Key points are not available for this paper at this time.
Abstract Data‐driven artificial intelligence weather prediction (AIWP) models show great potential in weather forecasts, facilitating paradigm shift of prediction from a deductive to an inductive inference. However, this shift raises concerns regarding the performance of the AIWP models in severe weather forecasting. Tropical cyclones (TCs) are one of the most typical cases of severe weather prediction. In this study, we compare Western Pacific TCs in 2023 produced by the AIWP model, Pangu‐Weather, with those generated by numerical weather prediction (NWP) models, specifically the European Center for Medium‐Range Weather Forecasts (ECMWF) and the National Centers for Environmental Prediction (NCEP), in the operational context. We analyze the impact of different initial conditions (ICs) on AIWP models, representative by Pangu‐Weather, in TC forecasting. Our analysis includes statistical evaluation of forecast skill related to TC activity, track, intensity, and a case study on the physical structure of a TC. The Pangu‐Weather model exhibits superior forecast skills compared to the NWP model regarding TC tracks and environmental variables within TC activity domains, particularly at longer forecast lead times. However, the overly smooth forecasts of Pangu‐Weather and the coarse‐resolution ICs with reduced information of TCs potentially lead to the underestimation of intensity and a weakened dynamic‐thermodynamic structure of TCs. Also, Pangu‐Weather shows low sensitivity to ICs concerning TC structure and intensity. Hybrid models combining physical processes with data‐driven approaches may enhance AIWP performance for severe weather forecasting.
Shi et al. (Fri,) studied this question.