基于GIS海南岛气象要素空间插值法比较分析

摘要

应用普通克里格插值法（0K）、反距离权重插值法（IDW）、样条插值法（Spline）以及混合插值法（MLR）对海南岛18个长期气象观测站1979~2008年年均气温、年降水量、年日照时数以及年积温进行空间栅格化，并进行结果对比分析。结果表明：通过交叉检验的MAE和RMSIE的误差分析表明，MLR插值精度最高，样条插值法插值精度最低，普通克里格与反距离权重插值法插值精度接近。总体来看，混合插值法（MLR）是海南岛气象要素最优空间插值法。%In this paper, based on regression analysis combined with spatial grid interpolation, the annual air temperature, annual precipitation, annual sunshine hours and annual accumulative temperature from 18 meteorological stations from 1979 to 2008 in Hainan Island were interpolated by using the spatial interpolation methods of Inverse Distance Weight （IDW）, Ordinary Kriging （OK）, Spline and combined method （MLR）, and the result are compared and analyzed. The result of the cross validation tests of MAE and RMSIE shows that the MLR gives lowest errors and Spline gives highest errors. On the whole, MLR is the best spatial interpolation for climate factor in Hainan Island.