Physics-Informed Deep-Learning Prediction of Tropical Cyclone Wind Size in the Western North Pacific

Accurate forecasting of tropical cyclone (TC) wind size remains a significant challenge in operational meteorology. This study develops a physics-constrained long short-term memory (LSTM) model to improve TC wind-field size prediction in the western North Pacific (WNP), leveraging multi-source data, including best-track records, global reanalysis datasets, and outputs from an operational regional TC numerical model, the China Meteorological Administration Typhoon Model (CMA-TYM). Key predictors such as TC internal characteristics (maximum wind speed and initial vortex size) and environmental variables (ocean heat content, relative humidity, and vertical wind shear) are diagnostically selected to train a bidirectional LSTM with an attention mechanism for forecasting the radius of maximum wind (R_max) and wind profile parameters. The model achieves a 24-h mean absolute error (MAE) of 8.3 km for R_max (91% of samples with error < 20 km) and 34.8 km for R₁₇, defined as the gale-force wind radius at 17 m s⁻¹(78% of samples with error < 50 km). For 120-h forecasts, the R₁₇ relative error stabilizes at 0.1–0.2, exhibiting robustness to data scarcity and slower error growth than conventional methods. Compared with CMA-TYM, the LSTM model reduces 6–48-h forecast errors on average by 25.8% for R₂₆ (damaging-force radius, 26 m s⁻¹) and 38.5% for R₃₃ (hurricane-force radius, 33 m s⁻¹), while also capturing asymmetric wind-field structures and improving the accuracy of long-axis wind distribution during TC intensification. Sensitivity analysis identifies initial vortex size, ocean thermal conditions, and environmental humidity as the dominant drivers governing TC size variability. The proposed framework demonstrates superior stability and resolution in TC size forecasting, providing a valuable tool for disaster risk assessment and early warning systems.