The current MLDA identifies the radar signature of wet snow at higher antenna elevations and projects this designation onto lower elevations. We propose to improve the existing MLDA by examining the radar signature of wet snow at lower elevations and then to refine that designation with wet bulb temperature data from NWP models. This allows for better performance in situations where the ML is horizontally non-uniform. The proposed 1scheme incorporates the ideas of decision tree logic, fuzzy logic, and object identification techniques to merge radar and model data. The radar designation is considered most reliable at ranges close to the radar, and in locations with wet bulb temperatures between 0 and 4 Celsius. The model information is used more aggressively at longer ranges where the quality of radar designation deteriorates due to beam broadening and overshooting the melting layer.
The presentation of the improved MLDA algorithm will include a detailed comparison with the current algorithm and an examination of the changes produced in the HCA and QPE outputs. Several cases have been examined with an emphasis on cold season events showing strong horizontal variability in the parameters of the melting layer. Furthermore, an analysis of the algorithms performance has been completed for a case where the model data are compromised.