The potential of near-infrared reflectance spectroscopy for soil analysis — a case study from the Riverine Plain of south-eastern Australia

Abstract

Environmental management in agricultural systems must be maintained while controlling costs and increasing productivity. To obtain a better response from inputs in agriculture, cost-effective soil analysis is needed to enable site-specific applications. Near-infrared reflectance spectroscopy (NIRS) technology has the potential to provide a rapid, low-cost analysis enabling within field variability to be identified. NIRS was evaluated for its ability to predict a range of soil properties in the Riverine Plain soils of southern New South Wales. Over 550 topsoil (0-10�cm) and 300 subsoil (40-50 cm) samples from a range of soil types were air dried and ground before scanning with a NIRSystems model 6500 scanning spectrophotometer. The Partial Least Squares (PLS) regression procedure was used to determine the best correlation (i.e. calibration) between the chemical reference data and spectral data for both topsoil and subsoil samples. A validation set of samples was used to test the predictive ability of NIRS for a number of soil properties. The results demonstrated that NIRS can successfully determine some soil properties in both the topsoil and subsoil. In the topsoil, cation exchange capacity (CEC), exchangeable Ca and Mg, pH and Ca : Mg ratio were predicted with a high level of accuracy and organic carbon and exchangeable sodium percentage (ESP) with an acceptable level of accuracy. In the subsoil, CEC, exchangeable Na, Ca, Mg, ESP, pH and Ca : Mg ratio were all predicted with a high degree of accuracy. The predictive ability of NIRS for many soil constituents may make it suitable for use in agricultural soil assessment for site-specific agriculture in the Riverine Plain soils of southern New South Wales.