Comparative analysis of soil physicochemical properties for tracing soil spatial origins using digital soil mapping products.

Abstract

This study evaluates the capability of basic soil physicochemical properties to optimally screen soil evidence for forensic investigations to find potential source areas. Simulated spatial source-tracing experiments were conducted on soil samples from a soil series database from Anhui Province, China using similarity-matching algorithms. The potential source area was represented by digital soil mapping products, with predicted values and their uncertainties in each grid cell. Properties considered as tracers were bulk density (BD), total Nitrogen (TN), total Potassium (TK), total Phosphorous (TP), cation exchange capacity of the fine soil (CEC), pH value, clay particles, silt particles, sand particles, and soil organic matter (SOM). The spatial indicative capability of each property was assessed in two dimensions: provenancing accuracy and provenancing convergence (the ability to narrow down the provenance range). These were based on three criteria: source point coverage, administrative division coverage, and spatial likelihood of provenancing ranges. The soil properties BD, TN and TK have high accuracy but poor convergence, while pH has strong convergence but the lowest accuracy. By contrast, TP and CEC showed robust and stable source-tracing performance for both dimensions. This method can be adapted to any region with a suitable digital soil mapping coverage and estimates of laboratory precision.