Prediction of organic carbon and nitrogen in forest soil using visible and near-infrared spectroscopy

The aim of this paper was to analyse the potential of laboratory Vis-NIR spectroscopy to determineorganic carbon and nitrogen in a representative forest area of the Calabria region (south Italy). To do that,calibration models based on laboratory Vis-NIR spectroscopy and PLSR analysis were developedseparately for soil organic carbon (SOC) and nitrogen (N). Soil samples (0-20 cm-depth) were collected at216 locations, oven-dried and passed through a 2 mm sieve and and analyzed to estimate SOC and Nconcentrations. Subsequently the Vis-NIR reflectance of each soil sample was measured in laboratory,under artificial light, using an ASD FieldSpec IV 350 - 2500 nm spectroradiometer (Analytical SpectralDevices Inc., Boulder, Colorado, USA). In order, to develop models based on soil spectra and referencelaboratory data of SOC and N, Partial least squares regression (PLSR) was used. Before applying thePLSR, spectra data were split into a calibration (144 samples) to develop the models and a validation set(72 samples) to assess the prediction accuracy of the calibration models. Results revealed a high level ofagreement between measured and predicted values with high R2 and RMSE; model validation withindependent data was satisfactory for both the studied soil properties.