Review on soil sensing technologies: devices, statistical models, and future perspective

Abstract

Soil sensors play a crucial role in agriculture and environmental monitoring, especially in the context of sustainability and related social benefits. Traditional soil analysis methods are typically costly, time-consuming, and rely on physical sampling and laboratory testing, which limits their ability to provide spatially continuous, field-scale information. In contrast, precision agriculture requires affordable, fast, and reliable sensing techniques that can deliver actionable insights at scale. Emerging indirect sensing technologies-based on electromagnetic, radioactive, and optical principles-are increasingly used for measuring individual parameters or estimating multiple soil attributes through sensor fusion. To enhance measurement accuracy, considerable efforts have been devoted to the development of statistical and machine learning models that account for interactions among soil properties and utilize multivariate data. The growing availability of computational resources has further emphasized the value of integrating large volumes of data from sensors, computer vision, and hyperspectral imaging into decision support systems for agricultural and environmental applications. This review summarizes the main technologies and statistical approaches for soil quality assessment, highlighting current capabilities, limitations, and future directions.