Machine Learning-Reconstruction Tropical Cyclone Precipitation Dataset (MLRTCP)

1. Dataset Overview Machine Learning-Reconstruction Tropical Cyclone Precipitation Dataset (MLRTCP) is a machine learning (ML)–based model that reconstructs TC-induced precipitation by integrating key TC characteristics (e.g. wind speed, track trajectory) and environmental factors (e.g. sea surface temperature, topography). This dataset was developed to investigate the spatial and temporal variability of tropical cyclone precipitation over coastal China from 1960-2020. It can be used for hydrological modeling, risk assessment, and climate studies. 2. Data Format and Structure This dataset provides gridded precipitation fields within the influence range of tropical cyclones, with a temporal resolution of 3 hours. The precipitation fields were generated based on the tropical cyclone tracks from the CMA Best Track Dataset. Information of these TC track are listed in info_xxxx files in infos folder. All data are stored in GeoTIFF (.tif) format. The filenames follow the convention CycloneID_YYYYMMDDHH. For example: 1_2000050603.tif represents the precipitation associated with Typhoon No. 1 in 2000, at 06 May 2000, 03:00 UTC. 101_2001051415.tif represents the precipitation associated with Typhoon No. 1 in 2001, at 14 May 2001, 15:00 UTC.