干旱区生态环境知识资源中心

Soil Visible-Near-Infrared (Vis-NIR) spectroscopy has become an applicable and interesting technique to evaluate a number of soil properties because it is a fast, cost-effective, and non-invasive measurement technique. The main objective of the study to predict soil erodibility (K-factor), soil organic matter (SOM), and calcium carbonate equivalent (CaCO3) in calcareous soils of semi-arid regions located in south of Iran using spectral reflectance information in the Vis-NIR range. The K-factor was measured in 40 erosion plots under natural rainfall and the spectral reflectance of soil samples were analyzed in the laboratory. Various soil properties including the CaCO3, soil particle size distribution, SOM, permeability, and wet-aggregate stability were measured. Partial least-squares regression (PLSR) and stepwise multiple linear regression (SMLR) were used to obtain effective bands and develop Spectrotransfer Function (STF) using spectral reflectance information and Pedotransfer Function (PTF) to predict the K-factor, respectively. The derived STF was compared with developed PTF using measurable soil properties by Ostovari et al. (2016) and the Universal Soil Loss Equation (USLE) predictions of the K-factor. The results revealed that the USLE over-predicts (0.030 t h MJ(-1) mm(-1)) the K-factor when compared to the ground-truth measurements (0.015 t h MJ(-1)) in the semi-arid region of Iran. Results showed that developed PTF had the highest performance (R-2 = 0.74, RMSE = 0.004 and ME = -0.003 t h MJ(-1) mm(-1)) to predict K-factor. The results also showed that the PLSR method predicted SOM with R-2 values of 0.67 and 0.65 and CaCO3 with R-2 values of 0.51 and 0.71 for calibration and validation datasets, respectively. We found good predictions for K-factor, with R-2 = 0.56 and ratio of predicted deviation (RPD) = 1.5 using the PLSR model. The derived STF (R-2 = 0.64, RMSE = 0.002 and ME = 0.001 thMJ(-1) mm(-1)) performed better than the USLE (R-2 = 0.06, RMSE = 0.0171 and ME = 0.0151 thMJ(-1) mm(-1)) for estimating the K-factor.