5种CMIP5模拟降水数据在中国的适用性评估

受气候变化影响,水资源短缺现象严重,自然灾害频发。对未来水资源状况进行有效的预测和评估,可为水资源合理配置工作提供理论及决策参考。全球气候模式是大尺度模拟预测未来降水的主要手段,但模式数据的模拟质量及其在不同研究区适用性需要进行系统评估。研究以1971—2000年2 474个气象站点监测降水数据为基准,对5种CMIP5模式模拟降水数据(0.5°×0.5°分辨率)进行系统性评估,分析CMIP5模式模拟降水数据在中国区的适用性。对比分析了1971—2000年间5种CMIP5模式模拟降水数据与中国地面监测降水数据所反映的中国降水趋势变化的时间和空间特征,对CMIP5模式模拟月降水数据进行经验校正,在数据相关性、误差和趋势一致性3个角度,采用相关系数(r),偏差均值(DAVE),相对绝对误差(ERA)和均方根误差(ERMS)4个指标和Seasonal Kendall(SK)趋势检验方法,从时间变化和空间分布特征两个角度,对校正后的CMIP5模式模拟降水数据在中国区域的数据质量进行系统评估。5种CMIP5模式模拟中国区年降水高于监测数据均超过20%,通过经验校正,在保证年降水不发生变化的情况下,使模拟降水数据与监测数据在逐月的相关性大大提高,误差大大降低。对5种模式模拟数据的综合评价显示,最适合模拟中国区降水的模式是HadGEM2-ES模式。对模式数据的系统性评估,为研究合理选择、校正和使用CMIP5数据集提供理论和技术参考。

Evaluation on the Applicability of 5 Kinds of CMIP5 Simulated Precipitation Data in China

Affected by climate change,water shortage is serious and natural disasters occurring frequently.Effective forecasting and assessment on the future state of water resources will provide theoretical and decision-making reference for the rational allocation of water resources.Global climate models are the main means of large-scale simulation and prediction for future precipitation.However,the quality of model data and the suitability of different research area need a systematic assessment.Based on the monitoring precipitation data from 2474 meteorological stations during the period from 1971 to 2000 in China,we performed the systematic assessment on the simulated precipitation data from five CMIP5 models (0.5°×＜ 0.5° resolution),and analyzed applicability of the simulated precipitation data of CMIP5 model in China.Comparative analysis was conducted to detect the characteristics of temporal and spatial variation trend of precipitation data of five CMIP5 model and meteorological stations in China from 1971 2000.Empirical correction on monthly precipitation data from CMIP5 models was conducted and systematic assessment of data quality of the corrected data was also performed using the correlation coefficient (r),the average deviation (DavE),relative absolute Error (ERA) and the root mean square error (ERMS) indicators in aspects of data consistency and the ability to describe the trend for China region.Annual precipitation simulated from five CMIP5 models in China was higher than the monitoring data by more than 20％,however,by empirical correction,the error of monthly precipitation data with monitoring data was greatly reduced.Results of evaluation for five CMIP5 models show that the most suitable model for China is HadGEM2-ES model.Systematic assessment for model data can provide theoretical and technical reference for the rational choice,correction and application of CMIP5 data set.