青藏高原被称为“亚洲水塔”， 其水资源的变化对下游的天气气候有重要的影响。降水是水循环的关键环节， 因此， 准确模拟青藏高原降水对我国水资源安全有重大意义。近年来， 一些研究发现对流解析模拟（即当网格尺度小于4 km时关闭对流参数化方案的模拟）能够提升青藏高原降水的模拟效果， 然而， 这些研究仅仅选取了1~3种对流参数化方案来进行对比研究， 对流解析模拟是否优于任意对流参数化方案仍然未知。本文评估了WRF模式中9种积云对流参数化方案与不使用对流参数化方案的对流解析模拟（Convection-Permitting Modeling， CPM）对2009年夏季青藏高原地区降水的模拟能力。结果表明： 模拟总体高估了青藏高原2009年夏季降水， 存在0.4~2.0 mm·d^-1的误差， 对青藏高原CAPE值和潜热通量的模拟过大可能是造成青藏高原降水模拟偏大的原因之一。在所有模拟中， G3积云对流参数化方案对平均降水和日变化的模拟效果最好， 能更好地模拟出平均降水的降水强度、 空间分布和降水落区以及降水日变化。CPM对降水整体的模拟效果次于G3积云对流参数化方案， 不能有效地改善对降水日变化的模拟， 但是可以改进对降水频率的模拟。在不同高原生态区内， 所有模拟都不能合理地模拟出荒漠区和喜马拉雅南麓的降水， 但相较于参数化方案， CPM可以大大地降低荒漠区的误差。在其他区域内， CPM和Tiedtke积云对流参数化方案的表现都较好。综合平均降水和降水频率， CPM、 Tiedtke和G3积云对流参数化方案对不同区域、 不同强度的降水模拟误差最小。因此我们建议： 模拟青藏高原夏季降水时可优先考虑G3和Tiedtke积云对流参数化方案， 在计算资源充足时， 可以考虑采用高分辨率的对流解析来提高青藏高原降水频率的模拟。

The Qinghai-Xizang （Tibetan） Plateau is known as the Asian water tower.The change of its water resources has an important impact on the weather and climate in the lower reaches.Precipitation is a key role in the water cycle.Therefore， it is of great significance to accurately simulate plateau precipitation for water resources security in China.In recent years， some studies have found that the convection-permitting model （the cumulus parameterization scheme could be turned off when grid scale is less than 4 km） could improve the precipitation simulation over the Qinghai-Xizang （Tibetan） Plateau.However， the previous studies only selected 1~3 cumulus parameterization schemes for comparison.It is still unknown whether convection-permitting model is superior to any cumulus parameterization scheme.In this paper， the ability of nine cumulus parameterization schemes in WRF and convection-permitting model （CPM） in simulating the precipitation over the Qinghai-Xizang （Tibetan） Plateau in the summer of 2009 was evaluated.The results showed that the simulations overestimate the summer precipitation over Qinghai-Xizang （Tibetan） Plateau in 2009， and the error was about 0.4~2 mm·d^-1.The over simulation of CAPE and latent heat flux in Qinghai-Xizang （Tibetan） Plateau may be one of the reasons for the overestimations.Among all simulations， the G3 cumulus parameterization scheme has the best simulation of the mean precipitation the mean precipitation and the dinural cycles precipitation and it can can better capture the precipitation intensity， spatial distribution， precipitation area and diurnal cycles of the mean precipitation.CPM showed overall the second best precipitation simulation following G3 cumulus parameterization scheme， which can not effectively improve the simulation of diurnal cycles of precipitation， but can improve the simulation of precipitation frequency.In different plateau ecological regions， all simulations cannot reasonably simulate the precipitation in the desert area and the southern foot of the Himalaya， but compared with the cumulus parameterization schemes， CPM can greatly reduce the error in the desert area.In other regions， CPM and Tiedtke cumulus parameterization scheme performed well.Considering the mean precipitation and precipitation frequency， the CPM， Tiedtke and G3 cumulus parameterization scheme have the minimum simulation error for precipitation in different regions and intensity.Therefore， we recommend to adopt G3 or Tiedtke cumulus parameterization schemes in simulating summer precipitation over the Qinghai-Xizang （Tibetan） Plateau， and when the computational resources are allowed， the high-resolution CPM can be considered to improve the precipitation frequency simulations.