Vis-NIR spectroscopy as an eco-friendly method for monitoring pedoenvironmental variations and pedological assessments in Lut Watershed, Central Iran

Abstract

Soil spectroscopy has become an applicable method for monitoring and management of the spatial variability of soil characteristics. However, no widely knowledge about the potential application of this method in arid regions is still available. This research aimed to ascertain the performance of visible and near-infrared spectroscopy (Vis-NIR) to detect soil variations related to pedogenetic processes and pedoenvironmental conditions and to show the perspective of the most important variables changing spectral behavior in Lut Watershed. Twenty-three pedons in three clusters, including site 1 (xeric-mesic soil moisture-temperature regimes), site 2 (aridic-mesic, and aridic-thermic), and site 3 (extreme aridic-hyperthermic), with different geomorphology and lithology, were selected. Physicochemical properties and different iron forms were analyzed; X-ray diffraction (XRD), magnetic susceptibility (χlf), and Vis-NIR spectroscopy were also performed. The most important attributes influencing spectral morphology were related to soil texture, organic carbon content, clay mineralogy, evaporitic minerals (carbonate, halite, gypsum, and anhydrite), and iron oxides (goethite, hematite, and magnetite). Geometric parameters (area and depth) of absorption features located in 400–1000 nm (iron oxide), 2200 nm (clay), and 2340 nm (CaCO3) were significantly decreased by organic carbon in site 1. Increasing χlf values masked the wavelengths in 1400, 1540, 1750, and 1900 nm (CaSO4.2H2O) in site 2. Moreover, increasing salinity and anhydrite content covered the features of iron oxides in 400–570 nm in site 3. The finger-shaped absorption in 1450–1475 nm (CaSO4.2H2O) disappeared with an increasing amount of anhydrite and its degree of dehydration. Principal component analysis (PCA) showed a suitable performance for soil discrimination in the three climatic sites with different pedogenetic processes. Validation results by the partial least squares regression (PLSR) models indicated excellent accuracy in predicting CEC, EC, χlf, and gypsum content (RPIQ = 4.17, 3.00, 3.00, and 2.58, respectively) and good accuracy in predicting Fed and CCE (RPIQ = 2.30 and 2.06, respectively). Overall, spectral sensing was recognized as a suitable pedoenvironmental monitoring indicator for pedological assessments in Lut Watershed.