An algorithm based on Extended Kalman Filter to process data from polarimetric X-band weather radars

Dual-polarization Doppler (also called polarimetric) weather radar systems provide measurements related to the amplitude and phase of the radar signal at the horizontal and vertical polarizations. The relationships between these measurements can be used to constrain the processing/retrieval of the radar variables. In the present contribution, an algorithm based on extended Kalman filtering is proposed. It takes into account the relationships between polarimetric variables and their spatial correlation to estimate the specific differential phase shift on propagation Kdp, the attenuation corrected radar reflectivity at horizontal (vertical) polarization Zh(v), and the differential phase shift on backscatter delta.

Simulated radar fields (from simulated 2D fields of raindrop size distributions) are used to parameterize and evaluate the proposed algorithm in a simulation experiment. Kdp and delta are retrieved with an excellent accuracy, outperforming existing estimators solely based on smoothed measurements of the total differential phase shift Psidp. In terms of attenuation correction, the proposed algorithm performs similarly to the commonly used ZPhi algorithm. By comparing the directly retrieved differential phase shift on propagation Phidp with the cumulated Kdp estimate, the algorithm can also be used to check the accuracy of the total calibration of the radar.

The extended Kalman filter method is also tested on rain data collected by a polarimetric X-band radar in the Swiss Alps. The rain rate values during one rain event are retrieved from the polarimetric radar variables estimated using the proposed method, and then compared to rain rate values obtained from a disdrometer considered as a reference. The good agreement between disdrometer and radar estimates confirms the reliability of the proposed polarimetric radar processing algorithm.