基于回归克里格法的土壤盐分采样点布局优化

Layout Optimization of Soil Salt Sampling Points Based on Regression Kriging

为提高土壤盐分分布图绘制精度或降低野外采样和室内分析成本,优化土壤采样点布局,本文以回归克里格法估计方差(RKV)最小和普通克里格估计方差(OKV)最小为目标函数,运用空间模拟退火(SSA)算法对土壤盐分采样点布局进行优化,其中回归克里格法(RK)中的协变量选择与土壤盐分分布关系密切且容易获取的环境变量;通过最邻近距离分布函数(G函数)、间隔距离分布函数(F函数)和K-S检验对优化结果进行评价。结果表明:荒地分布在大尺度上(400 m×400 m)与盐分相关性强,而在小尺度上(50 m×50 m)与盐分相关性弱;以荒地分布(与荒地最短距离)作为协变量,利用RK法优化采样点布局(采样数为55)后得到的研究区平均RKV为0.0807,相较平均OKV(0.1143)缩小了29.39%;优化后采样点布局的G函数曲线和F函数曲线全部落在95%置信区间内,即F函数和G函数均与其理论值无显著差异,采样点布局良好覆盖了地理空间;K-S检验表明,优化后采样点上与荒地最短距离的分布与其总体分布服从同一概率分布,即采样点布局良好覆盖了属性空间。本研究可为河套灌区土壤盐渍化监测点布局提供参考。

In order to improve the drawing accuracy of soil salt distribution map or reduce the cost of field sampling and indoor analysis,it is very necessary to optimize the layout of soil sampling points.Taking the minimum regression Kriging estimated variance(RKV)and the minimum ordinary Kriging estimated variance(OKV)as the objective function,the layout of soil salt sampling points was optimized by spatial simulated annealing algorithm(SSA),the selection of covariates in regression Kriging(RK)method was closely related to soil salt distribution and easy to obtain environmental variables.The optimization results were evaluated by the nearest distance distribution function(G-function),interval distance distribution function(F-function)and K-S test.The results showed that wasteland distribution on a large scale(400 m×400 m)was significantly correlated with salt,but weak correlated on a small scale(50 m×50 m).Taking the distribution of wasteland(the shortest distance from wasteland)as the collaborative variable and optimizing the sampling layout by RK method(sampling number was 55),the mean RKV of the study area was 0.0807,which was 29.39%smaller than the mean OKV(0.1143).The G-function curve and F-function curve of the optimized sampling set all fall within the 95%confidence interval,there was no significant difference between the F-function and G-function and their theoretical values,the layout of the sampling points was well arranged and covered the geographical space.K-S test showed that the distribution of the shortest distance from wasteland on the optimized sampling point and its population obeyed the same probability distribution,the layout of the sampling points was well arranged and covered the attributes space.The research results can provide reference for the layout of soil salinization monitoring points in Hetao Irrigation Area.