Prediction of Selected Soil Properties Using Visible and Near Infrared Spectroscopy in Bardsir Area, Kerman Province

Soil spectroscopy in the visible and near infrared (Vis-NIR) range has widely been used as a rapid, cost-effective, and non-destructive technique to predict soil properties. Since little data is available about soil properties determined by using this technique, the present research was carried out to evaluate the efficiency of Vis-NIR spectroscopy to estimate several soil properties in Bardsir area, Kerman Province. About 150 complex surface soil samples were collected from four different land uses from depth of 0-20 cm. Soil organic carbon, equivalent calcium carbonate, pH, and the amount of silt, clay and sand particles were measured by routine laboratory methods. Reflectance spectra were obtained from air-dried samples under controlled laboratory conditions using an ASD FieldSpec Pro spectroradiometer in 350-2500 nm wavelength range. Partial least squares regression was used for calibration of spectral and laboratory data using cross validation. Coefficient of variation for organic carbon, equivalent calcium carbonate, sand, silt, clay, and pH values were 0.68, 0.62, 0.64, 0.66, 0.3, and 0.01, respectively. Based on RPD values (Ratio of Prediction to Deviation), the precision of the prediction model for sand and silt contents was quite suitable, and for organic carbon and equivalent calcium carbonate it was suitable. [H1] However, the predictions of the model for clay content and pH were poor.Furthermore, standard normal variate (SNV) was the best pre-processing method to predict organic carbon, whereas, first derivative with SG smoothing (FD-SG) showed better estimation for carbonate, sand, and silt. Consequently, Vis-NIR spectroscopy is capable of predicting several soil properties at the same time. As the model accuracy is acceptable, it has the potential to substitute conventional laboratory analyses of selected soil properties.