Research progress and prospects of tropical cyclone precipitation modelling

Abstract

[Objective] The modelling and assessment of tropical cyclone precipitation serves as the foundation for tropical cyclone warning and risk evaluation. This paper aims to address issues such as unclear differences in tropical cyclone precipitation modelling method. [Methods] Through literature review, this paper provides a systematic review of the characteristics, progress, applicability, and representative models of four types of models, including numerical weather prediction models, statistical models, physical models, and machine learning models. A comparative analysis is conducted, followed by suggestions and prospects for the development of these four types of models. [Results] The results show that numerical weather prediction models have high reliability and are suitable for forecasting tropical cyclone precipitation. Statistical models can generate numerous simulated tropical cyclones, making them suitable for estimating precipitation return periods. Physical models, based on simplified calculations, provide a good explanation for the mechanisms of tropical cyclone precipitation. Machine learning models exhibit strong flexibility and can be integrated with other models, showing significant potential for future development. [Conclusion] In the future, in addition to further improving relevant models, it is essential to strengthen the synergy between precipitation and secondary disasters, as well as the application of new technologies in precipitation modelling. This can enable rapid and accurate estimation of tropical cyclone precipitation, providing better support for regional tropical cyclone warning and risk prevention.

Keywords

tropical cyclone / precipitation model / numerical weather prediction model / statistical model / physical model / machine learning / risk assessment