C- and Multiangle Ku-Band Synthetic Aperture Radar Data for Bare Soil Moisture Estimation in Agricultural Areas

A sensitivity analysis of C-band (5.3 GHz) and Ku-band Although conventional ground-based soil moisture mea(14.85 GHz) synthetic aperture radar (SAR) data to the surement techniques, such as the neutron probe, time bare soil moisture content of agricultural fields was condomain reflectometry, and gravimetric methods, are acducted in this study. The C-band data were obtained curate and permit measurements of the entire soil prowith a 238 incidence angle, whereas the Ku-band data file, they are essentially point-based measurements at a were obtained with 358, 558, and 758 incidence angles. specific time. On the other hand, remote sensing can The fields presented either a small-scale or an intermediprovide indirect estimates over large areas with frequent ate-scale periodic soil roughness components, associated repeat coverage. The microwave region (from 1 mm to with level-basin and furrow irrigation systems, respec1 m wavelength) has been the most favorable spectral tively. For fields with a small-scale roughness component, region for surface soil moisture studies (Schmugge, 1985; the SAR data were sensitive to soil moisture, particularly Engman and Chauhan, 1995). There is a large contrast at the C-band with a 238 incidence angle and Ku-band between the dielectric constant e of liquid water (~80) with a 358 incidence angle. For fields with a intermediand dry soil (3–5) within this spectral range, which makes ate-scale roughness component, both Cand Ku-band the synthetic aperture radar (SAR) data sensitive to soil data were nearly insensitive to soil moisture. By using a moisture content (Ulaby et al., 1986; Engman and Chautheoretical surface scattering model, this study also anahan, 1995). lyzed the effects of different soil roughness components For a bare soil surface, the radar backscattering pro[root mean square (RMS) height h, correlation length, cess is controlled by the dielectric property of the soil, and periodic row structure] in the SAR data. For fields which is related to the soil moisture content, and the soil with RMS height ,0.3 cm, a small variation in h (from random surface roughness (Ulaby et al., 1978; Bernard 0.1 to 0.3 cm) provoked a significant variation in the et al., 1982; Dobson and Ulaby, 1986; Bertuzzi et al., SAR data (up to 8 dB). Elsevier Science Inc., 1998 1992). One way to minimize the soil roughness influence is to use the copolarized ratio technique; that is, the ratio between HH and VV polarized waves. Autret et al. (1989) INTRODUCTION and Chen et al. (1995) reported that SAR data derived Estimation of soil moisture in agricultural areas is imporfrom this technique are almost completely independent of tant for improving yield forecasts, scheduling irrigations, soil roughness. However, these types of data have been and other farm management activities (Idso et al., 1975). collected by sensors mounted on airplanes or trucks and are available only at restricted experimental sites. No cur-