Quantitative analysis of runoff, total suspended solids, and total nitrogen dynamics, along with the identification of key factors within catchments, is essential for accurately addressing issues related to turbid and polluted water. Nevertheless, their implementation encounters significant challenges when applied to a mixed catchment containing mountain areas and lowland polder regions, due to the highly heterogeneous hydrological behaviors and consequently the lack of an appropriate approach. Faced with this problem, this study developed a framework by coupling the Soil and Water Assessment Tool (SWAT) and improved Polder Hydrology and Nitrogen modelling System (PHNS), and Random Forest analysis method to track the spatio-temporal changes in runoff, total suspended solids, and total nitrogen loading and identify their environmental determinants in a representative mountain-lowland mixed catchment, southeastern China. The coupled model performed very well for runoff (R