气候资料缺测插补方法的对比研究

采用均生函数正交筛选（ＭＧＦ）和一维车贝雪夫多项式展开（ＣＰ）进行了年降水量各种缺测情况下资料的插补试验，并计算各种统计量，结果表明：无论何种缺测ＭＧＦ法拟合或插补精度一般高于ＣＰ法，尤其对连续多年缺测和序列一开始便连续缺测更为明显；同一种方法１ａ缺测拟会精度高于多年缺测，连续４—５ａ缺测但在序列中处于不同的位置拟合结果差别不大，但多段同时缺测拟合精度低于一段缺测.

STUDY OF INTERPOLATION METHODS OF MISSING ANNUAL PRECIPITATION DATA FOR LONG-TIME SERIES

The interpolation experiments and statistics calculations of missing annual precipitation have been done using mean Generating Function (MGF )and one-dimensional Chebyshev polynomials(CP).The results have shown that MGF for fitting and interpolationaccuracy of missing annual precipitation are usually higher than CP,especially missing data with continuous many years and at beginning of series. In the case of using same method,accuracy of fitting for onlyone year missing is higher than that of many years. But there are not big difference between fitting results while 4-5a missing data in different location of data series.However accuracy of fitting forseveral parts missing at the same time is lower than that of only one part missing.