D

uring the summer, hikers and canoeists often

flock to Minnesotas Boundary Waters Canoe
Area Wilderness (BWCAW) to observe the
beauty and tranquility of this remote land that is
filled with picturesque lakes and forests. However,
on the July 4th weekend in 1999, the serenity so often
identified with the region briefly gave way to a terrify-
ing display of one of natures most violent stormsa
derecho.
On this holiday weekend, camp counselors Emily
Boyd and Maddy Bennett were guiding a group of
six teenage girls on a weeklong canoe exploration
out of Moose Lake near Ely, Minnesota. Unknown
to the group, a large windstorm producing gusts in
excess of 40 m s
1
had already cut a swath of dam-
age across northern Minnesota and was heading
straight for them. The group was canoeing on Knife
Lake in the BWCAW around noon when threatening
skies brought the group to shore. The group quickly
beached their canoes and took cover.
As the storm hit, Emily and Maddy quickly pulled
the girls together and huddled over them. Watching
trees crash around them, the two leaders hurried the
group to an area that appeared safer near the waters
edge. Although the storm seemed to last an eternity,
blue sky appeared on the western horizon a mere 30
min later.
After the storm, the group returned back to the
location where they had huddled to gather their life
vests. To their shock, felled trees had crushed the
vests. One can only imagine in horror what would
have happened to the girls if Emily and Maddy had
not moved them to the safer location. Emily later
recalled that it was a true miracle that no one [from
their group] died in the storm that day. Shocked,
yet uninjured, the group spent the next two days
paddling over 20 miles and making six portages for
home. Regrettably, 25 other hikers and canoeists
in the area were not as lucky and suffered injuries,
including broken backs and necks. During the next
several days, rescue crews searched 2,200 camp-
sites in the BWCAW and Superior National Forest,
evacuating about 20 campers by air (Breining 2000).
Unfortunately, the storm killed two campers farther
downstreamone in Quebec, Canada (Mainville
AFFILIATIONS: ASHLEYMeteorology Program, Department of
Geography, Northern Illinois University, DeKalb, Illinois; MOTE
Climatology Research Laboratory, Department of Geography,
The University of Georgia, Athens, Georgia
CORRESPONDING AUTHOR: Walker Ashley, Meteorology
Program, Department of Geography, Northern Illinois University,
DeKalb, IL 60115.
E-mail : [email protected]
DOI :10.1175/BAMS-86-11-1577
In final form 20 July 2005
2005 American Meteorological Society
DERECHO HAZARDS
IN THE UNITED STATES
BY WALKER S. ASHLEY AND THOMAS L. MOTE
Derechos are shown to be as hazardous as many tornadoes and hurricanes
that affect the United States.
1577 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
1999), and another in Vermont (NCDC 1999). The
Vermont death occurred as a person camping in a tent
on a raft was blown into the water, became entangled
in the tent, and subsequently drowned.
Convectively generated windstorms occur over
a wide array of temporal and spatial scales (Fujita
1978; Fujita and Wakimoto 1981; Golden and Snow
1999); however, the more widespread and longer
lived of these windstorms are termed derechos
(Hinrichs 1888; Johns and Hirt 1987). Though an
extreme derecho case, the 4 July 1999 windstorm
that swept through the BWCAW is an unforgettable
example of the ferocity that can be associated with
derechos.
Although, Johns and Hirt (1987), Johns (1993),
and Wakimoto (2001) have claimed that derechos
account for much of the structural damage and
casualties resulting from convectively induced non-
tornadic winds, no study has examined the damage
and casualty statistics from derechos. Changnon et al.
(2001) suggest the threat from weather disasters is
likely to grow because of a number of societal factors,
including rapid population growth and expansion,
together with increases in wealth, development, and
urbanization. In order to illustrate the future risk
and potential vulnerability of the U.S. population
to these extreme windstorms, the following study
reveals the hazards
1
associated with derechos by
examining the climatology, casualty statistics, and
damage estimates of events that occurred during the
18-yr period 19862003. It is initially hypothesized
that derechos can be as hazardous, due to relatively
high frequencies and large spatial extents, as some
hurricanes and tornadoes.
BACKGROUND. In 1888, Iowa physical scientist
Gustavus Hinrichs termed any convectively induced
straight-line windstorm, or straight blow of the
prairie, a derecho (a Spanish derivative meaning
straight-ahead or direct) in correspondence
with the Spanish derivative tornado (based on the
Spanish word tornarmeaning to to turn). Fujita
and Wakimoto (1981) illustrated many of the same
characteristics of Hinrichs definition, labeling these
events as either a downburst cluster (major damage
axis between 40 and 400 km) or for larger events, a
family of downburst clusters (major damage axis of
400 km or more). Broadly, Johns and Hirt (1987) de-
fined the derecho to include any family of downburst
clusters (Fujita and Wakimoto 1981) produced by an
extratropical convective system, but also developed
specific criteria to define derechos utilizing con-
temporary terms and datasets (cf. their section 2).
Later studies by Bentley and Mote (1998), Evans
and Doswell (2001), Bentley and Sparks (2003), and
Coniglio and Stensrud (2004) have utilized differing
criteria to define derechos.
United States derechos have been documented in
the literature since the early 1980s (e.g., Johns and
Hirt 1983); only recently have enough events been
documented over a sufficient period to begin to ana-
lyze the climatology of these windstorms. Johns and
Hirt (1987) were the first to present a climatology of
these events for the warm seasons from 1980 to 1983.
Bentley and Mote (1998) and Bentley and Sparks
(2003) recently analyzed the years 19862000, provid-
ing the basis for a 15-yr derecho climatology, includ-
ing both cool- and warm-season events. These stud-
ies identified the tendency for derechos to occur in
specific seasonal corridors in the eastern two-thirds
of the United States. A separate study by Coniglio
and Stensrud (2004) attempted to further interpret
the climatology of derechos by classifying systemati-
cally, based on specific intensity classifications, 244
derechos from 1986 to 2001 (cf. their Table 1). These
climatologies do contain minor differences but, in
general, substantiate each other in regard to the rela-
tive spatial and temporal distributions. Nevertheless,
none of the studies could be considered a definitive
climatology because of issues regarding the severe
storm wind-event database (e.g., Doswell and Burgess
1988; Weiss et al. 2002; and Schaefer et al. 2003),
the varying criteria utilized (Coniglio and Stensrud
2004), and the relatively short time period examined
(Bentley and Sparks 2003). Furthermore, no study
has examined thoroughly the hazards associated
with derechos. Therefore, this investigation examines
the casualties and damage produced by these events
in order to provide researchers, policy makers, and
emergency managers with valuable information re-
garding the often-overlooked extreme straight-line
windstorms known as derechos.
RESEARCH METHODOLOGY. Derecho dataset.
The contiguous U.S. derecho dataset utilized in this
study was compiled through several sources including
two long-term climatological studiesBentley and
Sparks (2003) and Coniglio and Stensrud (2004). First,
1
Derecho hazard encompasses some aspect of derecho cli-
matology, but also includes the effect these storms have on
people and the built environment. In terms of this study, a
derecho hazard is defined as any derecho that results in a
human casualty or any amount of reported economic dam-
age. This is analogous to the tornado hazard description
provided by Boruff et al. (2003).
1578 NOVEMBER 2005 |
a dataset containing 230 derechos identified by Bentley
and Mote (1998) and Bentley and Sparks (2003) for the
period 19862000 was obtained (M. L. Bentley 2003,
personal communication). Data utilized in the Bentley
and Mote (1998) and Bentley and Sparks (2003) stud-
ies were derived from the Storm Prediction Centers
(SPCs) online database of several convective wind
gusts and the SPCs SeverePlot software (Hart 1993).
Bentley and Mote and Bentley and Sparks (2003) modi-
fied existing derecho identification methods proposed
by Johns and Hirt (1987) in order to facilitate analysis
of the large dataset (cf. their Table 1 or Coniglio and
Stensrud 2004, their Table 1).
Second, the derecho database employed in stud-
ies by Coniglio and Stensrud (2004) and Coniglio
et al. (2004), consisting of 244 events from 1986 to
2001, was acquired (available online at www.nssl.
noaa.gov/users/mcon/public_html /derlist.htm).
Coniglio compiled the database (hereafter Coniglio
and Stensrud 2004 dataset) utilizing the SPCs severe
convective wind database, SeverePlot software, and
available radar data.
In addition to the Bentley and Sparks (2003) and
Coniglio and Stensrud (2004) derecho datasets,
the authors examined all derecho literature and
documented any missing derechos not revealed by
the aforementioned climatological studies. Finally,
several additional events (e.g., 13 March 1993) were
added where the Johns and Hirt (1987) length criteria
were not met (e.g., squall line crossing the relatively
narrow Florida peninsula), but were indeed derechos.
Finally, the authors documented derechos from 2002
and 2003 by examining the SPCs daily online severe
storm reports, SPCs severe thunderstorm event da-
tabase, Storm Data, and SeverePlot.
In order to be consistent with the derecho iden-
tification methodology outlined by Coniglio and
Stensrud (2004), all derechos that were previously
not identified utilizing radar data [namely, those
events identified by Bentley and Mote (1998) and
Bentley and Sparks (2003)] were verified using avail-
able radar resources from the National Climatic Data
Center (NCDC; DIFAX radar summaries, 198695),
SPC (2-km regional composites, 200003), and the
National Aeronautics and Space Administrations
(NASAs) Global Hydrology Resource Center (8-km
national composites, 19952003). All events in the
database were scrutinized in order to make sure that
multiple swaths of damage were a part of the same
MCS as indicated by the radar data. Nine events from
the Bentley and Mote (1998) and Bentley and Sparks
(2003) dataset did not verify using this radar criterion
and were removed from our dataset.
Every effort was made to remove from the dataset
cases in which swaths of damage were caused solely
by supercells rather than quasi-linear convective sys-
tems. However, the limitations of the composite radar
data made it difficult in some cases to distinguish
storm types, especially prior to 1995. Therefore, it
is possible that some supercell, particularly high-
precipitation supercell, cases remain in the dataset
prior to 1995.
Additionally, no attempt was made to include the
requirement of a minimum of three 33 m s
1
wind
gust reports as originally suggested by Johns and Hirt
(1987). There appear to be a number of unresolved
issues that suggest that the inclusion of the 33 m s
1

wind gust criteria may promote irregularities and
biases in the dataset. These unresolved issues include
population biases (Johns and Evans 2000), changes
in the verification practices (Schaefer et al. 2003),
significant irregularities in extreme wind gust data
when examined by National Weather Service (NWS)
county warning areas of responsibility (Weiss et al.
2002), and the fact that Hinrichs (1888) and Fujita
and Wakimoto (1981) made no reference to wind
gust magnitudes in defining a derecho or a series of
downburst clusters (Bentley and Mote 2000). Conse-
quently, a set of consistent criteria (Table 1) without
this secondary wind requirement was employed to
construct the derecho dataset. This set of criteria is
similar to that proposed and utilized by Bentley and
Mote (1998), Bentley and Sparks (2003), Evans and
Doswell (2001), Coniglio and Stensrud (2004, their
low end criteria), and Coniglio et al. (2004).
After all events were compiled, the datasets cross-
checked to form a unified dataset. The derecho tracks
were used to extract individual severe wind reports
from SeverePlot and Storm Data. In the case of dif-
fering start (end) times obtained from coinciding
Bentley and Sparks (2003) and Coniglio and Sten-
srud (2004) events, the earliest (latest) wind report
obtained was used to identify the derecho start (end)
time and location. The wind reports were examined
to ensure that each event had spatial and temporal
continuity. Through the procedures above, a derecho
dataset has been derived for the United States utiliz-
ing consistent criteria (Table 1). In all, 377 events were
identified for the 18-yr period 19862003, an average
of nearly 21 events per year (Table 2).
Wind damage and gust reports from each derecho
were mapped onto a 1 latitude 1 longitude grid
to examine the spatial distribution of derechos. The
distributions were determined by identifying grid
cells with at least one wind report for a given event
and then summing the number of events affecting
1579 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
each grid cell. Contour maps were created using
inverse-distance interpolation (Davis 1986). This
technique smoothes maximum values and therefore
may underestimate extreme values in the plots. Thus,
actual maximum values of derecho frequency are
indicated on maps.
Casualty data. Since 1959, Storm Data has been the
primary source of severe event data utilized by meteo-
rologists and climatologists for locating areas of storm
damage and determining the number of casualties
produced by significant weather events. The process
by which these data are gathered has been reviewed
by Curran et al. (2000, their section 2) and Bentley
et al. (2002).
Although Storm Data contains the best informa-
tion on storms affecting the United States, it is not
all-inclusive owing to the difficulties inherent in the
collection of these types of data. Several studies have
illustrated the problems associated with Storm Data,
although most have focused on the underreporting of
casualties or damage produced by lightning (Curran
et al. 2000) or hail (Changnon 1999). As with hail and
lightning casualty tallies and damage estimates, dere-
cho-related casualties and damage likely receive less
attention than large-impact events such as floods,
hurricanes, and tornadoes. Thus, absolute values
obtained from Storm Data should be considered with
caution (Curran et al. 2000). As with any significant
weather event, deaths and injuries are more likely
to be accurately reported than damage, which can
involve intricate estimating procedures (Changnon
2003). Therefore, there is more confidence in the
casualty data than the damage estimate data. Finally,
there are a number of casualties that are indirectly
related to the storm (e.g., casualties due to clean-
up operations). These fatalities and injuries are not
included in storm causality tallies.
Despite these inherent problems with Storm Data,
it is the only consistent data source for storm-induced
casualties for the period of record. In this study, the
casualty information from Storm Data was utilized
without alteration.
For a particular derecho event, Storm Data
casualty statistics were coordinated with the derecho
wind report data in a geographic information system
(GIS) to ensure that all casualties were a consequence
of the straight-line winds from a corresponding
derecho. No casualties due to tornadic winds within
derecho-producing convective systems were included
in the derecho-induced, straight-line wind casualty
analysis. Finally, fatality data were obtained and tabu-
lated for both hurricanes (Tropical Prediction Center
2004) and tornadoes (SeverePlot and Storm Data)
in an attempt to compare contiguous U.S. derecho
fatalities with those attributable to hurricanes and
tornadoes. Only hurricane fatalities produced by a
landfalling or near miss hurricane (i.e., in which
the outer bands of the hurricane made it onto the
contiguous U.S. shoreline) were included in the U.S.
hurricane fatality statistics.
Finally, a similar interpolation procedure was uti-
lized to reveal the spatial distribution of derecho ca-
sualties. In this case, the sum of all injuries on a 11
grid was interpolated to illustrate the geographic
patterns associated with derecho injuries.
Damage estimates. Storm Data is the primary U.S.
government source of estimated storm-induced
damage data available to researchers. Initially, Storm
Data was utilized to estimate damage totals for
derechos in this study. However, estimates extracted
from Storm Data were promptly ruled inadequate by
the authors because 1) damage estimates produced
by NWS offices and compiled within the publication
are arbitrary and subjective; 2) in few instances are
TABLE 1. Criteria used to identify derechos for this study.
Minimum length
There must be a concentrated area of convectively induced wind gusts greater
than 26 m s
1
that has a major axis length of 400 km or more (unless a land con-
straint necessitates using a shorter distance).
Chronological progression
The wind reports must have chronological progression, either as a singular swath
(progressive) or a series of swaths (serial), and nonrandom pattern of occurrence
by temporally mapping the wind reports of each event.
Temporal and spatial restriction
No more than 2.5 h can elapse between successive wind reports with no more
than 2 of latitude and longitude separating successive wind reports.
Origin of wind swath
Multiple swaths of damage must be part of the same MCS as indicated by examin-
ing available radar data.
MCS continuity
The associated MCS, as indicated by available surface pressure and wind fields
and/or radar data, must have temporal and spatial continuity.
1580 NOVEMBER 2005 |
NWS offices provided with legitimate damage esti-
mates from necessary parties, emergency managers,
or insurance companies; 3) reporting inconsisten-
cies and a difference in reporting policies between
NWS offices (R. L. Beasley 2004, personal com-
munication); and 4) lack of estimates for described
damages. Therefore, a second, independent record
of estimated losses to the insurance industry was
employed to provide a potential assessment of the in-
sured losses due to derechos and to illustrate further
the major drawbacks when estimating damage totals
using solely the resources of Storm Data. These
datathe Property Claims Services (PCSs) catas-
trophe databasewere obtained from the Insurance
Services Office, Inc. (ISO; G. Kerney 2004, personal
communication). The catastrophe database includes
estimated industry-wide
insurance payments for
property lines of insurance
covering fixed property,
building contents, time-
element losses, vehicles,
and inland marine diverse
goods and properties (ISO
2004). Crop losses due to
peri l s are not i ncluded
but insured farm build-
ings and equipment are
included in the catastro-
phe estimates.
From 1986 to 1996, the
PCS database identified an
event (or series of related
events) as a catastrophe if
that storm episode caused
over $5 million in total in-
sured property losses. In
1997, a shift was made from
$5 to $25 million in order
to adjust for the effects of
inf lation on the selection
of catastrophes (Changnon
2001; ISO 2004). These data
have been utilized in the
past to estimate property
losses resulting from thun-
derstorm perils (Chang-
non 2001) and hurricanes
(Tropical Prediction Center
2004). Furthermore, a re-
cent evaluation of loss data
from natural hazards iden-
tified these PCS data as the
nations best available data (National Research Council
1999; Changnon 2001).
For each qualifying thunderstorm-caused catas-
trophe, the PCS dataset included the event date(s), the
conditions/perils causing the losses, and the amount
of insured losses by state. Unfortunately, the insured
losses are estimated for all storm perils, making it dif-
ficult to separate the cost of each peril (e.g., hail from
wind). In effect, then, extracting only derecho wind
damage loss from the PCS estimates was not possible
except in a single case (20 November 1989) where the
only peril associated with the catastrophe was wind.
In all other cases, losses from flooding, hail, and/or
tornadoes were included in the catastrophe database.
In these situations, only insured damage from dere-
chos that lack considerable tornado, hail, and flood
TABLE 2. The number of derechos, fatalities, and injuries for the 18-yr
period of record. Included are deaths and injuries from thunderstorm
winds during years in which tallies were available in Storm Data. Italicized
statistics indicate that the summary figures provided are for an 11-yr
period. Asterisk indicates preliminary data.
Year
Derechos Thunderstorm winds
Events Deaths Injuries Deaths Injuries
1986 10 6 134
1987 14 8 113
1988 2 0 3
1989 15 13 126
1990 10 7 196
1991 11 6 157
1992 12 2 136
1993 13 2 154 25 461
1994 21 6 93 15 337
1995 31 18 212 38 473
1996 24 11 142 23 335
1997 25 13 72 37 425
1998 42 21 606 41 860
1999 32 14 146 29 325
2000 31 10 87 25 296
2001 26 9 123 17 341
2002 29 6 42 17 287
2003 29 1 63 19 226
*
Sum 377 153 2605 286 4366
Mean 20.9 8.5 144.7 26 396.9
Max 42 21 606 41 860
Min 2 0 3 15 226
1581 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
events were tallied utilizing the
PCS data. In making this subjec-
tive judgment, the use of detailed
verbal descriptions of damage that
often accompany the events in
Storm Data, the analysis utilities
of NCDCs Severe Storm Event
database, and the mapping utili-
ties of SeverePlot were employed
to ensure that the damage losses
were predominantly due to the
straight-line winds associated
with a derecho. Events that had
substantial damage due to a tor-
nado along the leading edge of a
bow-echo-producing system were
not excluded. Conversely, events
that had any significant (F2 or
greater) tornadoes that were sepa-
rate from the derecho-producing
MCS and produced considerable
damage were removed from the
estimating procedure. Only 42
derechos met these criteria and
were included in derecho insured-
loss analysis.
The aforementioned proce-
dure indicates the dif f icult y
in trying to estimate damage
losses strictly from derechos.
Estimating damage tallies from
any storm peril is a subjective
procedure (Changnon 2003). The
estimates provided are utilized
for comparative purposes and to
obtain a sense of the damage po-
tential from these extreme wind-
storms. The estimates are not
intended to be accurate enough
to provide precise, flawless dere-
cho damage tallies.
For comparison purposes,
the estimated loss data were
adjusted for inf lation (to 2003
dollars in this study), utilizing an
implicit price deflator for gross
national product, as reported
FIG. 1. The frequency of U.S. derechos, 19862003. Symbol indicates
the location of maximum value before interpolation.
FIG. 2. The seasonal frequency of
U.S. derechos, 19862003. Symbols
indicate the locations of maximum
values before interpolation.
1582 NOVEMBER 2005 |
in the Economic Report of the President (Office of
the President 2004). Because this study does not
examine trends in normalized damage tallies, it is
not as important to control for wealth or population
as was the case in previous research on hurricanes
and tornadoes (Pielke and Landsea 1998; Brooks and
Doswell 2001).
RESULTS. Climatology. Studies by Johns and Hirt
(1987), Bentley and Mote (1998), Bentley and Sparks
(2003), and Coniglio and Stensrud (2004) have
revealed the major geographical distributions and
high-frequency corridors of derechos. However,
each of the climatologies presented in the past have
employed differing datasets and mapping techniques
to reveal the distribution of derechos in the United
States. The jointly compiled dataset utilized in this
study [i.e., including events from two recent long-
term derecho climatologiesBentley and Sparks
(2003) and Coniglio and Stensrud (2004)] provides
a unique opportunity to extend (i.e., with addition of
2002/03 derechos) and briefly reevaluate the preexist-
ing climatology of derechos. This process highlights
the threat of derechos to any one particular region of
the United States and subsequently may be employed
to develop a risk assessment of derechos.
When examining derechos for the entire year for
the 18-yr period of record (Fig. 1), a similar derecho
spatial pattern to that produced by Bentley and Sparks
(2003, their Fig. 14) is evident. Prominent features in
the derecho geographical distribution include 1) a
Southern Great Plains maximum located over the
northeast portion of Oklahoma, 2) an axis of higher
derecho frequencies extending from the Oklahoma
maximum southeastward toward the southern
Mississippi Valley region, and 3) a high-frequency
axis that corresponds with Johns (1982, 1984) north-
west-flow severe weather events extending from the
upper Midwest into the Ohio Valley. The Southern
Great Plains maximum likely occurs since this region
is affected by both cool- and transition-season serial
(Johns and Hirt 1987) events associated with travel-
ing midlatitude cyclones (Fig. 2b) and warm-season,
progressive (Johns and Hirt 1987) events (Fig. 2a)
while more poleward locales are affected by primar-
ily warm-season, progressive events. Similar to the
results of Bentley and Mote (1998) and Coniglio and
Stensrud (2004), cool-season derecho maxima are
found along the southern Mississippi Valley, while
warm-season event maxima are found in the south-
central Great Plains and the Ohio Valley.
Temporally, derechos are primarily warm-season
events, with 69% of the events in the compiled data-
set occurring between May and August during the
18-yr period (Fig. 3). As established in Bentley and
Sparks (2003), May has the highest derecho monthly
frequency with, on average, nearly five events occur-
ring annually during this month.
Casual ti es. Summary studies of thunderstorm-
induced, straight-line wind casualties are limited.
In fact, most investigations that examine annual
summaries of weather-related fatalities or injuries
typically focus on f loods, lightning, hurricanes,
and/or tornadoes (e.g., Riebsame et al. 1986; Curran
et al. 2000). Therefore, it is important to illustrate the
hazards associated with derechos by first examining
the number of casualties these events have produced
in the past and compare those statistics with those of
other thunderstorm-related perils.
Fatalities and injuries were tallied for all 377
derechos in the study. There were 153 fatalities dur-
ing the 18-yr period of record (8.5 yr
1
) attributable
to damaging straight-line winds from derechos. The
number of fatalities per year is highly variablefrom
21 in 1998 to no fatalities in 1988. Examining the
derecho fatalities spatially (Fig. 4) indicates an in-
teresting distribution that does not necessarily cor-
respond with derecho frequencies across the United
States (Fig. 1). Three northern statesMichigan
(17 fatalities), New York (23), and Ohio (16)contain
nearly 37% of all derecho fatalities. This is somewhat
counterintuitive as one would expect the highest
fatality rates in regions with the greatest likelihood
of derecho occurrences (i.e., across the south-central
FIG. 3. Monthly U.S. derecho frequency, 19862003.
1583 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
Great Plains). Several possible explanations could ac-
count for this unusual distribution, including 1) the
tendency for more outdoor-related activities (e.g.,
camping) in state parks and wilderness areas of Mich-
igan and New York (increased vulnerability); 2) boat-
ing activities along the Great Lakes (increased vulner-
ability); 3) the increased likelihood of particularly
intense, warm-season derechos across the northern
tier of the United States (Coniglio and Stensrud 2004)
(increased vulnerability); 4) a heightened awareness
of severe storms by people in the Southern Great
Plains states due to the high frequency of extreme
thunderstorm-related perils in this region (reduced
vulnerability); and 5) the existence of better warn-
ing systems in the Southern Great Plains (reduced
vulnerability). Thus, it is possible that there is some
underlying integration of both physical and social
vulnerabilities attributable to the observed derecho
fatality distribution (Riebsame et al. 1986).
When examining derecho fatalities by type,
boating and vehicular deaths accounted for nearly
50% of all fatalities (Fig. 5). In the majority of cases,
vehicular fatalities occurred in one of three ways:
1) overturned tractor semi-trailer, 2) felled tree land-
ing on automobile, or 3) an automobile driven into a
felled tree. Marine fatalities principally occurred as
drownings when either sailing vessels or motorized
boats were overturned due to
high derecho winds.
The deadl iest derechos
on record include events oc-
curring on 16 August 1997
(8 fatalities), 31 May 1998 (6),
15 July 1995 (6), 12 July 1995
(5), and 20 July 1987 (5). In
terms of total injuries and fa-
talities, there are several dere-
chos that have caused over 100
reported casualties, including
31 May 1998 (210), 9 April
1991 (135), 4 June 1993 (110),
and 10 February 1990 (103).
On average, 145 injuries
per year were attributable to
derechos; however, annual
values were highly variable
with a maximum of 606 inju-
ries occurring in 1998 and a
minimum of 3 injuries occur-
ring in 1988 (Table 1). In terms
of spatial distribution (Fig. 6),
derecho injuries were clustered
around several specific regions
including 1) Lake Michigan, 2) the Interstate-95 cor-
ridor in the Northwest, 3) the Ohio River Valley, 4) the
interior of the Southeast, and 4) the south-central
Great Plains. Like fatalities, the higher frequencies of
injuries tend to occur outside of regions of the highest
derecho frequency maxima. The states of Kentucky
FIG. 4. U.S. derecho fatalities by state, 19862003. Circles indicate fatality
locations.
FIG. 5. 19862003 derecho fatalities as classified by
location of occurrence.
1584 NOVEMBER 2005 |
(333 injuries), Michigan (200), and
Illinois (187) are highest in terms
of derecho injuries.
Unlike fatalities, a considerable
number of the injuries reported in
Storm Data are not accompanied
by a description of how the inju-
ries occurred other than that they
were caused by thunderstorm-
related winds. In fact, nearly 40%
of all derecho injuries reported in
Storm Data have no description of
injury type. The remaining 60%
of injuries were classified accord-
ing to how or where the injury
occurred, revealing a different
distribution by type (Fig. 7) than
derecho fatalities (Fig. 5). Injuries
in mobile homes (23% of classifi-
able injuries) and vehicles (21%
of classifiable injuries) lead all other injury types by
a considerable margin. Other high-frequency injury
types (accounting for nearly 10% of classifiable inju-
ries each) include camping, flying debris, permanent
structures/homes, and temporary structures (i.e.,
recreational or special-event tents).
Comparing derecho casualties to those produced by
all thunderstorm winds for the period 19932003 indi-
cates that derechos do not account for the majority of
casualties due to nonhurricane and nontornadic winds.
In terms of fatalities (injuries), derechos accounted
for 38.8% (39.9%) of all fatalities (injuries) caused by
thunderstorm winds during this 11-yr period.
Derecho, hurricane, and tornado casualty compari -
sons. In order to illustrate that derechos can be as
hazardous as most hurricanes and tornadoes, a
comparison between derecho-induced fatalities and
those produced from contiguous U.S. hurricanes and
tornadoes was constructed for the derecho dataset
period of record (Table 3). Derecho fatalities exceed
fatalities from F0 and F1 tornadoes by a wide margin,
but account for fewer deaths than those produced by
F0, F1, and F2 tornadoes combined. If one considers
that F0 and F1 tornadoes account for nearly 88% of
all U.S. tornadoes from 1986 to 2003, then derecho
fatalities exceed the number of fatalities produced by
most tornadoes. Notwithstanding, significant (F2
or greater) tornadoes contribute a disproportionate
number of tornado deaths, with these strong to
violent events accounting for 92.7% of all tornado
fatalities occurring in the United States during the
period of study.
Hurricane fatalities surpass those caused by
derechos; however, if the anomalously high fatality
rates from Floyd (56) and Fran (34) are removed,
the fatality statistics are essentially comparable. It is
important to consider that hurricane fatality statistics
include deaths related to hurricane-spawned torna-
does, floods, and riptides. For this study, those deaths
attributable to tornadoes embedded within or floods
attributable to derecho-producing convective systems
were not included. Hence, it is possible that fatalities
from derechos exceed those fatalities induced solely
by hurricane winds.
FIG. 6. The distribution of U.S. derecho injuries. Symbol indicates the
gridded maximum value before interpolation.
FIG. 7. 19862003 derecho injuries as classified by
location of occurrence. Only those injuries that were
classifiable based on Storm Data descriptions are
provided.
1585 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
Damage estimates. Straight-line winds associated with
derechos have been officially measured at over 60 m s
1

(e.g., 64.4 m s
1
on 22 April 1997, 67.5 m s
1
on 16 July
1980), equivalent to the sustained winds of a category
4 hurricane. A number of damage surveys from Storm
Data suggest higher wind gusts have occurred with
some events. Although most wind speeds in derechos
never approach these levels of severity, the straight-line
winds meeting even minimal derecho criteria (26 m s
1
)
can topple trees onto automobiles, overturn tractor
semi-trailers, or damage mobile homes. Undoubtedly,
derechos have the ability to produce substantial dam-
age in regions impacted by these events.
Previous studies that have documented derechos
have very limited descriptions regarding damage
summaries or estimates. When estimates are present-
ed for events, they are typically deduced from Storm
Data. There is only one case in the literature that
highlights, in detail, the damaging potential of a dere-
cho. Fujita and Wakimoto (1981) provided extensive
documentation of the 16 July 1980 derecho that pro-
duced widespread damage across large areas of Michi-
gan, Illinois, Wisconsin,
and Minnesota. They in-
dicated t hat t his storm
produced approximately
$650 mil lion in damage
as it traversed the four-
state region. Accounting
for inflation (to 2003 dol-
lars), this storm produced
an estimated $1.3 billion
in damage from strictly
straight-line winds. This
est i mate exceeds many
damage tallies from U.S.
hurricanes and is larger
than the inflation-adjusted
damage esti mates from
all major tornadoes that
have affected the United
States since 1890 (Brooks
and Doswell 2001). This
single event illustrates that
derecho damage can exceed
the damage from most hur-
ricanes and tornado events
affecting the contiguous
United States.
Unfortunately, detailed
damage summaries such
as those provided by Fujita
and Wakimoto (1981) are
not available for other derechos. Therefore, the PCS
catastrophe dataset was employed to estimate the im-
pact derechos have had on the insured built environ-
ment. In total, 206 of the 377 derechos in the dataset
(54.6%) were associated with 129 separate PCS ca-
tastrophe events. All thunderstorm perils (flooding,
hail, tornadoes, and wind) associated with these 129
catastrophe events were responsible collectively for
nearly $33 billion (2003 dollars) in insured losses. In
some cases multiple derechos [i.e., derecho families
(Bentley and Sparks 2003; Ashley et al. 2004)] were a
part of the same PCS-defined catastrophe.
A detailed process was utilized to extract PCS
damage estimates for derechos. Estimated insured
losses due to derechos are provided to illustrate the
devastating impact these events can have on the built
environment and compare these statistics with analo-
gous data from U.S. hurricanes. A number of extremely
intense derechos or families (e.g., 8 July 1993, 15 May
1998 family, 47 July 2003 family) were excluded from
this analysis because these events were accompanied
by considerable damage produced by flooding, hail, or
TABLE 3. The number of fatalities due to derechos, hurricanes, F0 and
F1 tornadoes, and F0, F1, and F2 tornadoes for the 18-yr period utilized
in this study.
Year Derechos Hurricanes
F0 and F1
tornadoes
F0, F1, and F2
tornadoes
1986 6 8 0 10
1987 8 0 2 11
1988 0 4 3 16
1989 13 37 9 21
1990 7 0 2 9
1991 6 15 8 12
1992 2 23 3 5
1993 2 3 6 19
1994 6 8 2 9
1995 18 23 3 12
1996 11 48 3 9
1997 13 2 7 14
1998 21 5 2 17
1999 14 60 6 14
2000 10 1 1 5
2001 9 0 5 19
2002 6 0 6 22
2003 1 17 3 5
Sum 153 254 71 229
1586 NOVEMBER 2005 |
tornadoes and, therefore, could not be accurately as-
sessed for damages solely due to derecho winds. Hence,
the high end damage potential of the most intense
derechos is likely not illustrated in this analysis.
In all, nine derechos and derecho families pro-
ducing more than $100 million in insured losses
were identified from the methods utilized in this
study (Table 4). The costliest derecho identified was
TABLE 4. Estimated insured losses due to catastrophic derechos identified in this study. A derecho or series
of derechos is indicated for each catastrophe. Asterisks indicate that two separate derechos occurred on
the same day and are included in the same catastrophe damage estimate. Parenthetical state identifica-
tions designate states that were not included in the PCS or Storm Data estimates but were impacted by the
derecho. Estimated (insured and uninsured) property damage totals from all perils for the corresponding
PCS catastrophe data were constructed from Storm Data for 19962002 events. All estimates account for
inflation and are adjusted to 2003 dollars.
Derecho event(s) States affected
Damage estimate (millions of $)
PCS Storm Data
6 Jul 1987 IL, IN, KY, MO 43
4 May 1989 OK, TX (LA) 180
22 May 1989 AR, KS, MO, OK 20
2 Jul 1989 LA, OK, TX (AR) 61
20 Nov 1989 CT, DE, MD, NJ, NY, PA 74
7 Jul 1991 IN, MI, OH, WI (IA, NY, PA) 77
2 Jul 1992 IL, IN (IA, MO) 37
1 May 1993 TX (NM) 48
4 Jun 1993 IL, IN, KY, MO, VA, WV (NC, TN) 78
10 Jul 1993 OH (WV, PA) 42
31 May 1994 UT (CO, WY) 42
1 Jul 1994 KS, NE (OK) 48
4 Apr 1995 CT, MA, NJ, NY 69
15 Jul 1995 MA, NY, PA 75
24 Jul 1995 OK (AR, KS) 63
21 May 1996 CT, MA, RI (NJ, NY) 39 5
6 Aug 1996 MN, WI (IA, MI, NE, SD) 23 3
29 Oct 1996 IL, IN, OH (IA, MI, WI) 68 7
19 May 1997 PA (OH, NY) 28 1
31 May 1998 MI, MN, WI (IA, NY) 432 455
4, 5, 6 Jun 1998 AL, AR, GA, MS, TN, TX 191 18
18, 20, 21 Jul 1998 IN, MI, OH, WI (MN, PA) 98 58
7 Sep 1998 CT, MD, MI, NH, NJ, NY, PA, VT, WV 252 203
10 Nov 1998
*
IL, IN, MI, MO, OH (AR, MS, TN, TX, OK) 98 12
26 Apr 1999 AR, LA, OK, TX 162 2
23, 25 Jul 1999 MI, MN, WI (SD) 103 3
30 Jul 1999 MI, WI (MN) 92 4
9 May 2000
*
IL, IN, MI, OH (MO) 106 19
18 May 2000 CT, NJ, NY, PA (MA) 79 3
16 Feb 2001 AL, GA, LA, MS 176 27
27 May 2001 KS, OK (TX) 98 28
9 Mar 2002
*
IA, IL, KS, MO, NY, OH, PA, WV 137 24
7, 8 Apr 2002 LA, MS 20 1
1587 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
the 31 May 1998 event that affected the Great Lakes
region. This event was responsible for $432 mil-
lion in insured property losses across the states of
Minnesota, Wisconsin, and Michigan. Interestingly,
this is the only event in which Storm Data property
damage estimates exceeded values obtained from
the insured losses estimated from PCS. In all other
derecho cases from 1996 to 2002, PCS estimates ex-
ceed, in some instances, by an order of magnitude,
the estimates provided by Storm Data. Storm Data
accounts for only 39.6% of the PCS damage losses
from the 27 events in which both Storm Data and
PCS data were available. This is especially troubling
since Storm Data estimates account for both insured
and noninsured losses while PCS data report only
insured losses. This suggests that estimated damage
totals from derechos cannot be deduced from Storm
Data.
The PCS catastrophe database includes insured
losses from hurricanes and is the primary dataset
employed by NOAAs Tropical Prediction Center to
estimate damage from Hurricanes. [The Tropical
Prediction Center utilizes a 2:1 ratio to estimate total
damage from hurricanes and therefore multiplies the
PCS insured loss estimate by two to obtain a hur-
ricane total (J. L. Beven 2004, personal communica-
tion). Since there is no published research to support
this ratio, it was not utilized in this study.] The most
damaging hurricanes affecting the United States from
1986 to 2003 were compared to the estimated insured
damages from derechos obtained from this study
(Table 5). Clearly, certain hurricanes (e.g., Andrew,
Hugo, etc.) are in a category
unto themselves and result
in enormous insured dam-
age estimates. However, in-
dividual derechos or fami-
lies of derechos appear to
approach the damage po-
tential of some of the most
damaging hurricanes in the
18-yr period of record. For
example, the 31 May 1998
case exceeded estimates
from Hurricanes Erin and
Bonnie and approached the
insured losses produced by
Hurricane Lilly. It is likely
that other extreme derecho
events have exceeded the
31 May 1998 case (e.g., 16
July 1980) and are compa-
rable to the most damaging
hurricanes in U.S. record.
Forest blowdowns. Not in-
cluded in either PCS or
Storm Data estimates is the
impact derechos have on
both private and public for-
ests throughout the United
States. Several derechos
have produced extensive
forest blowdowns, includ-
i ng t he Independence
Day Downbursts 4 July
1977 (3440 km
2
of forest af-
fected; see Fujita 1985), the
two Minnesota derechos
TABLE 5. The most damaging hurricanes that have directly impacted the
contiguous United States from 1986 to 2003 in comparison to the most
damaging derechos (highlighted in red) identified in this study utilizing
the procedures outlined in the methodology. Category indicates the
intensity (using the SaffirSimpson hurricane scale) of each hurricane as
it made landfall or approached the coast (for non-landfalling hurricanes).
Asterisks indicate that two separate derechos occurred on the same day
and are included in the same catastrophe damage estimate. All estimates
account for inflation and are adjusted to 2003 dollars.
Year Storm Catagory
PCS estimate
(millions of $)
1992 Andrew 5 18,985
1989 Hugo 4 3,993
1995 Opal 3 2,411
1999 Floyd 2 2,117
1996 Fran 3 1,803
2003 Isabel 2 1,685
1998 Georges 2 1,264
1991 Bob 4 766
2002 Lili 1 437
1998 31 May 432
1995 Erin 1 431
1998 Bonnie 3 394
1998 7 Sep* 252
1998 4, 5, 6 Jun 191
1989 4 May 180
2001 16 Feb 176
1999 26 Apr 162
1996 Bertha 2 152
2002 9 Mar* 137
1999 Irene 1 108
1588 NOVEMBER 2005 |
of 13 and 14 July 1995 (810 km
2
destroyed; NCDC
1995), the Adirondack derecho of 15 July 1995 (3642
km
2
affected, 505 km
2
sustaining moderate to severe
damage; NCDC 1995), and the Boundary Waters
derecho of 4 July 1999 (2691 km
2
affected, 1934 km
2

destroyed; Parke and Larson 2004; Price and Murphy
2002). The meteorological community often labels
these blowdown events as Pakwashs after the Pak-
wash Provincial Forest in northwest Ontario, Canada,
that was impacted severely by a windstorm produced
by a high-precipitation supercell that occurred on
18 July 1991 (Cummine et al. 1992). Derecho-pro-
duced blowdowns have altered forest landscapes and
community dynamics by influencing tree mortality
rates, reducing tree size and structure, decreasing
forest diversity, and modifying species composition
by advancing succession status (Peterson 2000).
The financial impact these events have on for-
ests is difficult to approximate. Some assessments
have been provided, including a monetary estimate
of the impact the 15 July 1995 derecho had on the
Adirondack Park in New York. The New York
Department of Environmental Conservation indi-
cated that the timber damage due to the derecho was
estimated at 1 billion board feet with an estimated
value of $234 million (2003 dollars; NCDC 1995).
Unfortunately, damage to forests is not the only haz-
ard realized by these events. In some cases, campers
and hikers visiting these forests during the height
of the tourist season have been killed or seriously
injured by derecho-felled trees.
Derechos have even sparked intense political de-
bates at the state and federal levels regarding what to
do with the disturbed forests left in the wake of these
windstorms. Both the Adirondack and Boundary
Waters derechos were responsible for extremely
contentious debates on whether salvage logging
operations should be permitted in public forests
after large blowdowns. In the case of Adirondack
Park, salvage operations had been in place for nearly
50 yr on the grounds of fire prevention and forest
conservation. However, following the assessment of
state conservation officers and consulting ecologists,
the state of New York enacted to forego salvaging
operations in the Adirondack forest after the dere-
cho. The final assessment concluded that the large
forest disturbance produced by the derecho should
be treated as a normal ecosystem process (Robinson
and Zappieri 1999).
Finally, the potential for catastrophic fires in for-
ests affected by derechos is often extremely high due
to dramatic increases in the volatility and fuel load-
ing brought about by the felled timber. For example,
the U.S. Forest Service Fuels Risk Assessment report
detailing the 4 July 1999 derecho impacts on the
BWCAW suggested that it is not a matter of if, but
when, a large significant wildland fire will threaten
the wilderness (Leuschen et al. 2000).
SUMMARY AND CONCLUSIONS. Assessing
the human and economic impacts of derechos has
received considerably less attention in comparison to
large impact events such as floods, hurricanes, and
tornadoes. By consolidating and extending the record
of U.S. derechos, this investigation quantitatively and
qualitatively revealed the impacts derechos have had
on the nation from 1986 to 2003. Results indicate that
derecho hazards (defined as derechos that result in
injury, death, or economic loss) can be as substantial
as hurricanes and tornado hazards.
For the 18-yr period investigated, derechos were
responsible for 153 fatalities and over 2,600 injuries.
These casualties were highly variable yet appeared to
be coupled with the corresponding annual frequency
of derechos. Derecho fatalities are comparable to
those produced by recent U.S. hurricanes and ex-
ceed the fatalities caused by more than 88% of the
tornadoes that have affected the United States from
1986 to 2003. Individually, derechos were responsible
for up to 8 fatalities and 204 injuries, indicating that
these storms can have a substantial human impact.
Derecho fatalities have a propensity to occur more
frequently in vehicles or while boating, while injuries
are more likely to happen in vehicles or mobile homes.
Both fatalities and injuries are more inclined to oc-
cur outside of the region that possesses the highest
derecho frequency. An underlying synthesis of both
physical and social vulnerabilities is suggested as the
cause for the casualty distribution. Contrary to asser-
tions in previous studies (Johns and Hirt 1987; Johns
1993; Wakimoto 2001), results from this investigation
suggest that derechos do not account for a majority
of the casualties owing to convectively induced non-
tornadic winds.
Utilizing catastrophe statistics compiled from
the insurance industry, this investigation was able
to obtain a sense of the financial impact derechos
can incur. Damage estimates obtained from the PCS
insurance catastrophe database were compared with
similar figures from Storm Data to reveal inadequa-
cies in the U.S. governments only publication esti-
mating losses from natural disasters. Results suggest
that Storm Data severely undercounts derecho dam-
age totals (in some cases, by an order of magnitude)
and is therefore inadequate for estimating damage
totals from derechos. It is likely that Storm Data is
1589 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
not practicable for estimating damage totals for other
thunderstorm-related perils, as well.
A number of derechos were extracted from the
dataset with the intention of providing an estimate
of the insured loss due to these windstorms. Insured
losses from many derechos and derecho families were
found to exceed $100 million dollars, with one event
(31 May 1998) nearing $500 million in estimated in-
sured losses. These estimated losses are comparable
to some of the most noteworthy U.S. hurricanes in
the last 18 yr and are equivalent to the most damaging
tornadoes in the U.S. history (Brooks and Doswell
2001). According to Fujita and Wakimoto (1981),
derechos have exceeded the $1 billion dollar threshold
in the past; however, in this study no events inducing
this amount of economic impact were identifiable due
to constraints imposed by the data and procedures
employed. Derecho damage tallies are likely to be
even higher than those measured in this study as no
estimate was provided for noninsured damages or
governmental expenditures, since some states that
were impacted by derechos were not included in the
damage estimates, and because several of the most
extreme derechos in the past 18 yr were not included
in the damage analysis. Further, the economic impact
due to derecho-induced forest blowdowns is often
indeterminate and is therefore typically excluded
from the loss estimates. In one case (15 July 1995),
an official estimate suggested that over $230 million
in timber was lost due to a derecho.
Future study should try to extend the clima-
tological record to increase our understanding of
how these events form and where they are most
frequent. A considerable effort should be made to
increase the amount of in-depth loss investigations
into future derechos in order to obtain meaningful
estimates of actual economic impacts produced by
these windstorms. For example, detailed studies of
the 1988 drought (Riebsame et al. 1991), Hurricane
Andrew (Pielke 1995), and the Great Flood of 1993
(Changnon 1996) have assisted in creating a new
awareness of data problems and have improved the
estimating procedures associated with each of these
perils (Changnon 2003). As suggested by Changnon
(2003), the nation needs a concerted effort, and a con-
tinuing program, to routinely assess and measure the
losses from weather extremes, including derechos, in
order to adequately monitor the ever-growing impact
of weather disasters, improve mitigation efforts, and
make informed decisions on policies that address
hazard issues.
In summary, the results of this analysis indicate
that derechos can be as hazardous, and are compa-
rable in magnitude to, most U.S. tornadoes and hur-
ricanes. The study has attempted to inform scientists,
emergency managers, insurers, and the public about
the severity of derechos so that they may take steps
to mitigate the potential future hazards of these tre-
mendous windstorms.
ACKNOWLEDGEMENTS. Thanks to Dr. Mace
Bentley for providing insightful comments that improved
the initial version of the manuscript, Jamie Dyer for pro-
viding an analysis routine, Drs. Mace Bentley and Michael
Coniglio for the availability of their derecho datasets, and
Gary Kerney and Jim Welsh for supplying the PCS data and
answering many questions regarding the collection of these
data. The suggestions from three anonymous reviewers im-
proved substantially the quality of this manuscript. Finally,
sincere thanks to Emily Boyd who graciously shared her
Boundary Waters derecho account and answered numer-
ous questions about the event as well as provided pictures
of the damage. This research was supported by a coopera-
tive agreement between the USDA Forest Service and the
University of Georgia.
REFERENCES
Ashley, W. S., T. L. Mote, and M. L. Bentley, 2004:
Derecho families. Preprints, 22d Conf. on Severe
Local Storms, Hyannis, MA, Amer. Meteor. Soc.,
CD-ROM, P4.4.
Bentley, M. L., and T. L. Mote, 1998: A climatology of
derecho-producing mesoscale convective systems
in the central and eastern United States, 198695.
Part I: Temporal and spatial distribution. Bull. Amer.
Meteor. Soc., 79, 25272540.
, and , 2000: A reply to comments on A clima-
tology of derecho producing mesoscale convective
systems 198695. Part I: Temporal and spatial distri-
bution. Bull. Amer. Meteor. Soc., 81, 10541057.
, and J. A. Sparks, 2003: A 15 yr climatology of
derecho-producing mesoscale convective systems
over the central and eastern United States. Climate
Res., 24, 129139.
, T. L. Mote, and P. Thebpanya, 2002: Using Landsat
to identify thunderstorm damage in agricultural
regions. Bull. Amer. Meteor. Soc., 83, 363376.
Boruff, B. J., J. A. Easoz, S. D Jones, H. R. Landry, J. D.
Mitchem, and S. L. Cutter, 2003: Tornado hazards in
the United States. Climate Res., 24, 103117.
Breining, G., 2000: Boundary Waters: The fire next
time. Minnesota Conservation Volunteer, Minnesota
Department of Natural Resources, MayJune. [Avail-
able online at www.dnr.state.mn.us/volunteer/may-
jun00/blowdown.html.]
1590 NOVEMBER 2005 |
Brooks, H. E., and C. A. Doswell III, 2001: Normalized
damage from major tornadoes in the United States:
19801999. Wea. Forecasting, 16, 168176.
Changnon, S. A., 1996: Losers and winners: A summary
of the floods impacts. The Great Flood of 1993, S.
Changnon, Ed., Westview Press, 276299.
, 1999: Data and approaches for determining hail
risk in the contiguous United States. J. Appl. Meteor.,
38, 17301739.
, 2001: Damaging thunderstorm activity in the Unit-
ed States. Bull. Amer. Meteor. Soc., 82, 597608.
, 2003: Measures of economic impacts of weather
extremes. Bull. Amer. Meteor. Soc., 84, 12311235.
, A. D. Changnon, and G. Hewings, 2001: Losses
caused by weather and climate extremes: A national
index for the U.S. Phys. Geogr., 22, 127.
Coniglio, M. C., and D. J. Stensrud, 2004: Interpreting
the climatology of derechos. Wea. Forecasting, 19,
595605.
, , and M. B. Richman, 2004: An observational
study of derecho-producing convective systems.
Wea. Forecasting, 19, 320337.
Cummine, J., P. McCarthy, and M. Leduc, 1992: Blow-
down over Northwestern Ontario. A derecho event
18 July 1991. Preprints, 4th Workshop on Operational
Meteorology, Whistler, BC, Canada, Atmospheric
Environment Service/Canadian Meteorological and
Oceanographic Society, 311317.
Curran, E. B., R. L. Holle, and R. E. Lpez, 2000: Light-
ning casualties and damages in the United States
from 1959 to 1994. J. Climate, 13, 34483464.
Davis, J. C., 1986: Statistics and Data Analysis in Geol-
ogy. 2d ed. John Wiley and Sons, 656 pp.
Doswell, C. A., III, and D. W. Burgess, 1988: On some
issues of United States tornado climatology. Mon.
Wea. Rev., 116, 495501.
Evans, J. S., and C. A. Doswell III, 2001: Examination of
derecho environments using proximity soundings.
Wea. Forecasting, 16, 329342.
Fujita, T. T., 1978: Manual of downburst identification
for project NIMROD. SMRP Res. Paper 156, Univer-
sity of Chicago, NTIS Accession No. N78-30771/7GI,
104 pp.
, 1985: The DownburstMicroburst and Macro-
burst. University of Chicago Press, 122 pp.
, and R. M. Wakimoto, 1981: Five scales of airflow
associated with a series of downbursts on 16 July
1980. Mon. Wea. Rev., 109, 14381456.
Golden, J. H., and J. T. Snow, 1991. Mitigation against
extreme windstorms. Rev. Geophys., 29, 477504.
Hart, J. A., 1993: SVRPLOT: A new method of access-
ing and manipulating the NSSFC severe weather
database. Preprints, 17th Conf. Severe Local Storms,
St. Louis, MO, Amer. Meteor. Soc., 4041. [Software
available online at www.spc.noaa.gov/software/
svrplot2/.]
Hinrichs, G., 1888: Tornadoes and derechos. Amer.
Meteor. J., 5, 341349.
ISO, 2004: Catastrophe loss estimate procedure descrip-
tion. Property Claim Services, 7 pp.
Johns, R. H., 1982: A synoptic climatology of northwest
flow severe weather outbreaks. Part I: Nature and
significance. Mon. Wea. Rev., 110, 16531663.
, 1984: A synoptic climatology of northwest flow
severe weather outbreaks. Part II: Meteorological
parameters and synoptic patterns. Mon. Wea. Rev.,
112, 449464.
, 1993: Meteorological conditions associated with
bow echo development in convective storms. Wea.
Forecasting, 8, 294299.
, and W. D. Hirt, 1983: The derecho: A severe
weather producing convective system. Preprints,
13th Conf. on Severe Local Storms, Tulsa, OK, Amer.
Meteor. Soc., 178181.
, and , 1987: Derechos: Widespread convectively
induced windstorms. Wea. Forecasting, 2, 3249.
, and J. S. Evans, 2000: Comments on A climatology
of derecho producing mesoscale convective systems
198695, Part I: Temporal and spatial distribution.
Bull. Amer. Meteor. Soc., 81, 10491054.
Leuschen, T, T. Wordell, M. A. Finney, D. Anderson,
T. Aunan, and P. Tin. 2000: Fuels risk assessment
of blowdown in Boundary Waters Canoe Area
Wilderness and adjacent lands: Superior National
Forest, Duluth, MN, USA. [Available online at www.
superiornationalforest.org/july4thstorm1999/bw-
cara/bwcawra.html.]
Mainville, S., 1999: The derecho of 45 July 1999 in
southern Quebec. Environment Canada Tech. Note
99N-04, Quebec Region, Environment Canada,
29 pp.
National Research Council, 1999: The costs of natural
disasters: A framework for assessment. National
Academy Press, 68 pp.
NCDC, 1995: Storm Data. Vol. 37, No. 7, 236 pp.
, 1999: Storm Data. Vol. 5, No. 41, 352 pp.
Office of the President, 2004: Economic report of
the president. U.S. Government Printing Office,
Washington, DC, 417 pp. [Available online at www.
gpoaccess.gov/eop/.]
Parke, P. S., and N. J. Larson, 1999: The boundary waters
windstorm of July 4, 1999. NWS, Duluth, MN. [Avail-
able online at www.crh.noaa.gov/dlh/science/.]
Peterson, C. J., 2000: Catastrophic wind damage to
North American forests and the potential impact of
climate change. Sci. Total Environ., 262, 287311.
1591 NOVEMBER 2005 AMERICAN METEOROLOGICAL SOCIETY |
Pielke, R. A., Jr., 1995: Hurricane Andrew in South Florida:
Mesoscale Weather and Societal Responses. ESIG,
National Center for Atmospheric Research, 156 pp.
, and C. W. Landsea, 1998: Normalized hurricane
damages in the United States. Wea. Forecasting, 13,
621631.
Price, C. G., and B. P. Murphy, 2002: Lightning activ-
ity during the 1999 Superior derecho. Geophys. Res.
Lett., 29, 571
Riebsame, W. E., H. F. Diaz, T. Moses, and M. Price,
1986: The social burden of weather and climate haz-
ards. Bull. Amer. Meteor. Soc., 67, 13781388.
, S. A. Changnon, and T. Karl, 1991: Drought and
Natural Resources Management in the United States:
Impacts and Implications of the 19871989 Drought.
Westview Press, 174 pp.
Robinson, G. R., and J. Zappieri, 1999: Biodiversity
policy in time and space: Lessons from divergent
approaches to salvage logging on public lands.
Conservation Ecology, 3. [Available online at www.
ecologyandsociety.org/vol3/iss1/art3/.]
Schaefer, J. T., S. J. Weiss, and J. J. Levit, 2003: The
frequency of severe thunderstorm winds over the
contiguous United States. 11th Int. Conf. On Wind
Engineering, Lubbock, TX, International Association
of Wind Engineers and Wind Science Engineering
Research Center at Texas Tech University, 8 pp.
[Available online at www.icwe.ttu.edu/icwe.htm.]
Tropical Prediction Center, 2004: NHC/TPC archive of
past hurricane seasons: Hurricane season tropical
cyclone reports. [Available online at www.nhc.noaa.
gov/pastall.shtml.]
Wakimoto, R. M., 2001: Convectively driven high wind
events. Severe Convective Storms, Meteor. Monogr.,
No. 50, Amer. Meteor. Soc., 255298.
Weisman, M. L., 1993: The genesis of severe, long-lived
bow echoes. J. Atmos. Sci., 50, 645670.
Weiss, S. J., J. A. Hart, and P. R. Janish, 2002: An
examination of severe thunderstorm wind report
climatology: 19701999. Preprints, 21st Conf. Severe
Local Storms, San Antonio, TX, Amer. Meteor. Soc.,
446449.
1592 NOVEMBER 2005 |