CMIP5模式对中国西北干旱区模拟能力评价

Evaluation of CMIP5 modes’s simulation ability in the northwest arid areas of China

气候模式是研究气候系统和气候变化的重要工具,气候模式结果是进行气候预测和气候变化风险评估的重要依据。基于中国西北干旱区78个气象站点1960-2005年的观测数据,对最新公布的CMIP5的39个模式在中国西北干旱区1960-2005年平均气温、降水的模拟能力进行评估。结果表明:多个模式模拟年平均气温与观测值的相关系数达到0.39,夏、秋季节的相关系数好于春、冬季,年平均气温模拟大多偏低2℃以上,其中MIROC4h、CCSM4和CMCC-CM对年平均气温的模拟绝对误差较小。模拟的年、季降水量与观测值的相关系数很差,均不到0.1。年降水量模拟普遍偏高100 mm以上,其中CMCC-CM、CNRM-CM5和MRI-CGCM3对年降水量模拟绝对误差较小。年际变化趋势上,模拟的平均气温升高趋势和降水量增加趋势均比观测趋势要低,模拟的冬季平均气温升高趋势偏低最明显,达-0.21℃/10 a,模拟夏季的降水量增加趋势偏低最明显,相对误差达-99%。CMIP5模式对中国西北干旱区的模拟效果整体上偏差较大,未来无论从物理过程还是模式算法都需要进一步研究和改进。

Climate models are important tools to study the effects of climate system and climate change,climate model results is important basis for weather forecast and climate change risk assessment. The northwest arid area of China is sensitive to climate change,and is one of the regions where atmosphere and land is strongly coupled. However,the current climate models still remain a lot of uncertainties about the region＇s simulation results. Therefore,it is necessary to access simulated results of climate models,which could provide scientific basis for climate modes’improvement and simulation ability. Based on the observed average temperature and precipitation of seventy-eight weather stations from 1960 to 2005 in the study area,the simulation ability of annual average temperature and precipitation from the latest thirty-nine CMIP5 modes were evaluated,adopting the indices such as correlation coefficient,mean absolute error,root-mean-square error and interannual variation. The results indicated that the correlation coefficients between simulated and observed data reached to 0.39,and the correlations in summer and autumn were better than those in spring and winter,the simulated annual average temperatures were 2 ℃ less in the most cases,mean absolute error of annual average temperature simulated by MIROC4h,CCSM4 and CMCC-CM is lower. The correlation coefficients between simulated and observed precipitation on the yearly and seasonal scales were very poor and less than 0.1,and the precipitation was overestimated by 100 mm in the most cases, mean absolute error of annual precipitation simulated by CMCC-CM,CNRM CM5 and MRI-CGCM3 is lower. As for the interannual variation trend,the rising trends of simulated annual average temperature and simulated precipitation were lower than that of observation,the rising trend of simulated average temperature in winter reached to -0.21 ℃/10 a and was the lowest. The rising trend of simulated precipitation in summer was the lowest and the mean error was close to -99%. On the whole,there are the marked differences between simulated values and observed values,especially for simulated precipitation. Simulation ability of climate modes in the northwest arid area of China is in need of further improving,physical mechanism of climate system is worth doing a study in depth,and the modes should be modified reasonably.