AI for polarimetric radar QPE: Getting ready for APAR applications

Although polarimetric weather radar measurements provide rich information about precipitation, traditional approaches of using them can only extract part of the information due to limitations of the applied analytical tools. For example, the performance of conventional radar rainfall estimation algorithms, such as the radar reflectivity (Z) and rain rate (R) relations, is highly dependent on the local raindrop size distributions that may vary in different precipitation regimes. This research aims to develop an artificial intelligence (AI) based quantitative precipitation estimation (QPE) system for the NSF/NCAR Airborne Phased Array Radar (APAR) system. In particular, a deep learning model driven by a convolutional neural network and a vision transformer is designed to quantify precipitation from the multi-parameter multi-dimensional APAR observations. During the APAR development phase, the National Weather Service (NWS) Weather Surveillance Radar - 1988 Doppler (WSR-88D) observations are used to train and test the deep learning QPE model. In addition, simulated APAR observations and polarimetric measurements from a ground-based PAR are used to investigate the potential extension and problems for future APAR applications. Challenges in fine-tuning and adapting the deep learning QPE system for different radar systems with different view angles will also be detailed.