A technique for developing a US climatology of thunderstorms: The ThOR algorithm

As ubiquitous as thunderstorms are, a climatology that characterizes their spatiotemporal distribution does not exist. The development of such a climatology is confounded by the complexity of identifying and tracking thunderstorms and the size of the data set required to undertake this task. Early attempts to develop climatologies for small regions of the US have relied on data with poor spatiotemporal resolution and/or thunderstorm proxies with large errors. A robust thunderstorm climatology requires a method for identifying and tracking thunderstorms that uses data with a fidelity that is capable of resolving thunderstorm structure and lifecycle.

The Thunderstorm Observation by Radar (ThOR) algorithm was crafted to develop this climatology. ThOR identifies thunderstorms through synthesis of Level II radar data collected by NOAA's NEXRAD network of Doppler radars, cloud-to-ground lightning recorded by the National Lightning Detection Network (NLDN), and derived proximity vertical wind profiles from the NCEP North American Regional Reanalysis database. ThOR involves three principal steps: 1) cluster identification, 2) cluster tracking, and 3) lightning attribution. Cluster identification is the identification of coherent regions of precipitation apparent in regional mosaics of composite radar reflectivity. The w2segmotion tool within the Warning Decision Support System – Integrated Information (WDSS-II) system is used for this step. Cluster tracking is accomplished through a procedure developed by the authors. Lightning attribution is the final step toward identifying thunderstorms and is accomplished through a method developed by J. Lahowetz. The end result is a collection of thunderstorm tracks output in both tabular form and as GIS shapefiles.

The performance of both the cluster identification and the cluster tracking within ThOR were scored using thunderstorm tracks synthesized from the manual analysis of a group of analysts. The algorithm was then applied to Great Plains thunderstorms between 2001 and 2009. Comparison was then made to several common proxies for thunderstorms that are based solely on cloud-to-ground lightning. A description of ThOR, an explanation of the verification, and preliminary results will be presented at the conference.