Rapid measurement of moisture content in groundnut kernels using non-destructive near infrared reflectance spectroscopy

Abstract

The Near Infrared Reflectance Spectroscopy (NIRS), a non-destructive and robust tool was calibrated for rapid estimation of moisture content (MC) in whole groundnut kernels. A set of 8 groundnut genotypes were soaked overnight, followed by drying in hot air oven at 60⁰ C. Data were recorded after every 2 hrs drying using a moisture meter followed by scanning in NIRS, till constant MC was obtained. NIR absorption spectral data from 400 to 2500 nm in 2 mm intervals were collected. Modified partial least squares (MPLS) regression was applied to scatter-corrected spectra (SNV and detrend). Calibration equation with high values for the coefficient of determination (R2), the coefficient of determination for cross-validation (1-VR) and low values for the standard error of calibration or standard error of crossvalidation were estimated. Among the various models employed, model 2 with pretreatment 2,4,4,1 was best with an R2 of 0.99 in the calibration set, 1-VR value of 0.99 in the cross-validation set, lowest values for the standard error of calibration (0.33) and standard error of cross-validation (0.55). Calibration equations for moisture content showed a close relationship between NIRS predicted and lab values in this model. Thus, the selected model can act as the best models for prediction of moisture content in groundnut kernels with high accuracy. This study shows the potential of NIRS to predict the moisture content of groundnut seeds as a routine method in breeding programs, processing industries and for farmer’s advice.