Modelling infiltration and infiltration excess: The importance of fast and local processes

Abstract Physically based distributed hydrological models aim at an adequate representation of hydrological processes, including runoff generation. A significant proportion of runoff is generated through the subsurface, that is, by groundwater flow or unsaturated subsurface stormflow. However, in the case of high rainfall intensity and/or low soil‐surface infiltrability, surface runoff may strongly contribute to total runoff, too, either through saturation excess (“Dunne‐type surface runoff”) or infiltration excess (“Hortonian surface runoff”). Both types of surface runoff can be rather important if antecedent wetness is high and parts of the catchment area are saturated (leading to saturation excess), or if the maximum infiltration rate into the soil surface is less than the actual rainfall intensity (resulting in infiltration excess). Even though the latter process can be very important during high‐intensity rainstorms, both for flood generation and for matter transport linked with surface runoff, an appropriate consideration of this process in catchment models is still challenging. Actually, budgeting between the actual rainfall intensity and the soil surface infiltration capacity is required and there are a number of challenges in the details: First, the “real” rainfall intensity may vary tremendously in time increments much smaller than the time step of the model. The soil surface infiltrability can also be significantly reduced, for example, by crusting, compaction or sealing of the soil surface or through hydrophobic effects. Otherwise, soil infiltrability can be strongly enhanced as a consequence of preferential flow paths/macropores caused by, for example, bioturbations or other voids. Finally, there is a high variability of such soil surface features at a small spatial scale, below the typical spatial modelling unit. We present observational data and approaches to deal with these challenges. We show results from combined infiltration/infiltration‐excess experiments and observations at three spatial scales. Then, we present a model approach based on a double‐porosity soil enabling the combined modelling of high infiltration rates and dampened soil moisture distribution after termination of infiltration, as observable in the field. Furthermore, we present an approach to model the effects of soil surface conditions on actual infiltration capacity. These approaches improved the plausibility and explanatory power of the model concerning surface runoff generation and soil moisture dynamics. For instance, model results at the plot and hillslope scales show that it is possible to simulate high infiltration rates jointly with a relatively slow movement of moisture within the soil matrix, field phenomena often observed in the case of heavy rainfall. We also simulate non‐linear and space–time variability effects of soil surface conditions, which can be important for flood generation.