1ICAR-National Institute of Secondary Agriculture, Ranchi, Jharkhand, India
2ICAR-Research Complex for Eastern Region, Farming System Research Centre for Hill and Plateau Region (FSRCHPR), Ranchi, Jharkhand, India
*Corresponding Author’s E-mail Address: sakha_yogesh@yahoo.co.in
This study examined the physical and compositional characteristics of sapota fruits (variety Murabba) and developed predictive mass models based on physical attributes using linear, power, quadratic, and S-curve models. The pulp, peel, and seed contents of the fruits were found to be 85.66 ± 2.04%, 11.48 ± 1.70%, and 2.86 ± 1.02%, respectively. The average spatial dimensions, including the major (L), first minor (W), and second minor (T) axes, were 53.86, 52.06, and 50.20 mm, respectively. The average values for geometric mean diameter (Dg), sphericity (Φ), volume (V), ellipsoidal volume (Vellip), criteria projected area (CPA), and mass were recorded as 51.99 mm, 0.97, 70765.21 mm3, 74212.01 mm3, 2131.51 mm2, and 67.22 g, respectively. The results of predictive mass modelling showed that nine models, namely, Dg -based quadratic, Dg -based power, Vellip -based linear, Vellip -based linear, Vellip -based quadratic, Vellip-based power, CPA-based quadratic, CPA-based power, LWT-based linear, and PA-based linear performed better, with higher coefficient of determination (R2 >0.95), and lower root mean square error (RMSE ≤ 2.39 g), and lower mean relative deviation (MRD ≤ 2.48) values. These metrics compare favourably with established fruit mass models: kinnow mandarin (R2 = 0.93, RMSE = 4.2 g), guava (R2 = 0.94, RMSE = 3.1 g), and persimmon (R2 = 0.93, RMSE = 5.8 g), demonstrating superior predictive accuracy for sapota var. Murabba. The Dg-based power model was the best single-variable prediction model, whereas the Vellip-based quadratic model was the best volume-based one. Overall, this study recommends two multiple linear regression models for indirect mass estimation of sapota fruits: LWT-based linear model (-133.38+1.15L+1.91W+0.774T) and projected area-based linear model (-32.89+1.84PA1-0.114PA2+2.714PA3). These mass prediction models can be directly used for the design and calibration of non-destructive, size- and mass-based grading systems for sapota fruits in packhouses and processing units. Their adoption will standardise quality, reduce manual grading, and support mechanised pulp-processing of the Murabba variety.
criteria projected area, ellipsoidal fruit volume, non-destructive fruit grading, regression-based mass estimation, size-mass relationship
