1 Introduction

Downbursts and tornadoes are among of the most widespread, persistent, and destructive hazards to affect Argentina and several other countries (Rasmussen et al. 2014; Durañona 2016). Atmospheric scientists have given continuous attention to these phenomena for about four decades but, to our knowledge, no urban vulnerability studies have been conducted to date. Studies on tornado and downburst damage in Argentina are comparatively fewer and involve damage reconnaissance reports of individual buildings (Balbastro and Sonzogni 2008; Natalini et al. 2012a) and estimates of regional impact of severe storms in northeast Argentina (e.g., Natalini et al. 2012b).

A severe storm impacted the city of Bell Ville, Argentina, on December 6, 2012, causing widespread damage. In order to learn about the damage characteristics of the storm, two surveys were conducted to characterize the storm and quantify the damage, respectively: the first a few days after the storm (on December 15 and 16) and the second almost eight months after the event. In the first survey, numerous observations were conducted on site that helped to define the footprint and intensity of the storm. The second survey, conducted from August 2–4, 2013, was led by the authors jointly with Ecosud Foundation, Volunteer Fire Dept., and the State Police. Around 560 local volunteers, including high school students, boy scouts, and other citizens, administered around 4000 questionnaires. The survey was widely advertised on local media and was favorably received by remarkable social participation.

The information gathered from the survey afforded insights into the damage types and the interconnectedness of building and infrastructure in a city similar in many respects to numerous others in South America. Furthermore, since the observed intensities of the downburst and tornado are the most frequent in the central region of Argentina and neighbor countries, the knowledge gained and recommendations proposed herein can be applied in other cities (see Schwarzkopf and Rosso 1993).

Bell Ville is a typical small Argentine city of about 40,000 inhabitants and an urban area of 15 km2. The city is crossed by the Tercero River and is located within the most fertile agricultural and grazing region of Argentina known as Humid Pampas comprised of the provinces of Córdoba, Santa Fe, and Buenos Aires (Fig. 4). The Pampas are traversed by a network of small- to medium-sized towns and settlements with populations ranging from about 200 to 200,000. The cities are separated from each other by uniform distances of about 15, 30, or 60 km (Villarroel 1976). This specific region of Argentina has experienced the highest occurrence of tornados and downbursts as will be discussed below.

2 Aspects of the damage reconnaissance

Information on the extent of damage caused by the storm was gathered using questionnaires administered on a house-by-house basis by local volunteers who were trained by engineers and architects who explained the technical details. The damage was measured using repair cost ratios, with mild (≤33 %), moderate (34–66 %), and severe (>66 %) damage levels defined for roofs, walls, windows, trees, and other elements. Information about the property characteristics was also gathered, including general features of the building (age and typology, surroundings), elements in the vicinity of the property (sheds, walls, trees), and building features (dimensions, roof type, wall type, windows).

The reconnaissance was focused on the western and southern parts of the city where most of the damage occurred. The zone was divided into five sections and virtually every property there was visited. Additional phone surveys were made on November 2013 to January 2014 to supplement the damage characterization in the eastern part of the city.

Caution should be exercised when interpreting the damage results because clean up will have taken place in the eight-month delay between the storm and the damage reconnaissance. Faded memories from some participants may also have affected the validity of the data collected, especially in the amount of damage estimated. Some other aspects intrinsic to the administration of the questionnaires also add uncertainty to the results including: absence of participants at the time of the visit, missed instances of damage by interviewers or interviewees, and reduced sampling in the eastern part of the city. Notwithstanding the shortcomings just mentioned, the storm caused widespread city damage which had not occurred in the recent past and consequently made a lasting impression on the population. This along with the fact that the reconstruction was very slow might lend some weight to the representativeness of the data to the type and extent of damage.

3 Characteristics of downbursts

Downbursts were characterized meteorologically in the mid-1970s and early 1980s by Fujita and Caracena (Gedzelman 1990). A downburst is a strong downward draft of air that emanates from a cloud and becomes a rapid divergence of high winds on or near the ground (Fig. 1). In terms of dimensional scale, downbursts are classified as macrobursts and microbursts (Fujita 1985; Caracena et al. 1989; Hjelmfelt 1988). The former has an outflow diameter of damaging surface winds in excess of 4 km persisting from 5 to 30 min while the latter has a diameter 4 km or less and persist for 2–5 min (Wakimoto 1985). In general, microbursts can be more damaging than macrobursts reaching winds as high as 270 kph, as opposed to 215 kph in the latter (Fujita 1985; Rose 1996; Barnes 2001). Microbursts are further classified as dry or wet depending on the presence or absence of precipitation. The terms downburst and microburst are used interchangeably in this paper.

Fig. 1
figure 1

(Left) Conceptual model of a microburst that include streamlines, a ring vortex, the wind profile in light blue, and the stagnation point (+symbol). Based on Fujita (1985) and Wolfson (1988)

Full size image

The wind profile of microbursts differs from the profile of boundary layer winds used for design purposes (CIRSOC 1994; ASCE 2010) in that the profile of the former reaches the maximum at a certain height and then decreases, whereas in the latter the increase is monotonic with height (Fig. 2). The maximum wind speed in microbursts occurs at heights of <5 % of the downdraft diameter (10–50 m) and horizontally at a radius of about 1.5 downdraft diameters from the stagnation point where radial velocity is close to zero (1.5–3 km) (Kim and Hangan 2007; Wood et al. 2001; Holmes 2002). This wind distribution might explain the damage pattern of another storm that occurred on December 2014 in the city of Leones, Argentina, where a small microburst damaged the higher part of several grain silos (bucket elevators and conveyor systems) located 30 m high, but caused no damage at the silos themselves or other elements at a lower level. The question whether the current wind codes are apt for modeling the wind load of a downburst is a field of current development (Letchford and Lombardo 2015), although a recent study has shown that for some typologies, design codes might currently be non-conservative (Jesson et al. 2015).

Fig. 2
figure 2

Horizontal wind profile of a downburst contrasted with a typical boundary layer wind profile. The legend lists the atmospheric boundary layer, a profile developed in Kim and Hangan (2007), and an empirical function and experimental results referenced in the mentioned paper. [Graph reprinted from Journal of Wind Engineering and Industrial Aerodynamics, Vol. 95(4), “Numerical simulations of impinging jets with application to downbursts” by Kim, J., and Hangan, H. Copyright 2007, with permission from Elsevier]

Full size image

The damage caused by microbursts is measured with the same Fujita Scale (F-Scale) used for tornados (Fujita 1971, 1978, 1981; Forbes and Wakimoto 1983; Doswell 2003). The damage pattern of downbursts is characterized by a diffluent pattern, whereas tornado damage exhibits confluence and rotation (see Holmes and Oliver 2000). In open terrain, microburst damage exhibits a characteristic homogeneous pattern of straight and curved line outflows, whereas in urban areas surface wind speeds are reduced by buildings and trees. Urban roughness is, however, non-homogeneous, with abrupt variations in building heights allowing irregular pockets of high wind gustiness to occur, resulting in localized damage (Fig. 3).

Fig. 3
figure 3

Idealized representation of the microburst wind field over different terrains: a wind speeds in unobstructed terrain, b idealized reduction in wind speeds due to a city’s homogeneous roughness, c more realistic depiction of wind speed reduction and pockets of high gustiness due to singularities in the urban roughness generated by local changes in objects’ heights and density

Full size image

After considering the general aspects of the downbursts, we now turn to describing severe storm activity in Argentina.

3.1 Downburst and tornado activity in Argentina

The activity of severe storms in Argentina has been systematically studied since 1971 (Schwarzkopf et al. 1983–1999; Schwarzkopf and Migliardo 1976; Schwarzkopf and Rosso 1993; Soliño and Schwarzkopf 1982; see also Brooks et al. 2003; Dotzek et al. 2003, 2005) with descriptions of events going back to 1930. The approximate severity distribution of recorded downbursts in Argentina, as estimated in a sample collected in the period between 1971 and 1998, is shown in Table 1 in which F0 and F1 events predominate. In comparison, the approximate distribution of downbursts in the United States, sampled in the period 1973–1977, exhibits a larger share of F2 and F3 intensities. The reason for this difference can probably be attributed to the completeness of the series, the fact that the US is larger in size, and also because, for the same weather properties, the storms in the USA have more time and greater longitudinal extension to develop and re-develop. (Fujita 1978; Fujita and Wakimoto 1981; Wheatley et al. 2006).

Table 1 Intensity distribution of samples of downbursts in Argentina and United States
Full size table

With regard to tornado frequency, most of the tornados recorded in Argentina in the period 1930–1999 belong in the range of F0–F3 intensities (Table 2) (Schwarzkopf et al. 1993). The reason that the record for Argentina exhibits fewer F0’s when compared to the US record is that these might go unnoticed or because of the low damages that they might have produced.

Table 2 Percentage of tornados in Argentina from 1930 to 1999 and in the USA from 1950 to 1999
Full size table

The spatial frequency distribution of tornado occurrence in Argentina (1930–1999) has been divided into three zones (Fig. 4), namely Zone A (0.8 × 10−4 tornados/km2/year), Zone B (0.4 × 10−4 tornados/km2/year), and Zone C (0.2 × 10−4 tornados/km2/year). No similar map for downbursts has been developed, although severe storms that generate downbursts have been observed to occur throughout the territory from parallel 43ºS northwards.

Fig. 4
figure 4

Spatial frequency of tornados in Argentina (1930–1999) with name of the provinces in blue. Based on Schwarzkopf et al. (1993), Schwarzkopf and Rosso (1993)

Full size image

4 Characteristics of the severe storm of December 6, 2012

The formation of downbursts is governed chiefly by three conditions: continued upward air motion over a wide area, highly convective available potential energy (CAPE), and marked increase of the moisture content in the preceding 12 h. First, on December 6 afternoon, a closed system of cyclonic air circulation formed in the lower atmosphere (1.5 km) over the center of the country driven by a cold front over La Pampa, a warm front over Córdoba, Santa Fe, and Entre Ríos. In the medium and high atmospheric levels, a large trough approaching from the Pacific Ocean generated a jet with maximum winds of 140 and 160 kts. These atmospheric conditions led to a regular upward motion of the air over the central region of the country.

Second, on December 6 at 09:00 a CAPE value of about 2000 J/kg indicated the availability of energy for the formation of violent storms as the atmosphere over southeast Córdoba became highly unstable (Fig. 5). Finally, in the twelve hours between December 5, 21:00, and December 6, 09:00, there was an increase of moisture (15–20 g/kg) at the level of 850 hPa (1.5 km), and between December 6, 03:00 and 15:00 was of 30–50 g/kg (Fig. 5). As a result, the combination of these three atmospheric aspects indicated a high probability of severe localized storms. In addition, other atmospheric variables which are usually correlated with downburst activity had exceeded their threshold values: unstable values of the Showalter index (−4 to −6), large values of the equivalent potential temperature (θ e  > 350 K), potential instability (−24), and inertial instability (absolute vorticity ~ 0) (see Kunz 2007; Emanuel 1994).

Fig. 5
figure 5

(Left) CAPE (J/kg) (shaded), differential wind intensity between 500 and 925 hPa (magenta contour), convective inhibition CIN (J/kg) (black contour)— December 6, 2012-12GMT; (Right) Moisture convergence (g/kg) in 12 h and equivalent potential temperature θ e (K) contours at 850 hPa— December 6, 2012-18GMT. Analysis of Global Forecast System (GFS) produced by the National Centers for Environmental Prediction (NCEP). Charts courtesy of Matías Armanini

Full size image
Fig. 6
figure 6

Approximate footprint of the microbursts (red), tornado (blue), and the type of gustnado (green) with superimposition of location of one or more fallen trees (red dots), direction of some fallen trees, water tank covers, walls, and signs and location of the airport

Full size image

As a consequence of the foregoing atmospheric conditions, a supercell developed over southeast Córdoba, which produced the damaging storm over Bell Ville. As a result of the first damage reconnaissance conducted on December 15 and 16, it was observed that an F1 tornado with maximum gusts estimated at 160 kph from the WNW sector, track mean width of 600 m, and 20 km long in the rural area southwards of the city; two F0–F1 microbursts with wind speeds around 120–140 kph that developed from the SW; and a type of gustnado of 20–50 m wide that developed in the lateral contact interface between the almost simultaneous microbursts (see Burgess and Donaldson 1979; Forbes and Wakimoto 1983; Doswell and Burgess 1993; Grazulis 2001). The gustnado produced localized damages, more intense than the produced by the microburst mainly in the cemetery facilities, along the river coast, the bus station, 200 m west of the downtown bridge, and eastwards from a grain processing plant.

Following the description of the storm characteristics, we now turn to describe background aspects of the city infrastructure and next, the types and severity of damage observed.

5 Building and urban layout characteristics

Bell Ville is divided into four main areas: downtown (government, small shops), residential, green spaces, and industrial (medium to large metallurgical and agricultural warehouses on the outskirts of the city). The city’s layout is a regular grid street plan of square blocks of 130 m subdivided into parcels typically of 8–12 m wide and 15–30 m long (Fig. 7–Left). Electricity and telephone are aerially conducted on wooden, metallic, and concrete poles and serviced into the properties through a 1.5-m electric service entrance pillar located in front of the house.

Fig. 7
figure 7

(Left): surveyed parcel areas (m2). (Right): surveyed number of windows

Full size image

Residential houses are attached or detached (partially or completely) from its neighbors. Parcel backyards are delimited by 1.70- to 1.80 m-high brick walls (Fig. 8).

Fig. 8
figure 8

Aerial view of typical city blocks that shows the regular disposition of houses. Photograph taken during a 2014 riverine flood event. Courtesy O. Mori

Full size image

6 Types and severity of damage

This section describes the characteristics, extent, and intensity of the damages caused by the microbursts on the properties surveyed which sustained some kind of damage. The summaries are displayed in Fig. 9 and Table 3.

Fig. 9
figure 9

Summary of surveyed damage in reported number of affected buildings

Full size image
Table 3 Distributions of the damage by intensity
Full size table

6.1 Roof damage

The most prevalent roof type of dwellings is a reinforced concrete slab. Metal roofs of varying quality are also frequent, especially in warehouses, poor quality dwellings, and storage sheds. There are comparatively less wood roofs (garages and pavilions) and polycarbonate roofs (screened porches, awnings, and entrance canopies). Residential tile or metal roofs are usually supported by timber girders on masonry walls.

The maximum damage severity occurred in the southwest zone where winds were higher (Fig. 10). The damage severity sustained by roofs was mostly mild, but moderate and severe damage combined reached a comparable proportion (Fig. 9; Table 3). Metal and tiled roofs were the most damaged (Table 4). Moreover, metal roofs exhibited predominating mild and severe damage pointing out to the quality gulf between engineered warehouses and low-quality zinc roofs in small informal warehouses and poor quality dwellings. It must be noted that in general, construction practices of metal roofs focus more on the avoidance of water and moisture intrusions than the proper element fastening to resist extreme wind loads. Furthermore, houses or small warehouses are not required by the local building ordinance to meet the provisions of the wind standard CIRSOC 102. Anecdotal evidence indicated the existence of abundant debris from damaged roofs, e.g., contorted metal panels around trees, pieces traveling tens of meters, and in one instance even cutting off public telephone wires. Consequently, mitigation measures to strengthen metal roofs will result in a significant reduction of damage and debris.

Fig. 10
figure 10

Distribution of roof damage severity

Full size image
Table 4 Number of buildings with roof damage by type
Full size table

Concrete slab roofs sustained mostly indirect damage to the waterproofing membranes (e.g., asphalt-aluminum), but also direct damage caused by branches or tree toppling. Haphazard local wind flow conditions also exacerbated the damage of some buildings as discussed above (Fig. 3).

6.2 Damage of exterior walls

A quantification of exterior wall damages in the city could not be assessed. However, anecdotal evidence indicated that there were frequent cases of loose bricks falling at low wind speeds from the walls of unfinished buildings or fences.

A severe instance of wall damage occurred to the city’s airport non-engineered main hangar and other smaller buildings which sustained total damage due to the action of the tornado (Fig. 11). The unreinforced masonry walls collapsed out of plane as they lost the upper support provided by the roof. As a result, eleven airplanes were either partially or totally destroyed. Other buildings located in the vicinity of the airport also were destroyed.

Fig. 11
figure 11

The local airport hangar, before and after the storm struck. Photographs courtesy of O. Mori

Full size image

6.3 Damage of wall fences and electric service entrance pillar

The storm caused moderate to severe damage to a large amount of wall fences and electric service entrance pillars (Fig. 9; Table 3). The spatial distribution of damage severity to these elements (Fig. 12) suggests that these failed for less than maximum wind speeds and consequently, these elements are among the most vulnerable in a city. Multiple factors in addition to wind pressure contributed to wall fence damage including: problems with wall columns or intermediate pillars (absence of them, deficient spacing, faulty wall–foundation connections); poor foundations; excessively long walls; local conditions increasing wind speed (e.g., by surrounding buildings); and collapse caused by toppled trees, branches, or utility poles. The collapse of electric service entrance pillars was caused mainly because of faulty foundations and/or poor connection between the foundation and the pillar. These elements caused damage to vehicles, sheds, and trees.

Fig. 12
figure 12

Distribution of damage of walls fences and electric service entrance pillar

Full size image

6.4 Damage to windows and doors

Most houses in the city do not have shutters on all of their windows (average of 5.6 windows per house), and residential doors are usually made of metal, wood, or metal and glazing. The damage observed in windows was irregularly scattered, and the severity was mostly mild (averages of 1.8 breached windows per damaged building) (Fig. 13). However, a few breached windows may be enough to change the interior pressure of a building and compromise a weak roof. There were a few instances of appliances damaged by wind-driven rain entering through breached and poorly sealed windows with the consequent risk of electrocution (see Fig. 9; Table 3). Due to the potential damage associated with breached windows, these elements require close attention from the perspective of the owners and the municipality to reduce buildings’ vulnerability to downbursts. The damage sustained by doors and gates, caused by wind pressure, involved few instances but mostly recorded as severe because it involved the fall of these elements (Fig. 9; Table 3).

Fig. 13
figure 13

Distribution of damages to windows, doors, and gates

Full size image

6.5 Damaged trees

Trees are not reliable indicators of the wind speed at failure because of differences in species, growing conditions, and heights, but they are helpful to determine the overall damage pattern of the downburst and the directions of the winds (see Kuligowski et al. 2014). The downburst caused extensive damage to the tree canopy layer in Bell Ville. Many trees in the vicinity of the airport were completely uprooted. The approximate location of fallen trees is displayed in Fig. 6.

The species and number of fallen trees within the city’s natural reserve, Parque Tau, located in the southwestern area were: Tala (62), Pines (8), Algarrobo (5), Moradillo (3), Espinillo (3), Chañar (3). Aguaribay (2), Tipa (2), Acacia (1), Cedar (1), Paraíso (1), Mora (1), Grevillea (1), Poplar (1), Lapacho (1), Elm (1).

6.6 Damage to utilities infrastructure and other elements

Damage to utility infrastructure was widespread and disruptive (Fig. 14). A large number of electrical and telephone poles were downed resulting in service interruptions. Anecdotal evidence and visual inspections suggested that most of the them lacked adequate maintenance and fell at low wind speeds, while others were pulled down by trees or branches falling on the lines.

Fig. 14
figure 14

Direct and indirect damage sustained by diverse elements

Full size image

Non-engineered sign boards in the city’s commercial district were toppled by wind speeds which were significantly lower than the maximum, whereas the engineered sign boards performed satisfactorily. Countless water tank light concrete lids were detached from the top of the roofs by low wind speeds. Lids weigh about 15 kg and are fixed to the tank with metal straps. Newer water tanks have smaller and lighter plastic lids that cannot be pulled by wind. It was observed that many lids were either unfastened or precariously tied with a wire. A case was reported of a water tank cover that hit and killed a small animal: had it been a human, it is very likely the outcome would have tragically been the same. In many cases, water tanks stand on a structure on top of the roof and are confined within walls, but it is not infrequent that these water tanks are unprotected. Other elements that sustained damage include exterior HVAC units, water tanks, and awnings (Fig. 14).

7 Damage modes and damage sequences

The analysis of damage distribution maps, questionnaires, and anecdotal evidence along with engineering judgment were used to group components affected by similar wind intensities and to draw a preliminary picture of the damage sequences caused by the downburst.

The first group of elements to be damaged at low wind speeds were tree limbs, water tank lids, aerial utility lines, loose elements like roof metal sheets, and non-engineered sign boards. These in turn damaged contiguous elements such as wall fences, awnings, and vehicles. Poles fell in a domino effect which brought about interruptions in the power service over large portions of the city. It is important therefore from the preventive perspective of a municipality to assess the structural integrity of this group of elements in order to avoid a potentially significant amount of direct and indirect damage caused by lower wind speeds.

The group of elements damaged by moderate winds included uprooted small- and medium-sized trees, the uplifting of poorly constructed metal and timber roofs, damaged sheds, gates, poorly built electric service entrance pillars, windows, and doors. In rural areas, poorly built and/or grain silos, bucket elevators, and conveyor systems were also damaged with varying degrees of severity.

The elements damaged by the highest winds of this storm (F0–F1) included, in addition to some of the above elements, toppled large trees, failure of large non-engineered roofs, sign boards, exterior walls of unfinished buildings, grain silos, and their conveyor systems.

A graphical depiction of the foregoing observations is presented in Fig. 15 to roughly portray how damage might have propagated. Bars in the diagram indicate possible ranges of wind speeds that may initiate element damage and the links indicate damage propagation paths: for example, a tank lid can break a window changing the interior pressure compromising the metal roof of the house, the exterior walls, and the safety of people inside. The diagram represents that trees and branches have the highest potential for causing damage because of their prevalence across the city. Furthermore, if large quantities of trees are damaged by wind, the subsequent wind load pressures will be higher because the roughness decreases; consequently, adequate periodic maintenance of them is crucial.

Fig. 15
figure 15

Conceptual diagram of damage sequences. Color gray indicates the elements belonging to the city infrastructure, green buildings, and red injuries

Full size image

From anecdotal evidence, it was gathered that debris removal, cleaning efforts, and the restoration of some services in areas of the city were slow because the municipality or service companies did not have proper emergency plans. These shortcomings exacerbated the public perception of the severity of the storm.

Even though the diagram in Fig. 15 is by no means exhaustive, municipalities and service companies alike could benefit from developing city-specific diagrams like it to plan appropriate actions to mitigate the cascading damage and thus improve the city’s resilience. In what follows, some practical recommendations are presented to contribute to mitigation planning efforts.

8 Damage reduction measures

Some general recommendations are presented in this section that are deducted from the foregoing discussion and could contribute to increasing a city’s resilience to severe storms. Particularly, widespread damage was sustained by metal sheet roofs due to poor construction and inexistent inspection. Consequently, the involvement of engineers and architects in the design and construction along with an updated building code and a more stringent building permits process will reduce the downbursts vulnerability of the built environment. In the same vein, it will also be beneficial to raise public awareness about preventive securing of elements that are potential sources of projectiles (e.g., trees and branches, metal and polycarbonate roofs, loose tiles, and water tank lids), and providing proper shutters to windows.

Trees in private and public spaces produced significant indirect damage and flying debris; therefore, to prevent this damage cascading, the municipality should inspect trees in public spaces before the start of the severe storm season starts and also encourage the population to do the same with their trees. It must be noted however that tree topping (hatracking) or tree removal must be avoided in order to keep a healthy urban forest (Gilman et al. 2006). Subterranean power lines will also reduce the prospect of service interruptions caused by trees or falling poles, or water tank lids.

A preparedness plan should be jointly assembled and updated by public and private institutions to streamline operations, communication, and recovery efforts. In addition, service companies may conduct regular preventive and predictive maintenance studies of their networks and facilities. The municipality should have clear communication channels for coordination with regional, provincial, and national supporting agencies in the event of a disaster. Last but not least, it will be beneficial to plan periodical community training sessions to provide information on safety and individual preparedness.

9 Conclusions

Downbursts and tornados are recurrent and dangerous hazards that cause casualties and economic losses in many countries around the world. Downbursts in particular have received considerable attention from the meteorological community for at least four decades. Few studies have been carried out, however, to understand their damage potential on populated places. This study represents one effort in that direction. This paper presents the meteorological analysis of an F0–F1 downburst, a type of gustnado, and an F1 tornado that affected the city of Bell Ville, Argentina, and a vulnerability assessment of the resulting damages.

The damage data was collected by two surveys on more than 4000 properties, the first conducted few days after the storm and the second performed 8 months later. The analysis afforded insights into the spatial spread of the downburst, its damaging characteristics, and also on the damage modes observed.

The elements that sustained most of the damage were metal roofs, roofs with loose tiles, wall fences, and electric service entrance pillars. Several doors and windows sustained damage and allowed water intrusion. Fallen trees, branches, and poles were responsible for direct and indirect damage. Numerous water tank lids were turned into projectiles by low winds and were found around the city. A diagram that that represents the observed damage and possible ways in which it propagated was drawn from analysis data, anecdotal evidence, and engineering judgment.

The study findings presented herein are useful for mitigation and response and as such may also be applicable to other similar cities subjected to severe storm hazard. Additional work is needed to study the damage characteristics of more intense downbursts and tornados. Lastly, aftermath surveys are extremely useful and should be made part of standard practice after the occurrence of an event like the one studied. It is crucial that more similar surveys are conducted in the future to help assist efforts to understand and develop damage mitigation strategies.