A new time-space accounting scheme to predict stream water residence time and hydrograph source components at the watershed scale

Hydrograph source components and stream water residence time are fundamental behavioral descriptors of watersheds but, as yet, are poorly represented in most rainfall-runoff models. We present a new time-space accounting scheme (T-SAS) to simulate the pre-event and event water fractions, mean residence time, and spatial source of streamflow at the watershed scale. We use a physically based hydrologic model together with field data from the well-studied Maimai M8 watershed and HJ Andrews WS10 watershed to explore how catchment properties, particularly soil depth, controls the age and source of streamflow. Our model simulates unsaturated, saturated subsurface, and surface rainfall-runoff processes. We first demonstrate the ability of the model to capture hydrograph dynamics and compare the model flow component and age simulations against measured values at the two sites. We show that the T-SAS approach can capture flow and transport dynamics for the right dominant process reasons. We then conduct a series of virtual experiments by switching soil depths between the two watersheds to understand how soil depth and its distribution control water age and source. Results suggest that thicker soils increase mean residence time and damp its temporal dynamics in response to rainfall inputs. Soil depth influenced the geographic source of streamflow, whereas pre-event water sources became more concentrated to near stream zones as soil depth increased. Our T-SAS approach provides a learning tool for linking the dynamics of residence time and time-space sources of flow at the watershed scale and may be a useful framework for other distributed rainfall-runoff models.