Remote sensing of soils

Abstract

Global environmental changes are currently altering key ecosystem services that soils provide. Therefore, it is necessary to have up to date soil information on local, regional and global scales to monitor the state of soils and ensure that these ecosystem services continue to be provided. In this context, digital soil mapping (DSM) aims to provide and advance methods for data collection and analyses tailored towards detailed large-scale mapping and monitoring of soil properties. In particular, remote and proximal sensing methodologies hold considerable potential to facilitate soil mapping at larger temporal and spatial scales as feasible with conventional soil mapping methods [Mulder, 2013]. Existing remote and proximal sensing methods support three main components in DSM: (1) Remote sensing data support the segmentation of the landscape into homogeneous soil-landscape units whose soil composition can be determined by sampling. (2) Remote and proximal sensing methods allow for inference of soil properties using physically-based and empirical methods. (3) Remote sensing data supports spatial interpolation of sparsely sampled soil property data as a primary or secondary data source [Mulder, 2013]. Overall, remote and proximal sensed data are an important and essential source for DSM as they provide valuable data for soil mapping in a time and cost efficient manner. This document provides general insights into diverse aspects of soil related remote sensing, including DSM, remote sensing technologies and soil properties. In this context, we present the underlying concept of DSM and introduce approaches to predict the spatial distribution of soil properties. Furthermore, we introduce remote and proximal sensing technologies and the methodologies to extract soil properties in support of DSM. In this overview we consider established techniques within active, passive, optical and microwave remote sensing as well as proximal sensing that use key soil properties as proxies for soil conditions and characteristics. In addition, we discuss the opportunities, progress and limitations of remote and proximal sensing data in support of DSM and conclude by a gap analysis of current remote sensing technologies and products. Proximal sensing has been successfully used to derive quantitative and qualitative soil information [Viscarra Rossel et al., 2006b]. Most reported studies revealed the high potential of proximal sensing to estimate soil properties based on clear absorption features at the laboratory and local scale [Ben-Dor et al., 2008]. However, for large-scale mapping of soil properties, methods need to be extended beyond the plot scale. Important qualitative and, to a lesser extent, quantitative soil information can be obtained from remote sensing data. Airborne and spaceborne remote sensing provides qualitative information on soil properties having clear diagnostic absorption features at a regional to global scale. However, remote sensing-derived information has a lower accuracy and feasibility to obtain information compared to proximal sensing. The main limiting factors are (1) the coarse spatial and spectral resolution, (2) the low signal-to-noise ratio of high-resolution remote sensing data and (3) the bands of multispectral satellite sensors have not been positioned at diagnostic wavelengths. Future improvement to detect soil properties on a regional to global scale with high accuracy can be expected from recently launched Sentinel-1 and upcoming Sentinel-2, SWAP, and EnMAP missions. Despite the large potential of using proximal and remote sensing methods in support of DSM, advances are necessary to fully develop large-scale methodologies and soil products. Currently, DSM-studies make limited use of existing analysis and geostatistical methods to exploit the full potential of proximal and remote sensing data [Ben-Dor et al., 2009; Dewitte et al., 2012]. Improvements may be expected in the fields of developing more quantitative methods, enhanced geostatistical analysis that allow working with large remote sensing datasets. Further research priorities involve the development of operational tools to quantify soil properties, multiple sensor integration, spatiotemporal modelling and improved transferability of soil mapping approaches to other landscapes. This will allow us in the near future to deliver more accurate and comprehensive information about soils, soil resources and ecosystem services provided by soils at regional and, ultimately, global scale.