Soil Moisture Mapping Introduction

Soil moisture is an important parameter for many natural resource applications such as hydrological modeling, stream flow forecasting, and flood forecasting. SAR data are well suited for estimating soil moisture due to the dependence of the dielectric constant on moisture at these frequencies. As described in the work of Dobson et al.Dobson & Ulaby , for a given soil condition (roughness or texture) radar backscatter was found to be linearly dependent on volumetric moisture (mv ) in the upper 2 to 5 cm of soil with a correlation r double tilda 0.8 to 0.9 (sigma, zero degrees = A+Bmv) .

The presence of vegetative cover introduces another level of complexity to soil moisture mapping due to the interaction of the microwaves with the vegetation and soil. Depending on the amount of vegetation present, its dielectric properties, height and geometry (size, shape and orientation of its component parts) the sensitivity of microwave backscatter to volumetric soil moisture may be significantly reduced. The ability to effectively map soil moisture can be improved by judicious selection of imaging parameters such as incidence angle, wavelength and polarization.

Imaging at steep incidence angles is often chosen to minimize the contributions to backscatter of soil roughness and attenuation associated with above ground biomass. Backscatter is significantly affected by surface roughness at incidence angles beyond about 40°. Hence, imaging at lower incidence angles is recommended for soil moisture estimation.

C- Band in the HH polarization was found to be most sensitive to soil moisture and least sensitive to surface roughness in the presence of low biomass. In agricultural fields, as the vegetative component over the soil increases, longer wavelengths (e.g., L-band ) are needed to permit continued monitoring of soil moisture during the growing season Schmullius & Furrer Bouman & Hoekman. For shrubby or forest covered areas only longer wavelengths such as L-band or better yet P-band provide the penetration necessary for soil moisture estimation. Polarimetric data may help by reducing and/or accounting for the effects of roughness and/or vegetation on the soil moisture estimate. Polarization Dependence

The ability to estimate surface soil moisture (for depths from 0 to 2.5 cm) using various polarizations and polarimetric parameters of SIR-C data was reported in Sokol et al. The data were collected over bare soil surfaces in Southern Manitoba during April and October 1994. Evaluation of the polarimetric data was confined to data at incidence angles from 33 o to 38o. Polarimetric parameters examined included synthesized linear and circular polarizations, total power, Co- and Cross-polarization ratios, pedestal height, and Co-polarization Phase Differences (Table 9-1)

Data acquired in both the HH and VV polarizations were highly correlated with soil moisture (r = 0.86 - 0.87), whereas those at HV were more poorly correlated (r =0.71). Multiple regression analysis using various combinations of linear polarizations showed no significant improvement in soil moisture estimation.

The co- and cross polarization ratios were not as effective for soil moisture estimation as the data in HH or VV polarizations, although they have been used successfully elsewhere to help reduce the impacts of soil roughness and vegetation for data acquired at shallower incidence angles Oh et al 1992.

The mean Co-polarized Phase Difference (r = -0.35) was not significantly correlated to soil moisture. This parameter is often used to differentiate between scattering mechanisms, which in this case were relatively invariant and indicative of surface scattering (single bounce). Therefore, the low correlation was not unexpected.

The SIR-C data obtained in southern Manitoba show that information in the images at various polarizations and polarimetric parameters are highly inter-correlated (Table 9-2). Backscatter in HH, VV, and RL showed highest correlation with soil moisture.

Table 9-1. Correlation between radar backscatter and surface (0-2.5 cm) soil moisture (from Sokol et al).

C-Band Polarimetric Parameter Correlation Coefficient (r)
Simple Linear Correlation Results
HH Backscatter 0.86*
VV Backscatter 0.87*
HV Backscatter 0.71*
Total Power 0.87*
Co-Pol Ratio (HH/VV) 0.53*
Cross-Pol Ratio (VV/HV) -0.79*
Cross-Pol Ratio (HH/HV) -0.74*
Co-Pol Pedestal Height 0.82*
RL Backscatter 0.88*
RR Backscatter 0.68*
Co-Pol Phase Difference -0.35
Multiple Linear Correlation Results
HH + VV 0.87*
HH + HV 0.86*
VV + HV 0.87*
HH + VV + HV 0.79*

* statistically significant at p < 0.05

Table 9-2. Correlations between field averaged backscatter recorded for each linear and circular polarization on bare fields (from Sokol et al).

  HH VV HV Pedestal Height RL
VV 0.99*        
HV 0.86 0.86      
Pedestal Height 0.94 0.92 0.94    
RL 0.98 0.98 0.85 0.91  
RR 0.73 0.77 0.90 0.76 0.78

* correlation (r) coefficients

Examples of Co-polarization Signatures for backscatter from wet and dry soils are shown in Figure 9-13. It was found that for wet soils, where little penetration into the soil occurs, the intensity is highest in VV and the pedestal height is low (0.2) indicating a smooth surface with surface scattering predominating. For drier soils the maximum at VV is no longer present although the pedestal height is still 0.2 indicating a smooth surface. The increased microwave penetration of the soil under dry conditions accounts for the similarity of the responses in the HH and VV polarizations.

Figure 9-31

Figure 9-31


Figure 9-31. Co-polarization signatures from SIR-C data for, a) wet soils (30.5% moisture, surface roughness (rms) 17.4 mm) and, b) dry soil (17.7% moisture, surface roughness (rms) 13.2 mm (from Sokol et al).