Optimal GIS interpolation techniques and multivariate statistical approach to study the soil-trace metal(loid)s distribution patterns in the agricultural surface soil of Matehuala, Mexico

Soils in the past mining areas are susceptible to trace metal(loid)s deposition and pose a health risk to humans. The purpose of this work is to evaluate the distribution patterns and contamination characteristics of trace metal(loid)s in the agricultural surface soils of past mining regions. The contaminated site is near an abandoned mining area, which is surrounded by land used for maize cultivation. The multivariate statistical approach and GIS interpolation techniques are often used in spatial distribution mapping to predict metal(loid)s concentrations for arsenic (As) and other trace metals (i.e., Al, Fe, Mn, and Sr) in areas that have not been sampled. The mean relative error (MRE) and root mean square error (RMSE) values were used to evaluate and correlate the efficiency of deterministic interpolation (Inverse Distance Weighting-IDW, Local Polynomial-LP, and Radial Basis Functions-RBF) as well as geostatistical interpolation methods (Ordinary Kriging-OK and Empirical Bayesian Kriging-EBK). The results revealed that all interpolation techniques predicted the mean concentration of trace metal(loid)s in soil with moderate accuracy. It was found that agricultural soils contained arsenic enrichment (up to 185 mg/kg), up to five times the background concentrations (35 mg/kg), and 8.5 times the Mexican guidelines (22 mg/kg). The statistical analysis with the cross-validation method revealed that IDW and LP consistently provided the most accurate predictions of trace metal(loid)s concentrations while OK, EBK, and RBF techniques are less accurate. Overall, these GIS interpolation techniques help in the prediction of trace metal(loid)s concentrations at unexplored sites and establish the requirement for the amelioration of agricultural surface soil.