Chapter 6 - Machine learning for soil moisture assessment

Abstract

Soil moisture plays a key role in the Earth’s hydrological cycle and meteorological and climatic processes. The information on soil moisture content is required for irrigation scheduling, crop yield prediction, studies on weather and climate change, monitoring and forecasting extreme weather events like floods and drought, and estimation of runoff and soil erosion. The accurate and timely estimation and forecasting of soil moisture are necessary for these applications. Machine learning (ML) algorithms, like artificial neural networks, support vector machines, decision trees, random forest, and so on, are widely used for soil moisture assessment due to their ability to model nonlinear and complex relationships between variables. These algorithms are used to develop pedotransfer functions that can predict soil hydraulic properties, like available water capacity, hydraulic conductivity, soil water retention curve, and more. These algorithms are also used for the retrieval of soil moisture through remote sensing. By providing meteorological, vegetation, topographic, and historical input data about soil moisture variation, these ML algorithms can accurately forecast soil moisture after a few days. This information can be used for scheduling irrigation in the automated smart irrigation system. These algorithms are also extensively used for downscaling coarse resolution satellite-derived soil moisture products to finer spatial resolutions so that these products can be applied at the regional or watershed level. ML algorithms are contributing significantly to the progress of soil moisture research. In this chapter, an overview of the applicability of ML algorithms for soil moisture assessment in the various domains of soil moisture research is presented.