WARNING RESPONSE AND RISK BEHAVIOR IN THE OAK GROVE - BIRMINGHAM, ALABAMA, TORNADO OF 08 APRIL 1998
Matthew D. Biddle, M.A.
The Tornado Project - St Johnsbury, VT
c/o University of Oklahoma
Department of Geography
Norman, Oklahoma, USA
mbiddle@ou.edu
Return to Hazards Center Home Page
Return to Quick Response Paper Index
This material is based upon work supported by the National Science Foundation under Grant No. CMS-9632458. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
The tornado destroyed 1,100 homes and damaged at least 1,000 more. In addition, one school complex, two fire stations, 15 churches, 13 commercial buildings, 20 apartment units, and 4,000 acres of woodland were destroyed. Significant to severe damage occurred in the communities of Oak Grove, Concord, Rock Creek, Pleasant Grove, Sylvan Springs, Maytown, Wylam Heights, Edgewater, McDonald Chapel, Minor, Sandusky, West Ensley, and Pratt City, as well as a small part of the north-western fringes of Birmingham proper. The event is officially referred to as the "Oak Grove - Birmingham Tornado" by the National Oceanic and Atmospheric Administration (NOAA). Damage on the Fujita Damage Intensity Scale (Appendix F) was F-3 to F-4 in many places, with significant areas of F-5 damage, primarily in Oak Grove, Wylam Heights, and Edgewater (NWS 1998c).
The consensus, including surveys we conducted, is that this event was successfully forecasted with most of the public warned early and updated often for this tornado. The Storm Prediction Center (SPC) in Norman, Oklahoma issued numerous Tornado Watches that afternoon, and the National Weather Service (NWS) - Birmingham Weather Forecast Office (WFO), located in suburban Alabaster, issued numerous Tornado Warnings for Jefferson County, with lead-times ranging from five to 30 minutes (NWS 1998c). Additionally, radio and television (TV) dissemination of warnings took high profiles, as numerous TV stations provided live breaks in programming with extensive coverage. Tornadoes also caused major damage shortly before this disaster to the southwest in Tuscaloosa County, and shortly after to the northeast in St. Clair County, where two people were killed when their mobile home was destroyed near Moody.
Residents normally have ready access to emergency medical services and a "Level One Trauma Center," owing to their proximity to Birmingham. Emergency response organizations include both paid and volunteer fire departments. The area is served by all major television networks, cable TV including CNN and The Weather Channel, and numerous AM and FM radio stations. NOAA Weather Radio (NWR) is available in the area, and is broadcast from VHF transmitters located in Birmingham (KIH-54) and Tuscaloosa (KIH-60).
By any measure, the Oak Grove - Birmingham Tornado of 08 April 1998 (hereafter "OGB-98") was one of the worst ever to strike Alabama, both in terms of intensity and damage. Some recently reported damage totals approach 300 million dollars, which if accurate would, according to Grazulis et al. [1998], rank it as the tenth most damaging tornado on record in North America. We suspect the damage total is closer to 200 million than 300 million dollars, but reliable final totals remain unavailable.
Area media responded to the approaching storms with most radio and television stations relaying the watch and warning information in routine fashion. As the severity of the situation became increasingly evident, most major local TV stations broke regular programming with increasing frequency as the storms neared metropolitan Birmingham. Two TV Stations were identified in our survey as the primary stations consulted by those who received televised warning information. Both provided continuous severe weather programming that included radar summaries and projected storm paths, starting well before the tornado touched down at Oak Grove. Despite NWR availability, few survivors reported having access to an NWR receiver, and none reported it as their primary source of information. Some survivors noted that there are intermittent reception problems with NWR, allegedly related to topography and atmospheric conditions.
Emergency response was immediate, though compromised in most locations by downed trees and power lines that impeded access to the scene by search and rescue (SAR) teams with their usual vehicles. There is some evidence from recordings of the 911 Telephone System and public safety agency radio transmissions that first responders tended to see the damage in their given sector of operations as the worst in terms of damage magnitude. Understandably, this resulted in a brief time-lag before the overall response was coordinated, and neighboring jurisdictions understood the severity and areal extent of the disaster. At approximately 8:55 pm CDT (0135 UTC), the Jefferson County Emergency Management Agency declared a full-scale emergency and the Alabama State Emergency Management Agency (AEMA) activated the State Emergency Response Plan, dispatching mutual aid teams and National Guard units to the scene to augment local SAR teams already engaged.
In Concord, two of three fire stations were destroyed by the tornado. Fire rescue teams resorted to entering the scene on foot, often walking three miles or more to get to victims and bringing them out on stretchers crossing similar distances. Helicopters and repelling teams were required to remove some of the bodies in Edgewater and Wylam Heights. "Emergent" SAR volunteers became an important part of the response due to the difficulties of site access. Personally-owned pick-up trucks were used to transport both official SAR teams and ad hoc stretcher teams. Other volunteers became directly involved by giving first aid, while others cleared paths with chainsaws or heavy equipment.
Mutual aid from most area fire departments included manpower and equipment from Birmingham as well as from as far away as Montgomery. Shortly after the tornado, myriad local, state, federal and private organizations were on scene to render assistance. The Federal Emergency Management Agency (FEMA), the Department of Health and Human Services (HHS), and the Small Business Administration all set up field offices as did the American Red Cross, Salvation Army, and numerous other relief groups. A few days after the tornado, President Bill Clinton declared Tuscaloosa, Jefferson, and St. Clair counties Federal Disaster Areas.
Additional scenarios include failure to heed a warning due to being completely unaware of the risk owing to preformed attitudes about the hazard, perceptions that it "won't happen to me" (cognitive), or failing to understand the warning due to a language or age barrier (situational). To elicit mitigative response, a warning must convey that the threat is real. If there is significant doubt about its validity, it will be ignored. If it is only given marginal credibility, it is likely to evoke Confirmation Behavior -- by which recipients seek other sources (such as often happens with tornadoes) -- by dangerously venturing outside or to a window to look for the tornado (Drabek and Stevenson 1971).
Although these theories persist in contemporary hazard literature, Biddle [1994] found that Southerners had similar access to warning systems and equal understanding of tornado risk, when compared to residents of other tornado-prone areas of the country. In comparing survey data taken from Alabama college students to that of college students from the "Midwest" (Illinois) and the "Great Plains" (Oklahoma), religious and cultural differences that might exist were not expressed in perceptions of the tornado threat, understanding of weather terminology, mitigative strategies, or knowledge of, access to, and use of warning systems. In particular, Sims and Baumann [1972] argued that few Alabama respondents utilized warning information disseminated via television. Conversely, Biddle [1994], found that most Alabama respondents utilized TV warning information extensively. Data collected from OGB-98 further suggests that Alabamians largely embraced available warning technologies, and the existence of a unique southern culture of fatalism and warning ambivalence has continued to erode, or never existed.
It is more likely that local, rather than regional factors, such as the serendipitous nature of tornado paths and the misfortunes of vulnerable communities, play more tangible roles, than regional cultural attitudes. Community vulnerability is not only a function of the probability of impact, but also the degree to which the community can endure the impact. As with most disasters, the poor remain at highest risk due to factors related to cultural architecture, education, shelter availability, community infrastructure, family logistics, and access to services. Ultimately, risk is not connected solely to a lack of awareness among the poor, disenfranchised, and elderly. Rather, the more common concerns of daily life usually overshadow considerations of the low probability event, no matter how catastrophic or lethal its potential. This is especially true if individuals perceive little control over its probability (Drabek 1987).
The warning chain is no stronger than its weakest link. The reduction of tornado casualties is dependent upon several variables that constitute the warning system: early and accurate detection of tornadoes by meteorologists; efficient and clear communication of warnings to the public via multiple media; and expedient and successful mitigation reactions among the warned community. Consequently, this research aims to provide insight on the relationships between human behavior, and warning system characteristics that affect the capability and efficacy of warnings. It is hoped that analyzing the human ecology of this tornado disaster, combined with a growing body of similar studies will yield details about what works and what does not, and facilitate the design of future warning systems.
Did survivors receive the tornado warning(s), and if so, by what means and with how much lead-time?
What did survivors do when they received a warning or learned of the tornado?
What were survivor attitudes about tornado risk, tornado preparedness, and warning systems?
The purpose of the field operations were to gather information on casualties and survivors, with an emphasis on determining attitudes about tornado risk, warning sources and lead-time, shelter availability, mitigative behavior, and damage parameters. A questionnaire was developed with the assistance of Thomas Schmidlin of Kent University, to produce data amenable to comparison with that of other similarly studied tornado disasters. Permits to enter the disaster area were received from the Jefferson County Sheriff's Office. Matt Biddle, assisted by local resident Jeff Wright, drove along the damage path stopping when residents were observed outside working on damaged homes or interacting with public safety or relief officials. No one declined to submit to a survey, and usually detailed information and photographs of damaged structures were obtained. We spread out the survey to include locations encompassing the entire spectrum of damage severity, with gaps in areas where housing was sparse. It is acknowledged that this did not represent a true random sample, as we spent more time in the areas most severely damaged, representing areas of higher morbidity and mortality.
Relationships among interval/ratio data (e.g. age), were evaluated using a Standard t-test for significance. In cases where fatality data were used, the variance from fatality data were not used as a population estimate, but rather a pooled variance was computed from both fatality and survivor data, owing to the small number of fatalities. Population proportions (e.g. gender and race), were evaluated using a Standard Z-test (Burt and Barber 1996). Statistical significance for other comparisons (e.g. place) was attempted (Chi-square Test) where fatality data represented expected frequencies. Copies of the survey questionnaire are available from the authors.
The differences in mean age between respondent and fatality groups were not statistically significant at any level. Gender differences between the two groups were statistically significant at the 0.001 level. This is because more females than males were killed, but more males were interviewed as respondents. This is assumed to be a sampling bias due to the fact that more males were contacted outside their property (usually working in the debris) by our survey team. Racial differences between the two groups was also statistically significant at the 0.001 level, but the difference is quite small and believed related to the relatively different sample sizes. Responses of "unsure" or "refused" were treated as missing. Percentage values expressed in our discussions may not add up to 100% due to rounding.
Age VICTIMS RESPONDENTS SURVIVORS (Mean) 54.1(n=32) 55.2 (n=50) 31.5 (n=140) (High) 89 (2) 97 97 (Low) 04 22 01 00-03 00 (0%) 00 (0%) 12 (9%) 04-11 02 (6%) 00 (0%) 16 (11%) 12-17 01 (3%) 00 (0%) 16 (11%) 18-35 02 (6%) 08 (16%) 38 (27%) 36-50 10 (31%) 17 (34%) 25 (18%) 51-65 03 (9%) 14 (28%) 20 (14%) 66-80 09 (28%) 08 (16%) 10 (7%) 81+ 04 (12%) 03 (6%) 03 (2%)
There is no significant difference between victim and respondent ages, and the large drop in mean survivor age is due to the large number of children included in this population. There are no children in the respondent population since we did not interview minors as a head-of-household. The most striking aspect of age data is the fact that 40% of the fatalities were at least 66 years of age, compared to only 9% and 22% for the respondents and survivors, respectively.
Gender VICTIMS RESPONDENTS SURVIVORS Female 20 (62%) 26 (40%) 87 (55%) Male 12 (38%) 39 (60%) 70 (45%) Race VICTIMS RESPONDENTS SURVIVORS Black 09 (28%) 15 (21%) N / A White 23 (71%) 49 (75%) Other 00 (0%) 01 (02%)
Two-thirds (six) of all Black fatalities occurred in one home. Nonetheless, the racial make-up of the victim and respondent sample populations is quite similar.
Marital Status VICTIMS RESPONDENTS SURVIVORS (n=31) N / A N / A Never 04 (13%) Married 20 (65%) Divorced 01 (3%) Widowed 06 (19%)
While we were able to obtain marital status information for all but one of the fatalities, obtaining such from respondents proved difficult. Attempts to use closed-ended questions regarding marital status resulted in some ambiguity, as various cultural and legal interpretations of "married" and "divorced" seemed to result in significant confusion to the respondents. We did not wish to cause duress among those we surveyed, but rather to define any situational relationships of family structure to risk. We soon abandoned the collection of marital status information as it became apparent that the above four categories did little to define the myriad family structures and attitudes that exist. Therefore comparing fatality and respondent data is not warranted, and cross-comparisons to the marital status findings of other studies should be done with caution.
All four victims that were never married were juveniles. Our results from OGB-98 were significantly different than for ARK-97 (Schmidlin and King 1997) who found that of 26 deaths, 17% were never married and 22% were divorced. The mean age in ARK-97 was 43, with 52% female, whereas in OGB-98 the numbers were 54 and 62% respectively.
Place VICTIMS RESPONDENTS SURVIVORS Sector W 09 (28%) 11 (17%) (17%) Sector M 10 (31%) 14 (22%) (22%) Sector E 13 (41%) 40 (62%) (62%)
The tornado track was divided into thirds from west to east based on breaks in areas of housing density and spacing. In general: Sector W = Oak Grove / Concord / Rock Creek; Sector M = Pleasant Grove / Maytown / Sylvan Springs; Sector E = Wylam Heights / Edgewater / McDonald Chapel / Pratt City. We were able to obtain detailed demographic information for portions of each sector by using 1990 Census data based on Zip Codes. The areas do not match up exactly, but do provide a general idea of the situation for each sector of the disaster area.
W M E ZIP Code 35023 35118 35214 Females (%) 50 53 53 Black/White (%) 16.4 / 83.1 13.1 / 86.7 56.2 / 43.4 Median Age 35.4 38.1 35.6 Avg. Family Size 2.66 2.73 2.74 Median Family Income ($K) 29.6 27.2 26.0 Below Poverty Level (%) 9.2 11.5 16.9 High School Grad (%) 36.4 37.8 29.4 Unemployment (%) 5.0 7.2 7.9 Med. Yr, Home Built 1968 1965 1966 Mobile Homes (%) 21.1 9.8 3.2
All victims 16 years of age or younger died in Sector M. All Black fatalities occurred in Sector E, where they accounted for 69% of Sector E's deaths. All those killed in Sector W and Sector M were White. Six victims died in one Edgewater (Sector E) home, and two in another two doors down.
For fatalities, as stated, six people (20) died in one home (two survived at this site) in Edgewater, for the most at one location. There were two locations where 3 people (20%), both involving at least one parent and one child, were killed. Two were killed at five different locations (31%), with single fatalities for the remaining ten deaths (31%). Cumulatively, fatalities occurred at 18 different locations, and 44% involved multiple fatalities.
Sector E was dominated by single story residences (68%). Dominant styles were shotguns, hall and parlors, bungalows, and pyramid-style company homes (Appendix G), with very few two story homes encountered in this area. In Sector W, respondent homes were highly variable in architecture, as was Sector M, though there 30% of respondent homes were split-level styles of composite materials. In Schmidlin and King [1997], only 16% of fatalities in the ARK-97 event were killed in a house, with 56% in mobile homes. None had access to a below ground or windowless shelter. For the PSO-94 event, 75% were in mobile homes (Schmidlin and King 1995).
In the Edgewater and McDonald Chapel areas of Sector E, damage ranged from light to complete within the same block (See Appendix H). Many of the homes in this area were mining company pyramids on pier and beam foundations, all made of wood. Most were of sub-standard construction in spite of the era of construction (circa 1930's), often with exposed eaves and pine stud framing only one inch in thickness (BHS 1993). Often, to compensate for unexcavated natural topography at the plat site, one end of the home is raised up several feet, while the other end is nearly on the ground. The end result is that in such situations, certain homes may have increased or decreased wind failure thresholds owing to these angled elevations. The neighborhoods with significant numbers of pyramids were among the most devastated, yet we were unable to attach any patterns for risk of death to this style of architecture.
Split-level houses, while offering some below-ground shelter, were particularly prone to losing only their roof. This is apparently due to the attached garage, which feature large doors prone to failure at lower wind speeds, as the wind gains access to the interior walls of the structure and blows off the roof from the inside out. (Marshall 1999).
While many were destroyed, a substantial number survived. We surmise that this variability was due to the highly variable topography and differing orientations of the residences, resulting in each dwelling having different "faces to the wind." We feel that the circumstances surrounding the "angle of attack" of the wind and missiles versus the situation of the structure (foundation, slope, surface area, etc) are very important and can be significantly different for two building build nearly identical, but on different lots. We feel similar factors are involved in the vulnerability and damage patterns for automobiles and other vehicles.
Female survivors took shelter in basements 17% of the time, and first floors without windows half the time. Males found shelter 27% of the time in basements, with 35% in first floors with windows. Black survivors had shelter that was generally limited to first floors (67%), of which only half were without windows. In Sector M, most survivors were in a basement (50%), compared to first floor shelters that predominated in Sectors W and E.
In Sector W, fewer homes were assessed as "completely destroyed," with most whose owners were interviewed having only "significant damage." Access difficulties precluded getting to many sites in these sectors. Most residences where a fatality occurred were assessed by us as "blown away" (69%) or generally as having F-5 damage (e.g. no walls standing or swept foundation). A small number of victims (19%), died in homes that simply collapsed or suffered less than complete damage. Brick, at least upon cursory inspection, seemed to present lower F-scale damage than wooden structures. While this may seem logical, it likely represents a bias on our part based on the difficulties of F-scale assessment. Unanchored brick and metal structures tended to fit lower F-scale criteria, while anchored masonry structures tended to fit higher F-scale criteria.
Most of the unanchored structures that were in the F-4 or F-5 damage areas were often disintegrated and not readily observable in our spot surveys. Aside from the intrinsic mass and resistance of brick and masonry-walled homes, they also tended to be anchored to the foundation more often than homes built of other materials. Failed brick walls were observed both with and without foundation anchors. Very few wooden or metal "manufactured homes" had adequate anchoring. Thus, it is probable that brick structures offered more protection not only because of their material strength, but because they were more often anchored. This is in keeping with the observations of Marshall [1993].
Extremely high access ratings were assigned to 15% of the survivors. High access was given to 62%, and moderate access to 17%. Two percent of survivors received no warning, and only 2% were assessed as having low access. Most who took shelter in a first floor location at the time of tornado impact had no secondary warning source, whereas most of those who took shelter in basements received warnings (and confirmation information) from multiple warning sources and had higher access. Though Blacks tended to have slightly less access to warnings, especially based on quantity (none were rated "very high"), they still had 67% rated "high." Whites had 83% rated as "high" or "very high." No females were rated with the highest access, but 23% were rated with high access. Males tended to have access to multiple warning sources (84%), and were more apt to consult secondary sources (41)%.
A strong majority (85%) of survivors reported becoming aware of the approaching tornado via TV. Only 9% first learned of the tornado by seeing or hearing it. This varies considerably from previously studied events. In particular, Schmidlin and King [1995], found that roughly half of the survivors of PSO-94 became aware of approaching tornadoes by seeing or hearing them, and only 16% via televised warnings. They also reported 61% and 68% learned of the tornado by seeing or hearing it, and 21% and 10% from TV, in ARK-97 (Schmidlin and King 1997) and FLA-98 (Schmidlin et al. 1998) respectively.
Two-thirds of Blacks (67%) and 80% of Whites used TV as the primary warning source. Of those who used TV as the primary source, 54% were male and 46% were female. There was a problem with the data in assessing the telephone as primary warning source. However, most of those who received their warning via the telephone were females. There were no significant trends in primary warning source based on location (sector).
Less than five percent (4%) stated they did not know the difference between a watch and a warning. Those who claimed to be aware of the tornado watch also usually claimed to be diligent in monitoring forecasts and weather awareness in general. Those who rated themselves as regular monitors of severe weather forecasts and/or watches, not surprisingly had longer lead-times. A few (6%), reported they regularly monitor severe weather forecasts for information prior to the issuance of watches.
There were two foundation types identified with failure, damage, morbidity, and mortality -- the concrete slab, and the pier and beam system -- which are also common to the South as well as rural areas. Wooden single-story structures on either type of foundation comprised the majority of locations where deaths occurred (88%). These foundations were commonly unanchored, or anchored ineffectively (Marshall 1999). With the masonry pier system, houses are situated on blocks for the purpose of raising the floor joists and wall bases off damp or flood-prone earth, and to thwart termites. This not only made for frequent wall failure, but gave the wind easy access to the underside of flooring. With the slab, often the only thing holding the structure to the foundation is gravity. Strong winds on large flat surfaces, such as walls, easily negate this. In both cases, the result was often that these houses became airborne at low wind thresholds, perhaps in many cases at wind speeds less than 100 mph (160 km/hr).
Given this, the violent rating of the tornado becomes less important than the construction and architectural characteristics of the buildings. It is suggested here that even in a significantly weaker tornado, most of those killed would have still likely perished, but simply would not have been thrown as far downstream from their original home sites. This parallels the findings of Schmidlin and King [1995], who found that when flooring was removed from a foundation, the risk of death and injury rose dramatically. We were not able to determine the extent of anchoring involved at the lone mobile home fatality site.
Three people from one family (9%), were killed after taking shelter in the basement of their contemporary split-level home, when the wind disengaged an entire wall from the foundation, collapsing it down upon them. It is important to note however, that the overall protection offered from basement shelters remains significant. Only 9% of fatalities took shelter in a basement, and assessments at CTX-97 showed 100% of those killed at Jarrell, Texas had no below ground shelter (NWS 1998a). Tragedies of this type can usually be avoided by training occupants to take cover in the center of the basement, with secondary shelter under a staircase or heavy furniture.
Only one person was killed in an apartment, but we were unable to determine what floor the person lived. We noted a lack of multiple family housing in most of the damage path. This may have helped to keep the fatality rate down as most apartment complexes lack significant shelter. In ILL-90, 33% of the deaths were in apartments (NWS 1991). It is also significant to note that none of the 32 OGB-98 fatalities were killed in a motor vehicle and only 3 (9%) were in a mobile home (NWS 1998). This is anomalous compared to both historical and recent records. SPC [1999] data for the period 1985 - 1997 indicate 38% of all tornado deaths have been in mobile homes, and 11% in vehicles. In KAN-91, 13 of the 19 killed (68%) were killed in mobile homes in the same mobile home park. This was in spite of an available shelter and at least a seven minute NWS warning (NWS 1991a). Schmidlin [1995] reported that some studies have counted as much as 60% of all tornado fatalities in vehicles for a given event, and Schmidlin et al. [1998b] reported 15% fatalities in vehicles from 1975 - 1995. In ILL-90, 10% of the deaths were in vehicles, as were 6 of 7 (87%) fatalities in a 1993 tornado at Catoosa, Oklahoma (Biddle 1994).
It was determined that most of the OGB-98 victims died instantly. In two other studied events involving violent tornadoes, OPO-85 (CDC 1986) and ILL-90 (Brenner and Noji 1995), most of the victims were also thought to have died instantly. The historically high fatality rate for violent tornadoes (Biddle 1998), is accountable for about two-thirds (67%) of all tornado deaths, and is again identified as a major factor for mortality. However, it seems the more violent the tornado, the more "blurred" the signals for situational and behavioral elements of risk factors for death become.
Obviously, the high percentage of those who received warning and the general length of lead-time played major roles in reducing the number of casualties. The success of televised warnings was paramount, as was the performance of the NWS warning system. We speculate that this is at least partially due to the high level of cooperation between WFO - Birmingham and the Birmingham area media, various recent public relations efforts by WFO - Birmingham and AEMA, and the generally high severe weather awareness among much of the area public, that is likely due to recent highly publicized tornado disasters that have occurred in the region.
We are unable to construct any explanation for the apparent paradox related to warning awareness and compliance heightened by heavy TV severe weather coverage among those we surveyed in the Birmingham area, with that of the "numbing effect" reported from heavy TV coverage among survivors of other events. The high level of weather awareness and knowledge of mitigative action also were important, though with the exception of previously noting the potential impact of area tornado disasters of recent history, we found nothing that stands out in terms of NWS approach or media coverage that is substantially different from other areas of "Tornado Alley." The time of the tornado (early evening), may have resulted in more people being able to readily receive warnings, as did its proximity to a large metropolitan area.
The potential of school auditoriums and gymnasiums to be the site of tragic mass casualty events has not gone away with modernized warnings, and public awareness campaigns at schools must stress this continually. When a tornado watch is issued it is imperative that someone of authority with a school system monitor the weather and any warnings, and be in a position to notify whomever is supervising children to either refrain from activities in certain school facilities, or to take shelter in the designated areas. It was also disturbing to note that the hall ways between the classrooms, which we were told were the designated shelter areas for many of the students for tornado drills, were filled with bricks, concrete block, and other "chunky" debris. Were school in session, many casualties would have been likely.
At about the same time at the Bethel Baptist Church in Moody (St Clair County), what was to be a group of about 100 people were gathering at the church for Easter Pageant practice. At about 8 pm CDT (2100 UTC), church administrators who were aware of media reports of tornadic activity near Birmingham, made a decision to send everyone home, many whom had reported in their costumes. At approximately 9 pm CDT (2200 UTC), the church complex was destroyed. The mobile home where the two St Clair County fatalities occurred was located near the church.
Unlike the situation at Oak Grove School, church administrators showed exceptional insight in assessing the weather risk and making decisions to move people on to lower risk locations. Both of these close calls demonstrate the importance of the media in disseminating warnings and in having someone monitor the weather at such gatherings when severe weather is possible. We are unable to confirm this, but we surmise that awareness was high at least in part due to publicity surrounding the tragedy of PSO-94, where 20 people were killed in the collapse of the Goshen United Methodist Church near Piedmont, Alabama (NWS 1994).
Grazulis et al. [1998], suggest this is all too common a scenario and raise concerns about the probability of a mass casualty event that is the result of a tornado tangling with a major sporting event. The Barons and the community should consider themselves very lucky that this was not that event. It is imperative that those supervising and organizing public events have contingency plans in place, and consider weather information as important on a daily operational basis as parking and traffic or TV timeouts. Progress is made from time-to-time, as it was reported in the newspaper that at the 09 May 1998 Barons' game, 7000 people were evacuated to the concourse during a tornado warning, in accordance with the stadium "tornado policy."
Vehicle access to the scene remained impossible well into the night until crews with heavy tree removal and utility pole moving equipment from Alabama Power and Light (APL) and the Alabama Department of Forestry (ADF) cleared the way. CFD teams eventually got one rescue truck out of the debris of the west station, but Chief Miller noted that their response was only marginally affected by the damaged fire stations, as even if the equipment were not trapped in the buildings it would not have been able to progress down Warrior River Road a significant distance for some time. He added that he was generally pleased with the overall response given the conditions, that mutual aid was extensive, and that the department has replaced the damage CFD vehicles and rebuilt one of the stations with construction of the third pending.
At times among various factions involved in weather forecasting or product use, there has been discussions of a radical overhaul of the watch system, based on speculations that people do not pay attention to watches anymore. We do not believe this to be the case, and are aware of no contemporary scientific evidence to support this notion, at least for tornado watches (it may be true to some extent for severe thunderstorm watches). We reiterate the apparent overwhelming success of the tornado watches for the OGB-98 event, as more than 85% of survivors reported being aware of the tornado watch, and this was associated directly with longer warning lead-times. We also report that over 90% stated they understood the difference between a watch and a warning. We believe any future major modifications of the watch system should not occur without significant national studies of watch perception and utility. This includes any replacement of the current "box" format to other spatial constructs such as county outlined areas.
Biddle, MD, 1994: Tornado Hazards, Coping Styles, and Modernized Warning Systems. Thesis, Department of Geography, The University of Oklahoma - Norman. 143 p.
Biddle, MD, 1996: Re-examining Tornado Risk Coping Styles and Other Tornado Warning Issues Under the Modernized National Weather Service. Preprints: 18th Conference on Severe Local Storms, San Francisco, CA. American Meteorological Society. pp. 605-608.
Biddle, MD, 1998: Some Geographical and Climatological Aspects of North American Violent Tornadoes. Preprints: 19th Conference on Severe Local Storms, Minneapolis, MN. pp. 297-300.
Birmingham Historical Society (BHS), 1993: The Birmingham District - A Survey of Cultural Resources. Birmingham, AL.
Birmingham News - April 9 - 17, 23, 30, May 10, 1998. Birmingham, AL.
Binder, S and LM Sanderson, 1987: The Role of the Epidemiologist in Natural Disasters. Annals of Emergency Medicine 16:9:1081-1084.
Birmingham Regional Planning Commission (BRPC), 1991: Socioeconomic Data for Counties in the Birmingham Region. Birmingham, AL. 47 p.
Brenner, SA and EK Noji, 1995: Tornado Injuries as Related to Housing in the Plainfield Tornado. International Journal of Epidemiology 24:1:144.
Burnham, MW, 1992: Cry Wolf - Repeated Warnings Without Events. Hydrologic Engineering Center, University of California-Davis. 2 p.
Burt, JE and GM Barber, 1996: Elementary Statistics for Geographers, 2nd ed. Guilford Press, New York, 640 p.
Centers for Disease Control (CDC), 1986: Tornado Disaster - Ohio/Pennsylvania. Morbidity and Mortality Weekly Report 35:233-235.
Carroll, R, 1998: A Killer Tornado at Night. Southern Building Jul/Aug: 4-5.
Carter, AO, ME Millson and DE Allen, 1989: Epidemiologic Study of Deaths and Injuries Due to Tornadoes. American Journal of Epidemiology 130: 129-1218.
Duclos, PJ and RT Ing, 1989: Injuries and Risk Factors for Injuries from the May 29, 1982 Tornado, Marion, Illinois. International Journal of Epidemiology 18: 213-219.
Doswell, CE, III and AR Moller, 1998: Storm Spotting and Public Awareness Since the First Tornado Forecasts of 1948. Weather and Forecasting (submitted).
Drabek, TE, 1987: Human System Responses to Disaster: An Inventory of Sociological Findings. Springer-Verlag: New York, NY.
Drabek, TE and S Stevenson, 1971: When Disaster Strikes. Journal of Applied Social Psychology 2:187-203.
Fenneman, NM, 1946: Physical Divisions of the United States. (Map). US Geological Survey: Washington DC.
Glass, RI, RB Craven, DJ Bregman, BJ Stoll, N Horowitz, P Kerndt, and J Winkle, 1980: Injuries From the Wichita Falls Tornado: Implications for Prevention. Science 207: 734-738.
Grazulis, TP, 1993: Significant Tornadoes, 1680-1991. Environmental Films: St Johnsbury, VT. 1326 p.
Grazulis, TP, CA Doswell III, HE Brooks, and MD Biddle, 1998: A New Perspective of the Societal Impacts of North American Tornadoes Covering Two Centuries. Preprints: 19th Conference on Severe Local Storms, Minneapolis, MN. American Meteorological Society: pp. 196-199.
Grazulis, TP, 2000 (in press): Tornadoes of the United States. University of Oklahoma Press: Norman.
Gruntfest, E, 1987: "Warning Dissemination and Response With Short Lead-Times," in Hazard Management - British and International Perspectives. J Handmer (ed.). Geo Books: Norwich, UK. pp. 191-202.
Kessler, E and JT Lee, 1978: Distribution of the Tornado Threat in the United States. Bulletin of the American Meteorological Society 59:61-62.
Lillibridge, SA, 1997: "Tornadoes," in The Public Health Consequences of Disasters. EK Noji (ed.). Oxford University Press: New York, NY. pp. 228-244.
Linehan, U, 1957: Tornado Deaths in the United States. US Weather Bureau: Washington, DC.
Liu, S, L Quenemoen, J Malilay, E Noji, T Sinks, and J Mendlein, 1996: Assessment of a Severe Weather Warning System and Disaster Preparedness, Calhoun County, Alabama, 1994. American Journal of Public Health 86:87-89.
Marshall, TP, 1993 : "Lessons Learned From Analyzing Tornado Damage," in The Tornado: Its Structure, Dynamics, Prediction, and Hazards. C Church et al., (eds.). Monograph # 79. American Geophysical Union: Washington, DC. pp. 495-500.
Marshall, TP, 1999: "Wind Speeds and Damage in Tornadoes," presented at N W S - Lubbock / Texas Tech University Severe Weather Conference 1999. Lubbock, TX.
McAlester, V and L McAlester, 1988: A Field Guide to American Houses. Alfred Knopf Co: New York City. 525 p.
McCarthy, DM, 1993: Warning Operations in a Modernized NWS Forecast Office: Experiences at WSFO Norman, OK. Preprints - 17th Conference on Severe Local Storms, St. Louis, MO. American Meteorological Society. pp. 371-374.
McDonald, JR, 1993: "Damage Mitigation and Occupant Safety," in The Tornado: Its Structure, Dynamics, Prediction, and Hazards. C Church et al., (eds.). Monograph # 79. American Geophysical Union: Washington, DC. pp. 523-528.
Missouri State Census Data Center, 1999. University of Missouri Website for 1990 Alabama Census Standard Extracts http://www.oseda.missouri.edu/
NWS (National Weather Service), 1986: Natural Disaster Survey Report - Ohio - Pennsylvania Tornadoes of May 31, 1985. US Department of Commerce: Silver Spring, MD.
NWS, 1991a: Natural Disaster Survey Report - The Wichita - Andover Kansas Tornadoes of April 26, 1991. US Department of Commerce: Silver Spring, MD. p.
NWS, 1991b: Natural Disaster Survey Report - The Plainfield - Crest Hill Tornado of August 28, 1990. US Department of Commerce: Silver Spring, MD.
NWS, 1994: Natural Disaster Survey Report - Southeastern United States Palm Sunday Tornado Outbreak of March 27, 1994. US Department of Commerce: Silver Spring, MD.
NWS, 1997: NWS Service Assessment - March 1, 1997 Arkansas Tornado Outbreak. US Department of Commerce: Silver Spring, MD.
NWS, 1998a: NWS Service Assessment - Central Texas Tornadoes of May 27, 1997. US Department of Commerce: Silver Spring, MD.
NWS, 1998b: NWS Service Assessment - Central Florida Tornado Outbreak, February 22/23, 1998. US Department of Commerce: Silver Spring, MD.
NWS, 1998c: NWS Service Assessment - Southeastern United States Tornadoes, April 8, 1998. US Department of Commerce: Silver Spring, MD.
NWS, 1998d: NWS Service Assessment - Spencer, South Dakota Tornado, May 30, 1998. US Department of Commerce: Silver Spring, MD.
Perry, RW, MK Lindell, and MR Greene, 1982: Crisis Communications: Ethnic Differentials in Interpreting and Acting on Disaster Warnings. Social Behavior and Personality 10: 97-104.
Ruffner, JA and FE Bair (eds.), 1981: The Weather Almanac (3rd ed.). Gale Research: Detroit MI. p. 336. Sanderson, LM, 1989: "Tornadoes," in The Public Health Consequences of Disasters (2nd ed.). M Gregg, (ed.). Centers For Disease Control: Atlanta, GA. pp. 39-49.
Schmidlin, TW, 1993: "Tornado Fatalities in Ohio 1950-1989," in The Tornado: Its Structure, Dynamics, Prediction, and Hazards. C Church et al., (eds.). Monograph # 79. American Geophysical Union: Washington, DC. pp. 529-534.
Schmidlin, TW and PS King, 1995: Risk Factors for Death in the 27 March 1994 Georgia and Alabama Tornadoes. Disasters 19:170-177.
Schmidlin, TW and PS King, 1997: Risk Factors for Death in the March 1, 1997 Arkansas Tornadoes. Quick Response Report # 98. Natural Hazards Research Applications and Information Center: Boulder, CO. 6 p.
Schmidlin TW, PS King, BO Hammer, and Y Ono, 1998a: Risk Factors for Death in the 22-23 February Florida Tornadoes. Quick Response Report # 106. Natural Hazards Research Applications and Information Center: Boulder, CO 8 p.
Schmidlin TW, PS King, BO Hammer, and Y Ono, 1998b: Tornado Effects on Motor Vehicles. Preprints: 19th Conference on Severe Local Storms. Minneapolis, MN. pp. 428-430.
Sims, JH and DD Baumann, 1972: The Tornado Threat: Coping Styles of the North and South. Science 176:1386-1392.
Storm Prediction Center (SPC), 1999: Archival Data Website - http://www.nssl.noaa.gov/~spc/archive/products/
Tobin, GA and BE Montz, 1997: Natural Hazards: Explanation and Integration. Guilford: New York, NY 387 p.
Wilson CR and W Ferris (eds.), 1989: Encyclopedia of Southern Culture. University of North Carolina Press: Chapel Hill. 1634 p.
White, GF and JE Haas, 1975: Assessment of Research on Natural Hazards. MIT Press: Cambridge, MA. 242 p.
APPENDIX A
: Definitions
Convective Outlook - A forecast containing the area(s) of expected thunderstorm occurrence and expected severity over the contiguous United States, issued several times daily by the SPC. The terms approaching, slight risk, moderate risk, and high risk, are used to describe severe thunderstorm potential. Local versions sometimes are prepared by local NWS offices.
Doppler Radar - Radar that can measure radial velocity, the instantaneous component of motion parallel to the radar beam (i.e., toward or away from the radar antenna).
High Risk (of severe thunderstorms) - Severe weather is expected to affect more than 10 percent of the area. A high risk is rare, and implies an unusually dangerous situation and usually the possibility of a major severe weather outbreak (see convective outlook.)
Moderate Risk (of severe thunderstorms) - Severe thunderstorms are expected to affect between 5 and 10 percent of the area. A moderate risk indicates the possibility of a significant severe weather episode (see convective outlook).
Slight Risk (of severe thunderstorms) - Severe thunderstorms are expected to affect between 2 and 5 percent of the area. A slight risk generally implies that severe weather events are expected to be isolated (see convective outlook).
Tornado - A violently rotating column of air in contact with the ground and extending from the base of a thunderstorm. A condensation funnel does not need to reach to the ground for a tornado to be present; a debris cloud beneath a thunderstorm is all that is needed to confirm the presence of a tornado, even in the total absence of a condensation funnel.
Tornado Warning - issued by the National Weather Service or local emergency agencies when a tornado has been sighted by spotters or is indicated by NWS weather radar.
Tornado Watch - issued by the Storm Prediction Center or National Weather Service when tornadoes are possible in your area. Remain alert for approaching storms. This is time to remind family members where the safest places within your home are located, and listen to the radio or television for further developments.
APPENDIX B
: Track and Area Map
APPENDIX C
: Convective Outlook
CONVECTIVE OUTLOOK...REF AFOS NMCGPH940
REF WW NUMBER 0183...VALID TIL 1900Z
THERE IS A HIGH RISK OF SEVERE THUNDERSTORMS THIS AFTERNOON AND EARLY TONIGHT ACROSS PARTS OF NORTHERN MISSISSIPPI...NORTHERN AND CENTRAL ALABAMA...NORTHERN GEORGIA AND SOUTHERN TENNESSEE. THIS AREA IS TO THE RIGHT OF A LINE FROM 35 SSE MKL 40 WSW CSV 35 WSW TYS 35 SSW 50 NW AND 20 S AND 30 E LGC 20 W AUO 35 SW CKL 45 ESE GWO 30 SE UOX 35 SSE MKL.
THERE IS A MODERATE RISK OF SEVERE THUNDERSTORMS THIS AFTERNOON AND TONIGHT ACROSS MUCH OF MISSISSIPPI...SOUTHERN ALABAMA...CENTRAL GEORGIA...CENTRAL AND WESTERN NORTH CAROLINA AND SOUTH CAROLINA...MUCH OF EASTERN AND CENTRAL TENNESSEE NORTH OF THE HIGH RISK AREA. THIS AREA IS TO THE RIGHT OF A LINE FROM 35 N NCB 30 SW GWO 15 ESE MEM TO MKL TO BNA 20 ENE HSS 30 NNW GSO 10 NW RDU 15 E FAY TO FLO 35 ENE DHN 35 NNE MOB 35 ESE MCB 35 N MCB.
(SLIGHT RISK AND GENERAL THUNDERSTORM OUTLOOK AREA OMITTED)
...SIGNIFICANT SEVERE WEATHER OUTBREAK EXPECTED TODAY OVER MUCH OF THE SOUTHEASTERN US. A PUBLIC SEVERE WEATHER OUTLOOK WILL BE ISSUED AROUND 16Z.
...SEVERE THUNDERSTORM DISCUSSION...
POTENT SEVERE THUNDERSTORM PATTERN HAS EVOLVED OVER SOUTHEASTERN US WITH MARKED INCREASE IN INSTABILITY OVERNIGHT INTO THE AREA AHEAD OF STRONG UPPER SHORTWAVE TROUGH AND ASSOCIATED MID AND UPPER JET. SITUATION COMPLEX WITH SEVERE THUNDERSTORMS UNDERWAY WITH IMPRESSIVE BOW ECHO CURRENTLY MOVING INTO NORTHERN MISSISSIPPI AND ISOLATED SEVERE STORMS ALONG GULF COAST.
MODELS AGREE SOMEWHAT THAT A MORE ORGANIZED SURFACE LOW WILL DEVELOP THIS AFTERNOON VICINITY OF TENNESSEE / ALABAMA BORDER AND DEEPEN NORTHEASTWARD TONIGHT. WITH WIDESPREAD MODERATE INSTABILITY AND STRONG MID AND UPPER WINDS, CONDITIONS WILL DEVELOP BY MID AFTERNOON FOR SUPERCELLS AND TORNADOES ALONG AND AHEAD OF THE RAPIDLY MOVING LINE OF STORMS NOW MOVING INTO NORTHERN MISSISSIPPI. THIS IS REFLECTED IN THE HIGH RISK WHICH WAS SHIFTED EASTWARD INTO GEORGIA AND NORTH-WESTWARD INTO SOUTHERN TENNESSEE FOR THIS AFTERNOON AND EVENING.
...HALES / REHBEIN...04/08/98
APPENDIX D
: TORNADO WATCH
BULLETIN - IMMEDIATE BROADCAST REQUESTED
TORNADO WATCH NUMBER 194
STORM PREDICTION CENTER NORMAN OK
653 PM CDT WED APR 8 1998
THE STORM PREDICTION CENTER HAS ISSUED A TORNADO WATCH FOR PORTIONS OF
NORTHERN AND CENTRAL MISSISSIPPI
NORTHERN AND CENTRAL ALABAMA
EFFECTIVE THIS WEDNESDAY NIGHT AND THURSDAY MORNING FROM 730 PM UNTIL 200 AM CDT.
THIS IS A PARTICULARLY DANGEROUS SITUATION WITH THE POSSIBILITY OF VERY DAMAGING TORNADOES. ALSO. . . LARGE HAIL TO 2 INCHES IN DIAMETER. . . THUNDERSTORM WIND GUSTS TO 80 MPH . . . AND DANGEROUS LIGHTNING ARE POSSIBLE IN THESE AREAS.
THE TORNADO WATCH AREA IS ALONG AND 95 STATUTE MILES EITHER SIDE OF A LINE FROM 20 MILES WEST-NORTHWEST OF GREENWOOD MISSISSIPPI TO 35 MILES EAST-NORTHEAST OF ANNISTON ALABAMA.
REMEMBER . . . A TORNADO WATCH MEANS CONDITIONS ARE FAVORABLE FOR TORNADOES AND SEVERE THUNDERSTORMS IN AND CLOSE TO THE WATCH AREA. PERSONS IN THESE AREAS SHOULD BE ON THE LOOKOUT FOR THREATENING WEATHER CONDITIONS AND LISTEN FOR LATER STATEMENTS AND POSSIBLE WARNINGS.
OTHER WATCH INFORMATION . . . THIS TORNADO WATCH REPLACES TORNADO WATCH NUMBER 188. WATCH NUMBER 188 WILL NOT BE IN EFFECT AFTER 730 PM CDT. CONTINUE . . . WW 187 . . .WW 189 . . . WW 190 . . . WW 192 . . . WW 193. . .
DISCUSSION . . . THUNDERSTORM CLUSTER WITH A HISTORY OF TORNADOES AND SEVERE WEATHER WILL CONTINUE OVER WATCH AREA AS AIR MASS VERY UNSTABLE WITH CAPES ABOVE 3000 J/KG. 90 KT MID LEVEL SPEED MAX WILL MOVE ACROSS REGION AHEAD OF SURFACE FRONT AND VIGOROUS UPPER TROUGH THUS MAINTAINING POTENTIAL OF TORNADIC SUPERCELLS.
AVIATION . . . TORNADOES AND A FEW SEVERE THUNDERSTORMS WITH HAIL SURFACE AND ALOFT TO 2 INCHES. EXTREME TURBULENCE AND SURFACE WIND GUSTS TO 70 KTS. A FEW CUMULONIMBI WITH MAXIMUM TOPS TO 500. MEAN STORM MOTION VECTOR 24040.
. . . ROGASH
APPENDIX E
: NWS WARNING TEXT
THE NATIONAL WEATHER SERVICE IN BIRMINGHAM AL HAS ISSUED A TORNADO WARNING EFFECTIVE UNTIL 835 PM CDT FOR PEOPLE IN THE FOLLOWING LOCATION. . .
IN CENTRAL ALABAMA . . .
. . . JEFFERSON COUNTY . . .
AT 743 PM CDT. . . WEATHER RADAR SHOWED A POSSIBLE TORNADO MOVING OUT OF TUSCALOOSA COUNTY INTO SOUTHWEST JEFFERSON COUNTY. THE STORM WAS MOVING EAST NORTHEAST AT 40 MILES PER HOUR AND WILL MOVE THROUGH THE ALLIANCE - OAK GROVE AREAS. THIS STORM HAS BEEN ASSOCIATED WITH SIGNIFICANT DAMAGE IN TUSCALOOSA COUNTY.
«Call to Action»
THIS IS A DANGEROUS STORM SITUATION. ACT QUICKLY. IF YOU ARE IN THE PATH OF THIS TORNADO MOVE TO A PLACE OF SAFETY BELOW GROUND IF AVAILABLE. OTHERWISE . . . GO TO A SMALL INTERIOR ROOM ON THE LOWEST FLOOR POSSIBLE. AVOID WINDOWS. ABANDON CARS AND MOBILE HOMES FOR A REINFORCED BUILDING OR GET INTO A DITCH OR CULVERT.
*************************************************
BULLETIN - EAS ACTIVATION REQUESTED
TORNADO WARNING
NATIONAL WEATHER SERVICE BIRMINGHAM ALABAMA
832 PM CDT WED APR 8 1998
THE NATIONAL WEATHER SERVICE IN BIRMINGHAM AL HAS ISSUED A TORNADO WARNING EFFECTIVE UNTIL 905 PM CDT FOR PEOPLE IN THE FOLLOWING LOCATION. . .
IN CENTRAL ALABAMA . . .
. . . JEFFERSON COUNTY . . .
AT 831 PM CDT. . . WEATHER RADAR SHOWED A TORNADO LOCATED ABOUT 3 MILES WEST OF THE BIRMINGHAM AIRPORT MOVING EAST AT 40 MILES PER HOUR. THIS WARNING EXTENDS THE PRIOR WARNING FOR JEFFERSON COUNTY. THERE HAVE BEEN NUMEROUS REPORTS OF DAMAGE FROM THIS STORM.
«CALL TO ACTION»
THIS IS A DANGEROUS STORM SITUATION. ACT QUICKLY. IF YOU ARE IN THE PATH OF THIS TORNADO MOVE TO A PLACE OF SAFETY BELOW GROUND IF AVAILABLE. OTHERWISE . . . GO TO A SMALL INTERIOR ROOM ON THE LOWEST FLOOR POSSIBLE. AVOID WINDOWS. ABANDON CARS AND MOBILE HOMES FOR A REINFORCED BUILDING OR GET INTO A DITCH OR CULVERT.
APPENDIX F
: FUJITA DAMAGE INTENSITY SCALE
Fujita Scale (or F Scale) - A scale of wind damage intensity in which wind speeds are inferred
from an analysis of wind damage:
F0 (weak) - 40- 72 mph, light damage - chimneys downed, tree branches broken
F1 (weak) - 73-112 mph, moderate damage - mobile homes pushed off foundation or overturned
F2 (strong) - 113-157 mph, considerable damage - mobile homes demolished, trees uprooted
F3 (strong) - 158-206 mph, severe damage - roofs and walls torn down, trains, cars thrown
F4 (violent) - 207-260 mph, devastating damage - walls leveled, homes off foundations
F5 (violent) - 261-318 mph, (rare) incredible damage - homes, autos thrown > 100 meters
Return to Hazards Center Home Page