Review and evaluation of remote sensing methods for soil-moisture estimation

Abstract

Soil-moisture information plays an important role in disaster predictions, environmental monitoring, and hydrological applications. A large number of research papers have introduced a variety of methods to retrieve soil-moisture information from different types of remote sensing data, such as optical data or radar data. We evaluate the most robust methods for retrieving soil-moisture information of bare soil and vegetation-covered soil. We begin with an introduction to the importance and challenges of soil-moisture information extraction and the development of soil-moisture retrieval methods. An overview of soil-moisture retrieval methods using different remote sensing data is presented-either active or passive or a combination of both active and passive remote sensing data. The results of the methods are compared, and the advantages and limitations of each method are summarized. The comparison shows that using a statistical method gives the best results among others in the group: a combination of both active and passive sensing methods, reaching a 1.83% gravimetric soil moisture (%GSM) root-mean-square error (RMSE) and a 96% correlation between the estimated and field soil measurements. In the group of active remote sensing methods, the best method is a backscatter empirical model, which gives a 2.32-1.81%GSM RMSE and a 95-97% correlation between the estimated and the field soil measurements. Finally, among the group of passive remote sensing methods, a neural networks method gives the most desirable results: a 0.0937%GSM RMSE and a 100% correlation between the estimated and field soil measurements. Overall, the newly developed neural networks method with passive remote sensing data achieves the best results among all the methods reviewed.