Using RADARSAT-2 polarimetric and ENVISAT-ASAR dual-polarization data for estimating soil moisture over agricultural fields

There are several challenges in estimating soil moisture from radar remote sensing over agricultural fields in eastern Canada. To begin, snow cover or frozen ground is observed from November to April. From April to May, agricultural activities (e. g., ploughing and sowing) change the surface roughness from week to week thereby limiting the applicability of multitemporal and multi-incidence angle approaches. Techniques using a priori information on surface roughness are difficult to apply since the type of crop often changes from year to year. Here, we present an approach using effective roughness parameters (i.e., effective root mean square height and effective correlation length) that are obtained using an empirical relationship (independent of the crop type) between the root mean square height and the correlation length. The effective parameters allow us to resolve the Integral Equation Model for observed incidence angle and backscattering coefficient in HH and VV polarizations (sigma degrees(HH) and sigma degrees(VV)) using a look-up table. An additional challenge is posed by the growth of vegetation that begins in May. Three-component decompositions and radar vegetation indices were used to characterize the vegetation in agricultural fields. Surface backscattering coefficients in HH and VV polarizations (sigma degrees(SURF_HH) and sigma degrees(SURF_VV)) were calculated using the decompositions. An improvement in estimates of soil moisture was observed with the use of surface backscattering coefficients for bare soil and sparsely vegetated fields instead of the total backscattering.