Tornado Debris Data April 2014

WSI Corporation has recently developed a Tornado Debris Signature (TDS) algorithm that automatically detects the existence of debris lofted by tornadoes in real-time. The goal of this algorithm is to enable improved decision-making for deployment of insurance CAT team resources.

The success of the TDS algorithm is dependent upon the incorporation of data from the new dual- polarization (dual-pol) capability of the National Weather Service (NWS) radar network. These dual- pol data complement the more traditional radar-based Tornado Vortex Signature (TVS) detections to provide a better assessment of whether debris is lofted from a tornado.

About Dual-Polarization Radar Before the advent of dual-polarization (dual-pol) capabilities in recent years, the Weather Surveillance Radar – 1988 Doppler (WSR-88D) operated by the NWS was a non-polarimetric radar that only transmitted horizontally-polarized electromagnetic waves. This limited the measurement of refl ected energy to only the horizontal direction. Dual-pol radar sends simultaneous electromagnetic waves in both the horizontal and vertical directions. This enables the refl ected energy to have not only a horizontal measurement but also a vertical measurement. This two-dimensional representation of returning wave energy offers more detailed insight into the types and characteristics of the particles in the atmosphere (NSSL website).

Example of conventional versus dual-pol radar. Image courtesy of NOAA. Tornado Debris Signature

About Dual-Polarization Radar Continued... With the completion of the dual-pol WSR-88D upgrade, each of the radars now produce 14 new products that help to not only differentiate precipitation types and intensities but also non-meteorological particles. Some of the new radar products include Differential Refl ectivity (Zdr), Correlation Coeffi cient (CC), Specifi c Differential Phase (Kdp) and Hydrometeor Classifi cation Product (HC). Of these products, the CC is most important to the TDS algorithm because it refl ects the similarity between horizontal and vertical returns within a radar bin (Istok et al 2009). Since tornado debris is typically less uniform in shape than meteorological targets, CC can be a clue that a tornado is on the ground, provided a vortex signature is simultaneously present.

Example of low correlation coeffi cient as a result of lofted debris from May 19, 2013 Shawnee, OK tornado. Image courtesy of NWS Tulsa. Tornado Debris Signature

Characteristics of Tornado-Producing Storms Tornado-producing storms have several identifi able characteristics in radar data. None of these characteristics alone can demonstrate a tornado is present. When more than one is present, we can be more confi dent that a tornado is on the ground, lofting debris into the air.

A weak echo region (WER) is an area of markedly lower refl ectivity, resulting from an increase in updraft strength (Glickman, 2000) that can develop during the growth phase of a tornadic thunderstorm. A WER may also be surrounded on the top and sides by an area of higher refl ectivity (Glockman, 2000), referred to as a bounded weak echo region (BWER). The BWER is suggestive of a strong updraft that is preventing precipitation from reaching the ground, generally associated with a mesocyclone that is capable of producing a tornado.

Couplets in the velocity fi elds are another indication that a tornado is either occurring or imminent. A couplet occurs when signifi cant velocities towards and away from the radar are found in close proximity to one another. A couplet detected at the lowest elevation angles of the radar volume is evidence of rotation close to the ground. These couplets seen in the radar velocity data are the primary triggers for a TVS.

A hook echo is also a good indication that a tornado is either occurring or imminent. A hook echo is caused by the precipitation being pulled into the mesocyclone (Glickman, 2000). This typically shallow hook-shaped area of refl ectivity is often associated with a tornado.

Example of a hook echo (left) and velocity couplet (right) from May 20, 2013 Avant, OK tornado. Image courtesy of NWS Tulsa.

A debris ball is an area of extremely high refl ectivity that is generally located near a hook echo. Large (compared to precipitation), irregularly-shaped non-meteorological debris is often lofted into the atmosphere during a tornado (Bodine et al, 2013). The characteristics of this debris ball are the basis for using the correlation coeffi cient in the calculation of the TDS. Tornado Debris Signature

Raw Calculations The calculations for the TDS are based on the correlation coefficient and the corresponding reflectivity value at or near the location of a TVS. To have a TDS, a sufficiently low CC, sufficiently high reflectivity and a location within 130 km of a radar site are needed. TDS values can range from 0-10, but there are only a subset of discrete values that are produced in the current version of the algorithm (below).

TDS Value Criteria

0.0 No data or location beyond range

1.0 Reflectivity of area below threshold

2.0 Reflectivity is above threshold, but no CC data available for the area

3.0 Reflectivity is above threshold, but CC is above threshold

7.0 Reflectivity is above threshold, but CC is near but slightly above threshold

10.0 Reflectivity is above threshold, CC is below threshold. Tornado debris likely

A TDS detection is only triggered for a value of 10; lower non-zero values represent “failure modes”, or those TVS detections that do not have TDS detections associated with them. Tornado Debris Signature

Examples A few recent examples of TDS detections are now described in more detail.

Moore, OK On May 20, 2013, an EF5 tornado with estimated winds of 200-210 mph devastated the community of Moore, OK. The tornado developed at 2:45 PM CST, 4.4 miles west of Newcastle, OK. The tornado produced EF0 and EF1 damage for the fi rst 10 minutes. The fi rst TDS was detected at 2:55 PM CDT (below). During the 50 minutes the tornado was on the ground, it traveled 17 miles and had a maximum width of 1.3 miles. It destroyed homes and businesses in Moore and the surrounding communities (from NWS Norman). The tornado caused 24 fatalities. The evolution of the TDS during the event is shown below.

The fi rst tornado debris signature is detected near Newcastle, OK at 2:55 CDT on 5/20/13.

A tornado debris signature continues to be indicated near the end of the hook echo on the south side of Moore, OK at 3:15 CDT on 5/20/13.

The tornado debris signature passes Moore, OK at 3:25 CDT on 5/20/13. Tornado Debris Signature

Examples Continued...

Cleburne, TX On May 15, 2013, an EF3 tornado with estimated winds of 140 mph hit the town of Cleburne, TX. The tornado developed in the late evening, 9 miles south-southwest of Cleburne. The tornado produced pockets of EF3 damage during the 8.5 miles it traveled, and had a maximum width of 1 mile. It caused signifi cant damage to dozens of homes in Cleburne (from NWS Dallas/Fort Worth). An image of the TDS during the event is shown below.

The tornado debris signature near Cleburne, OK at 8:49 PM CDT on 5/15/13. Tornado Debris Signature

Performance The TDS has shown to be extremely effective at detecting tornadoes on the ground. During a test in late winter and spring 2011-12, 18 tornadoes were confirmed by spotters within 130 km (81 miles) of WSR-88D radars that had completed the dual-pol upgrade. In all 18 cases, a TDS was detected during part of the lifespan of the tornado. The tornadoes ranged in strength from EF0 to EF4. During the test period, TDS signatures occurred in 47 WSR-88D volume scans. Confirmed tornadoes correlated with 45 (96%) of those volume scans with detections. The strength breakdown of the The breakdown of TDS detection 18 tornadoes is as follows: is as follows:

EF0 3 # of TDS 47

EF1 2 # of TDS with a Tornado 45

EF2 5 # of False TDS 2

EF3 7

EF4 1

Shortcomings The data above clearly show that the TDS algorithm is quite useful in the detection of damaging tornadoes; however, there are a few shortcomings. The most obvious is that the storm needs to be sufficiently resolved by a WSR-88D dual-pol radar. Storms that are far from a radar site may not be well resolved due to the curvature of the Earth. Thus, the radar beam will intercept the storm at a higher altitude, making it more likely that the tornadic debris is underneath the beam. For this reason, the algorithm caps any potential debris signatures to within 130 km (or 81 miles) of the radar location. The second shortcoming is that a TDS relies on the detection of a TVS. While many tornadoes, especially strong, long-lasting ones, coincide with a TVS, weaker tornadoes have a lower tendency to trigger TVS. Finally, the TDS will not detect all tornadoes near the radar since many do not loft enough debris to trigger the algorithm, especially those tornadoes over open grasslands or other areas without significant population.

Summary The Tornado Debris Signature algorithm represents an evolutionary step in real-time tornado detection. While there are the typical limitations that radar imposes -- such as the tornado’s distance from the radar and how high the debris is lofted -- it aids those who need to make quick resource allocation decisions with greater confidence than ever before. When users receive notification of a tornado debris signature, it is likely damaging weather is occurring close to or over the specified location and that action is more likely to be required. Our findings also suggest the false alarm rate is considerably lower than National Weather Service Tornado Warnings; however, limitations in the algorithm and the radar itself means some tornadoes will go undetected when merely considering the algorithm alone. Additional information, such as spotter and chaser reports and confirmations from trained meteorologists, are still required to definitively say that a tornado has or has not touched down. Tornado Debris Signature

References Bodine, David J., Matthew R. Kumjian, Robert D. Palmer, Pamela L. Heinselman, Alexander V. Ryzhkov, 2013: Tornado Damage Estimation Using Polarimetric Radar. Weather Forecasting, 28, 139–158. [Available online at http://dx.doi.org/10.1175/WAF-D-11-00158.1]

Glickman, Todd S. (ed.) (2000). Glossary of Meteorology (2nd ed.). American Meteorological Society.

Istok, Michael J., M. A. Fresch, S. D. Smith, Z. Jing, R. Murnan, A. V. Ryzhkov, J. Krause, M. H. Jain, J. T. Ferree, P. T. Schlatter, B. Klein, D. J. Stein, G. S. Cate, and R. E. Saffle. “WSR- 88D Dual Polarization Initial Operational Capabilities.” 25th Conference on International Interactive Information and Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology(2009). [Available online at https://ams.confex.com/ams/pdfpapers/148927. pdf]

“NWS Damage Survey for 5/20/2013 Newcastle/Moore Tornado Event - Update 3”. National Weather Service Norman, Oklahoma. May 21, 2013. [Available online at http:// mesonet.agron.iastate.edu/wx/afos/p.php?pil=PNSOUN&e=201305211950]

“RESEARCH TOOLS: DUAL-POLARIZED RADAR.” NSSL Research Tools: Dual Polarized Radar. Web. 09 Sept. 2013. [Available online at http://www.nssl.noaa.gov/tools/radar/dualpol/]