Near-infrared spectroscopy as a tool for monitoring the spatial variability of sugarcane quality in the fields

It is known that Near-infrared spectroscopy (NIRS) is a reliable technique used in industrial laboratories to measure sugarcane quality. However, its use as a proximal sensing technology for monitoring the spatial variability of attributes in the fields has not yet been evaluated. The aim of this research was to examine the potential of NIRS for predicting and mapping Brix, Pol and Fibre content in a commercial sugarcane field. The quality attributes models were adjusted considering the spectral reflectance from the 1100-1800 nm wavelengths by using partial least squares regressions (PLSR). A total of 350 samples were collected in a sugar mill laboratory for calibration and cross-validation models development. For the external validation, 91 georeferenced samples were obtained from a commercial field. The results indicated that the developed models are capable of predicting Brix and Pol, with a coefficient of determination (R-P(2)) of 0.71 for both parameters, and with a root mean square error of prediction (RMSEP) of 0.80% and 0.58%, respectively. In contrast, the results for Fibre were unsatisfactory (R-P(2) of 0.24 and RMSEP of 1.15%). Predicted values showed spatial dependence of the sugarcane quality attributes. Predicted and observed values of Brix and Pol presented a coefficient of correlation of 0.85. Results showed that NIRS has potential to be applied as a proximal sensing method supporting crop management based on the spatial variability of the quality attributes. (C) 2021 IAgrE. Published by Elsevier Ltd. All rights reserved.