Identification of Precipitation-Clouds Based on the Dual-Polarization Doppler Weather Radar Echoes Using Deep–Learning Method

Abstract

The dual-polarization Doppler weather radar is a kind of radio-frequency sensor that can provide abundant information about atmospheric particle scattering behavior. The identification of the precipitation cloud type based on dual-polarization Doppler weather radar echoes is a study that classifies precipitation clouds based on the scattering theory of precipitation cloud particles to polarized electromagnetic waves. In recent years, the Doppler weather radar has been widely used in quantitative precipitation estimation, and the accurate identification of precipitation cloud types plays an essential role in improving the accuracy of precipitation estimation. The accuracy of the conventional precipitation cloud identification method relies on the number of features that are identified by human eyes, and it greatly reduces the operation efficiency. In order to improve the accuracy and efficiency of the precipitation cloud identification, a methodology of precipitation cloud identification based on deep learning is proposed in this paper. The method mainly consists of three major parts, which are constant altitude plan position indicator data inversion, zero-layer bright band identification, and precipitation-cloud classification by using the deep learning network model. At last, this paper evaluates the identification effect of this method through a real precipitation process. The results show that this method can distinguish the stratiform clouds and convective cloud precipitation in the precipitation area in real time, and it is in good agreement with the ground observation data. This method is very useful for improving the accuracy of the quantitative precipitation estimation of the Doppler weather radar.