华北平原保定气象站百年降水序列的建立

Construction of the Hundred-Term Precipitation Time Series at Baoding Station in North China Plain

基于多源的降水月值资料,在数据整合和初步质量控制基础上,同时采用标准化序列法和多元线性回归法对华北平原保定气象站1913~2014年月降水量资料进行了插补。通过交叉检验法分析发现,多元线性回归法插补得到的降水量序列效果较好。利用惩罚最大F检验(PMF)对插补后序列的均一性进行了检验,结果表明,通过插补得到的保定站百年降水量月值序列的均一性相对较好,进而建立了保定站百年降水月值序列。同时,通过与北京站和天津站均一化的百年降水量序列的综合对比得出该序列是相对合理的。

Using multi-source of monthly precipitation data on the basis of preliminary integration and quality control, monthly precipitation time series covering 1913-2014 at Baoding station are interpolated by both two approaches of standardized method and multivariate linear regression. The interpolation results analyzed by cross validation are that the method of multivariate linear regression is more right for precipitation data. Then, the better interpolated time series are homogenized by the penalized maximal F test (PMF), and results indicate that those monthly hundred- term precipitation data are relatively continuous. Moreover, we compare the data here with the homogenized monthly precipitation data during 1913-2014 at Beijing and Tianjin stations from some aspects of trend change, showing that the data we constructed in this paper are relatively logical and errorless.