Sentinel-1 Backscatter and Interferometric Coherence for Soil Moisture Retrieval in Winter Wheat Fields Within a Semiarid South-Mediterranean Climate: Machine Learning Versus Semiempirical Models

Abstract

This work aims to assess the effectiveness of machine learning (ML) algorithms and semiempirical models for surface soil moisture (SSM) retrieval by exploring the Sentinel-1 backscatter and interferometric coherence data. First, three commonly used categories of ML algorithms are evaluated using data gathered from diverse rainfed and irrigated wheat fields located in Morocco and Tunisia. Specifically, these algorithms include artificial neural network (ANN), deep neural network, three support vector regression (SVR) models [radial basis function (SVR_rbf), linear (SVR_linear), and polynomial (SVR_quad) kernels], and two tree-based methods [random forest and eXtreme Gradient Boosting (XGBoost)]. The comparison between predicted and measured SSM showed that the best retrieval results were obtained using Sentinel-1 data at VV polarization with R ranging between $ 0.68$ and $ 0.76$ and root-mean-square error (RMSE) of $ \text{0.05}\,\text{m}^{3}/\text{m}^{3}$ and $ \text{0.06}\,\text{m}^{3}/\text{m}^{3}$. Second, to further assess their transferability, the ANN, SVR_rbf, and XGBoost, which demonstrated the most favorable results from each category, were evaluated and compared against the coupled Water Cloud and Oh models (WCM), using a second dataset collected over a drip-irrigated wheat field in Morocco. Overall, the best retrieval results were achieved by ANN and SVR_rbf with R and RMSE of $ 0.81$ and $ \text{0.034}\,\text{m}^{3}/\text{m}^{3}$, respectively. In addition, their performances were consistent with that of WCM, which yielded R and RMSE values of $ 0.81$ and $ \text{0.04}\,\text{m}^{3}/\text{m}^{3}$, respectively. Finally, due to its good compromise between retrieval accuracy of SSM, processing time, and simplicity, SVR_rbf was chosen to generate high-resolution SSM maps from Sentinel-1 data over irrigated wheat fields.

Keywords

• Backscattering models
• irrigated and rainfed winter wheat
• machine learning (ML)
• semiarid regions
• Sentinel-1 backscatter and interferometric coherence
• surface soil moisture (SSM)