294 
Ecosystem Simulation System) is outlined. 
We explain and illustrate the flow of 
information from raw Thematic Mapper 
imagery, 30 meter digital elevation models 
(DEM) and digitized soils maps through the 
derivation of the important model parameter 
fields, extraction and definition of a terrain 
feature (object) model of a watershed and 
aggregation of the parameter fields into each 
discrete landscape object, and the final 
distributed simulation and reporting of 
results. 
As an example, we illustrate the 
parameterization and simulation of the North 
Fork of Elk Creek, a 17 sq.km, experimental 
watershed in the Garnet Range of western 
Montana. This watershed is in steep 
mountainous topography, with relatively thin 
Qolluvial soils. The mainstreams trend east- 
west, so that the major slopes are largely 
north or south facing, with the exception of 
more complex areas in the headwaters of the 
basin. The vegetation is dominated by dense 
conifer canopies on north facing slopes and 
more open canopies with a grass understory 
on south facing slopes, although previous 
logging activity prior to 1960 and more 
recent fires have influenced the forest pattern 
in some locations. 
MODEL DESCRIPTION 
As mentioned above, FOREST-BGC 
is a stand level model of forest ecosystem 
processes, driven by micrometeorologic and 
local soil conditions. While the main 
products of the model (in the form we are 
currently operating it) are ET and NPP, a 
host of intermediate and parallel products are 
computed, including the seasonal trajectories 
of soil water content, leaf water potential 
and runoff, which are very useful for 
validation purposes. In this respect the 
modeler has the opportunity to check the 
internal consistancy of the model with a 
suite of observations and measurements that 
can be made in the field, and not be limited 
simply to ’tuning’ the model to produce 
proper ET and NPP output. This sets the 
model apart from a number of empirical 
regression based approaches to the problem. 
FOREST-BGC has been previously 
described in detail by Running and Coughlan 
(1988) and is discussed here just in 
sufficient detail to present RESSYS. 
Required input data includes daily 
meteorogical conditions for a base station 
and site specific data for the forest stand, in 
this case, for each hillslope or portion of a 
hillslope. Site specific information includes 
topographic, soil and canopy information, 
specifically the aspect and gradient of the 
surface, SWC, LAI and elevation. A semi- 
empirical model, MT-CLIM, extrapolates the 
base station climatic data using accepted 
principles of mountain meteorology by 
considering the elevation, aspect and 
gradient of the landscape units relative to the 
base station (Running, Nemani and 
Hungerford, 1987). The model calculates 
canopy interception, evaporation, 
transpiration, runoff, photosynthesis, growth 
and maintenance respiration. Precipitation is 
routed through the canopy and into the 
snowpack or soil water where it becomes 
available for root uptake. Leaf transpiration 
is calculated with a Penman-Monteith 
equation based on micrometeorologic 
conditions, LWP and LAI, and drives the 
uptake and conductance of soil water under 
the constraints of physiologic conductance 
and soil water potential terms. Canopy 
photosynthesis is a function of C02 
diffusion gradient, a radiation and 
temperature controlled mesophyll C02 
conductance, the canopy water vapor 
conductance, LAI and the daylength. 
Average canopy radiation is computed from 
Beers Law for shortwave radiation canopy 
extinction, with species specific physiologic 
parameters. 
Of significance to our 
parameterization strategy is the strong 
nonlinearity of FOREST-BGC under certain 
conditions. This tends to occur most often 
when one or more model parameters 
becomes limiting, such as the effect of soil 
water at the onset of moisture stress. As