergency operations personnel, 
law enforcement agencies, and the general public. The 
NWS goal is to improve its ability to warn affected populations. 
Local governments and the news media routinely 
evaluate their capabilities and to identify areas 
where improvements are needed in the warning 
process. 
The research community surveys damaged areas to 
understand how structures perform. The purposes of 
this research are to reduce damage and save lives during 
future tornadoes, and to transfer information to critical 
sources. 
Based on post-event analyses, photogrammetric determinations 
of the motion of objects caught up in tornadoes 
documented on film and videotape, and measurement 
of speeds of objects using Doppler radar, it has 
been determined that the maximum wind speed in a tornado 
occurs 100 to 150 ft (30 to 50 m) above the ground 
(Golden and Snow, 1991). From their points of origin, 
54 percent of tornadoes travel northeasterly, 22 percent 
travel easterly, 22 percent travel southeasterly, 8 percent 
travel northerly, 2 percent northwesterly, and 1 percent 
travel southerly, westerly, and southwesterly, respectively 
(Twisdale, 1978). 
Dr. Theodore Fujita, professor of meteorology at the 
University of Chicago, conducted research on the complicated 
structure of tornado winds. He first documented 
the "suction spots" phenomenon from an aerial dam-
age survey of the 1965 Palm Sunday event. In addition 
to suction vortices, other asymmetric flow configurations 
have been documented (Golden and Snow, 1991). 

46 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Experts believe suction vortices may help explain why 
one structure is destroyed while an adjacent structure is 
basically untouched. 
Within the research community, there is no agreement 
regarding what causes the destruction of buildings. 
Some experts believe the difference in atmospheric 
pressure inside and outside causes buildings to explode. 
Others believe destruction is caused by wind-induced 
forces that tear structures apart from the outside. 
MITIGATION APPROACHES 
Mitigation opportunities for tornado winds are similar 
to mitigation measures for other wind hazards. 
However, the damage associated with violent tornadoes 
due to extreme wind speeds and pressures may be difficult 
to mitigate in a cost-effective manner. Attention to 
the type of structure used in tornado-prone areas may 
yield benefits, particularly by avoiding highly susceptible 
manufactured or mobile homes. 
The greatest protection is afforded by quality construction 
and reinforcement of walls, floors, and ceilings. 
Proper anchoring of walls to foundations and roofs to 
walls is essential for a building to withstand certain 
wind speeds. In tornado damage studies, the wind engineering 
research community has found considerable 
variability in construction quality and material (Golden 
and Snow 1991). Code adoption by local jurisdictions, 
compliance by builders, and local government inspection 
of new homes could reduce the risk of destruction 
in tornado-prone areas. 
Loss of life and injuries may be reduced if more individuals 
seek shelter in basements, small interior rooms, 
or hallways, and avoid rooms and buildings with large 
roof spans. An interior room or closet has less chance 
of collapse or failure than other areas. A reinforced, in-
residence tornado shelter may be constructed for 
approximately $6,000 (1995 value), while the cost of 
retrofitting an interior room or closet in an existing 
house would be approximately $3,000 (Golden and 
Snow, 1991). 
NOAA Natural Disaster Survey Reports highlight the 
need for local emergency managers, law enforcement 
officials, and local tornado spotters to provide information 
regarding tornado touchdowns. This allows NWS 
to quantify its radar findings and improve subsequent 
tornado watches and warnings. 
RECOMMENDATIONS 
Recommendations for reducing life safety risks associated 
with tornado events are identified in NOAA’s 
Natural Disaster Reports, including: 
• Improve radio and wire communications with the 
media and local emergency managers; 
• Equip gathering places with weather radios with an 
audible alert of warning and require testing of 
response and preparedness plans; 
• Continue awareness and preparedness efforts in 
schools; and 
• Make special efforts to inform mobile homes residents 
about the impacts of the tornado hazard as well 
as locations of safe shelters in times of emergency. 
BIBLIOGRAPHY AND REFERENCES 
Angel, W. 1993. National Summary of Tornadoes, 
1993. NOAA, National Environmental Satellite Data 
and Information Service, National Climatic Data 
Center. 
Byrant, E. 1991. Natural Disasters. New York, NY: 
Cambridge University Press. 
Coch, N.K. 1995. GEOHAZARDS: Natural and 
Human. Upper Saddle River, NJ: Prentice Hall. 
Galway, J.G. June 1981. "Ten Famous Tornado 
Outbreaks." Weatherwise Magazine. 
Golden, J.H., and J.T. Snow. 1991. "Mitigation Against 
Extreme Windstorms." Reviews of Geophysics. 
American Geophysical Union. Vol. 29, No. 4, pp. 477-
504. 
National Weather Service. 1994. Natural Disaster 
Survey Report: Southeastern United States Palm 
Sunday Tornado Outbreak of March 27, 1994. U.S. 
Department of Commerce, National Oceanic and 
Atmospheric Administration. 
National Weather Service. 1993. Natural Disaster 
Survey Report: The Widespread November 21-23, 
1992, Tornado Outbreak—Houston to Raleigh and 
Gulf Coast to Ohio Valley. U.S. Department of 
Commerce, National Oceanic and Atmospheric 
Administration. 

TORNADOES 47
National Weather Service. 1993. Natural Disaster 
Survey Report: Tampa Bay Tornadoes—October 3, 
1992. U.S. Department of Commerce, National 
Oceanic and Atmospheric Administration. 
National Weather Service. 1991. Natural Disaster 
Survey Report: Wichita/Andover, KS, Tornado, April 
26, 1991. U.S. Department of Commerce, National 
Oceanic and Atmospheric Administration. 
National Weather Service. 1991. Natural Disaster 
Survey Report: The Plainfield/Crest Hill Tornado— 
Northern Illinois—August 28, 1990. U.S. Department 
of Commerce, National Oceanic and Atmospheric 
Administration. 
National Weather Service. 1992. Tornadoes: Nature's 
Most Violent Storms. U.S. Department of Commerce, 
National Oceanic and Atmospheric Administration. 
Ramsdell, J.V., and G.L. Andrews. 1986. Tornado 
Climatology of the Contiguous United States. U.S. 
Department of Commerce, Pacific Northwest 
Laboratory, prepared for U.S. Nuclear Regulatory 
Commission. 
Twisdale, L.A., and W.L. Dunn. 1983. "Wind Loading 
from Multivortex Tornadoes." Journal of Structural 
Engineering. American Society of Civil Engineers. 
Vol. 109, No. 8, pp. 2016-2022. 
Twisdale, L.A., and W.L. Dunn. 1983. "Probabilistic 
Analysis of Tornado Wind Risks." Journal of 
Structural Engineering. American Society of Civil 
Engineers. Vol. 109, No. 2, pp. 468-488. 
Twisdale, L.A. 1978. "Tornado Data Characterization 
and Windspeed Risk." Journal of the Structural 
Division. American Society of Civil Engineers. 
Vol. 104, No. ST10, pp. 1611-1630. 


Photo of a windstorm
50 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
E
xtreme winds other than tornados are experienced 
in all regions of the United States and its territories. 
Areas experiencing the highest wind speeds are 
coastal regions from Texas to Maine, under the influence 
of North Atlantic Ocean and Gulf of Mexico hurricane-
related windstorms, and the Alaskan coast, under the influence 
of winter low-pressure systems in the Gulf of Alaska 
and North Pacific Ocean. Isolated wind phenomena in the 
Western States, such as the Chinook and Santa Ana winds, 
occur very locally along mountainous terrains. 
It is difficult to separate the various wind components that 
cause damage from other wind-related natural events that 
often occur with or generate windstorms. For example, 
hurricanes with intense winds often spawn numerous tornados 
or generate severe thunderstorms producing strong, 
localized downdrafts. 
Windstorms and wind-related events caused 63 fatalities in 
1993. Florida suffered the highest number of deaths, 10. 
From 1981 to 1990, the insurance industry spent nearly 
$23 billion on wind-related catastrophic events. Over the 
past 20 years, 193 Federal disaster declarations involved 
wind-induced damage. 
A project developed by the Wind Engineering Research 
Center at the Texas Tech University in Lubbock, TX and 
the Insurance Institute for Property Loss Reduction in 
Wheaton, IL created a new wind-resistance classification 
system for buildings (IIPLR, 1994). It allows owners to 
identify corrective actions needed to make buildings safer, 
and helps officials to pinpoint buildings for evacuation 
during windstorms. 
Improved building codes, retrofitting, and land use are 
some mitigation approaches used to limit exposure to 
windstorms. Recently, the National Research Council 
urged that a National Wind Science and Engineering 
Program be implemented to revitalize wind-hazard 
research (NRC, 1993). 

WINDSTORMS 51
Photo of a collapsed house.Photo: Red Cross 

52 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
Wind is defined as the motion of air relative to the 
earth's surface (Golden and Snow, 1991). The horizontal 
component of the three-dimensional flow and the 
near-surface wind phenomenon are the most significant 
aspects of the hazard. Extreme windstorm events are 
associated with extratropical and tropical cyclones, winter 
cyclones, and severe thunderstorms and accompanying 
mesoscale offspring such as tornados and down-
bursts (Golden and Snow, 1991). Winds vary from zero 
at ground level to 200 mph (89 m/s) in the upper atmospheric 
jet stream at 6 to 8 mi (10 to 13 km) above the 
earth's surface. 
The damaging effects of windstorms associated with 
hurricanes may extend for distances in excess of 100 mi 
(160 km) from the center of storm activity. Isolated 
wind phenomena in the mountainous western regions 
have more localized effects. 
In the mainland United States, the mean annual wind 
speed is reported to be 8 to 12 mph (4 to 5 m/s), with 
frequent speeds of 50 mph (22 m/s) and occasional wind 
speeds of greater than 70 mph (31 m/s). For coastal 
areas from Texas to Maine, tropical cyclone winds may 
exceed 100 mph (45 m/s). 
Large-scale extreme wind phenomena are experienced 
over every region of the United States and its territories. 
Additional wind hazards occur on a very localized level 
due to downslope windstorms along mountainous terrains, 
such as the Chinook winds along the eastern slope 
of the Rocky Mountains in Montana, New Mexico, 
Colorado, and Wyoming, and the Santa Ana winds of 
southern California. These regional phenomena, known 
as foehn-type winds, result in winds exceeding 100 mph 
(45 m/s), but they are of short duration and affect a relatively 
small geographic area. 
The Santa Ana winds only impact southern California. 
They generally occur during the late summer to early 
winter and are generated when the passage of dry, cold 
weather frontal systems is followed by a high pressure 
system developing over the Utah-Nevada area. The 
western flow of the cooler air mass loses moisture as it 
is forced to the west and funneled down through mountain 
passes along the western slopes of the mountains. 
With the rapid temperature increase as the winds 
descend, 5.5ºF for every 1,000 ft or 304 m of descent, 
the Santa Ana winds become very dry, hot, and fast. 
The Chinook winds along the eastern slopes of the 
Rocky Mountains occur during the winter as a result of 
large atmospheric movement over mountain ridges. 
The downslope winds along the eastern side gather 
strong near-surface speeds, with record wind gusts up to 
140 mph (63 m/s) measured in Boulder, Colorado. 
Other notable mountain windstorm events, such as 
Oregon's Columbia River Gorge winds and Utah's 
Wasatch Mountain winds, occur during cold fronts, 
when cold air masses funnel down through canyons. 
Severe thunderstorms also produce wind downbursts 
and microbursts, as well as tornados. Severe wind-
storms result in as many as 1,000 tornados annually 
(Golden and Snow, 1991). NWS uses the NEXRAD 
Severe Weather Potential algorithm as an automated 
procedure to detect severe local storms and to forecast 
the potential for tornados, hail, and heavy rainfall 
(Kitzmiller, McGovern, and Saffle, 1992). 
It is difficult to separate the various wind components 
that cause damage during a windstorm. For example, 
hurricanes have high winds rotating around the eye of 
the storm, spawn numerous tornados, and generate 
severe thunderstorms producing strong, localized down-
drafts (downbursts and microbursts) (Golden and Snow, 
1991). 
RISK ASSESSMENT 
Probability and Frequency 
Wind climatology and geographic distribution of wind-
speed variations are commonly associated with unique 
regional climatic and meteorological characteristics and 
seasonal patterns (Map 4-1). Coastal regions from 
Texas to North Carolina may experience intense winds 
from hurricanes and tropical storms. Most of Florida is 
susceptible to winds in excess of 100 mph (45 m/s) on a 
regular basis. 
Wind climatology depicted on Map 4-1 shows the 
fastest mile wind speeds expected to be encountered 
with a return period interval of 50 years, an annual probability 
of 0.002. Certain regions of the United States 
have been identified and specially designated because 
wind speeds faster than 70 mph (31 m/s) occur more 
frequently than the 50-year return period, or there are 
special high wind features such as downslope winds 
(western United States) or lake-effect winds (Great 
Lakes region). 
Exposure 
Areas experiencing the highest wind speeds are coastal 
regions from Texas to Maine, under the influence of 
North Atlantic Ocean and Gulf of Mexico windstorms 
associated with tropical cyclones, and the Alaskan 

WINDSTORMS 53
U.S. Map 4-1
54 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
coast, under the influence of winter low-pressure-system 
windstorms in the Gulf of Alaska and North Pacific 
Ocean. 
Property damage and loss of life from windstorms are 
increasing due to a variety of factors. Use of manufactured 
housing is on an upward trend, and this type of 
structure provides less resistance to wind than conventional 
construction. Uniform building codes for wind-
resistant construction are not adopted by all States, and 
population trends show rapid growth in the highly 
exposed areas of the coastal zone from Texas to Maine. 
Because of continued growth of the population in the 
coastal States susceptible to high winds from tropical 
cyclones, the deteriorating condition of older homes, 
and the increased use of aluminum-clad mobile homes, 
the impacts of wind hazards will likely continue to 
increase. The general design and construction of buildings 
in many high wind zones do not fully consider 
wind resistance and its importance to survival. 
Consequences 
Near-surface winds and associated pressure effects, 
positive, negative, and internal, exert pressure on structure 
walls, doors, windows, and roofs, causing the structural 
components to fail. Positive wind pressure is a 
direct and frontal assault on a structure, pushing walls, 
doors, and windows inward. Negative pressure affects 
the sides and roof where passing currents create lift and 
suction forces that act to pull building components and 
surfaces outward. The effects of winds are magnified in 
the upper levels of multi-storey structures. 
Just as positive and negative forces impact and remove 
a windward protective building envelope (i.e., doors, 
windows, walls), internal pressures rise and result in 
roof or leeward building component failures and considerable 
structural damage or collapse. Debris carried 
along by extreme winds can directly contribute to loss 
of life and indirectly to the failure of protective building 
envelope components. Upon impact, wind-driven 
debris can rupture a building, allowing more significant 
positive and internal pressures. 
The insurance industry spent nearly $23 billion on 
wind-related catastrophic events from 1981 to 1990 
(NRC, 1993). Of the three primary sources, hurricanes 
and tropical storms, severe thunderstorms, and winter-
storms, severe local windstorms accounted for 51.3 per-
cent of the expenditures. Windstorms in 1993 resulted 
in 40 fatalities, 129 injuries, and $231 million in dam-
age, while thunderstorm/wind events resulted in 23 
fatalities, 458 injuries, and $349 million in damage 
(NOAA, 1994). 
The NWS reported a significant increase in wind-related 
fatalities between 1992 and 1993, from 28 to 63. Of 
the deaths, 74 percent occurred in open areas and in 
vehicles. 
From June 1975 to May 1995, 193 Federal disaster declarations 
involved wind-induced natural hazards: 106 
for s, 40 for hurricanes and tropical storms, 25 for 
typhoons, and 22 for high winds. 
RESEARCH, DATA COLLECTION, AND 
MONITORING ACTIVITIES 
Key research on wind has been conducted by NWS 
branches and the Wind Engineering Research Center 
(WERC) at Texas Tech University, in conjunction with 
the Wind Damage Mitigation Committee of the 
Insurance Institute for Property Loss Reduction (IIPLR) 
and Dr. Theodore Fujita of the University of Chicago. 
Dr. Fujita's significant contributions to wind research 
include the development of the Fujita Scale to classify 
tornados, and the recent evaluation and assessment of 
downburst wind damages occurring with extreme thunderstorm 
and windstorm events, such as experienced in 
Kauai, HI during Hurricane Iniki in 1992. 
The NWS modernization program and the deployment 
and implementation of the wind-measuring system 
NEXRAD has improved the operational observation 
program and the climatology database. NEXRAD 
enhances the ability to predict downbursts and wind 
shears near airports. The data prompt development of 
frequency-of-occurrence data for mesocyclones, the 
typical wind circulation environment exhibited prior to 
the development of a tornado or severe thunder-
storm/wind event (NRC, 1993). 
A WERC/IIPLR project developed an alternative to the 
wind-resistance classification system used by the 
Insurance Services Offices which correlates a building's 
wind resistance with fire resistance. The WERC/IIPLR 
classification system identifies the wind resistance 
capabilities of a building based on building type (mate-
rial used and system employed during construction) and 
other related factors pertaining to environment, frame, 
roof envelope, wall envelope, and other considerations 
(IIPLR, 1994). The system allows for evaluation of the 
weak points of existing buildings, enabling owners to 
take appropriate corrective actions to make buildings 
safer. The system can be used to identify individual 
buildings for evacuation because of poor resistivity. 
Published reports on U.S. windstorm hazards discuss 
damages and include maps of the severity and extent of 
various types of extreme events (Golden and Snow, 

WINDSTORMS 55
1991; NRC, 1993). Updated information on the number 
of occurrences of windstorms is acquired continually 
by the NWS on a national and State-by-State basis. 
MITIGATION APPROACHES 
Improved and consistent building codes have been recommended 
as a key measure to mitigate life and property 
losses associated with windstorms (NRC, 1993). 
Adoption, enforcement, and compliance with a unified 
code for all regions of the United States would help 
ensure construction standards needed to build structures 
resistant to the lateral loads and uplift forces of severe 
winds. Improvements could be made to structural 
cladding, shuttering systems, and materials that are 
resistant to the penetration of wind-blown debris and 
projectiles. 
In addition to improved construction standards, land-
use regulations in the regions susceptible to windstorm 
hazards could limit exposure. Zoning as a form of land-
use management and control can regulate populations 
and residential, commercial, and industrial developments 
in locations of known high risk exposure. It also 
can be used to reduce building density, adjust timing of 
regional development plans, and define the type of 
developments allowable in these hazard areas (NRC, 
1993). 
RECOMMENDATIONS 
The National Research Council's publication on wind-
storm hazards, Wind and the Built Environment: U.S. 
Needs in Wind Engineering and Hazard Mitigation 
(1993), presents several categories of recommendations: 
wind hazards and related issues; nature of wind; 
wind engineering; mitigation, preparedness, response 
and recovery; education and technology transfer; and 
cooperative efforts. The report identifies the establishment 
of a National Wind Science and Engineering 
Program as a critical element necessary to "revitalize 
wind-hazard research." 
BIBLIOGRAPHY AND REFERENCES 
American National Standards Institute. 1982. Building 
Code Requirements for Minimum Design Loads on 
Buildings and Other Structures. ANSI A58.1. New 
York, NY: American Society of Civil Engineers. 
American Society of Civil Engineers. 1993. Minimum 
Design Loads for Building Other Structures. ASCE, 
July, 1993. New York, NY. 
Brinkmann, Waltraud A.R. 1975. Local Windstorm in 
the United States: A Research Assessment. Institute of 
Behavioral Science, The University of Colorado, 
Boulder, CO. 
Byrant, E. 1991. Natural Disasters. New York, NY: 
Cambridge University Press. 
Chagnon, S.A., and J.M. Chagnon. 1992. "Storm 
Catastrophes in the United States." Natural Hazards, 
Vol. 6, No. 2, pp. 93-107. 
Finkl, C. W., Jr. (ed.). 1994. Journal of Coastal 
Research, Special Issue No. 12. Coastal Hazards: 
Perception, Susceptibility and Mitigation. The Coastal 
Education and Research Foundation, Inc. 
Golden, J.H., and J.T. Snow. 1991. "Mitigation 
Against Extreme Windstorms." Reviews of Geophysics. 
American Geophysical Union. Vol. 29, No. 4, pp. 477-
504. 
Insurance Institute for Property Loss Reduction. 1994. 
Understanding the Wind Peril. 
Kitsmiller, D.H., W.E. McGovern, and R.E. Saffle. 
February 1992. The NEXRAD Severe Weather 
Potential Algorithm. NOAA Technical Memorandum, 
NWS TDL 81. Silver Spring, MD: Techniques 
Development Laboratory. 
National Research Council. 1993. Wind and the Built 
Environment: U.S. Needs in Wind Engineering and 
Hazard Mitigation. 
National Weather Service. 1994. Thunderstorms and 
Lightning: The Underrated Killers. U.S. Department 
of Commerce, National Oceanic and Atmospheric 
Administration. 
Ramsdell, J.V., D.L. Elliott, C.G. Holladay, and J.M. 
Hubbe. 1986. Methodology for Estimating Extreme 
Winds for Probabilistic Risk Assessments. 
Washington, DC: U.S. Nuclear Regulatory 
Commission. 
Twisdale, L.A., and P.J. Vickery. 1993. "Analysis of 
Thunderstorm Occurrences and Windspeed Statistics." 
In: Proceedings of the 7th U.S. National Conference 
on Wind Engineering. Los Angeles, CA: University of 
California. 
Twisdale, L.A. October 1988. "Probability of Facility 
Damage from Extreme Wind Effects." Journal of 
Structural Engineering. American Society of Civil 
Engineers. Vol. 114, No. 10, pp. 2190-2209. 


Photo of hail
58 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
H
ailstorms develop from severe thunderstorms. 
Although they occur in every State on the main-
land United States, hailstorms occur primarily in 
the Midwestern States. Only a localized area along the 
border of northern Colorado and southern Wyoming experiences 
hailstorms on 8 or more days each year. Most 
inland regions experience hailstorms at least 2 or more 
days each year. 
Hailstorms cause nearly $1 billion in property and crop 
damage annually, as peak activity coincides with the 
Midwest's peak agricultural seasons. Long-stemmed vegetation 
is particularly vulnerable to damage by hail impact 
and accompanying winds. Severe hailstorms also cause 
considerable damage to buildings and automobiles, but 
rarely result in loss of life. 
Efforts to reduce hailstorm damage generally are associated 
with mitigation activities for thunderstorms and wind-
storms, including building code improvement and enforcement 
and public awareness and education campaigns. 
Improvements in weather warning systems will enhance 
the ability to predict severe thunder and hailstorms. 

HAILSTORMS 59
Photo of a hailstormPhoto: Red Cross 

60 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
A hailstorm is an outgrowth of a severe thunderstorm in 
which balls or irregularly shaped lumps of ice greater 
than 0.75 in (1.91 cm) in diameter fall with rain 
(Gokhale, 1975). Early in the developmental stages of 
a hailstorm, ice crystals form within a low-pressure 
front due to warm air rising rapidly into the upper 
atmosphere and the subsequent cooling of the air mass. 
Frozen droplets gradually accumulate on the ice crystals 
until, having developed sufficient weight, they fall as 
precipitation. 
The size of hailstones is a direct function of the severity 
and size of the storm. High velocity updraft winds 
are required to keep hail in suspension in thunderclouds. 
The strength of the updraft is a function of the intensity 
of heating at the Earth's surface. Higher temperature 
gradients relative to elevation above the surface result in 
increased suspension time and hailstone size (Bryant, 
1991). 
RISK ASSESSMENT 
The areal extent and severity of hailstorm hazards are 
different from thunderstorms or tornados. Hailstorms 
occur more frequently during the late spring and early 
summer, when the jet stream migrates northward across 
the Great Plains. This period has extreme temperature 
changes from the ground surface upward into the jet 
stream, which produce the strong updraft winds needed 
for hail formation. 
Probability and Frequency 
Data on the probability and frequency of occurrence of 
hailstorms is limited, with little recent research. 
However, in Natural Hazards (1991), Byrant presented 
a map depicting the annual frequencies of hailstorm 
occurrences in the United States (from Eagleman, 
1983). Recreated in Map 5-1, it shows that only a localized 
area along the border of northern Colorado and 
southern Wyoming experiences hailstorms 8 or more 
days each year. Outside of the coastal regions, most of 
the United States experiences hailstorms at least 2 or 
more days each year. 
The areal extent and severity of the hailstorm hazard is 
not coincident with maximum thunderstorm or tornado 
activity. The middle areas of the Great Plains are most 
frequently affected by hailstorms. Multiple impacts of 
concurrent severe thunderstorm effects from extreme 
winds, tornados, and hail are very likely in this region, 
even though the Great Plains is not where each of these 
hazards, taken individually, occurs with the greatest frequency. 
Exposure 
Peak periods for hailstorms, late spring and early summer, 
coincide with the Midwest's peak agricultural sea-
sons for crops such as wheat, corn, barley, oats, rye, 
tobacco, and fruit trees. Long-stemmed vegetation is 
particularly vulnerable to damage by hail impacts and 
winds. Severe hailstorms also cause considerable dam-
age to buildings and automobiles, but rarely result in 
loss of life. 
The land area affected by individual hail events is not 
much smaller than that of a parent thunderstorm, an 
average of 15 mi (24 km) in diameter around the center 
of a storm (Pearce and others, 1993). 
Consequences 
The development of hailstorms from thunderstorm 
events causes nearly $1 billion in property and crop 
damage each year (NWS, 1994). Recent significant 
hailstorms include events in Denver, CO (1994) and in 
the eastern Texas-Oklahoma region (1995). 
The Property Loss Research Bureau indicates that the 
April-May 1995 hailstorm in the Texas-Oklahoma 
region may have been the worst on record in terms of 
non-agricultural property losses. However, more specific 
information is not readily available. 
The Midwest hailstorm and tornado event in April 1994 
lasted 4 days. According to Property Claims Services in 
Rahway, NJ, it produced 300,000 damage claims 
against insurers, more than Hurricane Andrew or the 
Northridge earthquake. 
RESEARCH, DATA COLLECTION, AND 
MONITORING ACTIVITIES 
Recent unpublished private research has used the hail 
dataset available from the National Climatic Data 
Center (NCDC) in Asheville, NC, to model hailstorm 
frequency and hailstone size. The model is geocoded 
into 6-mi (10-km) grid cells across the lower 48 States. 
When it is available to others, the hail model data may 
be integrated into existing geographic information systems 
to help develop a hail risk index methodology for 
identification of high-risk zones. 

HAILSTORMS 61
U.S. Map 5-1
62 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
MITIGATION APPROACHES 
Mitigation efforts to reduce hailstorm damage generally 
are similar to those associated with thunderstorm and 
windstorm hazards: building codes and public awareness 
and education campaigns. Weather warning system 
improvements and the modernization and implementation 
of NWS monitoring systems such as 
NEXRAD will improve prediction of the severe weather 
potential associated with thunderstorms and hail-
storm development. 
Seeding clouds with supercooled water containing silver 
iodide nuclei has been a hail suppression technique 
used around the world. Although it has reduced crop 
damage, the technique was discontinued in the United 
States in the early 1970s due to political controversy 
(Bryant, 1991). 
RECOMMENDATIONS 
Recent research on hailstorm hazards in the United 
States has not been published. Information collected by 
NCDC documents the annual incidence rate and dam-
age. Recommendations on research, hazard mitigation, 
or hailstorm control were not found in available published 
reports. 
In a June 1995 magazine article in Business 
Geographics, G. Mertz of DataMap, Inc., recommended 
the integration of the NCDC hail dataset into a geographic 
information system. A hailstorm exposure map 
and a hail index system could identify high- and low-
risk areas. The index was discussed as a useful tool for 
FEMA and the insurance industry to evaluate exposure 
and risk for residential, automobile, crop, and business 
insurance ratings. 
BIBLIOGRAPHY AND REFERENCES 
Byrant, E. 1991. Natural Hazards. New York, NY: 
Cambridge University Press. 
Eagleman, J.R. 1983. Severe and Unusual Weather. 
New York, NY: Van Nostran Reinhold. 
Gokhale, N.R. 1975. Hailstorms and Hailstone 
Growth. Albany, NY: State University of New York 
Press. 
Golden, J.H., and J.T. Snow. 1991. "Mitigation Against 
Extreme Windstorms." Reviews of Geophysics. 
American Geophysical Union. Vol. 29, No. 4, pp. 477-
504. 
Pearce, L., H. Hightower, B. Konkin, S. Megalos, and J. 
Pernu. 1993. British Columbia: Hazard, Risk and 
Vulnerability Analysis, Vol. 1. The Disaster 
Preparedness Resources Center, The University of 
British Columbia. 

HAILSTORMS 63


Photo of a snow covered mountain
66 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
S
now avalanches are not considered a major natural 
hazard because they impact relatively small areas of 
the United States. Compared with other hazards, 
snow avalanches have localized impacts and individually 
do not affect large numbers of people. However, the total 
number of deaths attributable to snow avalanches each 
year is exceeded only by those associated with floods, 
lightning, and tornadoes, and extreme heat. 
Most of the 10,000 reported avalanches that occur each 
year are in remote, unpopulated mountainous areas, along 
recognized avalanche paths in previously identified hazard 
zones. The threat is most severe in the mountainous 
Western United States, including Alaska. 
The sliding snow or ice mass in an avalanche moves at 
high velocities. It can shear trees, completely cover entire 
communities and highway routes, and level buildings. The 
primary threat is loss of life of back country skiers, 
climbers, and snowmobilers. 
Since 1970, an average of 144 persons have been trapped 
in avalanches annually: on average 14 were injured and 14 
died. The estimated annual average damage to structures 
is $500,000. 
Recent avalanche mitigation approaches have included 
avalanche hazard zoning, evacuation, artificial release, and 
avalanche-control structures. Artificial release is the most 
common measure used in the United States. Where other 
methods are ineffective or cannot be used, control structures 
may be installed. 
Photo of an avalanche
SNOW AVALANCHES 67
Photo
68 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
A snow avalanche is a slope failure composed of a mass 
of rapidly moving, fluidized snow that slides down a 
mountainside. The flow can be composed of ice, water, 
soil, rock, and trees (Armstrong and Williams, 1992; 
NRC, 1990; Coch, 1995). The amount of damage 
depends on the type of avalanche, the composition and 
consistency of the material contained in the avalanche, 
the velocity and force of the flow, and the avalanche 
path. 
The slope failure associated with an avalanche is caused 
by several factors, but is primarily due to large accumulations 
of snow on steep slopes. Avalanches occur on 
slopes averaging 25 to 50 degrees, and the majority start 
on slopes between 30 and 40 degrees. They are triggered 
by natural seismic or climatic factors such as 
earthquakes, thermal changes, and blizzards. Human 
factors, including snowmobiles, skiers, hikers, vehicle 
traffic, and elastic sound waves created by explosions 
can trigger events, as can direct dynamic loading from 
ice slabs falling off cornices (NRC, 1990). 
Natural and human-induced snow avalanches most 
often result from structural weaknesses within the 
snowpack. They are caused by changes in the type and 
thickness of the snowcover layer resulting from thermal 
fluctuations or multiple snowfall events. The potential 
for a snow avalanche increases with significant temperature 
influences, which cause metamorphic crystal 
change in the snow layer, and with accumulations of dry 
and wet snow over time. 
At the point where the shear forces of the overburdened 
upper layer overcome the resistant forces of the under-
layer, the mass slips and begins sliding downslope. The 
intensity and impact of the resulting avalanche depend 
on the volume of snow accumulated in the upper layer, 
the density of the material, the slope of the starting area, 
the avalanche path, and the runout zone at the bottom of 
the slope. 
Dry, low density avalanche releases can reach maxi-
mum velocities of 45 mph (20 m/s) to 157 mph (70 
m/s), depending on the vertical fall distance (Mears, 
1992). Slower estimated maximum velocities of 22 
mph (10 m/s) to 78 mph (35 m/s) for similar vertical fall 
distances are reached by wet, high density avalanches 
which can be more destructive and exert greater impact 
forces on structures. The weight of moving, water-saturated 
snow can be 25 to 31 lb/ft³ (400 to 500 kg/m³) 
throughout the entire depth of an avalanche flow 
(Mears, 1992). The moving mass entrains rock, soil, 
and other solid debris. 
The most common types of snow avalanches are loose-
snow and slab avalanches. A loose-snow avalanche is 
composed of dry, fresh snow deposits that accumulate 
as a cohesionless and unstable mass atop a stable and 
cohesive snow or slick ice sublayer. A loose-snow 
avalanche releases when the shear force of its mass 
overcomes the underlying resistant forces of the cohesive 
layer. 
A slab avalanche generally is composed of a thick, 
cohesive snowpack deposited or accumulated on top of 
a light, cohesionless snow layer or slick ice sublayer. At 
the starting surface or top of the slab, a deep fracture 
develops in the slope of well-bonded, cohesive snow. 
Fractures of more than 2 mi (3 km) in length have been 
observed, and they can reach across several starting 
zones. A slab avalanche release is usually triggered by 
turbulence or impulse waves. Release also occurs when 
the internal cohesive strength of the slab layer is greater 
than the bonding at the basal and lateral slab boundaries. 
As a release occurs, the slab accelerates, gaining mass 
and speed as it travels down the avalanche path. 
An avalanche path is determined by the physical limitations 
of the boundaries of the local terrain and manmade 
features. An avalanche may follow a path along a channelized 
or confined terrain, similar to debris flows or 
streams, before spreading onto alluvial fans or gentle 
slopes. In other instances, avalanches follow unconfined 
or planar slopes of a mountain side down to the 
abrupt slope change of the valley bottom (Mears, 1992). 
The avalanche path itself varies in width as it transitions 
along the path, depending on the confinement of the terrain 
and the velocity of flow. 
Mears (1992) describes the avalanche path as having 
three specific transition zones: 
• The Starting Zone is typically located near the top of 
the ridge, bowl or canyon, with steep slopes of 25 to 
50 degrees; 
• The Track Zone is the reach with mild slopes of 15 to 
30 degrees and the area where the avalanche will 
achieve maximum velocity and considerable mass; 
and 
• The Runout Zone is the gentler slopes of 5 to 15 
degrees located at the base of the path, where the 
avalanche decelerates and massive snow and debris 
deposition occurs. 

SNOW AVALANCHES 69
U.S. Map 6-1Map 6-1. Qualitative indicator of the severity of snow avalanches in the United States. 
Data not available for Puerto Rico, U.S. Virgin Islands, and Pacific Territories. 
Source: Personal Communications, Knox Williams, 1995. 
RISK ASSESSMENT 
Snow avalanches in the United States have varying levels 
of severity, depending on slope steepness and snow 
accumulation. As shown on Map 6-1, the threat is most 
severe in the mountainous western States, including 
Alaska. New England mountains have a low level of 
avalanche risk, and the rest of the United States has no 
risk. 
Snow avalanches occur throughout the winter and may 
flow down the same paths several times during a sea-
son. In avalanche-prone areas, most State and local 
officials maintain records of previous events and have 
documented repetitive paths. As a result, avalanche 
hazard zones can be identified in many areas. 
Probability and Frequency 
The geophysical processes that contribute to snow 
avalanches during a particular year are statistically 
independent of past events. However, like other similar 
natural processes, a return period and probability of 
occurrence can be developed from historical records. 
Avalanche occurrence is not directly attributed to a specific 
major meteorological event, such as the 1-percent-
annual-chance or 100-year snowfall. It is more commonly 
a result of a combination of weather and snow-
pack conditions (Mears, 1992). Unfortunately, the short 
period of recorded and observed avalanches and associated 
conditions that contribute to the risk make it difficult 
to develop return periods for each avalanche-prone 
area in the United States. 
The classification and delineation of avalanche hazard 
zones for Vail, CO, Ketchum, ID, and Juneau, AK, are 
similar to the Red and Blue Zones developed and used 
in the Swiss Alps. The Swiss zones are based on quantitative 
techniques for determining avalanche impact 
pressures and return periods. The Red Zone identifies 
areas of highest risk (NRC, 1990). Avalanches in Red 
Zones are either powerful (impact pressures greater 
than 630 lb/ft² (30kPa)) with a return period of 300 
years or less, or all avalanches irrespective of intensity 
with return pereiods of up to 30 years. The Blue Zone 
identifies areas subject to less frequent avalanches with 
30- to 300-year return periods, and impact pressures 
less than 630 lb/ft² (30 kPa). 
Although the same zone designation system is used in 
the United States, the definition of each zone requires 
modification due to the short period of historical record. 
This limitation makes it more difficult to determine 
return periods for avalanches in most locations (Mears, 
1992). 

70 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
In the Colorado Geological Survey report on avalanches, 
Snow Avalanche Hazard Analysis for Land-Use 
Planning and Engineering, Mears discusses the techniques 
available to determine a return period for 
avalanches based on historic records, recent observations, 
aerial photography analyses, and vegetation characteristics 
(Mears, 1992). A snow avalanche will inflict 
damage and changes to the forests and vegetation along 
its path, unless they are protected by deep snowpack. 
Thus vegetation is a valuable tool in determining when 
previous events occurred. Vegetation indicators for 
avalanche frequency are listed in Table 6-1. 
Avalanche activity along a path may be continuous 
throughout the season, in which case the annual probability 
of occurrence is constant regardless of the estimated 
return period of individual incidents. An 
encounter probability based on the relationship of the 
return period and annual probability was developed by 
LaChapelle (1966) to calculate the risk of an avalanche 
with a given return period occurring within a certain 
period of years. Even though the intensity and severity 
of an avalanche cannot be directly associated with a 
given frequency, return period, and encounter probability, 
the concepts are useful in land-use planning and 
engineering design. 
Exposure 
Back country skiers, backpackers, and snowmobilers in 
rural areas are at greatest risk of loss of life due to suffocation 
when buried in an avalanche. Avalanches can 
damage or destroy vehicles, highways, utilities, and 
buildings. 
Most avalanches that affect people occur in and around 
mountain resorts used by winter recreational enthusiasts, 
while others affect downslope transportation 
routes and communities. Although the damage from 
avalanches is not as widespread as other natural hazards, 
on an annual basis fatalities exceed the average 
number of deaths due to all other hazards, except 
floods, lightning, tornadoes, and extreme heat. 
The risk of avalanche loss is greatest on the flatter slope 
of the runout zone which is more conducive to development, 
transportation routes, and infrastructure. 
Exposure to the hazard has risen due to growth in winter 
recreational activities and resort facilities, mountain 
residences, highways, telecommunication lines, utilities, 
and mines in avalanche hazard zones. 
Consequences 
During the past two decades, snow avalanches have not 
prompted any federally-declared disasters, primarily 
because they tend to occur in areas with little development. 
Other reasons are improved and modernized 
monitoring techniques, awareness of the hazard, and 
remedial measures implemented to mitigate disaster 
potential. However, avalanches are still considered a 
prevalent hazard throughout the Rocky Mountain, 
Pacific Northwest, and Alaskan regions, with numerous 
incidents of property damage, injury, and death each 
year. From 1990 to 1995, 640 avalanches in the United 
States impacted individuals and property. 
Only approximately 10,000 of the estimated 100,000
snow avalanches that occur each year are reported.
Since 1970, an average of 144 persons have been
trapped in avalanches annually,
causing an annual average of 14
TABLE 6-1.—Vegetation as an avalanche-frequency indicator injuries and 14 deaths. Skiers, 
snowmobilers, and climbers in 
wooded and sloped areas in 
avalanche zones are most vulnerable 
(Armstrong and Williams, 
1992). For many avalanches in 
which people have suffered injuries 
or loss of life, the triggering mechanism 
for the avalanche has been the 
victims themselves. 
In the winter of 1994–95, California
had more snow avalanches than any
other State, with 4,787 reported at
seven locations. However, over the
10 years from the winter of 1985–86
through the winter of 1994–95,
Colorado's 65 fatalities were the
Source: Mears, 1992. highest. The activity with the high-
Return Period Vegetation Indicators 
1-10 years Track supports grasses, shrubs, flexible trees 
up to 6.5 ft. (2m) high; broken timber on 
ground and at path boundaries 
10-30 years Predominantly pioneer species; young trees 
similar to adjacent forest; broken timber on 
ground at path boundaries 
30-100 years Old uniform-aged trees of pioneer species; 
young trees of local climax species; old and 
partially decomposed debris 
100-300 years Mature, uniform-aged trees of local climax 
species; debris completely decomposed; 
increment core data required. 

SNOW AVALANCHES 71
est fatality rate for the same 
TABLE 6-2.—Avalanche Impact pressures related to damage.
period was back-country skiing, 
with 45 fatalities, followed by 
climbing, with 28 (Colorado 
Avalanche Information Center, 
1995, unpublished). 
Possibly the worst U.S. 
avalanche disaster occurred on 
March 1, 1910, in Wellington, 
WA. The avalanche derailed 
two trains, killing 96 persons 
(Armstrong and Williams, 
1992). 
Property damage associated Source: Mears, 1992.
with avalanches is a function of
several factors. Large external lateral loads can cause
significant damage to structures and facilities. Table 6-
2 indicates the estimated potential damage that can be
expected for a given range of impact pressures.
A breakdown of costs associated with all types of property 
damage due to snow avalanches is not available. 
The estimated average annual damage to buildings 
alone is $500,000. However, a tally of the impact on 
public property, includes forests and parkland, transportation 
routes, and utilities, coupled with the cost of 
litigation following injuries or deaths, raises the estimated 
annual cost of avalanches to more than $5 mil-
lion. 
NRC (1990) offers examples of avalanche damage 
described below: 
• Avalanches cost the Washington State Department of 
Transportation an estimated $330,000 each year for 
avalanche control, snow removal, and plowing, not 
including salaries and expenses of State employees. 
• Between 1977 and 1986, avalanche damages in 
Alaska were estimated to be $11.4 million. 
• On March 31, 1981, an avalanche at California's 
Alpine Meadows ski resort area resulted in seven 
deaths and caused approximately $1.5 million in 
property damage. Litigation and out-of-court settlements 
sent the cost spiraling to $14 million. 
RESEARCH, DATA COLLECTION, AND 
MONITORING ACTIVITIES 
Snowstorms and ice storms are monitored daily by 
weather observation centers and a network of State and 
regional avalanche forecast centers for Colorado, Utah, 
Montana, Washington, Oregon, and Wyoming. Data 
collection methodologies on the number and location of 
snowfalls improve with each passing storm. Prevailing 
wind direction and temperature fluctuations are analyzed 
to assess the probability of avalanches in high-
risk areas. 
Impact Pressure 
kPa lbs/ft2 Potential Damage 
2-4 40-80 Break windows 
3-6 60-100 Push in doors, damage walls, roofs 
10 200 Severely damage wood frame structures 
20-30 400-600 Destroy wood-frame structures, break trees 
50-100 1000-2000 Destroy mature forests 
>300 >6000 Move large boulders 
These methods are not completely effective in identifying 
potential avalanche hazard zones. In areas with 
known historical avalanche paths, the impact of the hazard 
may be mitigated by local experience and under-
standing of the hazard and triggering mechanisms. This 
knowledge makes the forecast and prediction process 
more reliable and enhances the effectiveness of warnings. 
Forecasting avalanches can be accomplished by different 
methods with varying degrees of success. As war-
ranted by climatic conditions, various methods are 
used. Weather forecasting of snowfall amounts, winds, 
and temperature are important tools to analyze snow-
pack conditions and avalanche potential for large geographic 
areas. 
Local area avalanche forecasting involves snow pit 
analysis. This technique requires assessing sublayer 
conditions and the composition of each layer to detect 
the existence of unstable boundaries between cohesive 
and cohesionless snow and ice layers. 
Large-scale computer-based forecasting methods 
require a historical and statistical database. Regional 
avalanche centers collect data on snow avalanches from 
a network of measuring sites. Nationwide programs for 
computer-based forecasting do not exist. 

72 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
MITIGATION APPROACHES 
Hazard mitigation efforts have been successful in 
reducing the exposure of people and property to snow 
avalanche hazards. Warnings are issued through radio 
and television, and posted, if possible, on bulletin 
boards in avalanche hazard areas. 
The potential for property damage has prompted local 
avalanche zoning ordinances in California, Colorado, 
Idaho, Utah, and Washington. For example, hazard 
zones have been established in Vail, CO, and Ketchum, 
ID, based on avalanche studies and assessments 
(Armstrong and Williams, 1992; Mears, 1992). They 
are delineated and mapped based on historic evidence, 
recorded seasonal events, and terrain features that indicate 
the potential for event occurrence. In Vail, 
avalanche zones have been adopted in the City's comprehensive 
land-use plan and are used to evaluate pro-
posed residential and commercial development. 
To regulate land use effectively and to reduce the risk of 
damage from avalanches, hazard zones may be adopted 
in local ordinances. As a result of legislation passed in 
1974, many avalanche-prone counties in Colorado have 
hazard plans or land-use rules that specifically address 
avalanche hazards. Even though Juneau, AK had deter-
mined its avalanche hazard zones based on studies and 
a geophysical investigation in the 1970s, and despite 
numerous avalanches, as of 1991 it had not adopted a 
zoning ordinance to regulate at-risk development 
(Armstrong and Williams, 1992). 
The key mitigation recommendations in the Colorado 
Geological Survey's Bulletin 49 (1992) to either eliminate 
or reduce the hazard include avalanche hazard zoning, 
evacuation, artificial release, and avalanche-control 
structures. Artificial release is the most common measure 
used in the United States, but it is not 100-percent 
effective. 
Control structures are used in areas where snow 
avalanches are unavoidable and other measures are 
ineffective or cannot be employed. Structures currently 
in use include supporting structures, snow-drift 
fences, deflecting berms, catching berms, catching 
structures, and direct-protection structures. Mechanical 
compaction and disruption of the cohesive slabs compresses 
or densifies the snow to strengthen the slope. It 
is commonly employed in ski resort areas to stabilize 
snow in starting zones (NRC, 1990). 
RECOMMENDATIONS 
The recommendations presented by NRC have 
addressed issues related to national leadership, hazard 
delineation and regulation, control measures, forecasting, 
research, and communications (NRC, 1990). NRC 
recognized snow avalanches as the "most frequent catastrophic 
mass movement in the Nation" and one of 
many components leading to extensive ground-failure 
hazards. 
NRC concluded that, even though snow avalanches are 
the greatest natural hazard affecting winter activities in 
mountainous regions, the hazard receives little attention. 
Greater funding support is needed in order to better 
understand, predict, and mitigate snow avalanche 
hazards, and to educate local officials and emergency 
managers. 
BIBLIOGRAPHY AND REFERENCES 
Armstrong, B.R., and K. Williams. 1992. The 
Avalanche Book, Revised and Updated. Golden, CO: 
Fulcrum Publishing. 
Byrant, E. 1991. Natural Disasters. New York, NY: 
Cambridge University Press. 
Coch, N.K. 1995. GEOHAZARDS: Natural and 
Human. Upper Saddle River, NJ: Prentice Hall. 
Colorado Avalanche Information Center (unpublished 
report). 1995. Denver, CO: Colorado Department of 
Natural Resources. 
La Chapelle, E.R. 1966. Encounter Probabilities for 
Avalanche Damage. U.S. Department of Agriculture, 
Forest Service, Alta Avalanche Study Center. 
Miscellaneous Report No. 10. 
Mears, A.I. 1989. "Regional Comparisons of 
Avalanche-Profile and Runout Data." Arctic and 
Alpine Research. Vol. 21, No. 3, pp. 232-238. 
Mears, A.I. 1992. "Snow Avalanche Hazard Analysis 
for Land-Use Planning and Engineering." Colorado 
Geological Survey Bulletin No 49. Denver, CO: 
Department of Natural Resources. 
National Research Council. 1990. Snow Avalanche 
Hazards and Mitigation in the United States. 
Washington D.C.: National Academy Press. 
Perla, R., T.T. Cheng, and D.M. McClung. 1980. "A 
Two-Parameter Model of Snow-Avalanche Motion." 
Journal of Glaciology. Vol. 26, No. 94, pp.197-208. 

SNOW AVALANCHES 73
Perla, R., K. Lied, and K. Kristensen. 1984. "Particle 
Simulation of Snow Avalanche Motion." Cold Regions 
Science and Technology. Vol. 9, pp. 191-202. 
Salm, B., A. Burkard, and H. Gubler. 1990. 
Berechnung von Fliesslawinen eine Anleitung fur 
Praktiker mit Beispeilen: Mitteilungen des 
Eidgenossischen Istituts fur Schnee-und 
Lawinenforschung. No. 47, pp. 37. 
Williams, K. Personal Communication, 1995. Fort 
Collins, CO: Colorado Avalanche Information Center. 


Photo 
76 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
W
interstorms consisting of extreme cold and 
heavy concentrations of snowfall or ice affect 
every State in the continental United States and 
Alaska. Areas where such weather is rare, such as the 
extreme South, are disrupted more severely by winter-
storms than are regions that experience severe weather 
more frequently. 
Winterstorms are known to spawn other natural hazards, 
such as coastal flooding and erosion, severe thunderstorms 
and tornados, and extreme winds. These effects disrupt 
commerce and transportation and often result in loss of life 
due to accidents or hypothermia. 
Between 1988 and 1991, a total of 372 deaths, an average 
of 93 each year, was attributed to severe winterstorms. 
The Superstorm of March 1993, considered among the 
worst non-tropical weather events in the United States, 
killed at least 79 people, injured more than 600, and caused 
$2 billion in property damage across portions of 20 States 
and the District of Columbia. 
Experience has shown that no area can fully prepare for 
severe winterstorms. The March 1993 Superstorm was 
one of the most widely forecast severe winter events, yet it 
devastated many parts of the Eastern United States for 
more than a week. 
Photo: Red Cross 

SEVERE WINTERSTORMS 77 
Photo: Red Cross 

78 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
Winterstorms and blizzards originate as mid-latitude 
depressions or cyclonic weather systems, sometimes 
following the meandering path of the jet stream (Bryant, 
1991). A blizzard combines heavy snowfall, high 
winds, extreme cold, and ice storms. The origins of the 
weather patterns that cause severe winterstorms, such as 
snowstorms, blizzards, and ice storms, are primarily 
from four sources in the continental United States. 
In the Northwestern States, cyclonic weather systems 
from the North Pacific Ocean or the Aleutian Island 
region sweep in as massive low-pressure systems with 
heavy snow and blizzards. In the Midwestern and 
Upper Plains States, Canadian and Arctic cold fronts 
push ice and snow deep into the interior region and, in 
some instances, all the way down to Florida. In the 
Northeast, lake effect snowstorms develop from the pas-
sage of cold air over the relatively warm surfaces of the 
Great Lakes, causing heavy snowfall and blizzard conditions. 
The Eastern and Northeastern States are affected 
by extra-tropical cyclonic weather systems in the 
Atlantic Ocean and Gulf of Mexico that produce snow, 
ice storms, and occasional blizzards. 
Many winter depressions give rise to exceptionally 
heavy rain and widespread flooding and conditions 
worsen if the precipitation falls in the form of snow. 
Snow volume exceeds that of rain by a factor of 7 to 10. 
Affected regions may be subjected to heavy snowfall. 
The winterstorm season varies widely, depending on 
latitude, altitude, and proximity to moderating influences. 
Severe winterstorms have affected every State in the 
continental United States and Alaska. Hawaii, Puerto 
Rico, the U.S. Virgin Islands, and the Pacific territories 
are not affected by this hazard. 
RISK ASSESSMENT 
Probability and Frequency 
Analysis of recorded snow level (depth) data can yield 
probability and frequency of occurrence associated with 
severe winterstorms. Snow level measurements are 
recorded each day at sites within the conterminous 
United States and Alaska. Snow level is an indicator of 
severity of winterstorms based on accumulation and 
associated snow loading. The weight of accumulated 
snow or ice results in snow load forces that can cause 
building collapse and damage to infrastructure. 
Map 7-1 shows the probability of a snow of given depth 
(in centimeters) occurring within geographic areas of 
the conterminous United States with a 5-percent chance 
of being equaled or exceeded in a given year. The map 
is based on daily recorded data by the U.S. Department 
of Commerce's National Climatic Data Center in 
Asheville, NC and the U.S. Department of Energy's 
National Renewable Energy Laboratory in Golden, CO. 
The data are available in a 1993 CD-ROM entitled 
Solar and Meteorological Surface Observation Network 
1961-1990 (NCDC, 1993). 
Regional studies of the mean annual snowfall for the 
Northeastern United States and the Great Lakes States 
show that the mean distribution of seasonal snowfall 
depends on latitude. Seasonal mean snowfall amounts 
range from 5.85 in (15 cm) in Virginia to more than 98 
in (250 cm) in the New England States, New York, and 
West Virginia (Kocin and Uccellini, 1990). To establish 
the relationship between probability and frequency of 
occurrence for snowfall, historical records are analyzed 
for snowfall totals, ice accumulation, and the number 
and location of severe winterstorms. 
Exposure 
Nearly the entire United States, except the extreme 
southern States, Hawaii, and the U.S. territories, are 
considered at risk for severe winterstorms. The degree 
of exposure depends on the normal severity of local 
winter weather. 
All of Alaska and some areas of the continental United 
States tend to be more susceptible than others to severe 
winterstorms, including the Upper Midwestern and 
Northeastern States. Generally, these regions are more 
prepared for severe winter weather. Those areas where 
such weather is rare are more likely to experience dam-
age and disruptions when winterstorms hit. 
Heavily populated areas are particularly impacted when 
severe winterstorms disrupt communication and power 
due to downed distribution lines. Snow and ice removal 
from roads and highways is difficult when accumulations 
build faster than equipment can clear. Debris 
associated with heavy icing may impact utility systems 
and transportation routes. 
Damage to buildings occurs especially in areas where 
normally anticipated snowfall depths do not warrant 
recognition in building codes. Roof collapse damages 
residential, commercial, and industrial structures. 

SEVERE WINTERSTORMS 79
U.S. Map 7-1Map 7-1. Snowdepth (in centimeters) with a 5% chance of being equaled or exceeded in any given 
year, from Solar and Meteorlogical Surface Observation Network 1961 - 1990 
Data not shown for Alaska, Hawaii, Puerto Rico, U.S. Virgin Islands, and Pacific Territories 
(Source: Data from National Climatic Data Center, 1993) 
The Superstorm of March 1993 illustrated that a severe 
winterstorm can be devastating. Table 7-1 presents the 
number of deaths and injuries and the estimated amount 
of damage resulting from the storm. In terms of reporting, 
NOAA found that problems arose in forecasting 
(before the event) and damage assessment (after the 
event). For example, NWS could have improved the 
coastal flood watches and warnings for the Florida Gulf 
Coast. An insufficient number of coastal observations 
and water-level measurements and a lack of storm-surge 
guidance products hampered effective forecasting. 
Consequences 
The occurrence of large snow storms, ice storms, and 
severe blizzards has a substantial impact on communities, 
utilities, and transportation systems, and often 
results in loss of life due to accidents or hypothermia. 
In addition to the impacts on transportation, power 
transmission, communications, agriculture, and people, 
severe winterstorms can cause extensive coastal flooding, 
erosion, and property loss. 
Between 1988 and 1991, NWS recorded 372 deaths that 
could be attributed to snowfall, ice storms, or extreme 
cold weather, an average of 93 deaths per year. In 1991, 
winter snows and blizzards were responsible for the 
deaths of 37 people and injuries to 350 nationwide. 
The Superstorm of March 1993 was among the worst 
non-tropical weather events in the United States, 
according to the Natural Disaster Survey Report published 
by NOAA (1994). The Superstorm caused more 
than $2 billion in property damage across portions of 20 
States and the District of Columbia. It slowed commerce, 
snarled traffic, disrupted communications and 
power, and drove tens of millions of people indoors for 
extended periods of time during the worst of the storm. 
The Superstorm is notable for its impact on a vast geographic 
area spanning virtually the entire eastern half of 
the United States. The NOAA report documents the 
impacts in six geographic regions, described below: 
• The Florida Gulf Coast was struck by an unprecedented 
extra-tropical storm surge of 9 to 12 ft (3 to 4 
m), damaging or destroying thousands of residences 
and businesses. 

80 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
TABLE 7-1.—Superstorm of March 1993: deaths, injuries, and damages by State 
State 
Deaths 
Direct/ 
Indirect Injuries 
Estimated 
Damages 
(in millions) 
Alabama 14/0 0 $100.0 
Delaware 0/2 0 $0.5 
District of Columbia 0/1 2 $0.5 
Florida 28/22 150 $1,600.0 
Georgia 15/0 420 $355.0 
Maryland 1/0 0 $22.0 
New York 8/0 4 $25.0 
North Carolina 2/7 13 $13.5 
Ohio 0/0 8 $5.0 
Pennsylvania 4/48 0 $10.0 
South Carolina 2/2 4 $22.2 
Tennessee 2/13 0 $0.5 
Virginia 0/11 0 $16.0 
West Virginia 3/6 0 $0.5 
Totals 79/112 601 $2,170.7 
Information unavailable for Connecticut, New Jersey, Rhode Island, 
Massachusetts, Vermont, Maine, and New Hampshire 
• The Middle Atlantic and 
Northeastern States were hit 
hard by blizzard conditions. 
The greatest incidence of 
power outages was reported in 
the Washington, D.C., area. 
Outages resulted from the coupled 
effects of urban density 
and severe ice accumulation. 
RESEARCH, DATA 
COLLECTION, AND 
MONITORING 
ACTIVITIES 
Snow, ice storm, and ice concentration 
data are collected at the 
Great Lakes Environmental 
Research Laboratory (GLERL) 
in Ann Arbor, MI. Data are analyzed 
and stored at the Snow and 
Ice Data Center at the 
Cooperative Institute for 
Research in Environmental 
Sciences, University of 
Colorado at Boulder. 
Snow and ice research is conducted 
at the U.S. Army Cold 
Regions Research and 
Engineering Laboratory in Hanover, NH. Research 
addresses engineering and design for cold regions, cold 
weather hydrology, climatology of winterstorms, and 
studies on snow, ice, and permafrost. 
The Snow Survey and Water Supply Forecasting 
Program operated by the Natural Resources 
Conservation Service in Portland, OR, determines the 
relationship of snowpack and potential runoff from 
mountain watersheds through field surveys. The net-
work of data sites provides daily data on streamflow 
potential that are useful for predicting snowmelt runoff 
and preparing water supply forecasts. 
MITIGATION APPROACHES 
Experience has shown that no area can prepare fully for 
severe winterstorms. The Superstorm of March 1993 
was among the most widely forecast winter events. 
Nonetheless, the unprecedented southward penetration 
of blizzard conditions paralyzed parts of the South for 
more than a week. 
Source: From NOAA, National Weather Service, 1994 
• Damage to the Southeastern States, particularly in 
southeastern Georgia and central eastern North 
Carolina, centered on beachfront property and marinas, 
although heavy inland agricultural damage 
caused by extremely high winds was reported. The 
storm caused considerable flooding along the Outer 
Banks of North Carolina. 
• Areas of the southern Appalachians as far south as 
Alabama reported blizzard conditions, with heavy 
snow combining with rapidly falling temperatures 
and very high winds. Snow accumulation and frigid 
temperatures in northern Georgia collapsed roofs, 
downed power lines, and resulted in at least 27 fatalities, 
most of which were due to exposure. 
• Blizzard conditions in the central Appalachians were 
less severe than those reported to the south, primarily 
due to reduced wind speeds. The worst snowfalls 
in five decades were reported in portions of eastern 
Kentucky and West Virginia. A state of emergency 
was declared in 25 counties in eastern Ohio due to 
snowfall and heavy winds. 

SEVERE WINTERSTORMS 81
Specific snow and ice storm mitigation approaches and 
program information currently in use throughout the 
United States were not identified for review and incorporation 
into this report. However, measures may 
include enhanced building codes, planned deployment 
of resources, underground utility lines for critical facilities, 
and increased tree trimming along utilities. 
RECOMMENDATIONS 
Recommendations pertinent to severe winterstorm hazards 
are included in the 1993 Natural Disaster Survey 
Report: Superstorm of March 1993 (NOAA, 1993). 
BIBLIOGRAPHY AND REFERENCES 
Byrant, E. 1991. Natural Disasters. New York, NY: 
Cambridge University Press. 
Coch, N.K. 1995. GEOHAZARDS: Natural and 
Human. Upper Saddle River, NJ: Prentice Hall. 
Kocin, P.J., and L. W. Uccellini. 1990. Snowstorms 
Along the Northeastern Coast of the United States: 
1955 to 1985. American Meteorological Society, 
Meteorological Monographs. Vol. 22, No. 44. 
National Climatic Data Center. 1993. Solar and 
Meteorological Surface Observation Network 
1961–1990, Version 1.0. (CD-ROM). U.S. Department 
of Commerce, National Climatic Data Center, and U.S. 
Department of Energy, National Renewable Energy 
Laboratory. 
NOAA, National Weather Service. 1994. Natural 
Disaster Survey Report: Superstorm of March 1993, 
March 12-14, 1993. 


Photo
84 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
Chapter Summary
E
xtreme summer weather is characterized by a combination 
of very high temperatures and exception-
ally humid conditions. When persisting over a 
period of time, it is called a heat wave. Many areas of the 
United States are susceptible to heat waves. 
The major threat of extreme summer weather is heatstroke, 
a medical emergency that can be fatal. Most at risk are 
outdoor laborers, the elderly, children, and people in poor 
physical health. The combined effects of high temperatures 
and high humidity are more intense in urban centers 
than in rural areas. The States that have experienced a 
higher degree of exposure to this hazard are Louisiana, 
Arkansas, Illinois, South Dakota, Arizona, Florida, and 
Pennsylvania. 
Approximately 200 deaths a year are attributable to 
extreme heat. In 1980, when summer temperatures 
reached all-time levels in most of the Central and Southern 
States, more than 1,700 deaths were diagnosed as heat-
related. A July 1995 heat wave caused 670 deaths—375 in 
the Chicago area alone. 
Extreme summer weather is also hazardous to livestock 
and agricultural crops. It can cause water shortages, exacerbate 
fire hazards, and typically prompts excessive 
demands for energy. Roads, bridges, and railroad tracks 
are susceptible to damage from extreme heat. In the summer 
of 1988, a drought/heat wave in the Central and 
Eastern States resulted in $40 billion in damage as well as 
many fatalities. 
Air conditioning effectively mitigates the effects of excessive 
heat on humans. However, a study by Kilbourne 
(1989) found that, at temperatures above 99ºF, the 
increased air movement produced by fans may actually 
exacerbate heat stress. 

EXTREME SUMMER WEATHER 85
Photo of a sunny day
86 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
HAZARD IDENTIFICATION 
Generally, heat stress is divided into four categories 
(Table 8-1). Steadman (1979) developed a heat index 
that includes the combined effects of high temperature 
and humidity. The NWS gathers and compiles information 
used to estimate the index, and index values are distributed 
to the public and the weather broadcasting 
industry. The heat index is a measure of the severity of 
extreme summer weather. 
An estimation of the heat index is a relationship 
between dry bulb temperatures (at different humidities) 
and the skin's resistance to heat and moisture transfer. 
Because skin resistance is directly related to skin temperature, 
a relation between ambient temperature and 
relative humidity versus skin (or apparent) temperature 
can be determined. If the relative humidity is higher (or 
lower) than the base value, then the apparent temperature 
is higher (or lower) than the ambient temperature. 
Steadman (1979) presented data relating air temperature 
and relative humidity to the heat index or the apparent 
temperature using various factors. To indirectly arrive 
at a "heat index equation" using more conventional variables, 
a multiple regression analysis was performed on 
the data from Steadman's table. 
The major human risks associated with extreme heat are 
described below. 
• Heatstroke. Heatstroke, considered a medical emergency, 
is often fatal. It occurs when perspiration and 
the vasomotor, hemodynamic, and adaptive behavioral 
responses to heat stress are insufficient to pre-
TABLE 8-1.—Heat index/heat disorders 
Source: From National Weather Service, 1997. 
vent a substantial rise in core body temperature. 
Although standardized diagnostic criteria do not 
exist, a medical condition is usually designated as 
heatstroke when rectal temperature rises above 105ºF 
as a result of environmental temperatures (Kilbourne, 
1989). Patients may be delirious, stuporous, or 
comatose. Rapid cooling is essential to prevent permanent 
neurological damage or death. The death-to-
care ratio in reported cases varies from 0 to about 40 
percent, and averages about 15 percent (Vicario and 
others, 1986). 
• Heat Exhaustion. Much less severe than heatstroke, 
heat exhaustion victims may complain of dizziness, 
weakness, or fatigue. Body temperature may be normal 
or slightly to moderately elevated. The primary 
cause of heat exhaustion is fluid and electrolyte (salt) 
imbalance due to increased perspiration in response to 
intense heat. Therefore, treatment is directed to the 
normalization of fluid and electrolyte status, and the 
prognosis is generally good (Knochell, 1974). 
• Heat Syncope. Usually associated with exercise by 
people who are not acclimatized, heat syncope refers 
to a sudden loss of consciousness. Consciousness 
returns promptly when the person lies down. The 
cause is thought to be circulatory instability in 
response to heat, and the condition causes little or no 
harm to the individual. 
• Heat Cramps. Heat cramps occur when people 
unaccustomed to heat exercise outdoors. The cramps 
are thought to be due to mild fluid and electrolyte 
imbalances and generally cease to be a problem after 
acclimatization (Knochell, 1974). 
RISK ASSESSMENT 
Probability and Frequency 
The apparent temperature or heat 
index is a quantitative measure of 
extreme summer weather that can be 
used to characterize the probability 
and frequency of heat hazards. Map 
8-1 shows one approach to describe 
the geographic distribution and frequency 
of extreme summer weather. 
It was developed from 30 years 
(1961-90) of relative humidity and 
temperature data at 48 climatic stations 
in the conterminous United 
States (NOAA and DOE, 1993). 
Danger Category Heat Disorders 
Apparent 
Temperature 
(0F) 
IV Extreme Danger Heatstroke or sunstroke imminent. >130 
III Danger Sunstroke, heat cramps, or heat 
exhaustion likely; heat stroke 
possible with prolonged exposure 
and physical activity. 
105-130 
II Extreme Caution Sunstroke, heat cramps, and heat 
exhaustion possible with prolonged 
exposure and physical activity. 
90-105 
I Caution Fatigue possible with prolonged 
exposure and physical activity. 
80-90 

EXTREME SUMMER WEATHER 87
U.S. Map 8-1Map 8-1. Severity and areal extent of extreme summer heat in the United States, based on NWS heat 
index. Data not available for Alaska, Hawaii, Puerto Rico, U.S. Virgin Islands, 
and Pacific Territories. 
Source: Data from U.S. Department of Commerce and Others, 1993. 
Because of the limited number of climatic stations, the 
map shows a broad generalization of trends in heat 
index, and may not show all areas that are subject to 
extreme summer weather. 
Data used to compile Map 8-1 included hourly readings 
for between 2 p.m. and 5 p.m. for June, July, and 
August, assuming that the annual maximum temperature 
and relative humidity occurs during summer after-
noons. The annual maximum values for the four time 
readings and the three months were used in a frequency 
analysis to determine the percent chance of a given heat 
index being exceeded in any year. The annual maxi-
mum values were ranked, and the annual percent chance 
of exceedance was computed by the Weibull plotting 
position formula to determine the heat index with a 5 
percent chance of being exceeded in any given year. 
Minor adjustments to the data were required to address 
apparent measuring or recording errors, and temperatures 
below 70ºF were not used. 
Exposure 
Each year, many areas of the United States and its territories 
experience periods of prolonged high temperatures 
combined with high humidity. In susceptible 
areas, people usually are aware of the hazard, anticipate 
it, and are accustomed to avoiding its potentially dangerous 
effects. However, extreme summer heat does 
strike areas not accustomed to the phenomenon, where 
people tend to be less prepared. 
The areas subject to heat index values in excess of 
115ºF for the 5-percent-annual-chance event are the 
Southeastern, Southwestern, and Midwestern States 
(Map 8-1). Areas near the coast experience a combination 
of high temperatures and relative humidity that 
results in high apparent temperatures. Southwestern 
States experience low to moderate relative humidity, but 
extremely high temperatures. 
Extreme summer weather poses the greatest danger to 
outdoor laborers, the elderly, children, people in poor 
physical health, and people residing in homes without 
air conditioning. During prolonged periods of extreme 
summer weather, local governments, voluntary organizations, 
and medical and health-care facilities may be 
burdened. 
Previous research indicates that people over age 75 are 
most susceptible to extreme summer heat. The elderly 
are more likely to have chronic diseases or to be taking 

88 NATURAL HAZARDS: ATMOSPHERIC HAZARDS
medications that can increase risks. Sweating is the 
body's natural mechanism for reducing high body temperature, 
and the body temperature at which sweating 
begins increases with age. People taking neuroleptic 
and anticholinergic drugs should be counseled regarding 
possible increased sensitivity to heat (Kilbourne, 
1989). 
More deaths from extreme summer weather occur in 
urban centers than in rural areas. One reason that the 
effects of hot weather may be more extensive in urban 
areas is that poor air quality may exacerbate severe conditions. 
The masses of stone, brick, concrete, and 
asphalt that are typical of urban architecture absorb 
radiant heat energy from the sun during the day and 
radiate that heat during nights that would otherwise be 
cooler. Tall city buildings may effectively decrease 
wind velocity, thereby decreasing the contribution of 
moving air to evaporative and convective cooling. The 
heat differential was documented during the 1980 heat 
wave in Kansas City, where there was a difference of 
2.5ºC in the daily maximum temperature and 4.1ºC in 
the daily minimum temperature between downtown 
and the suburban airport. 
Consequences 
In years during which a major heat wave does not occur 
in the United States, an average of approximately 200 
heat-related deaths are reported (Kilbourne, 1989). 
When prolonged periods of abnormally high temperature 
or heat waves affect large areas, the number of 
deaths attributed to heat rises greatly. In 1980, when 
summer temperatures reached all-time highs in much of 
the Central and Southern States, over 1,700 deaths were 
identified as heat-related. 
The Central Plains and Corn Belt States experienced a 
heat wave during July 15-19, 1995 during which apparent 
temperatures climbed above 120ºF. A significant 
portion of the Eastern States was in the danger category 
during that same period, with apparent temperatures 
ranging from 105ºF to 120ºF. 
The relative poverty of some urban areas may con-
tribute to the severity of the extreme summer weather 
hazard. Low-income people are less able to afford 
cooling devices and the energy needed to operate them 
(Kilbourne, 1989). 
The U.S. Department of Commerce reported that the 
1995 heat wave caused 670 deaths—375 in the Chicago 
metropolitan area alone. Many deaths were among 
low-income elderly in residential units not equipped 
with air conditioning. Local utilities were forced to 
impose controlled power outages because of excessive 
energy demands, and water suppliers reported very low 
levels of water in storage. This intense heat also caused 
the loss of tens of millions of cattle and poultry 
throughout the Midwest. 
When heat waves are accompanied by drought, agricultural 
losses can be high. A drought/heat wave during 
the summer of 1993 affected the Southeastern United 
States, causing approximately $1 billion in damage and 
an undetermined number of deaths. Another 
drought/heat wave during the summer of 1988 affected 
the Central and Eastern United States, causing approximately 
$40 billion in damage and many deaths. A 
drought/heat wave during the summer of 1980 caused 
an estimated $20 billion in damage and many deaths. 
Extreme summer weather can cause damage to roads, 
bridges, and railroads. High temperatures can be partially 
responsible for deflection of rails and related rail-
road accidents. 
RESEARCH, DATA COLLECTION, AND 
MONITORING ACTIVITIES 
Several scientists have attempted to develop mathematical 
models to quantify the increase in number of deaths 
expected for a given temperature increase. These models 
take into account such factors as the usual seasonal 
trends in mortality, acclimatization, the age of structures, 
and previous hot weather exposure of the at-risk 
population. However, they have not yet been useful in 
the prediction of adverse, heat-related health effects 
(Kilbourne, 1989). 
MITIGATION APPROACHES 
As with most natural hazards, public education about 
the effects of extreme heat and how to mitigate those 
effects is useful. NWS, through local television and 
radio announcements, alerts people about the onset of 
extreme summer weather and poor air quality. The 
alerts, advising high risk people to reduce physical 
activity and stay in air-conditioned buildings, have 
helped reduce fatalities and injuries. 
BIBLIOGRAPHY AND REFERENCES 
Kilbourne, E.D. 1989. "Heat Waves." In The Public 
Health Consequences of Disasters 1989. CDC 
Monograph. U.S. Department of Health and Human 
Services, Public Health Service, Centers for Disease 
Control, Atlanta, GA. 
Knochel, J.P. 1974. "Environmental Heat Illness: An 
Eclectic Review." Archives of Internal Medicine. Vol. 
133, pp. 841-864. 

EXTREME SUMMER WEATHER 89
National Weather Serivice. 1997. Online. Accessible 
at http://www.nws.noaa.gov/sr/ohx/heat.htm. 
Steadman, R.G. 1979. "The Assessment of Sultriness. 
Part I: A Temperature-Humidity Index Based on 
Human Physiology and Clothing Science." Journal of 
Applied Meteorology. Vol. 18, pp. 861-873. 
U.S. Department of Commerce, National Oceanic and 
Atmospheric Administration, National Weather 
Service. July 1995. Weekly Climate Bulletin. 
U.S. Department of Commerce, National Oceanic and 
Atmospheric Administration, National Weather 
Service. July 1995-September 1996. Climate Outlook. 
National Oceanic and Atmospheric Administration, 
National Climatic Data Center, and U.S. Department of 
Energy, National Renewable Energy Laboratory. 
September 1993. Solar and Meteorologic Surface 
Observation Network, 1961-1990, Version 1 on 3 CD-
ROMs. 
Vicario, S.J., R. Okabajue, and T. Haltom. 1986. 
"Rapid Cooling in Classic Heatstroke: Effect on 
Mortality Rates." American Journal of Emergency 
Medicine. Vol. 4, pp. 394-398.