2026-10-07T05:19:16Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33944188 2026-09-20T10:09:03Z category_25756 item_type_3 month_year_09_2026

Multi-Task Forecasting for Tropical Cyclone Size of Western North Pacific Using Multi-Source Satellite Data with Deep Learning ChangJiang Zhang (25075102) Climate change processes Typhoon forecast <ul><li>The input data were obtained from the ERA5 [26], GridSat-B1 [27], and CMORPH CDR [28] datasets. ERA5 provides specific humidity and the U- and V-components of wind speed. In this study, data at four pressure levels, namely 200 hPa, 500 hPa, 850 hPa, and 1000 hPa, were selected. The GridSat-B1 and CMORPH CDR datasets provide satellite observations in the infrared (IR) and passive microwave (PMW) bands, respectively. To ensure spatial consistency among the multi-source datasets, all the data were resampled to a uniform spatial resolution of 0.07°. Subsequently, the data were centered on the TC center and cropped into images of 201 × 201 pixels.</li><li>The observational data were obtained from the IBTrACS dataset. In this study, the original data were processed and reorganized, and the observations were temporally integrated according to the evolutionary process of TCs. Each sample contains the TC state information at nine consecutive time points (−24 h, −21 h, −18 h, −15 h, −12 h, −9 h, −6 h, −3 h, and 0 h). Specifically, the data include the TC center coordinates (longitude and latitude), quadrant-specific size parameters under different wind-speed thresholds, maximum wind speed, distance from the coastline, minimum sea-level pressure, and TC motion direction and translation speed.</li></ul><p></p> 2026-09-20T10:09:03Z Dataset Dataset 10.6084/m9.figshare.33944188.v1 https://figshare.com/articles/dataset/Multi-Task_Forecasting_for_Tropical_Cyclone_Size_of_Western_North_Pacific_Using_Multi-Source_Satellite_Data_with_Deep_Learning/33944188 CC BY 4.0