Synthetic tornado climatologies based on Doppler radar vortex detection algorithms: An update

Tornado verification can be problematic in certain locations, such as those with low population density, in mountainous and forested regions, and where storm verication is not as actively pursued (such as in Europe and other foreign countries). An earlier study attempted to answer the question of whether radar-observed storm-scale vortex signatures (e.g., mesocyclones and TVSs) can be used as a proxy to synthetize tornado climatologies where verification is problematic.

In that study, we started with a database of "well-verified" tornado events (43 cases with about 207 tornado reports), and the associated output from Doppler radar algorithms designed to detect mesocyclone- scale vortices in thunderstorms. The likelihood that particular detections were tornadic based on a single algorithm detection attribute (a three-dimensional "strength rank") were then determined. The posterior probabilities of the well-verified dataset were applied to a test data set that was assumed to be poorly-verified (the poorly- verified data set is actually a three-year nearly-conclusive set of all the storm-scale vortex detections collected from the Pittsburgh radar during 1996-1998). The resulting synthetic climatological values for number of tornadoes are quite similar to the actual number of reported tornadoes within the domain of the radar for the "poorly-verified" data set.

For this paper, the analysis is continued using a number of other algorithm detection attributes (such as rotational velocity, depth, etc.), and using another three-year "poorly-verified" data set from St. Louis, Missouri. Some spatial statistical analysis is also performed to reduce the range-dependencies and some data artifact problems, and to determine some geographical distributions of synthsized tornadoes.

These results offer some optimism that radar data could be used to determine synthetic climatologies of tornadoes wherever radar data are collected. This method, coupled with other methods to synthsize tornado climatologies, might lead to the a better understanding of the distributions of tornadoes worldwide.