Scaling up From Heterogeneous Soils to Heterogeneous Catchments.

Despite their central importance to hillslope hydrology, predictions of hillslope discharge to streams that are right ``for the right reasons" have proved exceedingly difficult. Finding an appropriate, general framework for doing so would have a transformative impact on hydrologic forecasting of stream flow; water supply; and fluxes of pollutants, nutrients, pathogens and sediment. A primary obstacle to such predictions is the heterogeneity in landscape properties within a hillslope, and its interaction with the boundary conditions operating on the hillslope, such as the spatiotemporally varying climatic inputs, and connection to receiving landscape units. In this talk I will discuss recent progress towards an understanding of these interactions in saturated subsurface lateral flow. In particular I will discuss A) efforts to develop a similarity framework that can be used to understand the relative roles of internal structure and boundary conditions in controlling hillslope dynamics B) observations of the role of internal heterogeneity in controling transient discharge from hillslopes and C) efforts to develop an upscaled model of hillslope flows that accounts for the effects of heterogeneity based on a continuous time random walk (CTRW).

The importance of understanding the landscape as a dynamic filter of hydrologic variability will be emphasized. The physical structure of landscapes, including unresolved spatial heterogeneity, alters the variability of water flows generated by the precipitation in a manner analogous to a signal filter. This is the origin of unit hydrograph and other linearized approaches to runoff prediction. However, in certain circumstances of subsurface lateral flow, the flow itself is able to generate potential gradients that alter the local flow direction towards areas of high conductivity, and these lead to the reorganization of flowpaths through the hillslope as a whole. It is shown that this reorgainization makes the hillslope response highly sensitive to the connection with the downslope boundary (as expressed in the boundary condition), and reduces our ability to make inferences about the variability of residence times in the hillslope from the discharge signal alone.