Schmitz, Christine Maria: Forecasting Tools and Risk Prediction Models for Decision Support in Fruit Production. - Bonn, 2026. - Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn.
Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5-87222
Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5-87222
@phdthesis{handle:20.500.11811/13815,
urn: https://nbn-resolving.org/urn:nbn:de:hbz:5-87222,
doi: https://doi.org/10.48565/bonndoc-754,
author = {{Christine Maria Schmitz}},
title = {Forecasting Tools and Risk Prediction Models for Decision Support in Fruit Production},
school = {Rheinische Friedrich-Wilhelms-Universität Bonn},
year = 2026,
month = jan,
note = {Tree fruit production in Germany is confronted with various risks that affect quality and reduce yields, such as disease infestation or weather events. The occurrence of late spring frosts often leads to cell damage, resulting in yield reduction and reduced fruit quality, e.g. due to 'frost rings' on the peel. With ongoing climate change, fruit trees tend to bloom earlier, which goes along with an increase in spring frost frequency during the last years in some regions.
In this thesis, I applied the decision analysis methodology to develop probabilistic models. Decision analysis allows for the inclusion of expert knowledge into the modelling procedure and accounts for uncertainty and variability of the model parameter values. Furthermore, I used Kriging Interpolation and process-based phenology modelling to produce countrywide maps of historic and future frost risk in apple production.
1. In chapter 2, I introduce ProbApple a model to make probabilistic forecasts of total apple yield and high-quality yield at four time points during the growing season (at full bloom, before fruit thinning, after June drop, and four weeks before harvest). Using a case study on 'Gala' apple in Rhineland, we showed the functionality and structure of the model and compared the high-quality yield in orchards with and without anti-hail netting.
2. In chapter 3, I analyze the frost frequency during and after apple bloom in the time periods 1993-2007 and 2007-2022 based on historic temperature and phenology data. We observed a trend toward earlier apple bloom and increasing frost frequency across nearly in the whole country. Frosts below 0°C occurred in all apple production regions, while only during a few frost events the temperatures fell below -2.2°C.
3. In chapter 4, I describe a decision support model to advise apricot producers on investments in frost protection. We compared yield effects and Net Present Value of candles, below-canopy irrigation, as well as mobile and stationary wind machines in relation to apricots production without frost protection. Despite an appreciable increase in yield, the additional income was not sufficient to cover the additional costs for frost protection.
4. In chapter 5, I present a comparison of eight frost protection measures (mobile or stationary wind machines, portable or tractor-mounted gas heaters, overhead or below-canopy irrigation, candles and pellet heaters) regarding their economic efficiency and yield effects in apple production using decision analysis. It turned out that overhead irrigation and stationary wind machines appear to be the most promising frost protection measures, but they do not necessarily increase farmers' revenues.
5. In chapter 6, I show maps for the future spring frost risk in German apple production from green tip stage until summer under four climate change scenarios for the years 2050 and 2085. The phenology forecast indicates that the current trend towards earlier apple blossom will continue in future. The effect of climate change of frost risks differs between regions and climate change scenarios. However, we showed that late spring frost will remain a challenge for apple production even at the end of the 21st century.
Overall, the forecasting tools, risk prediction models, and their results in this thesis may be useful for fruit growers to make informed decisions in short- and long-term planning. They can be used by policymakers as a reliable source of information and serve scientists as a base for further development of probabilistic models in fruit production.},
url = {https://hdl.handle.net/20.500.11811/13815}
}
urn: https://nbn-resolving.org/urn:nbn:de:hbz:5-87222,
doi: https://doi.org/10.48565/bonndoc-754,
author = {{Christine Maria Schmitz}},
title = {Forecasting Tools and Risk Prediction Models for Decision Support in Fruit Production},
school = {Rheinische Friedrich-Wilhelms-Universität Bonn},
year = 2026,
month = jan,
note = {Tree fruit production in Germany is confronted with various risks that affect quality and reduce yields, such as disease infestation or weather events. The occurrence of late spring frosts often leads to cell damage, resulting in yield reduction and reduced fruit quality, e.g. due to 'frost rings' on the peel. With ongoing climate change, fruit trees tend to bloom earlier, which goes along with an increase in spring frost frequency during the last years in some regions.
In this thesis, I applied the decision analysis methodology to develop probabilistic models. Decision analysis allows for the inclusion of expert knowledge into the modelling procedure and accounts for uncertainty and variability of the model parameter values. Furthermore, I used Kriging Interpolation and process-based phenology modelling to produce countrywide maps of historic and future frost risk in apple production.
1. In chapter 2, I introduce ProbApple a model to make probabilistic forecasts of total apple yield and high-quality yield at four time points during the growing season (at full bloom, before fruit thinning, after June drop, and four weeks before harvest). Using a case study on 'Gala' apple in Rhineland, we showed the functionality and structure of the model and compared the high-quality yield in orchards with and without anti-hail netting.
2. In chapter 3, I analyze the frost frequency during and after apple bloom in the time periods 1993-2007 and 2007-2022 based on historic temperature and phenology data. We observed a trend toward earlier apple bloom and increasing frost frequency across nearly in the whole country. Frosts below 0°C occurred in all apple production regions, while only during a few frost events the temperatures fell below -2.2°C.
3. In chapter 4, I describe a decision support model to advise apricot producers on investments in frost protection. We compared yield effects and Net Present Value of candles, below-canopy irrigation, as well as mobile and stationary wind machines in relation to apricots production without frost protection. Despite an appreciable increase in yield, the additional income was not sufficient to cover the additional costs for frost protection.
4. In chapter 5, I present a comparison of eight frost protection measures (mobile or stationary wind machines, portable or tractor-mounted gas heaters, overhead or below-canopy irrigation, candles and pellet heaters) regarding their economic efficiency and yield effects in apple production using decision analysis. It turned out that overhead irrigation and stationary wind machines appear to be the most promising frost protection measures, but they do not necessarily increase farmers' revenues.
5. In chapter 6, I show maps for the future spring frost risk in German apple production from green tip stage until summer under four climate change scenarios for the years 2050 and 2085. The phenology forecast indicates that the current trend towards earlier apple blossom will continue in future. The effect of climate change of frost risks differs between regions and climate change scenarios. However, we showed that late spring frost will remain a challenge for apple production even at the end of the 21st century.
Overall, the forecasting tools, risk prediction models, and their results in this thesis may be useful for fruit growers to make informed decisions in short- and long-term planning. They can be used by policymakers as a reliable source of information and serve scientists as a base for further development of probabilistic models in fruit production.},
url = {https://hdl.handle.net/20.500.11811/13815}
}





