987 
■adiative transfer code 
:re was no correlation 
surface temperatures 
still be expected. 
TC to retrieve surface 
thod, they found that 
as within 0.02 for all 
d over grassland and 
lly affected standard 
candidate vegetation 
effects. In a si m ilar 
rements to validate a 
AVI) computed from 
r ides the capability of 
lith angle. 
te the surface energy 
i Were combined with 
components. It was 
er on the radiometric 
easured sensible heat 
the measured values. 
. The remote sensing 
i 10 4 m. Differences 
antly with pixel size. 
boundary layer data 
al. (1994a) attempted 
only remotely-sensed 
uniform aerodynamic 
ere extrapolated to a 
3te sensing to provide 
means of estimating 
i to estimate the soil 
at TM measurements 
getation temperature. 
1 to estimate sensible 
use the infrequently- 
lable meteorological 
growing season with 
e, humidity and wind 
>th latent and sensible 
ited over the depth of 
ch were compared to 
:ed by the integrated 
mting for large scale 
3ES data with a solar 
al stations inside and 
stantaneous" satellite 
rences in daily means 
while five day means 
3_3. Assessment of Surface Characteristics (box 3 of Figure 1). 
3.3.1. Quality and Variability of Ground Truth Data. During Monsoon’90 and WG’92, latent heat flux was 
estimated by making measurements of three of the four energy balance components (the residual approach), 
namely net radiation (Rn), soil heat flux (G) and sensible heat flux (H). Rn and G were measured using 
conventional instrumentation at each site, and H was determined by the variance method based on measurements 
of air temperature and wind speed at one height. This technique satisfied the requirements for low-maintenance 
and continued operation during each experiment. Research was conducted to compare energy balance 
components measured using this approach with more traditional eddy correlation techniques. Stannard et al. 
(1994) found that soil heat flux (G) measured by the different systems showed significant variability which was 
mainly attributed to measurement errors and differences in technique used in estimating the heat storage in the 
top 5 cm. The ratio of sensible (H) to latent (LE) heat fluxes, H/LE or the Bowen ratio, showed increasing 
values going from eddy correlation tower to the gradient-measuring system. Kustas et al. (1994a) computed the 
latent heat fluxes by having estimates of three of the four energy balance components (the residual approach), 
namely Rn, G and H. The H-values were determined indirectly by the variance method. Comparison with the 
eddy correlation measurements of H and LE at two sites with good fetch showed agreement between the two 
techniques was within 20% for daytime conditions. 
3-3.2. Inferring Geophysical and Biophysical Parameters Via Remote Sensing Information. Several 
investigations addressed the relation between surface parameters and remotely sensed data acquired in the visible, 
near-IR, thermal and microwave wavelengths. One study utilized vegetation indices (including NDVI) and 
obtained surface information useful in modeling the water and energy balance. Moran et al. (1994a) found that 
a soil-adjusted vegetation index (SAVI) from ground-, aircraft- and satellite-based sensors was highly correlated 
with temporal changes in vegetation cover and biomass. On the other hand, quantifying the spatial variability 
in these quantities across WGEW was less successful. Batchily et al. (1993) analyzed the temporal sequence of 
eight TM images acquired during WG’92 to investigate existing and improved vegetation indices (Vis) to extract 
spatial and temporal vegetation information. They found that most Vis yielded different results and 
interpretations among the various ground covers and thus the choice of VI for vegetation monitoring and 
assessment in semiarid regions became critical. 
A detailed study of the variation in surface temperature and emissivity was conducted by Humes et al. 
(1994b). From ground observations, they found differences between vegetation and soil background temperatures 
were typically 10° to 25°C near midday. In addition, they discovered from ground and aircraft data that the 
spatial variability in surface temperature at local and basin scales were similar, suggesting under certain 
conditions spatial variability in surface temperature may be scale independent. The mixture of the soil and 
vegetation at most sites yielded an average emissivity of about 0.98. 
There were also techniques employed for estimating vegetation height and cover. Weltz et al. (1994) utilized 
an airborne laser system for measuring landscape patterns over large areas as a way to determine mean vegetation 
height and cover. Estimates of vegetation height and cover obtained from laser flights were compared to ground- 
based line-transect methods. The agreement was quite good. Furthermore, the laser data provided the ability 
to separate and map distinctly different plant communities. In another study, estimates of the local and regional 
scale aerodynamic roughness for WGEW came from the laser data analyzed by Menenti and Ritchie (1994). 
Values compared well to estimates obtained from techniques using micro-meteorological measurements. 
The utility of remote sensing for mapping surface soil moisture was tested by Schmugge et al. (1994) and 
Jackson et al. (1992) with passive microwave data collected from the PushBroom Microwave Radiometer 
(PBMR) and the Multifrequency Microwave Radiometer (MMR), respectively. The brightness temperatures were 
highly correlated to ground-based surface soil moisture measurements. They also discovered that changes in the 
microwave brightness temperatures after a rainfall were highly correlated to the amount of rain, up to a certain 
threshold value. Kustas et al. (1993) found a high correlation between microwave brightness temperatures and 
the evaporative fraction (EF) when the contribution of soil evaporation was significantly affecting the magnitude 
of EF. Under dry conditions, the spatial variability in EF was related to SAVI. 
3.4. Discussion of Monsoon’90 and WG’92 Results 
Preliminary results from the Monsoon’90 and WG’92 Experiments have addressed many of the issues involved 
in the development of a distributed hydrologic model (Figure 1). Substantial progress was made in the utilization 
of remotely-sensed data to infer surface characteristics (boxes 3 and 4). The effects of atmosphere and sun/sensor 
geometry on sensor signal response in optical spectrum were quantified and operational techniques for correction 
of these effects were proposed. With these corrections, optical data had potential for estimation of such surface