中国区域平均降水量序列构建方法比较研究

由于地面观测台站空间分布不均匀，运用不同区域平均技术方法研究中国降水特征和变化规律的结果存在显著差异，是区域降水变化研究不确定性的重要来源之一。本文以“中国地面与CMORPH（CPC Morphing Technique）融合逐日降水产品”作为参照值，基于中国地面2425 站观测资料，采用5种网格尺寸的经纬度网格面积加权平均方法、省面积加权平均方法、直接平均方法等计算中国区域平均降水量时间序列，比较所得序列统计属性与参照值的偏差，判别方法的优劣。分析表明，1998～2012年，2.5°网格和5.0°网格区域平均方案所得年降水量序列的变化速率和离散程度同参照值最为接近，分别是最优和次优的区域平均方案。省面积加权平均方法对多年平均年降水量的计算准确，但对年降水量变化趋势的估计效果稍差，距平百分率序列很不准确。网格过疏或过密都会使区域平均结果出现较大误差，直接平均方法的误差亦偏大，可靠性较低。

Construction Method for Regionally Average Precipitation Time Series in China

In China, using different technical methods for studying the regional average precipitation characteristics based on station data could yield significantly different results. In this paper, with "China ground and CMORPH (CPC Morphing Technique) fusion daily precipitation product (V1.0)" as the reference value data, different methods for calculating the regional average precipitation time series in China were compared using the precipitation data of 2425 stations. These methods included the latitude and longitude grid area weighted average method of five grid sizes, the provincial area weighted average method, and the direct average method. The results show that the values of linear trend and the standard deviation of annual precipitation obtained using the 2.5° grid and 5.0° grid area average methods are the closest to those of the reference value. Therefore, 2.5° grid and 5.0° grid area average methods are optimal and suboptimal methods, respectively. The average annual precipitation for 15 years obtained using the provincial area weighted average method, which is commonly used in China’s climate monitoring business, is closest to the reference value. However, its estimation for linear trend and standard deviation is lower than the vale obtained from the 2.5° and 5.0° grid area weighted average methods. The sequence of percentage anomalies is inaccurate. If the grid is too sparse or too dense, the regional average result is far from the reference value. The error in the direct average method is also large, with low reliability.