Advantages using the thermal infrared (TIR) to detect and quantify semi-arid soil properties

Abstract

Monitoring soil surface dynamics in semi-arid agricultural landscapes becomes increasingly important due to the vulnerability of these ecosystems to desertification processes. Observations using remote sensing via the traditionally used visible-near infrared (VNIR) and shortwave infrared (SWIR) wavelength regions can be limited due to the special characteristics of such soils (e.g. rich in quartz, poor in clay minerals, coarse textured, and grain coatings). In this laboratory-based work we demonstrate the capabilities of the thermal infrared between 8 and 14 μm (longwave infrared) to detect and quantify small ranges of the soil properties sand-, clay, and soil organic carbon (SOC) content, as they appear in the semi-arid agricultural landscapes of the Mullewa region in Western Australia. All of the three soil properties could be predicted using the longwave infrared (LWIR) spectra with higher accuracy and precision than from the VNIR-SWIR wavelength region. The study revealed the complex relationships between the soil properties and the VNIR-SWIR soil spectra, which were caused by the spectral influence of the soils' grain coatings (based on iron and clay minerals). These difficulties could be handled more appropriately by the prediction models based on the LWIR soil spectra. Our results indicate that in order to quantitatively monitor farming areas for such erosion-related soil properties; remote sensing using the LWIR wavelength region would produce better estimates than using the wavelength ranges in the VNIR-SWIR.