Retrieval of the soil parameters beneath a salt crust using Synthetic Aperture Radars (SARs) relies on the knowledge of the surface roughness configuration. It involves costly ground surveys or the use of informed guesses, whose accuracy is challenging to assess due to the lack of available information. We present here a fully remotely-sensed methodology that uses a combination of C- and L-band: (1) co-polarized channels at C- and L-bands for a roughness estimate and (2) dual polarized channels at L-band for the retrieval of the subsurface configuration. This SAR-based methodology is tested on an Andean salt flat where a dedicated ground survey was made. Results are promising since the use of roughness estimated for a synergy between C- and L-bands performed equally well (in terms of root mean squared error, RMSE) regarding the methodology using fieldwork estimates. The brine salinity and the water table depth both retrieved from the SAR-based methodology were overestimated with an RMSE error of 2 - 3 % and 4 - 6 ‰, respectively.