Enhancing apportionment of the point and diffuse sources of soil heavy metals using robust geostatistics and robust spatial receptor model with categorical soil-type data. (October 2020)

• Record Type: Journal Article
• Authors: Qu, Mingkai; Chen, Jian; Huang, Biao; Zhao, Yongcun
• Is Part Of: Environmental pollution. Volume 265(2020)Part A
• Publication Date: 2020-10
• Subjects: Point sources -- Diffuse sources -- Soil heavy meals -- Soil types -- RAPCS/RGWR-ST -- Robust geostatistics; Pollution -- Periodicals; Pollution -- Environmental aspects -- Periodicals; Environmental Pollution -- Periodicals; Pollution -- Périodiques; Pollution -- Aspect de l'environnement -- Périodiques; Pollution -- Effets physiologiques -- Périodiques; Pollution; Pollution -- Environmental aspects; Periodicals; Electronic journals; 363.73
• DOI: 10.1016/j.envpol.2020.114964
• Languages: English
• ISSNs: 0269-7491

Abstract

Abstract: Soil-type data usually contain valuable information about soil heavy metal (HM) concentrations; however, they were rarely considered in the apportionment of point or diffuse sources in previous studies. In this study, the spatial variations of the soil HM concentrations in Jintan County, China were partitioned into two portions – the soil-type effects and the corresponding residuals, using analysis of variance (ANOVA). Standardized robust kriging error (SRKE) with soil-type data as auxiliary information (SRKE-ST) was proposed to identify the high-value spatial outliers of soil HMs, and the performance of SRKE-ST was compared with that of commonly-used SRKE. Robust absolute principal component scores/robust geographically weighted regression (RAPCS/RGWR) with soil-type data as auxiliary information (RAPCS/RGWR-ST) was proposed to apportion the diffuse sources of soil HMs, and the performance of RAPCS/RGWR-ST was compared with those of RAPCS/RGWR and commonly-used absolute principal component scores/multiple linear regression (APCS/MLR). Results showed that (i) RSKE-ST effectively excluded high-value spatial outliers resulting from the effects of complex soil-type polygons on soil HM concentrations; (ii) RAPCS/RGWR-ST generated higher estimation accuracy in source contributions and less negative contributions than RAPCS/RGWR and APCS/MLR did. It is concluded that the proposed RSKE-ST and RAPCS/RGWR-ST could effectively use categorical soil-type data to enhance, respectively, the identification of high-value spatial outliers (i.e., potential point sources) and the apportionment of diffuse sources of soil HMs in large-scale areas. Graphical abstract: Image 1 Highlights: Categorical soil types affected the concentrations of multiple soil heavy metals. Soil-type data were incorporated in the apportionment of point and diffuse sources. RSKE-ST effectively excluded high-value spatial outliers caused by the soil types. RAPCS/RGWR-ST generated higher apportionment accuracy for the diffuse sources. The proposed models can use auxiliary soil-type data to enhance source apportionment. Abstract : SRKE-ST and RAPCS/RGWR-ST could effectively use categorical soil-type data to enhance, respectively, the identification of high-value spatial outliers and the apportionment of diffuse sources of soil HMs in large-scale areas.