Disentangling runoff generation mechanisms: Combining isotope tracing with integrated surface/subsurface simulation

[Display omitted] •An integrated approach involving HGS modeling and isotope analysis is developed.•The contribution of the subsurface water to the stream is 32% for one rainfall event.•Rainfall infiltration-mixing-exfiltration process is vital to runoff generation.•Exfiltration-induced overland flow is the dominant water source for a rainfall event. The majority of runoff studies to date have reported a similar pattern of rapid rain-to-runoff conversion via overland flow and shallow subsurface water flow, as well as the ubiquity of thresholds and hysteresis in rainfall-runoff response. But each of those approaches has limitations in terms of inference and process description. Processes of water transport and mixing and how they influence runoff generation are still not well understood or quantified. In this study, HydroGeoSphere, a fully-integrated surface/subsurface flow model, was used together with stable isotopes for disentangling runoff generation mechanisms in a headwater catchment. The study area (0.21 km2) is the first-level tributary of the Xin’an Jiang River located in a humid climate region of eastern China. Water flow simulation elucidated the spatial water transport process, and isotope tracing quantified the complex water mixing process between the shallow soil and hillslope surface. This study provides insights into the rapid transformation of rainfall infiltration and mixing in soil and exfiltration to a hillslope. Results show that the stream runoff was separated into rainfall-induced overland flow (9 %), exfiltration-induced overland flow (including the 53 % of mixed water from rainfall infiltration and 6 % of the stored soil water), and subsurface flow (32 %). Exfiltration-induced overland flow was the dominant water source in the headwater catchment during a rainfall-runoff event. The mutual influence of rainfall infiltration and soil water exfiltration caused the shallow soil to rapidly become saturated, thus forming saturation excess overland flow. The consistence of estimated soil water velocity based on the integrated isotope and numerical modeling approach, at about 0.4 m/d nearby the stream channel, improved the reliability of model visualizations. Further analysis indicates that water flow paths and velocities responding to the rainfall event were attributed to the soil moisture conditions and topography. This study demonstrates that identifying the rapid mixing processes between the rain and the soil water was crucial for under