Printer Friendly

Crowdsourcing the El Reno 2013 tornado: a new approach for collation and display of storm chaser imagery for scientific applications.

ABSTRACT

The 31 May 2013 El Reno, Oklahoma, tornado is used to demonstrate how a video imagery database crowdsourced from storm chasers can be time-corrected and georeferenced to inform severe storm research. The tornado's exceptional magnitude (~4.3-km diameter and -135 m s"! winds) and the wealth of observational data highlight this storm as a subject for scientific investigation. The storm was documented by mobile research and fixed-base radars, lightning detection networks, and poststorm damage surveys. In addition, more than 250 individuals and groups of storm chasers navigating the tornado captured imagery, constituting a largely untapped resource for scientific investigation.

The El Reno Survey was created to crowdsource imagery from storm chasers and to compile submitted materials in a quality-controlled, open-access research database. Solicitations to storm chasers via social media and e-mail yielded 93 registrants, each contributing still and/or video imagery and metadata. Lightning flash interval is used for precise time calibration of contributed video imagery; when combined with georeferencing from open-source geographical information software, this enables detailed mapping of storm phenomena. A representative set of examples is presented to illustrate how this standardized database and a web-based visualization tool can inform research on tornadoes, lightning, and hail. The project database offers the largest archive of visual material compiled for a single storm event, accessible to the scientific community through a registration process. This approach also offers a new model for poststorm data collection, with instructional materials created to facilitate replication for research into both past and future storm events.

**********

Storm chasers offer opportunities for researchers to develop unprecedented visual archives of severe storm and tornado characteristics that by precise temporal and location fixing are now useful for scientific application.

Over the past two decades, the increasing availability of meteorological data and mobile communications has led to a rapid growth in recreational storm chasing (Young 2010). Once mostly limited to severe storm researchers and a few hobbyists, storm chasing is now pursued by thousands of individuals from many countries, with the attention focused across the central United States (U.S.) during the peak of tornado season (April-June). The chaser community now includes tour groups led by professional guides, photographers, media groups, and researchers actively gathering data on storms. To date, storm chaser data resources, and particularly the visual observations that chasers routinely capture, have remained underutilized by the scientific community. Here, we present a case study demonstrating the high research value of storm chaser imagery after it has been located accurately in space and time, and its conversion to a scientifically valuable research database.

Scientists in field programs have at times commented on the interference presented by storm chasers (Young 2010). However, chasers' regular presence around supercellular and tornadic storms presents enormous potential to provide visual and meteorological observations useful to researchers. Several studies have leveraged storm-chasing observations (e.g., Pietrycha et al. 2009; Allen 2012; Wurman et al. 2013; Longmore et al. 2015), demonstrating their potential for providing visual observations and wind measurements. Larger field campaigns have also applied this technique of multiple-vehicle observations in an organized fashion, such as implemented during the two Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX) field programs [16 and 40 vehicles, respectively: Rasmussen et al. (1994); Wurman et al. (2012)], but have found the coordination of these large teams highly challenging. Another limitation to these programs was the number of potentially observable tornadic storms during the scheduled campaign periods and over a predetermined study domain. In contrast, field observations from small research groups that are flexible to observe storms during active periods [e.g., Tactical Weather-Instrumented Sampling in/near Tornadoes Experiment (TWISTEX): Karstens et al. (2010); Doppler on Wheels (DOW): Wurman et al. (1996) and Wurman and Kosiba (2013); radial X-polarized radar (RaXPol): Pazmany et al. (2013)] have shown considerable success obtaining research-grade observations. Most of these observations are coincident with independent storm chaser records of storm imagery and other data, which remain untapped.

The confluence of these two independent interests, storm chasers and field researchers, on the same tornadic supercell presents a research opportunity: provided that storm chaser video and still imagery can be collated and fixed precisely in time and location, researchers could potentially apply them in studies on storm hazards, thermodynamics, kinematics, microphysics, and electrification. Such a concept builds on the pioneering studies, predating the rise of storm chasing, that first incorporated crowdsourced imagery of tornadoes (e.g., van Tassel 1955; Hoecker 1960; Fujita 1960). Fujita (1960) sourced film and still photography from local residents and used these uncoordinated observations to produce composite analyses of a storm's tornadoes and mesocyclone structure. This technique was limited by available technology and relied on pattern matching of finescale visual characteristics and cloud tags. Hoecker (1960) used photogrammetry to derive velocity estimates of identifiable cloud tag and airborne debris features, which yielded motion estimates to 75 m [s.sup.-1], and conform well to modernday measurements of tornadic windfields (Hoecker 1960; Wurman et al. 2013).

Advances in tornado research in the decades following these photographically based studies have been registered primarily using remote sensing [e.g., mobile Doppler radars: Wurman et al. (2013), Tanamachi et al. (2012), Snyder and Bluestein (2014), etc.; lightning detection networks: Calhoun et al. (2013), etc.; and satellite platforms: Bedka et al. (2015)], mobile mesonets in field programs to sample the near-tornado environment (e.g., Karstens et al. 2010; Edwards et al. 2013), poststorm damage surveys (e.g., Atkins et al. 2014), and numerical modeling (e.g., Markowski and Richardson 2014; Orf et al. 2016). Advances also continue to be registered through single perspective photo or videogrammetry paired with radar observations (e.g., Wakimoto et al. 2011; Atkins et al. 2012) including for the 2013 El Reno tornado (Wakimoto et al. 2015).

Despite these important advances, there remains a need for applying visual observations to address several as yet unexplained aspects of tornadoes and their parent thunderstorms. The temporal characteristics of continuously evolving processes such as storm dynamics, tornadogenesis, suction vortices, structural failures, debris loading, and lightning activity can all be better documented and understood through imaging systems with high frame rates. In radar studies, the aliasing between successive radar scans can make tracking of rapidly moving storm-scale features problematic (Snyder and Bluestein 2014). Recreational storm chasers commonly utilize relatively high-resolution imaging systems, including digital single-lens reflex cameras, consumer-grade video cameras with high-definition resolution and framing rates from 24 to 60 [s.sup.-1], weatherproof miniaturized video cameras mounted on the exterior of vehicles, and mobile phones now capable of high-quality video. Consequently, many daytime tornadoes in the United States are now increasingly documented in high-resolution imagery from multiple perspectives over an extended time in the storm's life cycle. Such documentation becomes optimized when a synoptically evident meteorological setting occurs, which may result in storm chasers converging on a single target storm. This scenario creates the opportunity for parallel, albeit uncoordinated, visual data collection on a single tornadic supercell, or on multiple supercells during an outbreak. Until now, however, there has not been a complementary program structured to consolidate, organize, and analyze this imagery.

The El Reno storm. On 31 May 2013, a long-lived supercell produced a large, 40-min-duration tornado near El Reno, Oklahoma. The tornadic path width reached at least 4.3 km (NOAA 2013; Wurman et al. 2014). The tornado was assessed by the NOAA Survey and Wakimoto et al. (2015) at EF3 intensity on the Enhanced Fujita scale, based on observed damage, which contrasted to observed wind speeds of 135 m [s.sup.-1], recorded by mobile Doppler radars (Snyder and Bluestein 2014; Wurman et al. 2014). It is the first tornado known to have caused storm chaser mortality, including the TWISTEX research team of Timothy Samaras, Carl Young, and Paul Samaras. The tornado also caused multiple injuries, when numerous chase groups were caught in the tornado's path (Draper and Peter 2013; Wurman et al. 2014). The synoptic setting featured a stationary front oriented southwest (SW)-northeast (NE) from Oklahoma into Missouri, while a strong dryline extended southward into the Texas panhandle. Surface cyclogenesis in northwestern Oklahoma prompted the dryline circulation to surge eastward and moisture convergence to occur in the vicinity of the triple point near El Reno, Oklahoma. This led to a tornado-supportive environment by the time of storm initiation, characterized by strong instability and a strong potential for supercells and significant tornadoes (Wurman et al. 2014; Snyder and Bluestein 2014). For a more detailed examination of the El Reno supercells environment and evolution through its tornadic phases, see Bluestein et al. (2015).

The setting focused storm chasers and researchers around central Oklahoma prior to convective initiation. Shortly after storms began developing at 2100 UTC (UTC = CDT + 5 h; herein, all times are UTC), the dominance of the southernmost supercell west (W) of Oklahoma City concentrated image and data collection, both of precise time- and location-fixed remote sensing observations by researchers and uncontrolled visual observations of storm chasers. Following tornadogenesis, the rapid increase of tornado width and changes in forward speed and direction caused many storm chasers to lose situational awareness and become engulfed within or behind the tornadic circulation. Several teams were severely impacted, with storm chasers being at least five of the eight associated fatalities, all of which occurred in vehicles thrown from roadways by tornadic winds (Wurman et al. 2014).

In this study we describe the approach and methods of the El Reno Survey (www.el-reno -survey.net), a project developed to compile, analyze, and collate storm chaser imagery of the 2013 El Reno storm into a database and multiperspective visualization tool for research applications. The El Reno storm's structure, hazards, and radar-derived characteristics have already been described by Wurman et al. (2014) and Snyder and Bluestein (2014). The El Reno case has also been used to demonstrate the application of new technologies for high-intensity lightning events (Lang et al. 2015), rapid-scan polarimetric radar (Tanamachi and Heinselman 2016), and phased array radar (Kuster et al. 2015). Here, our focus will be on the approach utilized in the El Reno Survey to create the first-ever research database comprising storm chaser imagery from a high-intensity tornado event and derivative preliminary results.

The paper is structured as follows: In the next section, "Project methodology and data," describe the Survey approach along with the data used to produce a precisely synchronized and georeferenced visual archive; "Results" explores the data collection results and output products; and in "Discussion," we outline example applications to which the El Reno dataset is being or could be utilized, along with initial research findings. Finally, in "Future directions," we discuss challenges encountered and lessons learned during production of the El Reno dataset and illustrate how this study can be used as a pilot for future endeavors relating to crowdsourced visual observation archives of severe storms and other phenomena.

PROJECT METHODOLOGY AND DATA. Here we outline the procedure to gather and collate an imagery archive for the El Reno Survey, describing required engagement between the project team and the storm chaser community via social media and subsequent steps to process and catalog gathered material (Fig. 1).

Soliciting storm chaser observations. To obtain an initial sample of storm chasers present, we searched social media (e.g., YouTube, Vimeo, Facebook, Twitter), online forums such as Storm Track (www .stormtrack.org), archives of the Spotter Network (www.spotternetwork.org), and websites of individual chasers. This identified more than 250 separate chase units (one or more persons in a vehicle) who had been in close proximity to the tornadic mesocyclone, and therefore were of interest for the Survey objectives. These individuals include experienced "veteran" chasers, less-experienced and novice chasers, meteorologists, storm spotters conducting civil defense operations, researchers, media chasers, and some local residents. A project website was subsequently developed, with an online survey form (http:// el-reno-survey.net/storm-observer-reporting-form/) for participating parties to register with the Survey and submit metadata, including contact information, driving routes, types of media collected, and notable events observed. We polled experienced storm chasers on the suitability of our survey template; their helpful suggestions improved the format. On the survey form, we included a formal declaration indicating that the copyrights of all imagery submitted were to remain solely the property of the photo/videographer and that usage would be restricted to the Survey and researchers for scientific purposes only.

Following these preparatory steps, we issued a solicitation for participation through a range of social media channels. We explained the purpose of the data collection effort and invited all chasers and individuals who observed the storm to contact the project team about sharing their imagery and data resources. Targeted requests were also extended to individual chasers known to have notable footage. Following initial contact and registration, respondents who agreed to make their observations available were asked to upload unedited footage to a central repository. This generally required several gigabytes of video and/or still image file transfers per contributor. Material received varied greatly in quality and potential utility, but we accepted and archived all entries in the recognition that researchers may find useful information even in lower-quality imagery.

Video time correction. Each video sequence was accurately calibrated in time by comparing the time intervals between lightning flash occurrences, particularly cloud-to-ground (CG) flashes, to a reference dataset available in the remotely sensed National Lightning Detection Network (NLDN) flash archive (Cummins and Murphy 2009). This archive records the timing, location, and peak current amplitude of CG events with millisecond accuracy, providing a calibrated reference database. CG flashes with peak currents > 20 kA that were recorded on the NLDN matched video-confirmed CG events very well, so these were used as the calibration dataset. Lower-intensity flashes on CG flash detection networks may be misclassified intracloud flashes and thus were omitted (Fleenor et al. 2009; Cummins and Murphy 2009). Time calibration on each video segment is performed as follows:

1) Analyze video segment to identify all CG flashes and optical transients, recording frame number of each flash in relative time on a metadata spreadsheet (Fig. 2a).

2) If multiple CG events are present in the segment, match pattern to NLDN-detected CG flash timeline to set an accurate time.

3) If a single CG flash is present in the segment, match pattern of the lightning channel morphology and/ or duration of continuing current illumination to establish first-guess estimates of CG flash time, then cross-validate synchronicity with optical transients from other videos already set in time.

4) If no CG events are present, use pattern matching to cross-validate timing between optical transients recorded in other videos already set in time.

5) If all of the above lightning alignment methods are inconclusive, use other information in the video segment to establish placement in time. This may include distinctive cloud features also captured in other videos, audio tracks conveying National Weather Service warnings and live television station feeds, and time of passage relative to Survey participant vehicles with time-set video imagery captured in the footage. Since videoclip metadata includes sequential naming conventions, time-fixed segments preceding or following an unfixed segment provide initial bracketing for establishing the correct timeline.

Through this procedure, we temporally synchronized almost all video clips [greater than or equal to] 15-s duration to within one video frame (i.e., 0.033 s at 29.97 fps) of actual time. Video segments were excluded from analysis if they could not be synchronized.

Camera viewpoint geolocation. To fix camera locations along the video-clip timeline, either available GPS logs were used, or, more commonly, Google Maps (maps.google.com) spatial matches to video frames, to identify a precise latitude-longitude location. Most Survey participants did not record vehicle locations in real time. Therefore this information was determined during postanalysis and then added to chasers' respective metadata files. On their survey forms, most participants were able to provide some driving route information and positions relative to major events in storm evolution. This information was used when reviewing submitted videos to identify visible road intersections and other distinguishing features and landmarks using Google Maps, and where available, its ground-level Street View utility (Fig. 2b). Latitude-longitude coordinate pairs of vehicle positions were retrieved for an initial set of points and entered on each chaser's raw metadata spreadsheet, with accurate placement in time provided by the annotated time code already added to video segments. Additional location fixes are determined at 71-s time steps, coinciding with radar scans (see below), with vehicle/camera positions determined through careful observation of lesser features, such as farm buildings and other structures, counts of electrical utility poles and fence posts, gates, and driveways, all of which are identifiable using zoomed-in satellite views. For each locational fix, we also note the video camera's instantaneous viewing azimuth on the metadata form. The view direction is estimated with +10[degrees] precision, based on road orientations and landscape features identifiable in mapping software. The rural farm road grid in the area, mostly oriented in the cardinal directions at 1.6-km intervals, greatly facilitated azimuth determinations.

Database entry. Combining these two analytic approaches, video time correction and camera viewpoint geolocation, we are able to determine time and location parameters for each video segment with high precision, allowing scientific applications such as time-fixing features observed during storm evolution, photogrammetric analysis via multiperspective imagery, and image pairing with simultaneous radar imagery. For each videographer's time-fixed segments, imagery is arranged into 1-h segments covering the period 2230-2330, which encompasses the most important features in mesocyclone evolution, tornado formation, and evolution through the circulation becoming rain wrapped to the east of El Reno. The imagery is reprocessed at 1280 x 720 pixel resolution at 29.97 fps (a widely used standard in the United States), with on-screen annotations providing the videographer's name and accurate UTC time code. For temporal continuity, periods without video coverage are displayed as blank frames with the annotations retained. For unstable hand-held or moving vehicle sequences, we selectively performed image stabilization to improve the presentation and note this in metadata.

In the final database, each chaser unit's materials consists of the following: 1) the edited hour-long video containing time-fixed video clips, 2) all video files submitted by the contributor in their original, full-resolution format, and 3) an Excel file providing second-by-second metadata for the period of analysis. Thus, in the final database, for parties who wish to manually handle the imagery, we provide both original raw footage and a processed version, including accurate time and location attributes presented in a common format. We note that no attempt has been made to unwarp lens-distorted imagery from wide-angle lens perspectives to normal zoom and create a standard real-world view for analysis purposes.

Online visualization tool. One of this project's objectives is to facilitate visualization and navigation of the data resources through a web interface. The online tool developed, designated the Tornado Environment Display (TED), is accessible through the project website and provides simultaneous display of radar, CG lightning, chaser positions, and up to four simultaneous videos, against a Google Maps background (Fig. 3). The user can select between one and four chaser video displays, starting at any time point between 2230 and 2330. The background mapping platform has the major tornado track shaded gray, while arrowhead icons indicate chaser positions oriented according to camera direction at the time of the most current locational fix. Once set in motion, videos play synchronously as chaser icons and a circle denoting tornado position, scaled to the tornado diameter, move in real time on the map. The radar layers are from the S-band Multi-Function Phased-Array Radar (MPAR) at the National Weather Radar Testbed in Norman, Oklahoma, which updates at 71-s intervals (Kuster et al. 2015). Tornado positions are interpolated from 1-min centroid coordinates as reported by the National Weather Service (NOAA 2013). Standard Google Maps navigation options are available (zoom, drag, and a toggle between road map and satellite view). For the radar layer, users may select between 0.5[degrees]-tilt MPAR reflectivity and velocity fields in smoothed or unsmoothed format. Other radar archives, including from mobile research radars, may become available in later iterations. Several demonstrations of the display are presented in the next section, and online readers can access animations using links provided in figure captions.

RESULTS. Status of survey database. In November 2013, five months after the El Reno storm, we announced the Survey by submitting links to our online form on community forums and other social media. The database currently contains 93 registrants in total, of which 87% have followed through by uploading imagery. Our subjective characterization of the potential scientific value of each contributor's materials, ranks 64% as high value, 27% as medium, and 9% as low; the clarity and stability of a chaser's submitted imagery, and the content and storm features contained therein, were the basis of these determinations. The database can be accessed directly upon request at our project website (http://el-reno -survey.net/contact-us/), where users can gain access after registering and agreeing to respect copyrights on imagery usage terms.

Findings on storm phenomena. Representative examples of novel findings brought forth during database collation and analysis are presented below. Additional elaboration on applications in lightning research, cloud model simulations, and giant hail production, among other topics, will be detailed in subsequent publications.

Cloud-to-ground lightning: Network detection efficiency and flash frequency. The Survey database provides relatively complete documentation of lightning flash activity viewable from the surface, within a 15-km radius of the supercell's mesocyclone and encompassing storm development through the El Reno tornado's dissipation (2100-2343). For the period of focused analysis, between 2230 and 2330, 125 video-confirmed CG flashes are identified in our database. All but 14 (89%) can be matched temporally within a few milliseconds of NLDN-determined CG events with ground strike detections in the mesocyclone region. Of the 111 common events, 110 (99.1%) were indicated by the NLDN to be of positive polarity. A multitude of NLDN-detected negative polarity CG events over the same time (n = [+ or -] 1,200) could not be confirmed by video. All of these had low peak currents (<15 kA), except for a -27-kA CG flash at 2252:02, shown in Fig. 3, that is the only video-confirmed negative CG flash over the 60-min period of focused study. A detailed assessment of lightning network detection efficiency, based on El Reno Survey validation data, is presented in Coy et al. (2016).

Distinct patterns in lightning flash activity appear to phase with events in the El Reno supercell's dynamic evolution. Episodes of frequent positive CG discharges within and close to the mesocyclone, alternating with pronounced lulls in activity, precede the genesis of the El Reno tornado at 2202. The peak flash rate of 11 CG flashes [min.sup.-1] at 2233-34 occurs 29 min before the tornado. Conversely, multiminute periods of extremely low counts of any form of visible flash activity (both CG and intracloud) are equally apparent in the video analysis.

Tornado evolution and structure. As amply documented elsewhere (Wurman et al. 2014; Snyder and Bluestein 2014; Bluestein et al. 2015; Atkins et al. 2014), the El Reno tornado was uncommonly large, intense, and complex, with a structure that could be characterized as a multivortex mesocyclone (MVMC; Wurman and Kosiba 2013; Wurman et al. 2014). The El Reno Survey database and TED viewing utility facilitate time-specific investigations from many perspectives, revealing the temporal evolution and spatial aspects of both large and finescale features and enabling precise comparison of morphological characteristics with synchronous radar observations. Several examples of multiple-vortex structure as imaged in sequential time steps are described here.

Video analysis reveals that first ground contact by a condensation funnel occurs at 2302:15, somewhat earlier than has been previously reported in the literature [e.g., 2304 in Bluestein et al. (2015)]. By 2304, multiple vortices are present beneath a broad truncated funnel (Fig. 4a). These circulations are most clearly visible from viewing perspectives to the north (N) west-southwest (WSW) through northeast, where a bright background silhouettes the vortices.

The most significant of many internal features within the broader tornado is a large and long-lived (~10 min) subvortex that tracked in a quasi-cycloidal path from 2314 to 2324, in effect constituting a "tornado within a tornado" (Marshall et al. 2014; Wurman et al. 2014). On RaXPol radar, the origin of this circulation can be traced to a tornado vortex signature (TVS), first evident at 2313:54 about 2.8 km north-northwest (NNW) of the primary circulation, which propagates cyclonically around the parent circulation as a satellite vortex (Bluestein et al. 2015). Multiperspective video reveals that this subvortex has its origin as one of a series of small funnel clouds that develop from a low-to-the ground tail cloud at 2313:12. A large condensation funnel extending to the ground materializes rapidly within rotating rain curtains at 2314:30, then follows a decaying orbit into the primary tornadic circulation from the NNW (Fig. 4b). A similar "parade of vortices" structure (L. Orf 2014, personal communication) was observed by two Survey contributors (L. Bruns and D. Robinson; not shown) at 2319-20, with up to five ropelike vortices, some with ground circulations, feeding into the larger circulation from the NW.

The El Reno tornado achieves its broadest diameter around 2318-20 as it accelerates eastward across U.S. Highway 81 south of El Reno (Wurman et al. 2014). During this period, the major subvortex occupies the tornado core and causes a highly publicized incident, when a chase vehicle operated by a media group is lofted from the roadway and rolled some distance to the east (Wurman et al. 2014). As viewed from all quadrants beyond the radius of its outer wall, the tornado appears as a vast, rain-wrapped rotating cylinder. However, at this same time a Survey contributor (A. Gwyn) caught within this outer wall and positioned 0.5 km north of the core captured footage of the inner subvortex itself crossing US-81. This footage reveals that this subvortex was itself a multiple-vortex circulation at this time, with numerous vortices rotating about a common center (Fig. 4c).

In our final example, we explore wind damage occurrences on the periphery of the tornado, while it was close to its maximum intensity. During this period, extremely strong inflow winds and associated hazards were particularly challenging to in situ chasers attempting videography. Despite an absence of CG lightning between 2319:30 and 2326:30, pattern matching of weak optical transients from intracloud discharges enabled precise time calibration, and video stabilization facilitated coherent tracking of features despite often-severe camera motion during filming. The visual characteristics of damage-causing features can now be examined at high temporal resolution. At 2326:07, three videographers located close together at a RaXPol deployment site east-northeast (ENE) of the tornado captured the disintegration of one or more large structures, as a rapidly moving subvortex passed to its rear in line with the broader vortex (Fig. 5). Successive video frames reveal strongly divergent debris motions, as large objects loft upward and outward, including to the south (S) in apparent opposition to extremely strong cyclonic flow (videos by K. Butler, C. Asselin, and V. Deligny). At 2325:55, the subvortex featured volume-averaged groundrelative winds of 119.5 m [s.sup.-1] just above the surface and an exceptionally rapid forward translation speed of 78 m [s.sup.-1] (Snyder and Bluestein 2014; Wurman et al. 2014). Simultaneous to this structural failure, another videographer (J. Bishop), driving east on Interstate 40, captured the sudden disintegration of a steel framed highway billboard 2.5 km ENE of the tornado centroid. In addition, a highway exit sign is destroyed and thrown across Interstate 40, also at 2326:07 in the Asselin video. The presence of the extremely intense subvortex implicates it in these damage events, but the video evidence suggests otherwise. Synchronized frame-by-frame analysis of the video sequences reveals that starting around 2326:01 an intense wet microburst impacts the surface in the area of the three damage incidents and appears to be responsible for all. A RaXPol 1.0[degrees] velocity scan at 2325:59 shows both the intense subvortex 3.8 km west-southwest (WSW) of the radar site, and a strong divergence signature consistent with a microburst exhibiting inbound flow of 35-40 m [s.sup.-1] at 0.5 km and outbound flow of 40-45 m [s.sup.-1] at 1.5 km range just SW of the radar (Fig. 5i; image contributed by H. Bluestein and J. Snyder, University of Oklahoma).

Giant hail occurrence. The Survey video analysis enables precise determination of time and location of hail fall in several sectors of the storm. The El Reno supercell produced hailstones with dimensions approaching and possibly exceeding the largest [a list of the largest recorded hailstones is summarized in Blair and Leighton (2012)] on record. Observations suggest that hail size reached a maximum along a NE-SW-oriented corridor 6-8 km northwest of the tornado centroid around 2318-20, and photographic documentation by the general public of fallen hailstones confirms diameters reached at least 16 cm (Witt et al. 2015). Videos from two contributors (H. Farrar and G. Rhinehart) establish the precise time and location of several exceptionally large hailstones as they reached the surface. Photogrammetry of video frames that show one falling hailstone suggests an estimated diameter that may have exceeded 20 cm prior to impact and a fall velocity of ~67 m [s.sup.-1] (Fig. 6). In a follow-up site visit, the Survey team interviewed local property owners, who reported that all buildings experienced severe roof damage from hailstones, including two hailstone penetrations directly through 26-gauge corrugated metal roofing. Large hail fall in progress was also documented on video in several unconventional locations of the supercell. This includes estimated 4-7-cm-diameter hailstones at 2317-18, falling at highly oblique angles well within the tornado's broad circulation in its eastern quadrant (video by M. Gribble, M. Gotl, and R. and S. Thompson). Other hailstones, estimated to be 4-7 cm, rebounded at least 5 m vertically after ground impact, 5 km east-southeast (ESE) of the tornado centroid at 2320 in an area of weak low-level radar reflectivity (video by S. Talbot).

DISCUSSION. The 31 May 2013 El Reno storm was an exceptional tornadic supercell of interest to researchers, particularly for its subvortex features and characteristics, extreme width of radar-derived EF1 winds, and giant hail production. This storm provides an ideal case to illustrate the utility of a visual database for incorporation into analysis with radar, lightning network data, and other spatiotemporally precise data. A nonexhaustive list of research applications of the El Reno Survey database and TED visualization tool include multiperspective visual reconstruction of tornadogenesis, multiplevortex evolution and long-lived subvortex behavior, spatiotemporal measurements of giant hail occurrence for integration into radar studies, verification of lightning network detection efficiency and confirmation of systemic biases, CG lightning morphology and duration characteristics, hazards presented to storm chasers and the public at large, and validation of high-resolution cloud model simulations of super-cell thunderstorms. As discussed above, tornado research has developed mostly through distinct modes of investigation, framed by technological advances and observational approaches. Our database compilation is of potential application in all of these research approaches. It may foster greater integration in tornado research looking forward, initially for the El Reno case, but also by extension of this approach to other storms from the past and the future.

The most critical factor enabling the creation of the project database was the success of the El Reno Survey in procuring voluntary contributions from storm chasers. The high degree of participation likely relates to the often-stated but mostly unrealized intent expressed by storm chasers that their observations could contribute to advancing tornado science, along with a desire to share their experience of the storm that took the lives of members of the chasing community.

Two additional factors were essential to developing a scientific database from contributed imagery: the availability of geospatial mapping and time calibrated lightning data resources. Cost-free online geographic information systems, including spatially matched high-resolution satellite imagery, provided the needed resource for georeferencing storm chaser video. Google Earth software contains imagery from multiple dates, offering both pre- and poststorm views that enable scene matching, even where structures and vegetation were severely reconfigured or destroyed by the storm. The time-calibrated NLDN data are also available upon request and at no cost to universities for education and research (Unidata N.D.). Our project data, in turn, refines how researchers might apply filtering, when analyzing NLDN CG flash data. The mischaracterization of the El Reno supercell's low amplitude (i.e., <15 kA) CG lightning production by the NLDN is now unambiguously established with multiperspective videography. Misclassification of low-amplitude CG lightning by the NLDN has been noted in previous studies of central U.S. supercell storms (Emersic et al. 2011; Strader and Ashley 2014) and particularly for those containing an inverted polarity charge structure (e.g., Calhoun et al. 2013). In the El Reno case, this also causes the NLDN's unfiltered CG flash representation to greatly distort the flash polarity distribution. As displayed in real time and in the NLDN network archive, ~10% of CG flashes close to the El Reno supercell's mesocyclone were indicated as positive polarity, whereas the reality was the diametric inverse (99.1%). Previously, filtering thresholds of 10 kA have been used by investigators as limits on flashes for inclusion in analysis (e.g., Calhoun et al. 2013; Strader and Ashley 2014). However, minimum peak current of 146 video-confirmed El Reno CG flashes is 19.2 kA, suggesting that a higher threshold set closer to this value might be considered in future studies.

Visual representation of storm phenomena. The storm features presented in the "Results" section as examples of the scientific utility of the El Reno storm visual database elucidate both previously studied and newly identified features. Storm videography between 2300 and 2305, available from more than two-dozen perspectives, offers further details into the tornado-genesis process and early structure of the nascent tornado. These views indicate the formation of the condensation funnel occurred earlier than previously stated in the literature. Numerous small-scale vortices not resolved by the beam binning of radar owing to their small size and distance from the radar (Bluestein et al. 2015) are also identifiable in the visual archive.

The multivortex structure of the long-lived subvortex, revealed in video at 2319-20, may be the unique observation of a process normally hidden from view when it occurs in other large tornadoes. Wurman and Kosiba (2013) presented a case captured by mobile Doppler radar of a subvortex within a broad tornado that itself contained subvortices. The hierarchy of circulation scales in the El Reno tornado, whereby a multivortex mesocyclone harbors a multivortex subvortex, is therefore revealed to be even more complex than previously reported (Wurman et al. 2014; Snyder and Bluestein 2014; Bluestein et al. 2015). The complexity of damage in this area, with EF3 structural damage registered up to 700 m outside the track of the subvortex center (Wakimoto et al. 2015), may relate to this multivortex structure.

In the final example from 2325 to 2326, severe wind damage occurrences can be examined in the context of two intense wind features, tracking within the broader tornadic circulation. Frame-by-frame video analysis from different perspectives, and corroborating evidence from a strong divergence signature in the Doppler velocity field (see Fig. 5), reveal that the simultaneous failure of structures separated by hundreds of meters is likely caused by an intense, wet microburst on the periphery of the tornado. It is apparently unrelated to the passage of a high-intensity subvortex, passing 2.5 km away closer in to the center of the broad tornadic circulation. Fujita inferred that damaging downdrafts, which he termed twisting downbursts, may accompany tornadoes in the presence of a strong hook echoes and parent mesocyclones (Fujita 1981; Forbes and Bluestein 2001). This case, from the El Reno storm, is associated with damage of at least EF2 intensity and offers perhaps the first visual documentation of this phenomenon in progress.

Challenges and opportunities encountered. The El Reno Survey approach is based on a modern adaptation of a data collection method pioneered in the 1950s. The resulting Survey also promotes a two-way interaction between the Survey team and storm chasers, allowing us to report back, share project deliverables, and build a sense of partnership between chasers and researchers in scientific developments. As part of this collaborative process, we have developed an online reference document (http://el-reno-survey.net /category/findings/) that aims to inform storm chasers on the best practices for scientific data collection. Over time, adoption of these may benefit the analysis in other cases. Also provided on our project website are templates and instructions for soliciting chaser observations, compiling chaser metadata, and building edited videos from raw footage.

CHASER DATA PROVISION. Responses to data requests for the El Reno Survey were mostly enthusiastic, but follow-through in providing promised data by some chasers remains incomplete. As the Survey approach becomes better known, and products such as the TED visualization tool increase its visibility, chasers may be more willing to overcome hesitation over participation.

IMPROVING STORM CHASE METADATA ARCHIVING. The limited interest in scientific applications of storm chase data leads many chasers to not record data for archival purposes. Storm chasers have the means to automatically log GPS positions on mobile phones and in-vehicle navigation devices, yet results from the El Reno Survey has shown that only a small fraction do so routinely. We also note that while numerous chase vehicles are equipped with meteorological instrumentation, no Survey contributors provided archived meteorological data. In at least one case, however, this was due to destruction of instrumentation after encountering vehicle-relative winds measured to at least 70 m [s.sup.-1] (estimated ground-relative 56.2-59.4 m [s.sup.-1] accounting for vehicle motion; reported by B. Hendrickson and D. Rodriguez).

AUTOMATION OF TIME CORRECTION FOR VIDEO FOOTAGE. One of the limitations of this approach is the need for time-consuming manual analysis to time-calibrate and geo-calibrate raw video footage. We now encourage storm chasers to perform day-of-chase camera-clock calibration, using an authoritative measure (e.g., http://time.gov). For future storm video collation efforts, we recognize the need for automating temporal synchronization and are currently exploring options to leverage flash magnitude and interval between lightning flashes. For georeferencing imagery, we encourage chasers to adopt GPS-track archival as standard practice to reduce the need for manual pattern-matching analysis with satellite maps. An optimal solution would be cameras with calibrated time, GPS, and azimuthal parameters incorporated directly into metadata. Without such field data collection, manual pattern matching to Google Maps and similar software will remain necessary to establish georeferencing.

REMOVING MOTION AND IMAGE DISTORTIONS. For analysis purposes, storm chaser footage would ideally be taken by a vehicle- or tripod-mounted camera with minimal lens distortion. In contrast, much of the footage we received was hand held and often highly unstable. We used Final Cut Pro video editing software to stabilize imagery, which often succeeded in rendering poor quality video footage usable for scientific application. Many chasers now use video cameras mounted on their vehicles, and some even operate camera arrays covering all quadrants. With attributes of continuity and relative stability, such footage has proven among the most valuable collected by the Survey. Although not performed for the current project, unwarping imagery affected by lens distortions would be of value, particularly for photogrammetry and multiperspective analysis. However, this requires provision of metadata barrel characteristics of the recording device, focal length, type of sensor, and resolution.

FUTURE DIRECTIONS. The introduction of multiperspective and calibrated storm chaser video with full metadata and the TED visualization tool will enhance analysis options in research on tornadic supercells and other storm related phenomena. Beyond research applications, the TED tool also offers an educational opportunity for both students and chasers to explore the structure, evolution, intricacies, and hazards of a high-magnitude tornadic supercell. Looking forward, if voluntary contribution of storm chaser resources can be maintained, replication of this program for other storm cases of research interest should be possible, from both the recent past and in the future. For this broader objective, we offer several recommendations leveraging the experience gained during the El Reno Survey effort.

* Centralizing crowdsourcing appeals. We anticipate that crowdsourcing calls for storm chaser material are likely to increase. The unrestricted crowdsourcing opportunities afforded by social media mean that any party could issue such appeals. However, in order not to fatigue the pool of potential respondents, we suggest that such calls be limited for major events of high research interest and/or public impact. Also beneficial would be adoption of a standard call for participation tailored to the rapidly evolving social media environment.

* Rapid response. To maximize potential utility by researchers, storm database solicitation and preparation should be completed as soon as possible, ideally within three to six months following the event under study.

* Centralizing databases and data access. The El Reno Survey, and other storm case databases developed hereafter, ideally would be housed permanently in a common repository, such as a government laboratory or academic institution, where open-access provision would be maintained indefinitely, with archived events accessible by users through a single online portal.

* Standardizing search and display tools. Although developed to serve the purposes of the El Reno Survey project, the TED visualization tool is designed to serve as a standard display tool for browsing visual resources in any severe storm database. The current version is designated as El Reno-TED, and any subsequent versions created for other storms should follow this naming convention.

* Real-time data provision by storm chasers. Almost all of the chasers who observed the El Reno storm operated autonomously, with independent decision-making. Yet it is striking how the road grid and storm visual presentation frequently caused the random assemblage to self-organize into spatially distributed formations (see, e.g., Figs. 3, 4, and 6). Such a collection of observation points, moving in formation with a target storm, has in the past been the objective for deliberately deployed mobile mesonets of instrument-equipped research vehicles (Straka et al. 1996). This suggests the potential for a crowdsourced mesonet for real-time, in situ observations, possibly by making available to storm chasers a common microweather station, in effect, a permanent, self-deploying "VORTEX armada," provided scientific calibration of the observational devices can be maintained in an organized manner (e.g., Pietrycha et al. 2009; Wurman et al. 2012). The current Spotter Network of storm observers and chaser volunteers (Pietrycha et al. 2009) or an accessible venue for chasers to contribute their material similar to the approach used by mPing (Elmore et al. 2014) offer useful frameworks from which this type of approach could be developed.

ACKNOWLEDGMENTS. The El Reno Survey acknowledges the research legacy of Tim Samaras, Carl Young, and Paul Samaras and their contributions to tornado data collection. We warmly thank the multitude of storm chaser contributors to the El Reno Survey, without whose generosity this project would not have been possible; and Elke Edwards, who managed the survey form collections and created our project website. Phased Array Radar data were provided courtesy of NOAA/National Severe Storms Laboratory (NSSL), while NLDN data was provided courtesy of the University at Albany, State University of New York. Thanks also go to Charles Kuster, Don Burgess, and Arthur Witt at NSSL for an introduction to MPAR and providing data layers for use in the TED tool and to David Vollaro and Kevin Tyle at the University at Albany who kindly fielded our NLDN data request. Kristin Calhoun at NOAA/NSSL contributed valuable discussions and analysis on lightning with assistance from James Coy of Texas A&M University through his participation in the 2015 Research Experiences for Undergraduates (REU) Program at the National Weather Center. We thank Howard Bluestein and Jeffrey Snyder at the University of Oklahoma for contributing the RaXPol image shown in Fig. 5. The development of the TED visualization tool was funded by the National Geographic Expeditions Council (Grant EC0692-14). Gifts to Appalachian State University from the Weissman Family Foundation and several private donors supported the initial organizational effort for the El Reno Survey.

REFERENCES

Allen, J. T., 2012: Supercell storms: Melbourne's white Christmas 2011. Bull. Aust. Meteor. Oceanogr. Soc., 25, 47-51.

Atkins, N. T., A. McGee, R. Ducharme, R. M. Wakimoto, and J. Wurman, 2012: The LaGrange tornado during VORTEX2. Part II: Photogrammetric analysis of the tornado combined with dual-Doppler radar data. Mon. Wea. Rev., 140, 2939-2958, doi:10.1175/MWR -D-ll-00285.1.

--, R. Wakimoto, K. M. Butler, H. B. Bluestein, K. J. Thiem, J. C. Snyder, J. Houser, and J. Wurman, 2014: Aerial damage survey and high resolution dual-polarization radar analysis of the 2013 El Reno tornado. 27th Conf. on Severe Local Storms, Madison, WI, Amer. Meteor. Soc., 13.3. [Available online at https://ams.confex.com/ams/27SLS/webprogram /Paper254162.html.]

Bedka, K. M., C. Wang, R. Rogers, L. D. Carey, W. Feltz, and J. Kanak, 2015: Examining deep convective cloud evolution using total lightning, WSR-88D, and GOES-14 super rapid scan datasets. Wea. Forecasting, 30, 571-590, doi:10.1175/WAF-D-14-00062.1.

Blair, S. F., and J. W. Leighton, 2012: Creating high-resolution hail datasets using social media and post-storm ground surveys. Electron. J. Oper. Meteor., 13, 32-45.

Bluestein, H. B., J. C. Snyder, and J. B. Houser, 2015: A multiscale overview of the El Reno, Oklahoma, tornadic supercell of 31 May 2013. Wea. Forecasting, 30, 525-552, doi:10.1175/WAF-D-14-00152.1.

Calhoun, K. M., D. R. MacGorman, C. L. Ziegler, and M. I. Biggerstaff, 2013: Evolution of lightning activity and storm charge relative to dual-Doppler analysis of a high-precipitation supercell storm. Mon. Wea. Rev., 141, 2199-2223, doi:10.1175/MWR-D-12-00258.1.

Coy, J. J., K. M. Calhoun, and A. Seimon, 2016: A comparison of lightning flashes observed by groundbased detection networks and video. [Available online at https://ams.confex.com/ams/96Annual /webprogram/Paper283144.html.]

Cummins, K. L., and M. J. Murphy, 2009: An overview of lightning locating systems: History, techniques, and data uses, with an in-depth look at the U.S. NLDN. IEEE Trans. Electromagn. Compat., 51, 499-518, doi:10.1109/TEMC.2009.2023450.

Draper, R., and C. Peter, 2013: The last chase. Natl. Geogr. Mag., 2013 (November), 2-37. [Available online at http://ngm.nationalgeographic.com/2013/ll/biggest -storm/draper-text.]

Edwards, R., J. G. LaDue, J. T. Ferree, K. Scharfenberg, C. Maier, and W. L. Coulbourne, 2013: Tornado intensity estimation: Past, present, and future. Bull. Amer. Meteor. Soc., 94, 641-653, doi:10.1175/BAMS -D-ll-00006.1.

Elmore, K. L., Z. L. Flamig, V. Lakshmanan, B. T. Kaney, V. Farmer, H. D. Reeves, and L. P. Rothfusz, 2014: MPING: Crowd-sourcing weather reports for research. Bull. Amer. Meteor. Soc., 95, 1335-1342, doi:10.1175/BAMS-D-13-00014.1.

Emersic, C., P. L. Heinselman, D. R. MacGorman, and E. C. Bruning, 2011: Lightning activity in a hail-producing storm observed with phased-array radar. Mon. Wea. Rev., 139, 1809-1825, doi:10.1175 /2010MWR3574.1.

Fleenor, S. A., C. J. Biagi, K. L. Cummins, E. P. Krider, and X.-M. Shao, 2009: Characteristics of cloud-to-ground lightning in warm-season thunderstorms in the Central Great Plains. Atmos. Res., 91, 333-352, doi:10.1016/j.atmosres.2008.08.011.

Forbes, G. S., and H. B. Bluestein, 2001: Tornadoes, tornadic thunderstorms, and photogrammetry: A review of the contributions by TT Fujita. Bull. Amer. Meteor. Soc., 82, 73-96, doi:10.1175/1520 -0477(2001)082<0073:TTTAPA>2.3.CO; 2.

Fujita, T. T., 1960: A detailed analysis of the Fargo tornadoes of June 20, 1957. U.S. Weather Bureau Rep. 42, U.S. Government Printing Office, 67 pp.

--, 1981: Tornadoes and downbursts in the context of generalized planetary scales. J. Atmos. Sci., 38, 1511-1534, doi:10.1175/1520-0469(1981)038<1511: TADITC>2,O.CO; 2.

Hoecker, W. H., 1960: Wind speed and air flow patterns in the Dallas Tornado of April 2, 1957. Mon. Wea. Rev., 88, 167-180, doi:10.1175/1520-0493 (1960)088<0167:WSAAFP>2,O.CO; 2.

Karstens, C. D., T. M. Samaras, B. D. Lee, W. A. Gallus Jr., and C. A. Finley, 2010: Near-ground pressure and wind measurements in tornadoes. Mon. Wea. Rev., 138, 2570-2588, doi:10.1175/2010MWR3201.1.

Kuster, C. M., P. L. Heinselman, and M. Austin, 2015: 31 May 2013 El Reno tornadoes: Advantages of rapid-scan phased-array radar data from a warning forecasters perspective. Wea. Forecasting, 30, 933-956, doi:10.1175/WAF-D-14-00142.1.

Lang, T. J., S. A. Cummer, D. Petersen, L. Flores-Rivera, W. A. Lyons, D. MacGorman, and W. Beasley, 2015: Large charge moment change lightning on 31 May to 1 June 2013, including the El Reno tornadic storm. J. Geophys. Res. Atmos., 120, 3354-3369, doi:10.1002/2014JD022600.

Longmore, S., and Coauthors, 2015: An automated mobile phone photo relay and display concept applicable to operational severe weather monitoring. J. Atmos. Oceanic Technol., 32, 1356-1363, doi:10.1175/JTECH-D-14-00230.1.

Markowski, P. M., and Y. P. Richardson, 2014: The influence of environmental low-level shear and cold pools on tornadogenesis: Insights from idealized simulations. J. Atmos. Sci., 71, 243-275, doi:10.1175 /JAS-D-13-0159.1.

Marshall, T. P., D. Burgess, G. Garfield, R. Smith, D. Speheger, J. Snyder, and H. Bluestein, 2014: Ground-based damage survey and radar analysis of the El Reno, Oklahoma tornado. Proc. 27th Conf. on Severe Local Storms, Madison, WI, Amer. Meteor. Soc., 13.1. [Available online at https://ams.confex.com /ams/27SLS/webprogram/Paper254342.html.]

NOAA, 2013: El Reno and Moore damage survey report. [Available online at www.srh.noaa.gov /oun/?n=events-20130531-stormdata.]

Orf, L., R. Wilhelmson, B. Lee, C. Finley, and A. Houston, 2016: Evolution of a long-track violent tornado within a simulated supercell. Bull. Amer. Meteor. Soc., doi:10.1175/BAMS-D-15-00073.1, in press.

Pazmany, A. L., J. B. Mead, H. B. Bluestein, J. C. Snyder, and J. B. Houser, 2013: A mobile rapid-scanning X-band polarimetric (RaXPol) Doppler radar system. J. Atmos. Oceanic Technol, 30,1398-1413, doi:10.1175 /JTECH-D-12-00166.1.

Pietrycha, A. E., S. F. Blair, T. J. Allison, D. R. Deroche, and R. V. Fritchie, 2009: Emerging technologies in the field to improve information in support of operations and research. Electron. J. Oper. Meteor., 10, 2009-EJ2. [Available online at http://nwafiles.nwas .org/ej/pdf/2009-EJ2.pdf.]

Rasmussen, E. N., J. M. Straka, R. Davies-Jones, C. A. Doswell, F. H. Carr, M. D. Eilts, and D. R. MacGorman, 1994: Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc., 75, 995-1006, doi:10.1175/1520-0477(1994)075<0995:VOTOOR>2,OCO;2.

Snyder, J. C., and H. B. Bluestein, 2014: Some considerations for the use of high-resolution mobile radar data in tornado intensity. Wea. Forecasting, 29, 799-827, doi:10.1175/WAF-D-14-00026.1.

Strader, S. M., and W. Ashley, 2014: Cloud-to-ground lightning signatures of long-lived tornadic supercells on 27-28 April 2011. Phys. Geogr., 35, 273-296, doi:10 .1080/02723646.2014.918527.

Straka, J. M., E. N. Rasmussen, and S. E. Fredrickson, 1996: A mobile mesonet for finescale meteorological observations. /. Atmos. Oceanic Technol., 13, 921-936, doi:10.1175/1520-0426(1996)013<0921:AMMFFM >2-O.CO; 2.

Tanamachi, R. L., and P. L. Heinselman, 2016: Rapidscan, polarimetric observations of central Oklahoma severe storms on 31 May 2013. Wea. Forecasting, 31, 19-42, doi:10.1175/WAF-D-15-0111.1.

--, H. B. Bluestein, J. B. Houser, S. J. Frasier, and K. M. Hardwick, 2012: Mobile, X-band, polarimetric Doppler radar observations of the 4 May 2007 Greensburg, Kansas, tornadic supercell. Mon. Wea. Rev., 140, 2103-2125, doi:10.1175/MWR-D-ll-00142.1.

van Tassel, E. L., 1955: The North Platte Valley tornado outbreak of June 27, 1955. Mon. Wea. Rev., 83, 255-264, doi:10.1175/1520-0493(1955)083<0255 :TNPVTC"2.0.CO; 2.

Wakimoto, R., N. T. Atkins, and J. Wurman, 2011: The LaGrange tornado during VORTEX2. Part I: Photogrammetric analysis of the tornado combined with single-Doppler radar data. Mon. Wea. Rev., 139, 2233-2258, doi:10.1175/2010MWR3568.1.

--, --, K. Butler, H. Bluestein, K. Thiem, J. Snyder, and J. Houser, 2015: Photogrammetric analysis of the 2013 El Reno tornado combined with mobile X-band polarimetric radar data. Mon. Wea. Rev., 143, 2657-2683, doi:10.1175/MWR-D-15-0034.1.

Witt, A., D. W. Burgess, A. Seimon, and J. T. Allen, 2015: Rapid-scan dual-polarization WSR-88D observations of an Oklahoma hailstorm producing extremely-large hail. Proc. 37th Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 134. [Available online at https://ams.confex.com/ams/37RADAR /webprogram/Paper275689.html.]

Wurman, J., and K. Kosiba, 2013: Finescale radar observations of tornado and mesocyclone structures. Wea. Forecasting, 28,1157-1174, doi:10.1175/WAF-D-12-00127.1.

--, J. M. Straka, and E. N. Rasmussen, 1996: Fine scale Doppler radar observations of tornadoes. Science, 272, 1774-1777, doi:10.1126/science.272 .5269.1774.

--, D. Dowell, Y. Richardson, P. Markowski, E. Rasmussen, D. Burgess, L. Wicker, and H. B. Bluestein, 2012: The second Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX2. Bull. Amer. Meteor. Soc., 93, 1147-1170, doi:10.1175 /BAMS-D-11-00010.1.

--, K. Kosiba, and P. Robinson, 2013: In situ, Doppler radar, and video observations of the interior structure of a tornado and the wind-damage relationship. Bull. Amer. Meteor. Soc., 94, 835-846, doi:10.1175 /BAMS-D-12-00114.1.

--, K. Kosiba, P. Robinson, and T. Marshall, 2014: The role of multiple-vortex tornado structure in causing storm researcher fatalities. Bull. Amer. Meteor. Soc., 95, 31-45, doi:10.1175/BAMS-D-13-00221.1.

Young, J. R., 2010: Tensions grow between tornado scientists and storm chasers. Accessed September 2015. [Available online at http://chronicle.com/article /Tensions-Grow-Between-Tornado/65985/?key =TWx2JwQybnBFNHYzL3MTfyVSYXJ6dUxwOXd HZHsaY1pc.]

AFFILIATIONS: A. SEIMON--Department of Geography and Planning, Appalachian State University, Boone, North Carolina; ALLEN--International Research Institute for Climate and Society, Columbia University, New York, New York; T. A. SEIMON--Wildlife Conservation Society, New York, New York; TALBOT--Springfield, Illinois; HOADLEY--Falls Church, Virginia

CORRESPONDING AUTHOR: Dr. Anton Seimon, Dept. of Geography and Planning, ASU Box 32066, Appalachian State University, Boone, NC 28608

E-mail: anton.seimon@gmail.com

The abstract for this article can be found in this issue, following the table of contents.

DOI: 10.1175/BAMS-D-15-00174.1

In final form 6 April 2016
COPYRIGHT 2016 American Meteorological Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Seimon, Anton; Allen, John T.; Seimon, Tracie A.; Talbot, Skip J.; Hoadley, David K.
Publication:Bulletin of the American Meteorological Society
Geographic Code:1U7OK
Date:Nov 1, 2016
Words:8974
Previous Article:Where are the lightning hotspots on Earth?
Next Article:NASA's hurricane and severe storm sentinel (HS3) investigation.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters