Non-destructive prediction of sweetness of intact mango using near infrared spectroscopy

Agriculture industries are continuously in search of new user friendly technologies to evaluate the intrinsic properties of fruits before it is put in the market for the consumer. In the current study the potential of near-infrared (NIR) spectroscopy in the wavelength range of 1200–2200 nm was evaluated to determine total soluble solids and pH for seven major cultivars of mangoes from seven states of India. NIR models were developed based on multiple-linear regression (MLR) and partial least square (PLS) regression employing preprocessing technologies (baseline correction, smoothening, multiplicative scatter correction (MSC) and second order derivatisation). The multiple correlation coefficients for calibration and validation were found to be 0.782 and 0.762 for total soluble solids and 0.715 and 0.703 for pH respectively. The standard errors of calibration, prediction, biases and differences in them were low which indicated the NIRS potential to predict internal quality parameters (TSS and pH) of mango non-destructively for both models.

► NIRS calibration models (PLS and MLR) were developed for different groups of wavelengths in the range of 1100–2500 nm to predict quality (TSS and pH) of different mango cultivars. ► PLS model was found more suitable for the prediction of TSS and pH as compared to MLR. ► Second order of derivatisation of spectral data improved prediction of TSS whereas pH could be predicted without any data treatments. ► The calibration models so developed will be helpful in designing portable offline and/or online instruments for quick grading of fruit.