Utilizing preharvest and packinghouse data in combination with storage trials to develop an intelligent logistic management system for 'Orri' mandarins

Abstract

'Orri' mandarin is the main citrus variety grown and exported from Israel. In the current study, we assessed the potential of utilizing large-scale preharvest and commercial packinghouse data to predict the postharvest storage performances of different sets of 'Orri' mandarins. For that purpose, we examined during 3 years (2022–2024), the postharvest performances of between 79 and 93 sets of 'Orri' mandarins per year after 2, 4 and 8 weeks storage at 5 °C and one-week at shelf life at 22 °C. The observed preharvest data consisted climatic features (amounts of rain and warm days) and horticultural features (harvest time, tree age, yield, and soil type), and the observed packinghouse data consisted ripening features (total soluble solids, acidity and juice contents) and packing line quality evaluation features (occurrences of various defects and packaging rates). Consolidating the observed preharvest and packinghouse data with the postharvest storage data using machine learning algorithms, allowed the prediction of decay percentages by an R² of 0.60 and RMSE of 10.56 %, and of fruit quality acceptance score by an R² of 0.74 and RMSE of 0.57 on a quality scale from 1 (very bad) to 5 (excellent). The development of fruit quality prediction models based on preharvest and packinghouse data can assist in the future implementation of intelligent logistic management system that will allow to direct specific fruit sets to different market destinations based on their predicted shelf life's, thus assuring that only high-quality produce will arrive at the recipient markets.