The objective of this research was to develop an improved understanding
of the requirements for remote sensing derived land cover classifications
and vegetation indices for modeling forest transpiration and net primary
production (NPP) in a heterogeneous area. Field measurements and models
were used to determine if species differentiation is required of remote
sensing data in support of modeling ecosystem processes. The study site
was in northern Wisconsin, USA (45.92°N 90.27°W) and covered an
area of 102 km. Much of the area is a mixture
of forest cover types dominated by trembling aspen (Populus tremuloides Michx.),
sugar maple (Acer saccharum Marsh.), red pine (Pinus resinosa Ait.),
speckled alder (Alms rugosa DuRoi), and white cedar (Thuja
occidentalis L.). A diurnal model of forest transpiration using 30-minute
time steps matched field measurements better than a daily model of transpiration.
However, when summed over the 57 days examined, both approaches matched
the field data well. Light use efficiency (LUE) derived from NPP and absorbed
radiation measurements differed significantly among forest cover types.
LUE was used to estimate NPP spatially based on aggregated land cover data
and spatial leaf area index estimates. There was a less than 10% difference
in mean NPP between using a species-specific land cover classification
and a biome level classification. Spatial NPP estimates were also sensitive
to spatial representation of leaf area index (LAI). The effects of land
cover heterogeneity and atmospheric aerosols on four different vegetation
indices were quantified. Differences between indices resulted in a variation
in NPP estimates of up to 21%. This dissertation shows that remote sensing
data should contain sufficient spatial and spectral information for species
identification and quantification of land surface parameters that reflect
local ecological functioning for use in validation and modeling studies.
Future research should extend the framework used here to other regions,
especially those with strong environmental gradients.