Profile Classification Module of GPM-DPR algorithm: performance of first dataset

Abstract:

The Global Precipitation Measurement (GPM) mission was successfully launched in February 2014. It is the next satellite mission to obtain global precipitation measurements following success of TRMM. The GPM core satellite is equipped with a dual-frequency precipitation radar (DPR) operating at Ku and Ka band. DPR is expected to improve our knowledge of precipitation. Profile classification module of GPM-DPR is a critical module in the retrieval system for space borne radar. It involves two aspects: 1) precipitation type classification;and 2) melting region detection. Dual-frequency classification method that has been implemented into DPR algorithm relies on the microphysical properties using the difference in measured radar reflectivities at two frequencies, a quantity often called the measured dual-frequency ratio (DFRm). There are two aspects that control DFRm vertical profile: a) the non-Rayleigh scattering; b) the path- integrated attenuation. The DFRm is determined by the forward and backscattering properties of the mixed phase and rain and the backscattering properties of the ice. It holds rich information to assist in precipitation type classification and melting layer detection. In order to quantify DFRm features, a set of indices are defined. V1=(DFRm_max-DFRm_min)/(DFRm_max+DFRm_min). Where DFRm_max and DFRm_min are DFRm local max and min values. V2 is the absolute value of the mean slope for DFRm below the DFRm local min point. To further enlarge the difference between rain types, a third DFRm index V3 is defined V3=V1/V2. V3 is an effective parameter and provides a separable threshold for different rain types. The criteria for the melting layer top is defined as the height at which the slope of the DFRm profile hits a peak value. Similarly, the melting layer bottom is defined as the height the DFRm profile has a local minimum value. These criteria show good comparisons with other existing criteria. Dual-frequency classification method has been evaluated using GPM DPR real data since it is available. Initial evaluation shows good comparison on rain type classification from both dual-frequency classification method and Ku only method at a statistical point of view. Reasonable comparison can also be seen from melting region detection.