中文摘要：

利用局部最小二乘配置法（LSC）借助＂分治法＂寻找待求点附近已知点,以空间地统计中的半变异函数计算已知数据余差的协方差矩阵,以与待求点距离小于变程的已知点为起算数据,实现空间数据模拟。数值试验分析表明,与传统LSC相比,局部LSC可以在保证模拟精度的同时,提高模型计算效率。

英文摘要：

We have developed a local least squares collocation（LSC） method to decrease the computational cost of the classical LSC.Local LSC employs the divide-and-conquer method to determine the neighbor sampling points of the estimated one.The covariance matrix of residuals after trend removal is obtained with semivariance in geostatistics.The sampling points with the distance from the estimated one smaller than the range is used for local LSC computation.Numerical tests indicate that under the same simulation accuracy,local LSC is much faster than the classical one.The real-world example of simulating the DEM of Dongzhi tableland shows that LSC is more accurate than IDW and ordinary kriging.