Accuracy Assessment of Sampling Designs for Surveying Heavy Metal Content in Soil Using SSSI Aihua Ma; Jinfeng Wang; Keli Zhang

Website : 2 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 1. Introduction 2. Methodology 3. Case Study 4. Conclusion 5. Discussion

Website : 3 Why to sampling? Funds are not sufficient The survey is too large Faster time Better accurate 1. Introduction State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved

Website : 4 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved EnvironmentalLand and Resources EconomicEcological Application areas 1. Introduction

Website : 5 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process 2. Methodology

Website : 6 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process Soil, food production, land cover type, etc

Website : 7 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process N=4×8

Website : 8 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process Overall information: variance, the relative error, absolute error Users on the accuracy of sampling results

Website : 9 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process

Website : 10 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process

Website : 11 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process variance Independent samples Non-independent samples

Website : 12 Determine the study area Determine the study objects Determine the population sizes Calculate sample sizes Layout of samples Field survey results Compare efficiency Sampling Design Process Traditional models Spatial models Relative error Coefficient of variation Design effect

Website : 13 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 3. Case Study County of Zhongyang County of Jiaokou two counties Zhongyang and Jiaokou of Shanxi Province were selected as research area, they are high incidence areas of birth defects.

Website : 14 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved I use the soil samples as sampling data, soil samples were collected in most of villages, there are 84 points in all.16 kinds of elements in the soil were measured: Al ， As ， Ca ， Cu ， Fe ， K ， Mg ， Mo ， Na ， Ni ， Pb ， Se ， Sn ， Sr ， V ， Zn. Mo element is selected. 3.1The spatial distribution graph of data 3. Case Study

Website : 15 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 3.2 Data Exploratory analysis Mo semi-variogram mainly semi-variogram analysis semi-variance function graph can detect whether they have been measured to be spatial dependent among the samples.

Website : 16 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 3.3 Choose stratified index This area has the complex and varied terrains and landforms, four stratified way: soil type, geological surface, geochronology, hierarchical cluster.

Website : 17 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved Choose stratified index

Website : 18 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved Five kinds of sampling model are selected to compare the sampling efficiency. simple random sampling model stratified random sampling model spatial random sampling model spatial stratified sampling model sandwich spatial sampling model 3.4 Choose sampling models Traditional models Spatial models Systematic model

Website : 19 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 3.5 Choose efficiency indicator relative error Relative error ( )compares the difference between sample mean and its true mean, so the estimated relative error is defined as : where = sample mean = observable population mean

Website : 20 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 3.4 Choose efficiency indicator Coefficient of variation design effect Design effect is the ratio of estimated variance obtained from the (more complex) sample to the estimated variance obtained from a simple random sample of the same number of units. The coefficient of variation is a statistical measure of the dispersion of data points in a data series around the mean. It is calculated as follows:

Website : 21 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 4. Conclusion With smaller sample sizes, the simple random sampling model <stratified sampling model, and the interval is large. With larger sample sizes, the stratified sampling model fluctuates within a certain range, but is more accurate than the simple sampling random model relative error

Website : 22 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved The sandwich spatial sampling model is the newest method in the SSSI software. It has the same accuracy to the spatial stratified sampling, but it refers to report layers, which can be any unit, for example, a county border, provincial boundary, watershed, or artificial grid Report layers Stratified by soil type stratified by geochronology Administrative villages grid Mo We can see from the table, the relative errors are small, the sampling accuracy are high.

Website : 23 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved Coefficient of variation MO 元素 Soil type Geological surface geochronology hierarchical cluster

Website : 24 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved Coefficient of variation It shows which stratified method is more efficient, Stratification by soil type yields higher accuracy than by geochronology in the case of smaller sample sizes, but lower accuracy in larger sample sizes.

Website : 25 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved design effect 1. 设计效应 models Sample sizes Srs StrRs (a) SStrs (a) StrRS (b) SStrs (b) models Sample sizes StrRs (c) SStrs (c) StrRS (d) SStrs (d) MO

Website : 26 State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved 6. discussion  Efficiency is up to: Sampling models Stratified method  Future work Sample with layout

Website : 27 Thanks! State Key Laboratory of Resources & Environmental Information System Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences All rights reserved