TCP-Diffusion: A Multi-modal Diffusion Model for Global Tropical Cyclone Precipitation Forecasting with Change Awareness

Deep learning methods have made significant progress in regular rainfall forecasting, yet the more hazardous tropical cyclone (TC) rainfall has not received the same attention. While regular rainfall models can offer valuable insights for designing TC rainfall forecasting models, most existing methods suffer from cumulative errors and lack physical consistency. Additionally, these methods overlook the importance of meteorological factors in TC rainfall and their integration with the numerical weather prediction (NWP) model. To address these issues, we propose Tropical Cyclone Precipitation Diffusion (TCP-Diffusion), a multi-modal model for forecasting of TC precipitation given an existing TC in any location globally. It forecasts rainfall around the TC center for the next 12 hours at 3 hourly resolution based on past rainfall observations and multi-modal environmental variables. Adjacent residual prediction (ARP) changes the training target from the absolute rainfall value to the rainfall trend and gives our model the capability of rainfall change awareness, reducing cumulative errors and ensuring physical consistency. Considering the influence of TC-related meteorological factors and the useful information from NWP model forecasts, we propose a multi-model framework with specialized encoders to extract richer information from environmental variables and results provided by NWP models. The results of extensive experiments show that our method outperforms other DL methods and the NWP method from the European Centre for Medium-Range Weather Forecasts (ECMWF).

Tropical cyclones (TCs), such as typhoons and hurricanes, often bring extreme rainfall that can lead to devastating floods. Predicting this rainfall in advance is crucial for disaster preparedness, but remains a major scientific challenge. Most deep learning models that predict regular rainfall don’t work well in the case of TCs — they often accumulate errors over time and ignore important weather patterns.In this work, we introduce TCP-Diffusion, a machine learning model designed specifically for forecasting rainfall caused by tropical cyclones. Our model looks at past rainfall and a wide range of environmental conditions to predict where and how much it will rain around a cyclone over the next 12 hours. A key innovation is that our model learns to predict how rainfall will change, rather than its exact value — this helps reduce long-term errors and ensures the predictions make physical sense. We also combine our model with outputs from traditional weather forecasting systems to boost its accuracy.Our method outperforms both standard deep learning models and advanced numerical weather prediction systems in forecasting TC rainfall.