摘要
从统计长江中下游平原地区57个自动气象站2002—2007年期间1,4,7,10月份的日平均气温结构函数着手,推算出各月线性内插精度与台站间距的关系,由内插标准误差不超过观测标准误差的原则,估算出该地区气温场的最大容许测站间距为29 km。并根据不同的内插精度要求增设站点,增设站点的气温数据由插值效果较好的克里金方法内插获得。最后采用基于交叉验证的支持向量机对气温数据进行预处理,优化气象站模型,合理化站网密度。分析表明:当内插标准误差分别小于等于0.045℃,0.037℃,0.033℃时,自动气象站的布站间距分别为16 km,12 km,8 km,所需的站点数分别为2038,4268,8597个。从增设站点的分布来看,对站点数量要求较高的地区大多是等温线比较密集的地区,即气温差异大的地区。
This study is made on the relational distribution of weather station networks by using the spatial structure function as the criterion for the design. Using the data of daily air temperature from 2002 to 2007 over Chi- na＇ s Yangtze Plain, the spatial structure function is created. Furthermore, according to the principle that the stand- ard error of interpolation should not exceed the standard error of observation, the maximum admissible spacing be- tween stations is 29 km. The article adds different stations with the changing of interpolation accuracy. The tempera- ture of the adding staions is obtained by the means of Kriging which is the best interpolation method. Finally, using the method of SVM process the data of simulation station and optimize the model of distribution of the automatic weather stations. Results show that whenE = 0. 045 ~C , l = 16 km ,the needed amount of staions is 2 038, when E = 0.037 ~C , l = 12 km ,the needed amount is 4 268,whenE = 0. 033 ℃ , 1 = 8 kmthe needed amount is 8 597. In summary, the amount of stations is larger in the areas of Intensive isotherm where the climate difference is big than those of sparse isotherm where the climate difference is small.
出处
《科学技术与工程》
北大核心
2013年第14期4050-4056,共7页
Science Technology and Engineering
基金
国家重大仪器设备开发专项经费(2012YQ170003)资助
关键词
结构函数
最优插值
支持向量机
站网密度
spatial structure function best interpolation method SVM density of automatic weather-Station
作者简介
唐慧强（1965-），男，浙江嘉兴人，南京信息工程大学博士研究生，研究方向：智能仪器、气象仪器、无线传感器网络。E—mail：723187763@qq．com。
通信作者简介：范海娟（1988-），女，江苏南通人，南京信息工程大学硕士研究生，研究方向：气象数据处理。E-mail：619468772@qq．com。
