区域气候模式不同积云对流参数化方案对新疆气候模拟的影响研究

使用NCEP-FNL全球分析资料作为WRF模式的初始场和边界场，利用该模式中7种积云对流参数化方案对新疆地区进行2006年10月1日至2008年3月1日的模拟积分试验，重点考察模式在水平分辨率为10 km下不同积云对流参数化方案对新疆地区气象要素模拟的敏感性。结果表明:1）采用7种积云对流参数化方案的模式都能较好地模拟出年、雨季总降水量、平均温度的空间分布及大气的垂直结构。2）对于不同区域来说，采用各种积云对流参数化方案的模式都能模拟出候降水及候平均温度随时间演变，模式候降水与观测的相关系数在0.20~0.85之间，而候平均温度与观测的相关系数在0.98以上。对于整个新疆地区来说，采用各方案模式模拟的低层偏干偏冷，大气层结较稳定导致降水较观测偏少，而其中天山地区模式模拟的低层较观测偏湿偏暖，大气层结偏向不稳定导致降水偏多。3）采用新的Grell和Kain-Fritsch（new Eta）方案模式模拟的效果综合来看较好。因此利用WRF模式开展新疆地区数值模拟研究时应该考虑不同积云对流参数化方案适用范围。

Impacts of Different Cumulus Convective Parameterization Schemes on Regional Climate Model Simulation over Xinjiang

The NCEP-FNL dataset is used to provide initial and boundary fields for the WRF model. Seven cumulus convective parameterization schemes are applied in this model to simulate climate over Xinjiang region with 10-km horizontal resolution. The study is focused on spatial distributions and temporal evolutions of meteorological elements. The analysis results are as follows:1) The model with the seven Cumulus Convection Parameterization Schemes (CCPS) can well simulate the vertical structure of the atmosphere and the spatial pattern of annual and rainy seasonal precipitation and temperature. 2) For different regions, the correlation coefficients between simulated and observed precipitation are within the range of 0.20 to 0.85 and the correlation coefficients between simulated and observed temperature are above 0.98. In Xinjiang region, the humidity in the lower layer is drier in the simulations with various schemes than in the observations, and the simulated temperature is cooler than the observations. As a result, the simulated atmospheric stratification tends to be more stable than observations, which leads to precipitation underestimation in the simulations. The opposite is true for the Tianshan area in Xinjiang. 3) Precipitation and temperature simulated by WRF with new Grell and Kain-Fritsch (new Eta) schemes are better that with other CCPS. Therefore, it is necessary to carefully select appropriate cumulus convective parameterization schemes when using WRF model to simulate climate in Xinjiang.