Table 1: Average Surface Soil Moisture Estimates for the WD38 Subwatershed. 
461 
Date 
PBMR (%) 
SAR (%) 
Ground Data 
July 10 
13 
14.5 
12.0 
July 13 
- 
22.9 
25.1 
July 15 
23 
24.0 
25.0 
July 17 
26 
25.1 
22.8 
July 18 
19 
- 
20.8 
July 19 
19 
- 
19.7 
July 20 
- 
- 
17.5 
The microwave sensors flown during the campaign included the passive push broom microwave 
radiometer (PBMR), and the active synthetic aperture radar (SAR). The PBMR operates at L-band 
(/ =1.42 GHz) and has four horizontally polarized beams pointing at ±8° and ±24° from nadir. 
The cross track resolution of the PBMR is approximately 90 m. The SAR is a full-polarization 
radar whose central frequencies are 0.44, 1.25 and 5.33 GHz respectively. The azimuthal and slant 
range resolutions of the processed images are 12.1 m and 6.662 m under the normal mode. To allow 
for intercomparisons among data from different sources, all images are registered with reference to 
the DEM by using a first-order polynominal registration procedure. Geo-referenced images are then 
resampled to the resolution of the DEM using a bilinear interpolation scheme (Lin et al., 1993). 
To extract soil moisture from remote sensing imagery, we employed the PBMR inversion algo 
rithm developed by Jackson (1992). The procedure used in estimating the vegetation correction 
parameters was described in Wood et al. (1993). For the SAR, we used the regression relationships 
developed by Lin et al. (1993) based on concurrent radar and ground measurements. Table 1 lists 
the dates of data collection and the estimated soil moistures from the two sensors over the WD38 
watershed. Averages from the ground measurements are also listed for comparisons. 
3.2 Model Parameters and Initial Conditions 
The soil-water retention properties within the simulated region are estimated from the soil map 
based on the results of Loague and Freeze (1985). The catchment average value of K 3 is equal 
to 0.062 m/hr. The parameter / is determined following Beven’s (1982) suggestion for silt loam 
soils which occupy the majority of the studied watershed. Distribution of the topographic index is 
calculated from the DEM data using an algorithm similar to that presented in Burrough (1986). The 
mean and standard deviation of the ln(a/tan/3 ) distribution are 7.5 and 1.71, respectively. Base flow 
parameter Q 0 is estimated to be 0.001 m 3 /s. Computational time step for this simulation is set to 
be 30 minutes. 
The determination of initial soil wetness condition has been handled in alternative ways by various 
researchers. Two different techniques are used in this study to estimate the initial average water table 
depth zb. The first method, developed by Troch et al. (1993a), is based on Boussinesq’s groundwater 
equation and uses the streamfiow records at the catchment’s outlet. Using a lower envelop excluding 
5 % of the data and an average drainable porosity of 0.04, the zb is estimated to be 2.31 m for the 
WD38 subwatershed. The second initialization technique estimates zb by using the July 10’s surface 
soil moisture measured by the PBMR, with the assumption of Salvucci’s (1993) steady state soil 
moisture profile and the seasonal mean percolation rate. The resulting zb is nearly 1 m deeper than 
that obtained by using the first initialization technique. 
Given the zb, the initial soil moisture profile in the unsaturated zone for each pixel can be 
determined from Eq.(9) and Eq.(12). The depth of the surface zone is related to the penetration 
depth of the remote sensors. For the PBMR and the SAR, the penetration depth is estimated to be 
approximately 5 ~ 20 cm, under the MAC-HYDRO’90 land cover and soil moisture conditions. A 
spatially-constant value of 15 cm is used in this simulation.