【Objective】Soil spatial traceability has significant application value in forensic soil science and judicial identification.【Method】This study, based on 265 surface soil samples from Anhui Province, compares two soil provenance strategies: (1) a similarity-matching and spatial clustering approach, which filters similar samples using spectral and physicochemical property similarities and applies the DBSCAN algorithm to determine the potential source area of unknown samples; and (2) an inverse inference approach based on pedogenic environmental factors, which employs a random forest model to predict environmental variables such as soil parent material, land use, topography, climate, and vegetation, and infers provenance by integrating spatial distribution maps. By simulating provenance analysis, the accuracy and applicability of the two strategies were evaluated. 【Result】The results indicate that the similarity-matching strategy achieves higher localization accuracy under conditions of strong spatial proximity and well-established databases, while the inverse pedogenic environment inference strategy demonstrates superior spatial constraint capabilities in regions with limited databases or strong spatial heterogeneity. 【Conclusion】Each strategy has its own advantages, and their integrated application holds promise for further improving the accuracy and resolution of soil spatial provenance analysis.