Comparison of interpolation models for estimating heavy metals in soils under various spatial characteristics and sampling methods. Issue 2 (2nd March 2018)

• Record Type: Journal Article
• Authors: Liao, Yilan; Li, Dongyue; Zhang, Ningxu
• Is Part Of: Transactions in GIS. Volume 22:Issue 2(2018)
• Page Start: 409
• Page End: 434
• Publication Date: 2018-03-02
• Subjects: Geographic information systems -- Periodicals; 910.285
• DOI: 10.1111/tgis.12319
• Languages: English
• ISSNs: 1361-1682

Abstract

Abstract: Heavy metal pollution in soils has become increasingly challenging, especially in developing countries. Estimating the spatial distribution of heavy metals in soils is essential to preventing their build‐up. This article aims to identify the effects of spatial scales, spatial autocorrelation, sampling methods, and proportion on interpolation models in estimating the distribution of heavy metals in soils. Six interpolation models (area‐and‐point kriging, AAPK; inverse distance weighting, IDW; local polynomial interpolation, LP; ordinary kriging, OK; simple kriging, SK; and thin plate spline, TPS), three sampling methods (random, stratified, and systematic sampling), and five sampling proportions (1, 5, 10, 15, and 20%) are considered in this study using sets of simulated data, and the real situation was tested for verification. The results show that, in general, with the increase of spatial autocorrelation or the sampling percentage, the accuracy and stability of different interpolation models gradually increase; however, the various interpolation models have their own specific characteristics and application conditions. The best application conditions of the interpolation models compared with other models under the same situation are summarized and explained in theory. These conclusions have implications for future work.