降水要素空间插值精度的比较——以东北地区为例  

Comparison of Precision of Spatial Interpolation of Precipitation Factors:A Case Study in Northeastern China

作　　者：蔡福[1] 于慧波[2] 矫玲玲[3] 唐凯[3] 明惠青[4] 刘兵[4] 

机构地区：[1]中国气象局沈阳大气环境研究所,沈阳110016 [2]辽宁省气象台,沈阳110016 [3]黑龙江省气象局,哈尔滨150001 [4]沈阳农业大学,沈阳110161

出　　处：《资源科学》2006年第6期73-79,共7页Resources Science

基　　金：中国气象局沈阳大气环境研究所课题启动项目(编号:2005046)

摘　　要：以1961年～2004年东北3省44年172个观测站的四季降水资料为数据源，采用反距离权重（IDW）（方法1）、趋势面模拟＋残差内插（方法2）、ANUSPLIN软件插值（方法3）、空间化气候值＋年际距平值（方法4）4种空间插值方法，以相同建模站不同检验站、相同检验站不同建模站以及不同检验站不同建模站3种形式分别对多年平均降水，单年丰水年和少水年的冬、夏、年降水数据进行空间插值精度比较，最终得出以下结论：无论是多年还是单年数据，方法3对降水数据空间插值相对误差都是最小的，是一种操作方便，易于批量运算的最优方法。对于每种插值方法，不同时间尺度夏季和年降水空间插值相对误差小于冬季，而夏季和年差异不大，不同季节降水空间插值相对误差丰水年明显小于少水年。在检验站相同情况下，建模站增加并不一定会使降水要素空间插值精度提高；由于空间分布的差异．在建模站相同情况下，检验站数目增加对空间插值的精度影响不大。Based on seasonal precipitation data of 172 weather stations from 1961 to 2004 in Northeastern China, three circumstances including the same quantity of modeling stations but different quantity of testing stations, the same quantity of testing stations but different quantity of modeling stations, and different quantity of both modeling stations and testing stations, Inverse distance weighted（ method Ⅰ）, three dimension-second order trend surface analysis （ method Ⅱ ）, Anusplin software （ method Ⅲ ） and spatial climatic value integrating with multi-annual deviation from normal interpolation methods （metho Ⅳ） are used to interpolate rainy and rainless and annual precipitation data with different temporal scale ranging from multi-year and single-year to compare precision among them. The conclusions are as follows： firstly, the interpolation precision of method m is the best among the other methods for both multi-year and single-year bases. At the same time, method Ⅲ is convenient to operate and able to run with a huge amount of data; secondly, to every interpolation method, spatial interpolation error of summer and annual precipitation is smaller than that of winter at different time scale. It is not evident for difference of interpolation precision between the summer season and the year. For different seasons, the interpolation precision is evidently higher in rainy period than in rainless period; thirdly, under circumstance of the same testing stations, the interpolation precision of precipitation data is not always elevated with increasing of number of modeling stations. Moreover, with increasing number of testing stations, the interpolation precision won＇t be evidently affected on the circumstance with same number of modeling stations.

关 键 词：东北地区 降水 空间插值 精度比较 

分 类 号：S161.6[农业科学—农业气象学]