A Physics-Informed-Deep-Learning Intensity Prediction Scheme for Tropical Cyclones over the Western North Pacific

Accurate prediction of tropical cyclone (TC) intensity is challenging due to the involved complex physical processes. Here, we introduce a new TC intensity prediction scheme for the western North Pacific based on a time-dependent theory of TC intensification, termed the energetically based dynamical system (EBDS) model, together with the use of the long short-term memory (LSTM) neural network. In the time-dependent theory, TC intensity change is controlled by both the internal dynamics of the TC system and various environmental factors in terms of environmental dynamical efficiency. The LSTM neural network is used to predict the environmental dynamical efficiency in the EBDS model trained using the best-track TC data, the global reanalysis data during 1982–2017, and the analysis data from the Global Forecast System (GFS) of the National Centers for Environmental Prediction during 2019–2021. The transfer learning and ensemble methods are used to train the scheme using the environmental factors predicted by the GFS. The new scheme is evaluated for TC intensity prediction in 2017 using reanalysis data and 2021-2022 using the GFS prediction data. The intensity prediction by the new scheme shows better skill than the official prediction from the China Meteorological Administration and those by other state-of-art statistical and dynamical forecast systems, particularly for long lead-time forecasts.