Identifying spatial drivers of soil heavy metal pollution risk integrating positive matrix factorization, machine learning, and multi-scale geographically weighted regression

Yujie Pan; Anmeng Sha; Wenjing Han; Chang Liu; Guowangchen Liu; Emily Welsch; Min Zeng; Shasha Xu; Yi Zhao; Shang Tian; Yiyi Li; Rui Deng; Xin Zhang; Huanhuan Shi; Yu Cui; Changsheng Huang; Hongxia Peng

doi:10.1016/j.jhazmat.2024.136841

Identifying spatial drivers of soil heavy metal pollution risk integrating positive matrix factorization, machine learning, and multi-scale geographically weighted regression

J Hazard Mater. 2024 Dec 10:485:136841. doi: 10.1016/j.jhazmat.2024.136841. Online ahead of print.

Authors

Yujie Pan¹, Anmeng Sha¹, Wenjing Han², Chang Liu¹, Guowangchen Liu¹, Emily Welsch³, Min Zeng⁴, Shasha Xu¹, Yi Zhao¹, Shang Tian¹, Yiyi Li⁵, Rui Deng¹, Xin Zhang⁶, Huanhuan Shi⁶, Yu Cui⁷, Changsheng Huang⁸, Hongxia Peng⁹

Affiliations

¹ College of Environmental Sciences and Engineering, Peking University, Beijing 100871, China.
² Geological Survey Research Institute, China University of Geosciences, Wuhan 430074, China.
³ Department of Geography and Environment, The London School of Economics and Political Science, London WC2A 2AE, UK.
⁴ Wuhan Center of Geological Survey of China Geological Survey, Wuhan 430205, China.
⁵ College of Electronic Science and Control Engineering, Institute of Disaster Prevention, Hebei 065201, China.
⁶ School of Environmental Studies, China University of Geosciences, Wuhan 430078, China.
⁷ International Institute for Applied Systems Analysis (IIASA), Laxenburg A-2361, Austria.
⁸ Wuhan Center of Geological Survey of China Geological Survey, Wuhan 430205, China. Electronic address: penghx@cug.edu.cn.
⁹ School of Geography and Information Engineering, China University of Geosciences, Wuhan 430074, China. Electronic address: penghx@cug.edu.cn.

PMID: 39689561
DOI: 10.1016/j.jhazmat.2024.136841

Abstract

Soil heavy metal (HMs) contamination poses significant ecological and health risks, yet the spatial drivers of HMs pollution remain poorly understood. This study integrates pollution risk assessment, positive matrix factorization, machine learning, and multi-scale geographically weighted regression to develop a framework for identifying the spatial drivers of soil HMs contamination risk in Yangtze River New City, China. Analysis of 7152 samples revealed that although average HMs concentrations were below national standards, As, Cd, Cr, Cu, Hg, and Ni exceeded local background levels. Four key factors were identified as drivers of HMs contamination: natural sources (30.36 %, influenced by soil type), mixed agricultural and transportation sources (29.56 %, driven by cropland, aquaculture, and road density), human activities (12.68 %, including population density and community activities), and industrial sources (27.42 %, linked to factories and enterprises). Regional variations indicated that industrial activities, transportation, and human activities primarily influenced health risks, while agriculture and natural factors had a greater impact on ecological and environmental capacity risks. These findings underscore the importance of considering spatial heterogeneity in HMs pollution risk assessments and offer insights for developing targeted, region-specific policies to mitigate pollution risks of soil HMs.

Keywords: Heavy metals; MGWR; ML; Pollution source identification; Spatial heterogeneity.