热带岛屿地区不同云微物理方案对强降水模拟性能研究——以海南岛为例
Study on the Performance of Different Cloud Microphysics Schemes in Simulating Heavy Precipitation over Tropical Islands—A Case Study of Hainan Island

在全球气候变暖加剧极端降水的背景下，热带岛屿地区的精准降水预报对防灾减灾和可持续发展具有关键意义。海南岛作为典型热带季风区，其强降水过程受多重天气系统影响，但现有研究多聚焦于中高纬度地区，针对热带岛屿云微物理机制的认知仍存在显著缺口。本研究基于WRF V4.2模式，选取12类典型强降水事件(含台风、季风和对流系统)，系统评估Kessler、Lin、WSM3/5/6、Ferrier和Thompson共7种云微物理方案在海南岛的适用性，结合地面观测、CMPAS融合降水产品和GSMaP遥感数据，通过相关系数(R)、均方根误差(RMSE)和Kling-Gupta效率系数(KGE)等多维度指标验证模拟性能。研究发现：(1) 方案表现呈现显著时空分异性，Ferrier方案在秋季降水模拟中相关系数最高(R = 0.77)，而Thompson方案在误差控制(RMSE = 1.67 mm/h)和台风降水峰值捕捉(20 mm/h)方面最优；(2) Thompson (MP8)和WSM6方案(MP6)在降水过程模拟中展现出较高的综合可靠性；(3) 简单暖云方案Kessler在3月季风转换前期(R = 0.53)和9月残余台风降水(R = 0.63)中表现突出，揭示了热带降水暖云主导机制与复杂冰相参数化的适应性矛盾；(4) 提出季节–天气型动态方案配置策略：季风转换前期(3~4月)采用Kessler方案(MP1)，主汛期(5~9月)优选Thompson方案(MP8)，台风中后期(10~11月)切换至WSM3方案(MP3)。
In the context of global warming exacerbating extreme precipitation, accurate precipitation forecasting in tropical island regions is crucial for disaster prevention, mitigation, and sustainable development. As a representative tropical monsoon region, Hainan Island experiences heavy precipitation influenced by multiple weather systems. However, existing studies predominantly focus on mid- and high-latitude regions, leaving significant gaps in the understanding of cloud microphysical mechanisms over tropical islands. This study employs the WRF V4.2 model to evaluate the applicability of seven cloud microphysics schemes—Kessler, Lin, WSM3/5/6, Ferrier, and Thompson—by selecting 12 typical heavy precipitation events, including typhoons, monsoons, and convective systems. Using ground-based observations, the CMPAS merged precipitation product, and GSMaP satellite data, the simulation performance is validated through multiple metrics, including correlation coefficient (R), root mean square error (RMSE), and Kling-Gupta efficiency (KGE). The key findings are as follows: (1) The performance of the schemes exhibits significant spatiotemporal variations. The Ferrier scheme achieves the highest correlation coefficient in autumn precipitation simulations (R = 0.77), while the Thompson scheme excels in error control (RMSE = 1.67 mm/h) and capturing typhoon precipitation peaks (20 mm/h); (2) The Thompson (MP8) and WSM6 (MP6) schemes demonstrate strong overall reliability in simulating rainfall processes; (3) The simple warm-rain Kessler scheme performs particularly well during the pre-monsoon transition in March (R = 0.53) and residual typhoon rainfall in September (R = 0.63), revealing a mismatch between warm-cloud-dominated tropical rainfall and overly complex ice-phase