多套格点降水资料在云南及周边地区的对比

利用1979—2006年云南及周边地区148个测站月降水资料 (简称为STN) 与APHRO (日本APHRODITE高分辨率逐日亚洲陆地降水数据集)、GPCC (全球降水气候中心的月降水合成数据)、CRU (英国East Anglia大学提供的月降水要素数据集)、CMAP (雨量资料与卫星估计及NCEP/NCAR再分析降水场合并分析月数据)、GPCP (全球降水气候中心研制的全球陆地雨量计观测分析月数据) 5套格点降水资料，分析了云南及周边地区气候特征。结果表明:5套格点降水资料空间分布与STN基本一致。EOF第1模态空间场分布也表明:这5套格点降水资料与STN空间分布特征较为一致，但5套格点降水资料在滇南、滇西北、滇川黔交界的3个区域的分布与STN有较大不同，各套资料的EOF第1模态时间序列、与STN的相关系数及均方根误差均随时间不同呈较为一致的波动性；在降水空间分布、相关系数及均方根误差3个方面，APHRO适用性最好，GPCC次之，CMAP与GPCP无明显差别，CRU最差，其中APHRO，GPCC在对降水估计偏低，CRU对降水估计总体略高，CMAP略低，GPCP对降水估计则明显偏高。

Datasets and Rain Gauge Precipitation over Yunnan and the Surrounding Areas

The precipitation over Yunnan and the surrounding areas are analyzed from spatial and temporal distributions aspects using several datasets, including data from meteorological stations, APHRO data from Asian Precipitation-Highly Resolved Observational data integrations towards evaluation of water resourced, GPCC data from Global Precipitation Climatology Center, CRU data from Climatic Research Unit, CMAP data from CPC Merged Analysis of Precipitation, and GPCP data from the Global Precipitation Climatology Project. Assessments are carried out to examine the quality of APHRO, GPCC, CRU, CMAP and GPCP precipitation in Yunnan and the surrounding areas from space distribution, inter-annual and monthly variation.Five grid precipitation datasets show similar distribution of precipitation amount to station data, which can reflect the distribution characteristics of spatial distribution of precipitation. There exists the maximum horizontal gradient center in the south of Yunnan, but CRU, CMAP and GPCP cannot represent it. The EOF analysis results of the five datasets show similar spatial distributions of precipitation amount to station data, the first eigenvector space distribution is positive, but in the northwest of Yunnan and the south of Sichuan is negative. The first eigenvector in January is basically positive, but in July, it is negative in the southeast and southwest of Yunnan, the south of Sichuan, and that of other regions is opposite. APHRO and GPCC distributions of positive and negative value are consistent with those of STN, there is a significant difference between the spatial distribution of CRU, CMAP and STN, negative area is not seen in January and July, GPCP is more significant different compared with STN. Correlation coefficients of five precipitation dataset to STN have better consistency, and for most regions, correlation coefficients pass the test of 0.05 level, the correlation coefficient in January is higher than that in July, and the mean square error in July is higher than that in January. APHRO and GPCC underestimate the trend of precipitation, but the weak amplitude of GPCC is less than APHRO, GPCP precipitation estimation is significantly higher, which reaches the highest 18.73% in April, the trend of CRU and CMAP is not very clear.Above all, the application effects of five precipitation datasets in south of Yunnan, northwest of Yunnan, boundary of Yunnan, Guizhou and Sichuan, and boundary of Yunnan, Guizhou and Guangxi are poor, waves of five precipitation datasets in EOF leading time series, correlation coefficients and mean square error is coincident, integral application effect of APHRO is the best, with GPCC, CMAP and GPCP followed, but CRU is the worst in terms of spatial distributions, correlation coefficients and square errors. In terms of the leading modes, the first-three-variance contribution of APHRO is the lowest, then is GPCC, CRU, CMAP, GPCP, the difference in the second mode is not clear, APHRO and GPCC data underestimate, but CRU overestimates the intensity, and GPCP overestimates the trend largely. When the precipitation become larger, the trends is more clear.