CAJ | 학술논문

最优采样密度的确定是土壤环境调查、监测及监测网建设过程中的热点和难点科学问题，但当前对土壤环境监测最优网格尺度识别的研究较少。文章在对珠江三角洲地区耕地土壤按照2 km×2 km的网格加密调查的基础上，采用经典统计方法、地统计法和空间插值等方法开展土壤镉最优监测网格尺度研究。结果表明：（1）2 km×2 km、4 km×4 km、8 km×8 km和16 km×16 km 4种网格尺度下，采样点间距离均小于空间相关距离，具有中等的空间自相关性；8 km×8 km网格的变程及拟合度最大，说明利用地统计学方法进行珠三角耕地土壤镉含量变量的空间分析，8 km×8 km网格的空间表达能力最强；（2）不论从交叉验证模型，还是独立验证模型以及空间分布模式均可发现，2 km×2 km、4 km×4 km和8 km×8 km等3种采样尺度均可较可靠地反映珠三角区域耕地土壤镉含量的空间变异，而16 km×16 km的采样尺度则无法保证可靠性；（3）在置信水平为95%，相对误差为15%的条件下，采用Cochran公式获取的最佳采样数量为169个，而2 km×2 km、4 km×4 km和8 km×8 km满足此条件的点位数均大于169个，此时误差小于《土壤环境监测技术规范》（HJ/T 166─2004）中实验室允许的最大误差25%。总之，在珠江三角洲区域尺度上，以土壤镉作为评价目标的土壤采样中，8 km×8 km是比较适宜的采样尺度。
최우채양밀도적학정시토양배경조사、감측급감측망건설과정중적열점화난점과학문제，단당전대토양배경감측최우망격척도식별적연구교소。문장재대주강삼각주지구경지토양안조2 km×2 km적망격가밀조사적기출상，채용경전통계방법、지통계법화공간삽치등방법개전토양력최우감측망격척도연구。결과표명：（1）2 km×2 km、4 km×4 km、8 km×8 km화16 km×16 km 4충망격척도하，채양점간거리균소우공간상관거리，구유중등적공간자상관성；8 km×8 km망격적변정급의합도최대，설명이용지통계학방법진행주삼각경지토양력함량변량적공간분석，8 km×8 km망격적공간표체능력최강；（2）불론종교차험증모형，환시독립험증모형이급공간분포모식균가발현，2 km×2 km、4 km×4 km화8 km×8 km등3충채양척도균가교가고지반영주삼각구역경지토양력함량적공간변이，이16 km×16 km적채양척도칙무법보증가고성；（3）재치신수평위95%，상대오차위15%적조건하，채용Cochran공식획취적최가채양수량위169개，이2 km×2 km、4 km×4 km화8 km×8 km만족차조건적점위수균대우169개，차시오차소우《토양배경감측기술규범》（HJ/T 166─2004）중실험실윤허적최대오차25%。총지，재주강삼각주구역척도상，이토양력작위평개목표적토양채양중，8 km×8 km시비교괄의적채양척도。
Confirming what would be the optimal sampling density is one of the difficulties for soil environment investigation, monitoring and the construction of monitoring station networks. There are only a few studies that have been conducted on the optimal grid scale for soil environment investigations. This paper is based on the investigation of a 2 km×2 km dense grid of arable soil in the Pearl River Delta, and describes the research carried out to determine the optimal monitoring grid for measuring soil cadmium content by classical statistical methods, geo-statistical methods and spatial interpolation. The results indicate that: (1) Under different monitoring sizes of cells (2 km×2 km, 4 km×4 km, 8 km×8 km, and 16 km×16 km), the distances between the sampling points with intensive spatial autocorrelation are all smaller than the distance of spatial dependence;with the fitting degree increasing, the expressive ability of spatial variability of soil Cd concentration strengthen, indicating the expressive ability of 8 km×8 km grid is strongest;(2) The monitoring grids, except for the 16 km×16 km one, was able to accurately reflect the spatial variability of the arable soil cadmium content in the Pearl River Delta;And (3) the confidence level of the grids (2 km×2 km, 4 km×4 km, 8 km×8 km) to the Cochran Equation is 95%and the relative error is 15%. Under these conditions, the sampling error of the grids (2 km×2 km, 4 km×4 km, 8 km×8 km) was smaller than the permissible laboratory error as specified in the Soil Environmental Monitoring Technical Specifications (HJ/T 166─2004). In conclusion, the 8 km×8 km grid is the theoretical sampling grid for measuring soil cadmium content as the evaluation target on the scale of the Pearl River delta.