1. Introduction
a. Background
Detailed tornado damage surveys have been performed since the 1950s (Fujita 1959), but became more quantitative with the introduction of the Fujita (F) scale in 1971 (Fujita 1971), and the enhanced Fujita (EF) scale in 2007 (WSEC 2006; hereafter W06; Edwards et al. 2013). Between the 1970s and now, many surveys have been performed by groups of meteorologists and engineers, as well as others. However, many if not most surveys are not detailed; they do not include all objects/structures within the path and do not map all damage locations. Because of resource restrictions, many tornado surveys done by the National Weather Service (NWS) are restricted to finding tornado maximum intensities and tracks, all being accomplished in a single day of surveying. The survey for the 3 May 1999 Moore, Oklahoma, tornado is an example of a highly detailed survey where all damage points were surveyed by an NWS-led survey team (Speheger et al. 2002) and three surveys were led by engineering groups (FEMA 1999; Gardner et al. 2000; Marshall 2002).
On 20 May 2013, a tornado affected the cities of Newcastle, Moore, and Oklahoma City, in Oklahoma. The official pathlength of the tornado was 23 km (14 mi) and damage was up to 1737 m (1900 yards) wide (Fig. 1). It traveled from approximately 250° at an average forward speed of 9.5 m s−1. The tornado killed 24 people, injured 212 others, and damaged over 4250 structures. A team of surveyors from the National Weather Center (NWC; University of Oklahoma, National Severe Storms Laboratory, and the NWS Forecast Office and Warning Decision Training Branch) and two private organizations began surveying tornado damage on 21 May 2013 and continued the survey process for the next several weeks. The motivation to perform a detailed survey for this event came from 1) an interest in documenting as fully as possible the damage produced by a well-observed violent tornado and 2) the need to produce detailed information that could be used by follow-on research projects analyzing different aspects of the event, including radar studies.
Path of the 20 May tornado. EF-scale damage areas are color coded. Boundaries and names of cities mentioned in the text have been added.
Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-14-00039.1
b. Ground surveys
Ground surveys were performed by four teams with each team responsible for a segment of the path. Following the completion of surveys on 21 May 2013, team members reviewed the collected data and aerial imagery. Areas that were still closed off due to ongoing emergency operations or that were determined to need further investigation based upon aerial photography were identified. Follow-up surveys were completed by the original surveyors and additional staff from the NWC starting on 22 May 2013 and continued for the next several days; in all about 25 people participated in ground surveys. The rating process used in the survey follows steps outlined in EF-scale training materials developed by the NWS Warning Decision Training Branch (WDTB 2013). A strength of the EF scale used in this survey was the large number of damage indicators (DIs) available to rate damaged objects/structures, each with a range of intensities/wind speeds for expected, lower-bound, and upper-bound construction practices or levels of quality. A limitation of the EF scale was finding damage points that were not documented DIs within the scale. In total, 4253 damage points were evaluated by the surveyors, of which 4222 were DIs within the EF scale (W06).
The survey team was able to document large numbers of structures through the use of detailed ground surveys of most areas along the tornado’s path and from aerial imagery. Knowledge about the ground-rated structures (construction quality and other factors) allowed surveyors to use the aerial imagery to remotely evaluate, and eventually rate, the small number of structures that were not inspected on the ground. Damage information collection was centralized using the NWS Damage Assessment Toolkit (DAT; Camp et al. 2014). The DAT consists of an application available for smartphones and tablets, and a web application for access to data. DAT interfaces allow surveyors to quickly make EF-scale ratings by marking the DI and the corresponding degree of damage (DoD) for each damaged structure while simultaneously geotagging the structure and potentially adding other metadata, such as a photograph. The use of the DAT during the survey allowed teams to send information back to the NWC, where the information could be used to update the track path and intensities as the survey was ongoing.
c. Aerial surveys and imagery
An aerial survey completed by one of the coauthors (JL) from an Oklahoma Highway Patrol helicopter during the morning of 21 May 2013, before significant clean-up operations had begun, greatly aided the survey process. Surveying was also assisted by using high-resolution satellite imagery that was available via the Google Crisis Map (Google 2013) and later within the DAT. Aerial imagery collected by the Civil Air Patrol was also used to help finalize ratings. Toward the end of the survey process, a third aerial survey with a separate independent set of EF-scale ratings (Atkins et al. 2014) became available to the survey team. The third aerial survey was checked against the existing ratings, and a few ratings (less than 10) were changed, based on the new information.
2. Survey results
a. Western path segment
Some wind damage began southwest of Newcastle (Fig. 2). Early damage was all minor, consisting of lightly damaged home and barn roofs, damage to small outbuildings, and tree damage. A mapping of the tree damage (Figs. 3a,b) showed a chaotic pattern, sometimes with intermittent evidence of rotation and convergence, but lacking the consistent tree-fall pattern first seen near Oklahoma Highway 37 (Fig. 3c). Available photographic evidence from several storm spotter and media sources did not reveal any condensation funnel or sustained surface debris swirl development during the same period as the early damage. Also, viewing of dual-polarization data from two central Oklahoma Weather Surveillance Radar-1988 Dopplers (WSR-88Ds)1 did not reveal a well-defined or consistent tornado debris signature during the early damage. Although it is possible that one or more weak tornadoes existed for short periods and, along with strong, nontornadic winds of the developing low-level mesocyclone, contributed to the early damage, not enough evidence was present to define a tornado damage path. Therefore, the official start of the tornado was determined to be in the northwestern portion of Newcastle just south of Oklahoma Highway 37, at 1956 UTC.
Western section of the path of the 20 May tornado. EF-scale damage areas are color coded. Each dot marks a DI. Locations shown in Fig. 3 are marked.
Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-14-00039.1
Tree-fall patterns (arrows) in three selected areas during the early portion of wind damage: (a) earliest portion of the early path, near 16th St. and County Line Rd.; (b) middle portion of the early path, near 18th St. and Bermuda Dr.; and (c) transition to the beginning of the tornado path. Locations are marked in Fig. 2. Arrow direction gives direction of tree fall; arrow length has no special meaning. [Background images from Google Earth.]
Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-14-00039.1
Crossing Oklahoma Highway 37, the tornado moved northeast into a subdivision and produced EF4 damage only 1.4 km from the start. As the tornado approached and crossed the Canadian River, the width of the damage grew from 400 m just after forming to 800 m at the river. As the tornado crossed the river, it displaced two sections (~70 m in length) of a steel highway bridge from their concrete pillars (Fig. 4a).
Damage images: (a) old Interstate-44 bridge with two sections downed (note no apparent damage to new Interstate-44 bridge to the left of the old bridge); (b) oil storage tank moved 2.1 km from tank farm near S. W. 149th St. and Western Ave. to neighborhood near Santa Fe Ave. [one of the coauthors (CK) can be seen standing in front of the tank]; (c) large propane tank moved 900 m from Orr Family Farm to Westmoor neighborhood east of Briarwood Elementary School (inset: tank cradle at Orr Family Farm); (d) one of the destroyed classroom buildings at Briarwood Elementary School; (e) aerial view of PTES and surrounding neighborhoods (note houses with slabs swept clean south of the school); (f) aerial view of MMC and surrounding area, blue arrows mark movement of vehicles from MMC parking lot and debris trajectories west of Telephone Rd., blue dashed lines mark selected vehicles; (g) EF5-rated house, based on use of proper anchor bolts, nuts, and washers (note bent anchor bolts) to attach bottom plate to foundation; and (h) EF4-rated house, based on use of nails to attach bottom plate to foundation (note no appreciable penetration of cut nails into concrete foundation).
Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-14-00039.1
Northeast of the river in an area with fewer DIs, tornado damage continued to increase in width, reaching its maximum width of 1737 m. Thereafter, the width began to slowly narrow to between 1000 and 1400 m where it remained steady for some time. Turning more to the east along S. W. 149th St., the tornado continued to intensify, producing areas of EF4 damage and vegetation/ground scouring. Just before crossing Western Ave., the tornado impacted an oil drilling and storage facility. Five tanks were displaced; two traveled as far as 1.4 km (near Briarwood Elementary) and 2.1 km (near Santa Fe Ave.; see Fig. 4b). The dimensions and weights of the tanks are not known.
b. Central path segment
The tornado crossed Western Ave. at 2012 UTC just north of S. W. 149th at EF4 intensity, destroying several buildings, including a small strip mall and two farms (Fig. 5). Two large fuel tanks (each weighing 10 tons) at one of the farms were dislodged and traveled 700 and 900 m (Fig. 4c). The tornado continued east from the farms, leaving an impressive swath of vegetation/ground scouring in an open area, before impacting Briarwood Elementary School (Fig. 4d). Briarwood was initially rated EF5, but more recent engineering investigation by the American Society of Civil Engineers (ASCE) has revealed construction flaws that necessitate reducing the EF rating to lower-bound/less-than-standard-construction rating of EF4.2 Two wings of the school were completely destroyed. In the Westmoor housing addition just east of the school, the tornado produced a 160-m-wide swath of EF4 damage with two houses rated EF5 (see section 3a for discussion of the definition of EF5 damage to homes used in this survey). The first fatalities caused by the tornado occurred in the Westmoor neighborhood (black circles in Fig. 5).
As in Fig. 2, but for the central path segment. Locations of fatalities are marked with black circles, and numbers of fatalities at each location are annotated. Locations of Briarwood Elementary School, PTES, and MMC are shown.
Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-14-00039.1
Crossing Santa Fe Ave., the tornado began moving more northeastward and entered the Santa Fe Plaza neighborhood. In this area, the width of the EF4 damage expanded to nearly 350 m (Fig. 4e). This is narrower than the maximum EF4 width of ~600 m for the Greensburg, Kansas, tornado (Marshall et al. 2008), and the maximum EF4 width of ~550 m for the Joplin, Missouri, tornado (Marshall et al. 2012). The majority of fatalities in the 2013 Moore tornado occurred near the maximum width of EF4. Included in the area is Plaza Towers Elementary School (PTES), which had large portions destroyed, but damage was rated at EF4, based on construction practice. Seven fatalities occurred at the school. Homes swept clean off their foundations were found south of Plaza Towers, but were rated EF4, not EF5, again based on construction practice. Most of the homes near Plaza Towers were built after the 3 May 1999 F5 tornado, yet did not have adequate base plate connections to the concrete foundations. This is consistent with Marshall (2002), who found poor construction practices in areas rebuilt after the 3 May 1999 tornado.
The tornado continued northeast and north through a green space and into a neighborhood south of the intersection of S. W. 4th St. and Telephone Rd. Four homes in the neighborhood were rated EF5. Near the S. W. 4th St. and Telephone Rd. intersection, the tornado executed a loop where a convenience store was destroyed with three fatalities.
Completing its loop and moving southeast, the tornado severely damaged the Moore Medical Center (MMC; EF4). The tornado impacted many vehicles that were parked at the medical center (Fig. 4f). Most of the vehicles were pushed or lofted southeast, but one was found to have been lofted back to the west and deposited in a field more than a block to the west-northwest of the hospital. One vehicle landed on top of the two-story medical center. The tornado continued southeast, damaging and destroying several businesses near the medical center, before turning eastward across Interstate 35 (I-35).
c. Eastern path segment
In neighborhoods east of I-35, starting at about 2025 UTC, tornado damage (Fig. 6) continued to narrow in width compared to west of the interstate, from approximately 1300 m over Plaza Towers to around 500 m as it entered the neighborhoods just east of I-35. The forward speed of the tornado began accelerating from this point until the end of the track, at times >15 m s−1. The damage path along this part of the track maintained an EF4 core that was usually only one or two houses wide. Only two homes in the neighborhoods east of I-35 were rated EF5. The last two fatalities produced by the tornado occurred at a small industrial building and at a home east of I-35. The southern portion of Highland East Junior High School was destroyed, but because of adequate sheltering, injuries were few.
As in Fig. 2, but for the eastern path segment.
Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-14-00039.1
Crossing S. E. 4th St. east of Bryant Ave., the tornado exited the higher-density residential areas of Moore. From this point on, the tornado was in low-density housing areas or in completely rural areas. One final home in the low-density housing area was rated EF5. The final structures damaged were on a farm along S. Air Depot Road. The tornado dissipated in a tree line about 230 m east-southeast of the farm at 2035 UTC.
d. Damage summary and comparison with other tornadoes
In total, the survey documented 4253 objects damaged by the tornado, of which 4222 were defined as DIs within the EF scale and 31 were not DIs (Table 1). It is interesting to note that roughly half of all damaged structures were rated EF0. Less than 40% of all structures with EF1 and greater damage were completely destroyed (EF3–EF5), and less than 1% of EF1-and-greater damaged structures were EF5. Comparing to other events where numbers of F- and EF-scale ratings for structures were tabulated, there is a suggestion that one-half or more of damaged structures in cities large enough for complete mapping will experience only EF0 damage. Of course, the EF0 total might also include rear-flank and inflow winds, and damage from lofted debris, some of which could be outside of what is typically thought of as tornado wind damage. For Moore, comparing the 1999 tornado (old F-scale ratings) and the 2013 tornado (new EF-scale ratings), there is more EF1 damage and less EF3 damage in 2013 although the paths were through similar areas. However, none of the differences could be directly linked to the use of older and newer rating scales. Differences in building practice limit comparison of structure damage in events in different areas, but there is a greater amount of EF1 damage in Moore in 2013 when compared to Greensburg and Joplin. This difference could be the result of several factors, including rotational wind profiles and tornado rating practice. For all events, EF5 damage is limited to 1% or less of the affected structures.
Total number of surveyed DIs for the 2013 Moore tornado and three other F5/EF5 tornadoes. Percentages are given with respect to F1/EF1–F5/EF5 DIs unless indicated otherwise. Data for 3 May 1999 are from Marshall (2002), for 4 May 2007 from Marshall et al. (2008), and for 22 May 2011 from Marshall et al. (2012).
3. Discussion
a. Determining what constitutes an EF5 rating for one-/two-family homes (EF-scale DI 2)
EF-scale documentation defines assignment of an EF5 rating when a “well constructed” home is swept clean from its foundation (W06). The definition of a well-constructed home can vary by regional building practice, and several other factors. For this survey, it was decided that an EF5 rating would be assigned to homes that had the following characteristics: 1) foundation swept clean with debris strewn some distance downwind; 2) foundation (generally slab) to base-plate connections with properly spaced bolts with properly sized, fitted, and tightened washers and nuts; 3) removal of a large percentage of the base plates from the foundation; and 4) some anchor bolts bent (see Figs. 4g and 4h for examples). Implicit in this definition is that (independent of load–path connections above) the wind load has been transferred to the foundation–base-plate connection and failed there. This definition has several advantages: 1) surveyors can examine foundations and make a determination even when knowledge of other connections cannot be made because they cannot be found in the mixed debris, 2) it complies with local building codes, and 3) it preserves context to previous surveys done in the same areas with the same basic definition, for example, during the Moore event in 1999 (Speheger et al. 2002).
b. Use of non-DI damage and damage context
Damaged objects (31 in all, including a bridge, ground/vegetation scouring, large storage tanks, and vehicles; see Fig. 4 for examples) were found that were not represented DIs on the EF scale. Currently, such objects can only be used to supply context to the existing documented DIs. At times in this survey, EF-scale contours (e.g., those found in Figs. 1–2 and 5–6) were extended or shaped by the existence of non-DI damage. This is certainly a useful technique to apply, particularly in rural areas such as the beginning and end of the Moore tornado path. In the future, it is hoped that additional objects susceptible to damage by tornado winds can be added to the EF scale to expand its quantitative use.
c. Improving future surveys
Although effort was made to completely survey all objects/structures within the path during the first pass through the damage track, it was later found that some points needed initial survey or a revisit to complete some aspect of the rating at a point. Unfortunately, since most damage was cleaned up or moved within days of the tornado, later surveys sometimes were not possible. When large numbers of points needed to be surveyed, a combination of 1) real-time data compilation and sharing (the DAT was used here), 2) early and frequent team meetings to review findings and identify locations/objects that needed additional attention, and 3) a framework for later reliance on geolocated aerial/satellite imagery was utilized. All the steps used in this case can contribute to improved future surveys. This is particularly important for up- or downgrading potential high-end damage points. More survey details including other recommendations can be found in Ortega et al. (2014).
4. Concluding remarks
This study was a multiagency team effort to document as completely as possible the devastating 20 May 2013 tornado that impacted Moore, Oklahoma, and to provide information useful to future research concerning the event. The survey team implemented a technique that combined detailed ground surveys and aerial/satellite imagery to survey all of the impacted structures within the tornado’s path. A total of ~4250 structures were rated. Using techniques outlined in this survey, it is hoped that in the future additional resources will be found to allow for more complete surveys of significant tornado paths, including documenting all rated objects/structures in the DAT database. Such documentation could lead to being able to determine differences in building practices/DI distributions that lead to EF-rating differences.
The final overall rating of EF5 relied on damage to homes. A definition of EF5 damage to homes, appropriate for Oklahoma building practices and building codes, was developed and used. Although the team-produced definition proved useful in this case to separate EF4 and EF5 housing damage, more engineering analysis is needed in the EF scale to define “well-built homes.” A concern with the definition used in this survey is the possibility that an EF5 tornado in Oklahoma might not be an EF5 tornado in some other place with different building codes and different building practices, as well as different rating practices. Variability in rating practices is a growing concern. Hopefully, the NWS can address the issue with enhanced training related to survey practices and use of the DAT.
Acknowledgments
We thank the other members of the survey team: Kristin Calhoun, Tanya Brown, Patrick Marsh, Jack Friedman, Chuck Doswell, Darrel Kingfield, Ashlie Sears, Jeremy Wesley, John Ferree, and Bruce Thoren. We especially thank Lans Rothfusz who coordinated the first few hectic days of the survey. Funding for CIMMS authors was provided by NOAA/Office of Oceanic and Atmospheric Research under NOAA–University of Oklahoma Cooperative Agreement NA11OAR430072, U.S. Department of Commerce.
REFERENCES
Atkins, N. T., Butler K. , Flyann K. , and Wakimoto R. M. , 2014: An integrated damage, visual, and radar analysis of the 2013 Moore, Oklahoma EF5 tornado. Bull. Amer. Meteor. Soc., doi:10.1175/BAMS-D-14-00033.1, in press.
Camp, J. P., Rothfusz L. P. , Anderson A. , Speheger D. , Ortega K. L. , and Smith B. R. , 2014: Assessing the Moore, Oklahoma (2013) tornado using the National Weather Service Damage Assessment Toolkit. Special Symp. on Severe Local Storms: The Current State of the Science and Understanding Impacts, Atlanta, GA, Amer. Meteor. Soc., 830. [Available online at https://ams.confex.com/ams/94Annual/webprogram/Paper233734.html.]
Edwards, R., LaDue J. G. , Ferree J. T. , Scharfeneberg K. , Maier C. , and Coulbourne W. L. , 2013: Tornado intensity estimation: Past, present, and future. Bull. Amer. Meteor. Soc., 94, 641–653, doi:10.1175/BAMS-D-11-00006.1.
FEMA, 1999: Midwest tornadoes of May 3, 1999. Building Performance Assessment Team Rep., FEMA Publ. 342. [Available online at http://www.fema.gov/media-library/assets/documents/647.]
Fujita, T. T., 1959: A detailed analysis of the Fargo tornado of June 20, 1957. SMRP Tech. Rep. 5, University of Chicago, 29 pp.
Fujita, T. T., 1971: Proposed characterization of tornadoes and hurricanes by area and intensity. SMRP Tech Rep. 91, University of Chicago, 42 pp.
Gardner, A. K., Mehta K. C. , Tanner L. J. , Zhou Z. , Coner M. , Howard R. , Martinez M. , and Weinbeck S. , 2000: The tornadoes at Oklahoma City on May 3, 1999. Wind Science and Engineering Center Tech. Rep., Texas Tech University, 28 pp.
Google, cited 2013: Moore, Oklahoma tornado. [Available online at http://google.org/crisismap/2013-oklahoma-tornado.]
Marshall, T. P., 2002: Tornado damage survey at Moore, Oklahoma. Wea. Forecasting, 17, 582–598, doi:10.1175/1520-0434(2002)017<0582:TDSAMO>2.0.CO;2.
Marshall, T. P., McCarthy D. W. , LaDue J. G. , Wurman J. , Alexander C. , Robinson P. , and Kosiba K. , 2008: Damage survey of the Greensburg, KS tornado. Preprints, 24th Conf. on Severe Local Storms, Savannah, GA, Amer. Meteor. Soc., 8B.3. [Available online at https://ams.confex.com/ams/pdfpapers/141534.pdf.]
Marshall, T. P., Davis W. , and Runnels S. , 2012: Damage survey of the Joplin tornado: 22 May 2011. 26th Conf. on Severe Local Storms, Nashville, TN, Amer. Meteor. Soc., 6.1. [Available online at https://ams.confex.com/ams/26SLS/webprogram/Paper211662.html.]
Ortega, K., and Coauthors, 2014: Damage survey and analysis of the 20 May 2013 Newcastle–Moore EF-5 tornado. Special Symp. on Severe Local Storms: The Current State of the Science and Understanding Impacts, Atlanta, GA, Amer. Meteor. Soc., 828. [Available online at https://ams.confex.com/ams/94Annual/webprogram/Paper241554.html.]
Speheger, D. A., Doswell C. A. III, and Stumpf G. J. , 2002: The tornadoes of 3 May 1999: Event verification in central Oklahoma and related issues. Wea. Forecasting, 17, 362–381, doi:10.1175/1520-0434(2002)017<0362:TTOMEV>2.0.CO;2.
WDTB, cited 2013: EF-scale training. National Weather Service/Warning Decision Training Branch. [Available online at http://www.wdtb.noaa.gov/courses/EF-scale/.]
WSEC, 2006: A recommendation for an enhanced Fujita scale (EF-Scale). Wind Science and Engineering Center Tech. Rep., Texas Tech University, 95 pp.
