CCSS COMPLEXATON 2021-2022
Challenge 1: The one and only - Develop a grapevine phenological model that works for all grapevine varieties
Representative:
Paul Petersik, VineForecast UG Background of the problem:
Grapevine phenological models (models for the life cycle of the vine) play a crucial role in grapevine disease prediction. This is because grapevine diseases are favoured by specific phenological stages.
Phenological dynamics can vary significantly between varieties. Therefore, in current approaches phenology is either modelled for a specific variety or for an “average” variety. This makes their implementation into real world applications a tedious endeavour.
Challenges:
Develop one grapevine phenological model that can accurately predict the phenology of multiple grapevine varieties based on meteorological data and the variety as input. The model should be able to predict all, so-called BBCH stages between BBCH 9 (budbreak) and BBCH 89 (berries are ripe for harvest grapes).
Main questions to be addressed:
•
Is it possible to model grapevine phenology as a dynamical system, learning the dynamics of the system from data with a model as suggested in Raissi et al. (2018), Qin et. al (2019) or Rudy et. al (2019)?•
Can the “one and only” approach give similar results as the variety-specific models, i.e.Leolini et al. (2020) for budbreak or Molitor et al. (2020) all BBCH stages?
•
How does the approach of modelling grapevine phenology as a dynamical system compare to a simple regression model?Complex Systems Science aspects:
Grapevine phenology is influenced by a vast amount of factors. The main driver for the progression of the phenology is temperature. However, subtle aspects such as the temperatures in the previous winter, frost events in spring or the available water during the growth period can significantly alter the dynamics. Up until now, no comprehensive approach was undertaken to combine all these aspects into one model.
Possible societal importance/impact:
Viticulture is the agriculture industry with the highest demand for fungicide application. In fact, about 25 times more fungicides are applied per hectare in vineyards compared to primary agriculture (Eurostat, 2007; Eisenmann et al, 2016). Disease forecasting systems can help farmers to reduce their fungicide usage. A precise estimation of the phenology is crucial for grapevine disease prediction because grapevine diseases are favoured by specific phenological stages.
Eurostat, 2007:
https://ec.europa.eu/eurostat/documents/3217494/5611788/KS-76-06-669-EN.PDF
Eisenmann et al, 2016:
https://www.rebschule-freytag.de/files/simpler/download-files/2016_11_Piwi_Info.pdf Initial literature:
Phenology
•
Leolini, Luisa, et al. "Phenological model intercomparison for estimating grapevine budbreak date (Vitis vinifera L.) in Europe." Applied Sciences 10.11 (2020): 3800.•
Molitor, Daniel, Helder Fraga, and Jürgen Junk. "UniPhen–a unified high resolution model approach to simulate the phenological development of a broad range of grape cultivars as well as a potential new bioclimatic indicator." Agricultural and Forest Meteorology 291 (2020): 108024.Learn dynamics from data
•
Qin, Tong, Kailiang Wu, and Dongbin Xiu. "Data driven governing equations approximation using deep neural networks." Journal of Computational Physics 395 (2019): 620-635.•
Raissi, Maziar, Paris Perdikaris, and George Em Karniadakis. "Multistep neural networks for data-driven discovery of nonlinear dynamical systems." arXiv preprint arXiv:1801.01236 (2018).•
Rudy, Samuel H., J. Nathan Kutz, and Steven L. Brunton. "Deep learning of dynamics and signal-noise decomposition with time-stepping constraints." Journal of Computational Physics 396 (2019): 483-506.Data
