Main Article Content
Analysis for soil organic carbon (SOC) content is a key component of sustainable soil fertility management. Estimating this parameter using traditional methods is time – consuming and may be inappropriate for large scale monitoring. This study assessed the potential of a combined application of diffuse reflectance spectroscopy and the partial least square regression (PLSR) to analyse and predict SOC in soils. Fifty three (53) topsoil samples were collected from areas with differing land use activities, and then analysed for SOC using a Flash 2000 organic elemental analyzer. Diffuse reflectance spectra of soil samples were measured in the visible near infrared (VNIR) and the mid-infrared (MIR) wavelength ranges using a Ger3700 VNIR spectrophotometer and a FT-IR spectrometer respectively. Partial least squares regression (PLSR) was used to develop prediction models. Models developed from both spectra predicted SOC with accuracy close to the elemental analyzer (R2 > 0.80) and offers a reliable alternative to the traditional laboratory analyses. On comparison using the coefficient of determination (R2), ratio of performance to deviation (RPD) and the root mean square error (RMSE), VNIR spectra offer better accuracy with an R2 = 0.90, RPD = 3.12 and RMSE = 0.07 Log10 %SOC compared to the MIR spectra with an R2 = 0.85, RPD = 3.09 and RMSE = 0.08 Log10 %SOC.