Gust front detection using template matching on fused and multi-resolution radar data sets

We describe a novel data fusion algorithm that is a part of a larger project focused on detecting and predicting tornados and heavy rainfall. Specifically, our described method determines gust fronts, that is, convergence zones where cold and warm air masses meet. Due to meteorological constraints, we must combine multiple data sets of different resolution: reflectivity and shear (both azimuthal and radial). We base our decision technique on a method of template matching [R. Delanoy, et al., March 1992]. In this paper, we describe how to interpret these data, what characteristics are being sought, and what techniques we use in our solution. Of particular interest, we examine the effects of the templates and their ancillary scoring functions. We find that for these data, a non-linear scoring function provides robust discrimination.