) provided two-dimensional maps of surface soil moisture
(Schmugge et al., 1992). A sloping slab model was
used to simulate the eight day recession flow after a storm.
The simulated soil moisture and streamflow compared well
to the observations. Engman et al. (1989) concluded that the
remotely sensed soil moisture has potential to verify model
performance as well as identify areas of the basin that are
mainly contributing to the baseflow. Famiglietti et al.
(1992) evaluated a water balance model for the King's Creek
catchment at FIFE. The catchment, which is approximately
10 km
in size, contained a network of 20 raingages and a
flux station. The model utilizes spatial information on
topography, soils and precipitation for computing spatially
distributed hydrologic fluxes. The heterogeniety in land
surface moisture conditions lead to a range in the magnitude
of the surface fluxes computed over the catchment. Comparison
with the flux station in the King's Creek watershed
looked promising, but the evaporation at the catchment scale
could not be validated since a network of surface flux
measurements were not made inside the experimental
watershed. However, an average evaporation of seven flux
stations within the FIFE domain showed reasonable agreement
with model-derived catchment scale values.
Remotely sensed soil moisture and ground observations
from the MAC-HYDRO '90 experiment were compared to
a distributed hydrological model by Wood et al. (1993) over
a 0.5 km
subwatershed. The distributed hydrological model
of Paniconi and Wood (1993) was used to simulate the
spatial and temporal variation in soil moisture. The average
soil moisture from the passive microwave radiometer and
synthetic aperture radar and ground observations for the
subwatershed were in acceptable agreement; but the model
estimates were consistently higher. The model, however,
did reproduce the temporal trend over a ten day period. The
cause of the bias in the model output could not be explained.
The sensitivity of runoff simulation to initial soil water
content was investigated by Goodrich et al. (1994) using
Monsoon '90 data. The KINematic Runoff and EROSion
model (KINEROS; Woolhiser et al., 1990) was run for a
small (0.044 km
) and a medium sized (6.31 km
)
subwater-shed in the USDA-ARS Walnut Gulch Experimental Watershed
using several estimates of pre-storm initial soil water
content. The impacts of various spatial averages of soil
water on runoff simulations were analyzed. At the scale of
a small and medium sized catchment, a basin wide average
value of initial soil moisture from passive microwave remote
sensing data was adequate for the runoff simulations. This
result suggests that it may be possible to use satellite-based
microwave measurements to define pre-storm soil moisture,
even though such L-band (21 cm wavelength) passive
microwave sensors would have low resolutions (Jackson and
Schmugge, 1989). However, at both the small and medium
sized drainage basins, a relatively detailed spatially distributed
input of rainfall is needed in this semiarid environment in
order to achieve reasonable runoff predictions.