Science Assessment

Although there have been a number of space borne microwave instruments since the late 1970's, antenna technology and the need to accommodate requirements of the atmospheric and ocean sciences have resulted in these sensors operating at frequencies higher than what is deemed optimal for soil moisture estimation (6.9, 10, 19, 37, and 85 GHz). Research in soil moisture remote sensing began in the mid 1970's shortly after the surge in satellite development (Barton, 1978; Eagleman and Lin, 1976; Idso et al., 1975; Njoku and Kong, 1977; and Schmugge, et al., 1977). Theoretical and empirical evidence suggest that the 1400-1427 MHz region (L-band) is most suitable for soil moisture radiometry because long wavelengths penetrate soil and vegetation to a greater extent than higher frequencies (Jackson, 1993; Njoku and Entekhabi, 1996) and transmission is prohibited in the band by the Federal Communications Commission so that it remains suitable for radioastronomy observations (cosmic hydrogen absorption occurs at 1420 MHz).

At 1400 MHz, there is a large contrast between the dielectric properties of liquid water (~80) and dry soil (< 4). The dielectric properties of wet soil were studied by several investigators (e.g., Wang and Schmugge, 1980; Dobson et al., 1985; and Ulaby et al., 1986). As the moisture increases, the dielectric constant of the soil-water mixture increases and this change is detectable by microwave sensors (Njoku and Kong, 1977). The microwave brightness temperature of an emitter of microwave radiation is related to the physical temperature of the source through the emissivity. The depth through which energy is emitted and sensed by microwave radiometers has been the subject of research and discussion for many years, but is on the order of about 5 cm at L-band.

Soil moisture can be estimated using radiometer or radar measurements. Both radio-brightness and backscatter measurements have been shown to be sensitive to soil moisture. From dry to saturated conditions, radiobrightness can vary by as much as 150 K and backscatter can vary by up to about 7 dB. Experiments conducted over the past two decades have developed a basic operational approach whereby brightness temperature is correlated with soil moisture through calibration experiments. The primary relationships between surface emissions and observed brightness temperatures T_Bp at polarization p (V or H) can be expressed as

where T_s and T_c are the physical temperatures (K) of the soil and vegetation canopy, τ_c is the vegetation opacity along the slant path, ω is the single scattering albedo, and r_p is the soil reflectivity (at look angle θ). The reflectivity is related to the emissivity by e_p=(1-r_p). One approach to estimating vegetation opacity is through its relationship to the columnar vegetation water content W_c (kg/m²) given by

where b is a coefficient that depends on vegetation type (Jackson and Schmugge, 1991; van de Griend and Wigneron, 2004). The surface reflectivity is related to the soil dielectric constant ε by the Fresnel equations, with modifications for surface roughness (Wang and Choudhury, 1981). This relationship has been referred to as the τ-ω model. This technique can be applied successfully if other factors known to affect brightness temperature, such as instrument configuration and target characteristics (i.e., soil texture and temperature, surface roughness, and vegetation), are invariant for a particular locality (Schmugge et al., 1980; Schmugge, 1983; Engman and Chauhan, 1995). The spatial variability of target characteristics from one locality to another, and even within a single instrument footprint, complicates the application of this technique.

More recently, polarization difference indices are proposed to explain the relationship between differences in polarization signals and the soil dielectric properties (and therefore soil moisture). Owe et al. (2001) and Meesters et al. (2005) proposed a relationship between the index, called microwave polarization difference index (MPDI) to the emissivities. The MPDI is given as

and the vegetation optical depth is given by

where a and d are given as

and

The advantage of this analytical technique is that it only depends on the observed V and H polarized observations thereby eliminating the need to estimate the vegetation water content and the associated b parameter. In active growing season, vegetation water content is a highly dynamic parameter and therefore any estimate (quantitative or inferred) is fraught with estimation error.

For radar, the total copolarized backscatter from the surface is the sum of three components

The first term is the soil surface backscatterer, σ ^s_pp, modified by the two-way attenuation through a vegetation layer of opacity τ_c (along the slant path at look angle θ, and assumed unpolarized as for the passive case). The second and third terms represent the backscatter from the vegetation volume σ ^vol_pp and the interaction between the vegetation and soil surface σ ^int_pp, respectively (Ulaby et al., 1996). For bare or surfaces with little vegetation, the σ ^s_pp contribution dominate the received signal and are influenced primarily by the soil moisture and surface roughness. The backscatter associated with vegetation is complex as it is influenced by the geometry and orientation of leaves, stalks and stems.

In an Observing System Simulation Experiment within the Red-Arkansas River basin in the Southern Great Plains, Crow et al. (2005) evaluated three different soil moisture algorithms and concluded that retrieval errors of less than 3% VSM would be consistently met using either of three algorithms. However, where simulated vegetation water content (VWC) values were greater than 3 kg/m², a significant bias in soil moisture estimates occurred. Also, we believe that some of the assumptions made in that study regarding accuracies in input parameters were overly optimistic so that the asserted accuracy of less than 3% VSM may be too low, particularly in wet areas and those with actively growing vegetation cover.