| Summary: | Soils play a key role in various hydrological and meteorological applications. The objective of this paper is to analyze the spatial variability of very high resolution (3.3 m) RISAT-1 (5.35 GHz) data with surface soil parameters to produce soil moisture retrieval model. The behaviors of the RISAT-1 signal were analyzed for two configurations, RH and RV at high incident angle (48.11°), with regard to several soil conditions: volumetric moisture content (Mv), root mean square surface roughness (r.m.s.) and soil composition (texture). The relationship between the backscattering coefficient (σ°) and the soil parameters (moisture, surface roughness and texture) was examined by means of satellite images, as well as ground truth measurements, of each of the 23 plots, recorded during several field campaigns in the January 2015. RISAT-1 images demonstrate high potential for the identification of local variations of soil dielectric constant (ɛ), texture and Mv. σ° has a positive relationship with Mv both for σ° (RH) and σ° (RV) with R2 = 0.588 and R2 = 0.525. The roughness component was derived in terms of r.m.s. having a positive correlation with σ° (RH) (R2 = 0.009) and σ° (RV) (R2 = 0.029). Dielectric constant (ε) has a positive relationship with σ° (RH) having R2 = 0.656 and σ° (RV) having R2 = 0.534. By considering all the major influencing factors (σ° (RH), σ° (RV), ε and r.m.s.) a semi-empirical model has been developed, where Mv is a function of σ° (RH), σ° (RV), ε and r.m.s. This model has adjusted R2 = 0.956 and RMSE = 0.010 at 95% confidence level.
|
|---|