Sentinel-1 Imagery Incorporating Machine Learning for Dryland Salinity Monitoring: A Case Study in Esperance, Western Australia

Abstract

Due to the lack of a suitable theoretical model for simulating radar backscatter of soil based on salt content, we investigated a new method to exploit Sentinel-1 radar backscatters and polarimetric decomposition information for dryland soil salinity monitoring. Soil electrical conductivity (EC) was estimated using Sentinel-1 SAR imagery and field survey data combined with five machine learning models in Esperance, located in the southwest of Western Australia (SWWA). The performance of the five machine learning models was assessed and compared using the root-mean-square error (RMSE), the mean absolute error (MAE), and the correlation coefficient ($r$). The results revealed that the Random Forest Regression model (RFR) yielded the highest prediction performance ($\text{RMSE}=2.89\ S/m,\ \text{MAE}=1.90 S/m$, and $\mathrm{r}=0.81$) and outperformed the other models. It can be concluded that the intensity images of VV and VH polarization of SAR imagery have the potential to predict EC of soils in SWWA.