热带岛屿多尺度降水系统的积云对流参数化方案模拟性能评估——以海南岛为例

Evaluation of Cumulus Parameterization Schemes in Simulating Multi-Scale Precipitation Systems over a Tropical Island—A Case Study of Hainan Island

摘要: 降水作为地球水循环与能量循环的核心环节，对全球气候演变、水文过程及人类社会活动具有深远影响。热带地区贡献了全球约三分之二的降水量，而海南岛作为典型热带海岛，其独特的下垫面条件和复杂的海陆相互作用导致降水呈现多尺度、多形态特征，对数值模式中积云对流参数化方案的适应性提出了更高挑战。本研究基于WRF模式(V4.2)，选取7种积云对流参数化方案(包括Kain-Fritsch、BMJ、Modified Tiedtke等)，结合ERA5再分析资料、地面观测站数据、CMPAS融合降水产品及GSMaP卫星反演数据，对海南岛2017年雨季13类典型降水案例开展高分辨率(5 km)模拟，并通过相关系数(R)、平均绝对误差(MAE)和Kling-Gupta效率系数(KGE)等多指标评估方案性能。结果表明：(1) Modified Tiedtke(cu6)和New Tiedtke (cu16)方案在热带海岛环境表现最优，尤其是Modified Tiedtke (cu6)在台风活跃期(9~10月)相关系数高达0.913，全案例平均达0.41 ± 0.26；(2) “灰区”分辨率(4~10km)下综合季节性和降水类型差异，提出分时段优化策略：传统的Kain-Fritsch方案(cu1)和改进的Kain-Fritsch方案(cu10)适合模拟降水增长过程，但需要注意其在时间和降水量上的系统性偏差；Modified Tiedtke方案(cu6)和BMJ方案(cu2)适合模拟弱降水过程，并且Modified Tiedtke方案(cu6)在强降水情况下表现也较好；Multi-scale Kain-Fritsch方案(cu11)、KSAS方案(cu14)和新版Tiedtke方案(cu16)则展现出较好的综合性能。(3) Multi-scale Kain-Fritsch方案(cu11)在降水中心位置、降水空间分布和虚假降水控制上表现出优异性能。(4)综合季节性和降水类型差异，提出分时段优化策略：台风季节和季风转换前期选择Modified Tiedtke方案(cu6)或新版Tiedtke方案(cu16)，雨季盛期采用Kain-Fritsch方案(cu1)和Modified Kain-Fritsch方案(cu10)的集合平均。研究结果为热带海岛地区降水模拟的参数化方案优化提供了关键科学依据。

Abstract: Precipitation plays a pivotal role in the global water and energy cycles, significantly impacting climate evolution, hydrological processes, and human activities. The tropics contribute approximately two-thirds of global precipitation, and as a representative tropical island, Hainan Island exhibits distinct precipitation characteristics due to its unique underlying surface conditions and complex land-sea interactions, posing significant challenges to the applicability of cumulus convection parameterization schemes in numerical models. This study employs the Weather Research and Forecasting (WRF) model (V4.2) to simulate 13 typical precipitation cases during the 2017 rainy season at a high resolution (5 km), using seven cumulus parameterization schemes, including Kain-Fritsch, BMJ, Modified Tiedtke, etc. The simulation results are evaluated against ERA5 reanalysis data, ground-based observations, CMPAS merged precipitation products, and GSMaP satellite-derived data using multiple metrics, including correlation coefficient (R), mean absolute error (MAE), and Kling-Gupta efficiency (KGE). The results indicate that (1) the Modified Tiedtke (cu6) and New Tiedtke (cu16) schemes perform best in the tropical island environment, with Modified Tiedtke (cu6) achieving a correlation coefficient of up to 0.913 during the typhoon-active period (September–October) and an overall case-average of 0.41 ± 0.26; (2) considering the “gray-zone” resolution (4~10 km), a time-dependent optimization strategy is proposed: the traditional Kain-Fritsch scheme (cu1) and its modified version (cu10) are suitable for simulating precipitation growth but exhibit systematic biases in timing and magnitude, while the Modified Tiedtke scheme (cu6) and BMJ scheme (cu2) excel in weak precipitation scenarios, with Modified Tiedtke (cu6) also performing well in heavy precipitation events; (3) the Multi-scale Kain-Fritsch scheme (cu11), KSAS scheme (cu14), and New Tiedtke scheme (cu16) demonstrate strong overall performance, with the Multi-scale Kain-Fritsch scheme (cu11) exhibiting superior accuracy in capturing precipitation center locations, spatial distribution, and reducing false precipitation; (4) based on seasonal variations and precipitation type differences, a phased optimization strategy is recommended: the Modified Tiedtke scheme (cu6) or New Tiedtke scheme (cu16) for the typhoon season and early monsoon transition, and an ensemble of the Kain-Fritsch (cu1) and Modified Kain-Fritsch (cu10) schemes for the peak rainy season. This study provides critical scientific insights for optimizing cumulus parameterization schemes in precipitation simulations over tropical island regions.