Object-based Filter Design for Spectral Polarimetric Weather Radar

Polarimetric Doppler weather radar which simultaneously measures microphysical and motion features of precipitation, is well recognized as an indispensable tool for atmospheric observation. However, due to the complex operation environment, weather radar may be interfered by some large-intensity clutter, which will bury weak precipitation signal. Aiming at removing moving and stationary large-intensity clutter while retaining weak signal in weather radar, a new clutter suppression method named the object-based spectral polarimetric (OBSPol) filter is put forward. Based on the spectral polarimetric feature and the spatial autocorrelation of precipitation, the OBSPol filter is implemented in the range-Doppler spectrogram to mitigate unwanted clutter. The methodology of the OBSPol filter is as follows. Step 1, the spectral polarimetric observable is utilized to generate one binary mask where “1” indicates the precipitation. Step 2, the mathematical morphology method is used to recover the missing areas of the obtained mask. Step 3, the connected components are selected and integrated into several separate objects based on the reconstructed mask. Step 4, whether the produced objects are signal or not will be further judged based on extra feature fusion. Thus, a final filtering mask can be obtained in the range-Doppler spectrogram. Data collected by the polarimetric Doppler IRCTR Drizzle Radar (IDRA) is used to assess the performance of the proposed filter. In the situation of IDRA which has been interfered by moving narrow-band clutter since its installation in 2007, the extra feature in Step 4 is the spectral width. It is foreseeable that the OBSPol filter can be extended to remove different types of weather radar clutter with the proper feature fusion. Additionally, this technique can be applied in real time due to its low computation complexity