Testing of a background classification algorithm for use with dual-polarized radars in determining precipitation type at the surface

A key component to the new dual-polarized radar winter surface hydrometeor classification algorithm is the background classification derived from numerical model analyses. The accuracy of the background class is limited by the choice of algorithm, the uncertainty in the analysis system, and the horizontal variability of the precipitation type at the surface. Consideration of output from different algorithms, shows that all schemes considered are very accurate at detecting snow and rain. However, freezing rain and ice pellets are quite poor by comparison. As will be demonstrated using high-resolution observations of precipitation type, freezing rain and ice pellets often occur in mixes rather than in isolation. Using a background algorithm that allows for a freezing rain/ice pellet mix greatly improves the results. Statistical analyses as well as results from individual events will be presented that demonstrate the utility of the different background classifiers in concert with polarimetric radar observations.