Assessment of Different Backscattering Models for Bare Soil Surface Parameters Estimation from SAR Data in band C, L and P

Abstract

Synthetic Aperture Radar (SAR) is the most widely used sensor for retrieving soil surface parameters. This study was performed in two steps. In the first step, estimated backscattering coefficients using three models, Oh, Dubois, and IEM, in three bands of P, L, and C and two polarizations of HH and VV were compared with those extracted from SAR data acquire from AIRSAR over LWREW experimental site located in southwestern Oklahoma with a sub-humid climate. The results showed that the Oh model in band C had the best accuracy in both polarizations (RMSE_HH=1.48 and RMSE_VV=1.1). Dubois and IEM models were appropriately accurate in band L; however, both were less accurate compared with the Oh model in band C. In the second step, ground truth measurements of soil roughness, dielectric constant, and correlation length were compared with the corresponding results of inversion backscattering models. Based on the findings, it was concluded that IEM performed better at estimating soil roughness with RMSE=0.37, while Oh more accurately assessed dielectric constant in all three bands and at depths of 0–3 cm and 3–6 cm. All results confirmed that band P was not appropriate for retrieving soil surface parameters using backscattering models compared with the bands of C and L.

Keywords:

• SAR remote sensing
• backscattering model
• soil moisture
• soil roughness
• correlation length