Abstract: This study focused on the farmland system in the mining area of Erpengdianzi Town and Xiangyang Township, Huanren Manchu Autonomous County, Benxi City, Liaoning Province, to identify the dominant sources and processes of heavy metal contamination in farmland, clarify the spatial differentiation characteristics of heavy metal contamination signals across different subareas, and develop governance zones for management units. Based on multi-media measured data of farmland soils, river sediments, pollution source zones, and background sampling sites collected from June to September 2023, a migration-identification framework integrating subdomain alignment, prototype extraction, and non-negative contribution decomposition was developed after unified data cleaning and sample reconstruction. Auxiliary interpretation was conducted based on the risk-screening value framework specified in GB 15618-2018, and management units were integrated according to spatial continuity based on point identification. After removing anomalies, a total of 239 valid soil records corresponding to 120 valid primary sampling sites were retained. Auxiliary interpretation based on 120 surface-soil samples showed that 116 sites had at least one indicator exceeding the risk screening value. Cd was the element of primary concern, while Zn, Pb, and Cu also showed high rates of exceeding the screening values; 42.50% and 14.17% of the sites exceeded the risk intervention values for Cd and Pb, respectively. The farmland soils in the study area were mainly governed by the hydraulic transport type (43.33%), followed by the background-dominated type (29.17%) and the pollution source zone-dominated type (27.50%), indicating that heavy metal contamination was mainly controlled by the process chain of source release, pathway transport, and re-entry into farmland. The results suggest that heavy metal pollution formation in the farmland was mainly governed by pathway processes. Combining auxiliary risk interpretation with dominant-prototype identification helps to simultaneously address which elements and which positions and paths to prioritize for control. Furthermore, integrating point-based control targets into management units can improve the engineering feasibility of zoning-based management.