摘要
气象要素是资源、环境和灾害以及全球变化等领域研究的数据基础,格点化数据在未来研究应用中显得日益重要。基于中国境内667个基本和基准地面气象观测站点的基本气象资料,使用ANUSPLIN专用气候插值软件对1961—2006年逐日气温、降水进行插值,并利用未参与插值的全国1 667个加密站点对插值结果的准确性进行检验,同时与反向距离权重法和普通克吕格法等插值方法的结果进行对比。结果表明:利用667个站点使用ANUSPLIN软件进行逐日平均气温插值有92.0%的误差在2.0℃以内,75.0%的误差在1.0℃以内,0.9%的误差在5.0℃以上,平均绝对误差为0.8℃;对逐日降水进行插值,75.0%的误差小于5.0 mm,85%的误差小于10.0 mm,平均绝对误差为6.4 mm,误差大小与降水量呈现出正相关性,对局地强降水的插值效果不好,这可能与参与局部拟合插值的样本数太少有关;同时,夏季的温度插值误差小于冬季,而冬季的降水误差小于夏季。将ANUSPLIN的局部薄盘样条插值结果分别与反向距离权重法和普通克吕格法的插值结果进行对比,显示ANUSP-LIN软件的插值误差最小。结果同样表明,适当增加站点数量和提高DEM精度可进一步提高ANUSPLIN软件的插值精度。
Meteorological elements are the basic data for the study on resources,environment,disaster and global change fields,and their gridding data become more and more important for the study at large scale.The daily air temperature and precipitation were interpolated by spline interpolating method from ANUSPLIN software based on the data from 667 weather stations of China from 1961 to 2006 in order to get the gridding data,and the interpolating results were validated by the data from other 1667 dense observation stations.The interpolating results were compared with those data based on intense distance weight method and ordinary kriging method.It indicates that the errors within 2.0 ℃ for daily average air temperature account for 92.0%,and those within 1.0 ℃ and over 5.0 ℃ are 75.0% and 0.9%,respectively.The mean absolute error is 0.8 ℃.On the other hand,the errors within 5.0 mm and 10.0 mm for daily precipitation account for 75.0% and 85.0%,respectively,and the mean absolute error is 6.4 mm.The relationship between the error ranges of gridding precipitation and observation values of precipitation are positive,and interpolating accuracy decreases for local strong precipitation,which maybe result from the insufficient observation values.Furthermore,the interpolating accuracy for air temperature is higher in summer than in winter,while that for precipitation is higher in winter than in summer.The validation results indicate that the accuracy of gridding data based on spline interpolation method of ANUSPLIN software is higher than that based on intense distance weight method and ordinary kriging method.It also indicates that more observation data and finer resolution DEM can improve the accuracy of the spline interpolation method of ANUSPLIN software.
出处
《气象与环境学报》
2010年第2期7-15,共9页
Journal of Meteorology and Environment
基金
"十一五"国家科技支撑计划重点项目"农业重大气象灾害监测预警与调控技术研究"(2006BAD04B09)资助
关键词
ANUSPLIN软件
气象要素
插值方法
插值评估
ANUSPLIN software
Meteorological elements
Interpolation method
Interpolation assessment
作者简介
钱永兰，女，1975年生，高级工程师，主要从事农业气象和农业遥感应用研究，E—mail：qianyl@cma．gov．cn。
