986 
remote sensing data. Perry and Moran (1994) combined radiosonde measurements with a radiative transfer code 
(RTC) to correct satellite-based surface temperature measurements. Results showed that there was no correlation 
between radiosonde location or time, and resulting temperature corrections. The corrected surface temperatures 
from aircraft and satellite altitudes suggest that errors in excess of 2°C can still be expected. 
In a similar study, Moran et al. (1994a) combined solar radiometer measurements with an RTC to retrieve surface 
reflectance from satellite digital data in the visible and near-IR spectrum. Using this method, they found that 
the difference between aircraft- and satellite-based measurements of surface reflectance was within 0.02 for all 
bands on most dates. 
Qi et al. (1994) demonstrated that the bidirectional reflectance distributions measured over grassland and 
desert shrubs were significantly influenced by sensor/sun geometry. This substantially affected standard 
vegetation indices like NDVI (Normalized Difference Vegetation Index). Yet, several candidate vegetation 
indices suggested by the authors demonstrated great potential for minimizing view angle effects. In a similar 
vein, Chehbouni et al. (1994) utilized ground-based multiple view direction/angle measurements to validate a 
semi-empirical model that normalizes the Modified Soil Adjusted Vegetation Index (MSAVI) computed from 
any view angle to nadir. Comparison with the observations suggested that the model provides the capability of 
predicting nadir viewing values of MSAVI from any off-nadir values at a given solar zenith angle. 
3.2.2. Modeling Hydrologic and Surface Energy Fluxes. Most of the efforts to evaluate the surface energy 
balance utilized data collected in the optical wavebands. Ground-based remote sensing data Were combined with 
conventional meteorological data by Moran et al. (1994b) to compute the energy balance components. It was 
found that an additional resistance term accounting for the effect of partial vegetation cover on the radiometric 
temperature was needed in order to obtain satisfactory agreement between modeled and measured sensible heat 
flux. In general, flux estimates from the remote sensing model were within 10-15% of the measured values. 
A similar approach was taken by Kustas et al. (1994b) with low flying aircraft observations. The remote sensing 
data were averaged over a range of length scales to represent pixel sizes of order 10 2 to 10 4 m. Differences 
between the modeled and measured fluxes were less than 20% and did not vary significantly with pixel size. 
Similar results for estimating regional scale energy fluxes were obtained with atmospheric boundary layer data 
and remote sensing data averaged over the study area. In a related approach, Humes et al. (1994a) attempted 
to extrapolate energy fluxes evaluated at a reference site to other locations in WGEW using only remotely-sensed 
inputs. The analysis indicated that significant errors can result due to the assumptions of a uniform aerodynamic 
resistance and incoming radiation, both of which were violated when flux estimates were extrapolated to a 
different ecosystem and when there were partly cloudy skies. 
Moran et al. (1993) utilized the WG’92 data set to combine optical and microwave remote sensing to provide 
simultaneous vegetation and soil information for input to an energy transfer model as a means of estimating 
surface energy flux. They found that synthetic aperture radar (SAR) data could be used to estimate the soil 
temperature of the rangeland. Using SAR-based estimates of soil temperature with Landsat TM measurements 
of surface temperature and spectral vegetation indices, it was possible to estimate vegetation temperature. 
Vegetation, soil and air temperatures were introduced in a two-layer energy balance model to estimate sensible 
heat flux with reasonable accuracy. Maas et al. (1993) showed that it was possible to use the infrequently- 
obtained remotely-sensed information (monthly during WG’92) and routinely-available meteorological 
observations to simulate daily evaporation rates and biomass production over the entire growing season with 
reasonable accuracy. 
Basin scale energy fluxes were also evaluated using atmospheric profiles of temperature, humidity and wind 
speed in the lower troposphere from radiosonde data analyzed by Hipps et al. (1994). Both latent and sensible 
heat fluxes were determined using the conservation equations for heat and moisture integrated over the depth of 
the atmospheric boundary layer from a series of soundings. The results with this approach were compared to 
averages given by the meteorological network. The agreement between fluxes estimated by the integrated 
conservation equations and the meteorological network was satisfactory only after accounting for large scale 
advection. 
Basin scale estimates of the insolation were computed by Pinker et al. (1994) using GOES data with a solar 
flux inference model. The model-derived values were compared to averages from several stations inside and 
outside WGEW. For a clear day, differences between 5-minute ground data and "instantaneous" satellite 
estimates were within 3%. For a partly cloudy case, the agreement was not as good. Differences in daily means 
derived from the satellite and measured by the meteorological network were within 10%, while five day means 
were within 3% of measured.