High-resolution typhoon precipitation integrations using satellite infrared observations and multisource data

Typhoon-related precipitation over land can result in severe disasters such as floods and landslides, and satellites are a valuable tool for estimating surface precipitation with high spatial-temporal resolutions. Accordingly, this study develops a surface precipitation integration framework to combine high-resolution observations from the radiometers of two geostationary satellites, Fengyun-4A (F4) and Himawari-8 (H8), with high-density rain-gauge observations or IMERG data and atmospheric reanalysis data based on a random forest (RF) algorithm. The RF algorithm integrates cloud and atmospheric features from radiometric observations and reanalysis information, and the intensity and spatial distribution of precipitation can be revealed by high-density rain-gauge or IMERG data. We take three typhoons that made landfall in South China during 2018 as examples. The F4-based and H8-based results using rain-gauge data as the predictand both show excellent results, yielding correlation coefficients (R) of similar to 0.75 and probabilities of detection (POD) of similar to 0.95. In contrast, when IMERG data are used as the predictand, the corresponding R and POD drop to similar to 0.5 and 0.93, respectively, due to the uncertainties related to IMERG retrievals. By carefully choosing the predictor, our RF algorithm successfully integrates the information from satellite observations, surface measurements and atmospheric reanalyses, resulting in precipitation estimates that are highly consistent with actual ground observations. Consequently, our proposed integration framework can recon- struct hourly surface precipitation estimates at high spatial-temporal resolutions for historical typhoon studies.