地理研究 ›› 2001, Vol. 20 ›› Issue (4): 446-452.doi: 10.11821/yj2001040008

• 论文 • 上一篇    下一篇

Kriging方法在区域土壤水分估值中的应用

李海滨, 林忠辉, 刘苏峡   

  1. 中国科学院地理科学与资源研究所, 北京 100101
  • 收稿日期:2001-01-08 修回日期:2001-07-20 出版日期:2001-08-15 发布日期:2001-08-15
  • 作者简介:李海滨(1977-),男,四川南充人,硕士研究生,从事水文水资源研究工作。
  • 基金资助:

    中国科学院创新工程重点项目(KZCX2-310-03-02);中国科学院地理科学与资源研究所知识创新项目(CX10G-C00-05-01)资助

Application of Kriging Technique in estimating soil moisture in China

LI Hai bin, LIN Zhong hui, LIU Su xia   

  1. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
  • Received:2001-01-08 Revised:2001-07-20 Online:2001-08-15 Published:2001-08-15

摘要:

土壤水分的观测对于地表参数化的发展以及气候变化的研究有着重要的作用。本文对大尺度区域土壤水分的估值进行了尝试:采用1987年中国102个气象站点1米土层四个季节的土壤水分值作为样本,运用KRIGING方法,通过对半变异函数的计算和分析,得出了所研究7个采集日的拟合函数,发现均符合球状模型,对模型有关的参数进行了拟合。并将插值结果与距离反比法进行了对比性检验,同时给出了KRIGING方法的估值精度。检验结果表明平均相对误差和标准偏差均以距离反比法较小,以样本量较大的f时段为例对检验结果进行了深入分析。由此得出了KRIGING方法内插估值的优势和不足,简要给出了提高估值精度的可能方案。最后对中国东半部f时段的土壤水分值进行内插成图。

关键词: 克立格法, 半变异函数, 土壤水分, 距离反比法

Abstract:

In situ observations of soil moisture are invaluable means in developing land surface parameterization and studying pattern of climate change. However, the existing observations have been done only at point scale. Hence as to how to get regional soil moisture is of especially important. In this paper, geostatistical method (Kriging) is used to estimate soil moisture unknown at site A based on known soil moisture data around A. The data set of soil humidity in the top 1 m of 102 agrometeorological stations over China in 1987 is used for the estimation. In order to test how well the method works, we estimate one station's soil moisture as unknown by using other station's data. The observational data from that station is then taken as the true value. We gave the RMSE of Kriging interpolation method. and compared it with inverse distance square method The accuracy of the estimation is not high in terms of average relative error and standard deviation index For further analysis ,we took fperiod which had the maximal samples as an example.The average relative error of both methods was 0.26, the standard deviation of Kriging was 8.77 , the standard deviation of 5.17. The better results of the latter method was maybe due to itshomogenization over all the data with difference between the maximum and the minimum observed soil moisture being 42.85 that by geostatistical method being 31.25 and that by inverse interpolation method being 24.09. It is suggested that the combination of the two estimation methodsmay give better results. The inverse distance interpolation method is suitable for data with general variation characteristicswhile the geostatistical method is good for regional variable tendency. The average range calculated in this paper, around 500 km, is in agreement with the result of Entin et al. (2000) from 49 stations and 11 year records and Liu et al. (2001) with 99 stations with 3 year records.

Key words: Kriging, semivariogram, soil humidity, inverse distance square

  • S152.7