基于GIS空间插值的降水分布模拟方法比较研究

为了获取区域最优插值方案,模拟降水空间分布信息,基于1960-2015年湖南省87个气象站点资料,利用ArcGIS 10.2软件平台,采用克里金法、反距离权重法、样条函数法和趋势面法对研究区近56 a年均降水量进行空间插值,通过比较各插值方法的插值精度获得适合区域的插值方法,并分析了研究区降水空间分布特征。结果表明:湖南省降水空间插值受多因素影响,具有不确定性。交叉验证显示幂值为4的反距离权重法插值精度较高,基于TENSION模型的样条函数法次之,基于三角函数模型的克里金法精度略差,阶为2的趋势面法插值精度最差,然而4种插值方法一定程度上均能反映湖南省年降水量的空间分布特征。此外,不同插值方法插值结果存在差异性,其中,普通克里金法插值面较为粗糙,反距离权重法和样条函数法插值面较为光滑,趋势面法插值结果较为单调。

Comparison of GIS-based Spatial Interpolation on the Distribution of Precipitation

In order to obtain the optimal interpolation methods and simulate the spatial distribution of annual average precipitation in Hunan Province,this paper selects ordinary Kriging(OK),inverse distance weighting(IDW),Trend surface analysis(Trend),Spline function method(Spline)to interpolate based on data from 87 meteorological stations in Hunan Province from 1960 to 2015 under the support of ArcGIS 10.2 platform,and compares the interpolation precision of each interpolation method using cross validation method.Results indicate that:①the spatial interpolation of precipitation in Hunan Province has uncertainties affected by many factors.②IDW4 has the lowest interpolation error and higher accuracy,followed by Spline based on TENSION model,OK based on the trigonometric function model has slightly lower accuracy,and Trend of Order 2 has the worst interpolation accuracy.Those four methods can reflect the characteristics of the spatial distribution of annual average rainfall in Hunan Province to some extent.③There are certain differences in local areas among different methods,for example,OK's interpolation surface is rough,IDW and Spline have relatively smooth surfaces,and Trend's interpolation results are monotonous.