Aiding Tropical Cyclone Forecasting by Creating Synthethic 89- and 37- GHz Imagery from Operational Geostationary Satellites

Passive microwave imagery from low-earth orbit provides tropical cyclone (TC) structure information, making it a key tool for tropical cyclone operations; however, this imagery is available only from low Earth orbit satellites, resulting in limited revisit times, frequent tropical coverage gaps, and increased observation latency. To overcome these limitations and supply continuous, high spatial resolution data, we have developed synthetic microwave imagery at 37 and 89-GHz from the latest generation of geostationary satellites using machine learning (ML). TC-following 89-GHz imagery from fully-connected neural networks (ANN) and convolutional neural networks (CNN), as well as a consensus product (CONSEN), was run in real-time for 2023 for the Atlantic, Eastern Pacific, and Western Pacific basins using GOES-16, GOES-18, and Himawari-9, respectively. Here we will provide a brief overview on its performance, investigating the strengths and weaknesses of the different models and focusing on the benefits and disadvantages to the synthetic imagery as a tool to improve TC monitoring and prediction. Next, we will discuss the creation of a new 37-GHz product, including an overview of the ML approaches being applied and an evaluation of sample 37-GHz synthetic imagery. Evaluation of both 89 and 37 GHz imagery concentrates on analyzing results for emissions versus scatter-based regimes, ocean versus land, day versus night, distance from storm center, and by intensity stratification. This work demonstrates the utility of using geostationary satellite imagery, available world-wide in real-time, to provide timely synthetic microwave imagery for operational tropical cyclone products.

Disclaimer: The scientific results and conclusions, as well as any views or opinions expressed herein, are those of the author(s) and do not necessarily reflect those of NOAA or the Department of Commerce.