【摘要】 以国家气象局1971年~2000年30年整编资料中的东北、华中地区1978年、1984年、1990年、1996年4年的1月份、7月份及年平均气温数据为数据源,采用直接插值法(反距离权重法和普通克里格法)、趋势面模拟+残差内插法、空间化气候值+年际距平空间插值方法、空间化气候值+年际距平趋势面模拟+残差内插等4种方法,进行了空间化精度的比较研究。通过平均绝对误差(MAE)、平均相对误差(MRE)以及交叉验证等几种评估标准的比较,认定在具有30年月平均气温栅格数据库作为背景的前提下,采用空间化气候值+年际距平值IDW内插的方法在东北、华中两个地区空间化的误差相对较小,并且其操作方便,是一种对平均气温这一要素的短时间序列空间化而言既方便,插值效果又相对较好的空间化方法。更多还原

【Abstract】 Based on the data of mean air temperature in January, July and whole year of 1978,1984,1990,1996 in Northeastern and Central China, the comparison of precisions of spatial methods were conducted using direct interpolation methods including Inverse distance weighted and Ordinary Kriging, three dimension-second order trend surface analysis and spatial interpolation method, spatial climatic value integrating with the multi-annual deviation from normal interpolation methods and spatial climatic value integrating with trend simulating to the multi-annual deviation from normal combining with residual interpolation methods. Taking mean absolute error, mean relative error and crossing validation as evaluation criterion, it is concluded that as far as mean air temperature is concerned, the method of spatial climatic value integrating with the multi-annual deviation from normal interpolation by IDW is not only a convenient but also relatively precise spatial method with smaller error in Northeastern and Central China based on multi-annual mean air temperature raster database. Furthermore, three dimension-second order trend surface analysis and spatial interpolation method, which is suitable for interpolate to multi-annual mean climatic data, is unsuitable for the interpolation of short time serial climatic data. It is worthily noticed that above methods cannot play a good role to all of climatic factors because the diversity exists between different climatic factors for the difference of spatial-temporal distribution, continuity and local natural conditions.更多还原