Cloud-based Analytic Framework for Precipitation Research (CAPRi) to Enhance the Spatial Resolution of GPM Data

Cloud-based Analytic Framework for Precipitation Research (CAPRi) is developing a Deep Learning architecture to enhance the resolution of Global Precipitation Measurement (GPM) satellites surface rain rate data. The GPM Validation Network (VN) provides temporally and spatially matched 3D comparisons of precipitation retrievals from the GPM satellite and ground-based weather radars at a horizontal resolution of ~5 km (footprint of GPMs onboard radar). On the other hand, Multi-Radar Multi-Sensor (MRMS) data is a 2D composite of operational weather radar, satellite, and rain gauge observations over the continental United States and Canada, providing surface precipitation rate information at approximately 1 km spatial resolution. It is of interest to precipitation scientists to develop a technique for enhancing the spatial resolution of GPM rainfall data. Hence, the higher resolution MRMS data is used to enhance the resolution of GPM using Convolutional Neural Networks (CNNs) for better space-borne observations, mainly outside the coverage of ground-based weather radar. The super-resolved GPM observations can be used to create 3D convective-scale precipitation features. By tracing the new super resolution data back to the Dual-frequency Precipitation Radar (DPR) data using the VN, valuable information can be reconstructed. One of the use cases is utilizing the Cesium (map), a web-application of CAPRi, to investigate GPMs depiction of precipitation along the northern Gulf Coast during Hurricane Sally. Using the VN data within CAPRi to compare satellite-based precipitation estimates versus those obtained from ground-based weather radar networks, allows the spatial structure of rain bands to be accurately depicted, for adequately assessing rain band hazards. Similarly, the vertical development of precipitation features in different regions and seasons can be observed to study convection and the resulting precipitation.