Probabilistic Precipitation Estimation with the GOES16 Advanced Baseline Imager: A Machine Learning Approach

This study introduces a new machine learning-based probabilistic quantitative precipitation estimation (PQPE) algorithm that uses observations from the GOES-16 Advanced Baseline Imager (ABI) across the CONUS. It is developed and evaluated using the Ground Validation Multi-Radar/Multi-Sensor (GV-MRMS) system as a benchmark, and features Convolutional Neural Network (CNN) machine learning. The key advances of the algorithm include (1) the design of the CNN model to retrieve the distribution of precipitation instead of a single deterministic value; (2) incorporating a comprehensive set of predictors derived from infrared ABI channels complemented by environmental conditions from Numerical Weather Prediction (NWP) models, and (3) introducing probabilities of occurrence of precipitation types (e.g., convective and stratiform) retrieved from a separate machine learning model as predictors in the CNN model. Precipitation type predictors allow the same model to be used seamlessly across precipitation types. The analysis reveals that combining satellite and NWP categories of predictors leads to improved performance, with the greatest improvement for the stratiform precipitation type. The probability of precipitation type predictors also contribute significantly to the skill of the retrievals. Furthermore, improvements in conditional biases are demonstrated for all precipitation rates when compared to a deterministic CNN model.