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 (R²_P) of 0.71 for both parameters, and with a root mean square error of prediction (RMSE_P) of 0.80% and 0.58%, respectively. In contrast, the results for Fibre were unsatisfactory (R²_P of 0.24 and RMSE_P 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.

众所周知，近红外光谱法（NIRS）是工业实验室中用于测量甘蔗质量的可靠技术。但是，尚未评估其作为近场传感技术来监视字段中属性的空间变异性的用途。这项研究的目的是检验NIRS在预测和绘制商业甘蔗田中白利糖度，波尔糖和纤维含量方面的潜力。通过使用偏最小二乘回归（PLSR），考虑到1100-1800 nm波长的光谱反射率，对质量属性模型进行了调整。在制糖厂的实验室中总共收集了350个样品，用于校准和交叉验证模型的开发。为了进行外部验证，从商业领域获得了91个地理参考样品。两个参数的² _P）均为0.71，预测的均方根误差（RMSE _P）分别为0.80％和0.58％。相反，Fiber的结果不令人满意（R ² _P为0.24，RMSE _P为1.15％）。预测值显示了甘蔗品质属性的空间依赖性。糖度和波尔糖的预测值和观察值的相关系数为0.85。结果表明，基于质量属性的空间变异性，NIRS有潜力用作支持作物管理的近端传感方法。