Developing a More Complete Tornado Climatology Using a Large Radar Dataset

A known problem in the tornado reports record is an under-reporting of tornadoes in sparsely populated areas owing to a lack of damage indicators. We attempt to develop a more comprehensive analysis of where tornadoes likely occur using a large dataset of low-level rotation in thunderstorms. The Multi-Year Reanalysis of Remotely Sensed Storms, a quality controlled set of radar data derived from the WSR-88D network, is mined to determine which rotation swaths likely coincide with tornadoes, even if no tornadoes reports are nearby. A spatial estimate of tornado frequency across the contiguous United States is developed, with particular focus east of the Rocky Mountains. We compare this method to several other methods used to generate tornado climatologies and explore methods for exploring how to best identify tornadoes at distances far from the radar site. Additionally, we explore land cover types that frequently coincide with estimated tornado locations to identify possible ranges of surface roughness lengths that may be influencing near-surface wind fields associated with tornadoes.