FOREST ECOSYSTEM PROCESSES AT THE WATERSHED SCALE - INCORPORATING HILLSLOPE HYDROLOGY

An approach to distributed modeling of watershed hydro-ecological processes over large spatial scales is described. A data and simulation system, RHESSys (Regional HydroEcological Simulation System), combines a set of remote sensing/GIS techniques with integrated hydrological and ecological models in order to automate the parameterization and simulation of a suite of ecological and hydrological flux and storage processes through the watershed. Specifically, we simulate forest canopy net photosynthesis (PSN) and total evapotranspiration (ET) through the year with a modeling package that integrates FOREST-BGC, a stand level model of forest carbon, water and nitrogen budgets, with TOPMODEL, a quasi-distributed hydrological model. The latter model introduces the effects of hillslope hydrological processes, incorporating surface redistribution of soil water by saturated throughflow processes. The maintainance of regular patterns of soil water by throughflow processes cause forest ecosystem activity to vary dependent on hillslope position. The distributed framework is based on a terrain partition in which each terrain object (hillslopes and stream reaches) comprising the watershed are separately parameterized and simulated. The location of each terrain object within the watershed is explicitly represented while the internal variability of each object is represented as a joint parameter distribution. Generalization of the surface into different numbers of terrain objects (by growing or shrinking the extent of the stream network) is automatically accomplished using digital terrain data. This allows us to flexibly alter the spatial representation of the watershed by shifting surface information either into the internal hillslope parameter distributions or into greater numbers of hillslopes (and stream reaches). Limited simulations of a mountainous watershed in western Montana indicate that incorporation of within hillslope throughflow and the soil, topography and vegetation distribution has the effect of significantly altering the seasonal PSN and ET trends in comparison with lumped surface representations without lateral water flux.