Regional and national scale calibrations of hyperspectral gamma-ray signals for soil monitoring

Abstract

There is an increasing demand for accurate, timely soil information to ensure the sustainable management of our limited land resources. This information is crucial for effective environmental modelling, essential for adapting to climatic changes and ensuring global food security. Traditionally, soil information has been attained through conventional soil sampling and laboratory analyses, which are time consuming and expensive. Consequently, soil maps typically lack the fine-scale spatial and temporal resolution required for computer simulations, soil monitoring and land management. Increasingly, this fine-scale information is being attained through the use of proximal and remote sensors, which generally rely on indirect, surrogate indicators of soil variability, such as electrical conductivity. In this study, the potential of γ-ray spectroscopy as a soil-monitoring tool is assessed. The underlying principle of γ-ray spectroscopy is that long-lived terrestrial radionuclides act as environmental tracers, reflecting changes in the mineralogical and textural composition of soil. Airborne radiometric surveys proved to be valuable tools for geological mapping and have led to the development of ground-based (proximal) sensors for soil sensing. A recent study by Viscarra Rossel et al. (2007) demonstrated that robust predictions of topsoil characteristics could be made through multivariate calibrations of proximal γ-ray signals, at the within-field scale. Adopting this chemometric approach, this study assesses whether similar predictions could be made at coarser scales, using a laboratory-based spectrum analyser. The results show that at a regional scale, fair predictions of cation-exchange capacity (CEC) can be made, despite changes in parent material, land use and topography. However, more tenuous results were found at the national scale, which suggests that local relationships between γ-ray activity and soil properties (such as soil texture) may not necessarily hold at coarse scales. The findings indicate that radiometric baselines vary between soil types and host geologies, which subsequently mask localised variations in physical and chemical soil properties.