Correction of environmental interference in field applications of handheld NIR spectroscopy for grape quality assessment

Abstract

Sugar accumulation and organic acid degradation are key indicators of grape quality and require monitoring throughout the breeding process in the field. However, traditional methods are labor-intensive and limited by destructive sampling and laboratory-based chemical analyses, which post a major challenge in the early breeding stage when younger vines produce limited quantities of grapes. Handheld near-infrared (NIR) spectroscopy offers a rapid, non-destructive alternative but its performance is strongly influenced by external environmental factors, particularly variable illumination, when applied in the field.

To address this challenge, the correction potential of External Parameter Orthogonalization (EPO) for grape quality assessment under vineyard conditions was evaluated in the present study. Paired dataset of NIR spectra were acquired under controlled laboratory conditions and in the vineyard from the same biological samples. Partial Least Squares Regression (PLSR) was used as a baseline modeling approach to quantify the improvement. The result showed that EPO significantly improved model performance for the field acquired data, reducing the root mean square error of prediction (RMSEP) from 142.18 g/L to 17.49 g/L and increasing the R² from −5.15 to 0.74. These findings demonstrate the effectiveness of EPO in correcting environmental interference, enabling accurate and reliable field-based grape quality assessment using calibration models developed under laboratory conditions, which can accelerate the evaluation process in the breeding program and allow select grape material from early-stage vines to be used for further evaluations.