Application of network-based scale-adaptive cloud fraction scheme in kilometer-scale nocturnal precipitation simulation over the southeast Tibetan Plateau
摘要
Kilometer-scale modeling has identified a significant underestimation of nocturnal precipitation during summer in the southeast Tibetan Plateau (SETP), remaining a challenge that needs to be solved urgently. This study investigates the performance of a neural Network-based Scale-Adaptive (NSA) cloud fraction scheme coupled with the Weather Research and Forecasting (WRF) model in improving this simulation performance. The results demonstrate that the NSA experiment exhibits the most significant enhancement during June and August within the Yangtze basin (within the SETP), while a pronounced underestimation appears in July. This discrepancy can be attributed to the impact of the NSA scheme on the frequency of moderate rainfall in different summer months. Regarding June and August, the improvement is mainly related to the atmospheric humidification resulting from enhanced water vapor transport (WVT) following the implementation of the NSA scheme. The weakened WVT in July is further diminished in the NSA experiment, contrary to expectations. Under drier atmospheric conditions, cloud cover exhibits a relative deficit—most pronounced for high-altitude clouds compared to other summer months—which significantly affects radiation and dynamic processes. Consequently, land-air interaction becomes enhanced and exhibits complex interactions with background fields, resulting in a more severe underestimation.