Abstract: A two-dimensional shallow water model coupled with a cellular automata framework was developed to simulate surface runoff characteristics and heavy metal pollutant diffusion dynamics in Nanjing’s urban area under extreme rainfall conditions. This integrated methodology establishes dynamic neighborhood interaction rules to quantify pollutant transport processes, enabling a systematic quantitative evaluation of pollution dispersion patterns and associated population exposure risks. Applying the potential ecological risk index (RI) and health risk assessment model, the study quantified temporal variations in soil heavy metal threats to ecosystems and exposed populations before and after precipitation events. The results showed that the extreme rainfall significantly accelerates the migration of Cu, Zn, Pb, Cr, Cd, and Hg in the soil, with their maximum concentrations increasing to 221, 222, 178, 221, 127, and 94 times initial values, respectively. The average Risk Index (RI) of soil heavy metals increases from 127.82 to 685.46 after rainfall erosion. In terms of population health risk exposure, the average Hazard Index (HI) for adults rises from 0.08 to 0.46 and their total carcinogenic risk (TCR) increases from 0.571×10⁻⁶ to 2.803×10⁻⁶, while for children, the average HI climbs from 0.44 to 2.62 and their TCR surges from 1.003×10⁻⁶ to 4.925×10⁻⁶. It indicates that the spread of heavy metal pollution exacerbated by extreme precipitation increases the health risks of exposed populations, and the risks for children are considerably higher than those for adults.