<![CDATA[Abstract. Wirawan RL, Khuriyati N, Pamungkas AP, Ramadani MI. 2026. Effect of sampling depth on °Brix prediction in melon (Cucumis melo) using Visible Near Infrared Spectroscopy (Vis-NIRS). Asian J Agric 10 (1): g100139. https://doi.org/10.13057/asianjagric/g100139. The accuracy of non-destructive melon sweetness (°Brix) prediction using Visible Near Infrared Spectroscopy (Vis-NIRS) depends strongly on the quality and representativeness of destructive reference measurements. However, standardized reference sampling depth for °Brix determination has not been clearly established. This study systematically evaluated the effect of reference sampling depth on the performance of Vis-NIRS-based prediction models in melon (Cucumis melo). Spectral data were collected from six surface points per fruit, while reference °Brix values were obtained from two different tissue depths to represent differences in internal chemical composition. Full flesh depth sampling represented the entire edible mesocarp from the outer flesh to the fruit center, while partial flesh depth sampling represented only the outer mesocarp close to the rind. Artificial neural network models were developed to compare predictive performance. The model calibrated with full-depth reference values achieved higher prediction accuracy (92.88%) and stronger correlation (R²=0.75) than the model calibrated with partial-depth references (86.66%; R²=0.50), with the difference statistically confirmed by Welch’s t-test (p<0.001). These results demonstrate that aligning destructive reference measurements with fruit internal heterogeneity is essential for improving the reliability of non-destructive sweetness assessment. The findings provide practical guidance for developing more robust non-destructive quality evaluation protocols of melon and other horticultural crops with spatially heterogeneous internal structures.]]>
Copyrights © 2026
