XGBoost-SGS Concentration Mapping and Vadose Zone Transport Toolkit

Software toolkit for hybrid machine learning-geostatistical prediction of heavy metal concentrations in urban soils, combining XGBoost regression for deterministic trend modelling with Sequential Gaussian Simulation (SGS) for spatially explicit uncertainty quantification, coupled with a Numba-accelerated 1D vadose zone transport model. The workflow processes eight heavy metals (As, Cd, Co, Cr, Cu, Ni, Pb, Zn) through covariate preparation, XGBoost training, residual variography, SGS with 100 realisations, grid alignment, and scenario-based transport modelling using Van Genuchten water flow and linear isotherm sorption.