Refine
Document Type
- Doctoral Thesis (1)
Language
- English (1)
Has Fulltext
- yes (1)
Is part of the Bibliography
- no (1)
Keywords
- Black box (1) (remove)
Institute
- Obstbau (1)
Year of publication
- 2023 (1)
Physiological storage disorders cause significant economic losses in a number of commercially important pome fruit varieties worldwide. Under the same storage conditions with the same cultivar, the incidence and frequency of disorders may vary in different years in a manner that must be explained by an interaction of pre-harvest and post-harvest factors. Major factors known to influence disorder incidence include annual weather pattern, management in the orchard, such as leaf-fruit ratio, calcium content of the fruit, harvest date and storage conditions. So far, the occurrence of the disorders cannot be predicted with certainty, and thus to adapt the management in the orchard or in the storage facility accordingly. Analysis of large amounts of data using Artificial Intelligence (AI) offers new opportunities to link large data sets into a meaningful context. The aim of the present work is the small-scale and non-destructive monitoring of fruit development in orchards under the above mentioned management practices. AI prediction models were created for the cultivar 'Braeburn'. 'Braeburn' is most susceptible to physiological storage disorders which cause browning in the fruit tissue below the skin. With tent-like constructions over the fruit trees, in which the temperature was controlled during two fruit physiologically important periods (petal drop to T-stage, three weeks before harvest), the temperature influence on disorder incidence was investigated. To the best of our knowledge, this was the first time that different temperature profiles could be generated in the orchard for mature 'Braeburn' trees. Thereby a positive influence on the reduction of internal browning caused by warm night temperatures (>10°C) before harvest was observed. Overall, each trial year showed different occurrences of physiological disorders. Orchard management and weather conditions resulted in significantly different fruit growth patterns and optical non-destructive point spectroscopy. The bi-weekly spectroscopy measurements on the same fruit and subsequent use in Partial Least Square Regression (PLSR) models were tested for their informative value. Additionally, the future fruit state was predicted in models based on weather data. The statistical evaluation of three years of data showed that the number of destructive soluble solids content (SSC) samples required to be collected each year for the {PLSR} models with an acceptable error rate was 100. These samples in particular needed to include the range of low and high SSC values. An unbalanced laboratory error of 0.1-1.0 °Brix had no influence on the modeled SSC values. Compared to destructive SSC determination in the laboratory, SSC could be non-destructively determined with an average deviation of 0.5 °Brix. Field measurements after a rain event had no influence on visible spectral indices. However, orchard covers such as rain protection over the fruit trees led to lower water absorption values at 975 nm. The separation of data on a tree-sector level in this work was performed manually, but initial steps were taken to display fruit growth and spectroscopy data via GPS signal location within a map. The acquisition of small-scale data at the tree-sector level revealed significant differences for SSC, chlorophyll and anthocyanin indices as well as significant differences in the incidence of physiological disorders. In future research and modeling approaches the tree sector information should be taken into account, even if this is not yet feasible to implement in a completely automated system under the given circumstances. The prediction models for the development of physiological disorders (core browning, cavities) were able to predict with two years of data the development of the disorders in the storage and the fruit flesh firmness at harvest 90 % correctly. Further research using non-destructive fruit measurements in orchards and the influence of weather conditions will provide further insights into apple quality improvement.