WHAT GOOD IS A VIRTUAL WATERSHED: STRIKING AN APPROPRIATE BALANCE BETWEEN COMPLEXITY AND REALISM WHEN SIMULATING FLOW AND TRANSPORT IN MOUNTAIN CATCHMENTS (Invited Presentation)

The transport of solutes in mountain catchments is difficult to simulate realistically because of the complexity of the systems. Homogenized models may offer low uncertainty but often cannot capture intricate features, like springs and seeps, or provide estimates of the spatial distribution of fluxes. Detailed heterogeneity models (either explicit or stochastic) add more realism to a model but data scarcity often means that the distributed parameter fields have a high degree of uncertainty. So, which is better and what good are they? The right answer always depends on the specific question being asked, but this presentation considers how the transport response from minimally calibrated, but physically realistic, “virtual watersheds” can be used to expand or condition site-specific models. The key concept is that surface and subsurface velocities exhibit a level of incrementally stationary correlation within a watershed, or its sub-basins. Correlations in a real system can be inferred from the virtual analogs, as long as plausible configurations of hydrogeological material are used, so a physics-based virtual watershed can be thought of as a statistical analog to a real system. A spatial Markov model (SMM) of transport built from the analog can then use measured streamflow in conjunction with other direct observations to “translate” a transport model from one site to another. Elements of the new SMM model are still under development, but results using its core components already imply that realistic estimates of transport from analogs may soon be possible when exhaustive site characterization is not available.