五种常用降水量插值方法误差时空分布特征研究——以深圳市为例  

On Temporal and Spatial Error Distributions of Five Precipitation Interpolation Models:A Case of Shenzhen

作　　者：邬伦[1] 吴小娟[2] 肖晨超[1] 田原[1,3] 

机构地区：[1]北京大学遥感与地理信息系统研究所,北京100871 [2]武汉大学遥感信息工程学院,湖北武汉430079 [3]香港理工大学土地测量及地理资讯学系

出　　处：《地理与地理信息科学》2010年第3期19-24,共6页Geography and Geo-Information Science

基　　金：国家科技支撑计划课题(2008BAJ11B04;2006BAJ14B04);国家自然科学基金项目(40701134;40928001);国家高科技研究发展计划项目(2007AA120502);香港理工大学研究基金(Project No.G-U632)

摘　　要：在洪水、滑坡等地质灾害预警预报中,通常需要对多个时点、多个站点的降水量观测数据进行高精度插值,雨量插值精度对灾害预警预报的可靠度具有很大的影响,因此研究降水量插值方法误差的时空分布特征具有重要的科研和实用价值。该文以深圳市2008年6月12日至14日百年一遇的强降水过程为例,采用交叉验证方法对反距离权重法、普通克里金法、全局多项式法、局部多项式法和径向基函数法五种常用空间插值方法误差的时空分布特征进行分析,研究成果可为根据雨量时空分布特点选取适用雨量插值模型提供相关依据,并为相关研究提供借鉴。Precipitation is an important meteorological element widely used in the forecasting of many kinds of natural disasters,such as floods and landslides.Typically,interpolation models are applied to calculate the precipitation distribution over the working area based on limited observation data.It is important to study the temporal and spatial error distribution of precipitation interpolation models to get precise precipitation data and to make the disaster forecasting more credible.Five typical interpolation models,Inverse Distance Weighted,Ordinary Kriging,Global Polynomial,Local Polynomial,and Radial Basis Function,are chosen to carry a case study of Shenzhen,based on observation data of a rainfall process between June 12 and June 14,2008.It can be concluded that the errors of all the five methods are highly relevant to precipitation.Considering both the time and space series distribution,the errors of Ordinary Kriging method are minimum of all and the errors of Local Polynomial and Global Polynomial are the maximum.At the moment of maximum precipitation and at the locations of maximum precipitation,Ordinary Kriging method has the minimum errors while Global Polynomial method has the maximum errors.But at the moment of minimum precipitation,the errors of Global Polynomial,Local Polynomial and Inverse Distance Weighted are smaller than that of Ordinary Kriging,while errors of Radial Basis Function is the maximum.On the location of minimum precipitation,the errors of Inverse Distance Weighted and Global Polynomial are maximum and minimum respectively.The conclusions of this study may provide useful guidance on choosing suitable interpolation model.

关 键 词：降水量 插值误差 误差时空分布特征 地质灾害 预警预报 

分 类 号：P208[天文地球—地图制图学与地理信息工程] P426.6[天文地球—测绘科学与技术]