Parameter Lumping Implications to Hydrologic Simulations of a Post-Glacial Landscape

Abstract:

Even hydrologic models commonly described as being “fully distributed” require some degree of parameter lumping. This is typically necessary because of limited availability and resolution of watershed data, including precipitation, surface flows, soil distributions, land cover and drainage network hydraulic characteristics. The extent to which parameters are lumped is most often a decision made by the investigator with attention to research objectives, data availability, and model capabilities. Here we summarize the implications of parameter lumping to model efficiency using a distributed hydrologic simulation platform (MIKE SHE) applied to the Northwest River watershed in southern Maine, U.S.A. We compare parameter lumping effects on surface flows at the watershed outlet by varying the spatial scale (grid size), precipitation time step, computational time step, and agglomeration of landscape characteristics. The outcomes of these analyses are summarized relative to model efficiency with discussion of the implications to watershed management decisions and strategies.