Hyperspectral remote sensing applications in soil: a review

Abstract Hyperspectral remote sensing (HRS) enables a precise recording of the spectrum and a detailed analysis of the spectral properties of soil. There is a wealth of published research on HRS applications in soil research from the past 40 years, therefore, a review and illustration of the developments is critical for a better understanding of its opportunities and challenges for future development. This study reviewed the state-of-the-art regarding the applications of HRS for the extraction of soil properties including mineral identification, nutrient, organic carbon, moisture, salinity, and soil texture, then identify the opportunities and challenges and provide suggestions of a new direction for advancing its development. The results obtained provide new insights into the theory and methodology of HRS for studying soil attributes, but there is much work to be done, both experimentally and theoretically, before the physical and chemical processes predicting these soil parameters can full be understood. At the end of the chapter, new directions of studying soil properties using HRS are suggested, for example, new data mining technologies such as deep learning provide opportunities for efficient processing of large amounts of HRS data; the interactional mechanism of vegetation and soil should also be integrated into the construction of HRS monitoring models; the development from the empirical model to physical model will improve the universality and robustness of a given model; more attention should be paid to the application of multivariate models involving multiple parameters, rather than to single-factor models; by combining the temporal and dynamic characteristics of soil attributes with the process of vegetation growth, the accuracy of detecting and monitoring soil attributes will be greatly improved; and multidata fusion and multiscale data assimilation will become another research hotspot in HRS monitoring of soil properties.