[Defending thesis] Yulin Zhang

[Defending thesis] Yulin Zhang: Retrieve Total Transpirable Soil Water for Grapevines from a Vine Shoot Growth Index Collected Through Farmsourcing

Thesis topic cofunded by #DigitAg Yulin will defend his PhD on 10 October 2024 at 2pm @Institut Agro (Amphi 206, bâtiment 9, Campus Gaillard, 2 Pl. Pierre Viala, Montpellier).

Retrieve Total Transpirable Soil Water for Grapevines from a Vine Shoot Growth Index Collected Through Farmsourcing

Y Zhang

Yulin will defend his PhD on 10 October 2024 at 2pm @Institut Agro (Amphi 206, bâtiment 9, Campus Gaillard, 2 Pl. Pierre Viala, Montpellier).
To attend the defence by visioconference

My name is Yulin. I did my PhD in the research unit ITAP of Institut Agro Montpellier, in Southern France. I received my doctoral degree in Process Engineering in 2024. I like to think and resolve problems, that’s why I chose to become a PhD candidate. As an agronomist, I wish to develop my data science skills and apply them in actual agronomic problems. I developed a method that uses data collected through crowdsourcing to estimate grapevine’s water availability by applying an inverse modeling approach. The data I used were mainly observations on vine shoot growth, weather, and satellite data. This method is very low-cost, which facilitates winegrowers and other stakeholders to adopt relevant farming practices to address challenges brought by climate change. I was quite motivated by the nature of crowdsourcing data and the originality of inverse modeling. The topic is honestly very attractive. In three years, I developed some practical solutions with what I have, but I also learnt aspects that need to be improved. After three years, I still found this topic is very interesting and filled with potentials.

  • Starting date : 1st October 2021
  • University: University of Montpellier
  • PhD school: GAIA (Biodiversité, Agriculture, Alimentation, Environnement, Terre, Eau)
  • Scientific field: Process Engineering
  • Thesis management: Bruno Tisseyre, UMR Itap, Institut Agro
  • Thesis supervisors: Léo Pichon, UMR Itap, Institut Agro
  • Funding: #DigitAg – Région Occitanie
  • #DigitAg : Thèse cofinancée – Axe 5 : Fouille de données, analyse de données, extraction de connaissances, Challenge 5: Les services de conseil agricole

Keywords: Inverse modeling, Total Transpirable Soil Water, Vine shoot growth, Crowdsourcing, Viticulture, Mediterranean regions, Uncertainty, Predictive modeling, Data Pre-processing, Digital Agriculture

Abstract: The Total Transpirable Soil Water (TTSW) is crucial for wine producers as it significantly influences vineyard management decisions, such as canopy control and irrigation planning. In the context of climate change, accurately evaluating TTSW is increasingly vital, especially in the Mediterranean regions, to assess vineyard resilience. However, direct measurement of TTSW poses logistical challenges, and there has been a notable lack of low-cost estimation methods suitable for commercial vineyards. This PhD dissertation explores the estimation of TTSW for grapevines using an Inverse Modeling (IM) approach, leveraging vine shoot growth data collected via the farmsourcing mobile application ApeX-Vigne. The research addresses two primary questions: 1) How can a TTSW retrieval method based on farmsourced data be developed? and 2) How accurate is this method? The dissertation is structured as follows: Chapter 1 identifies a promising inversion pathway linking TTSW to a vine shoot growth index. Chapter 2 develops a transfer function to implement this pathway. Chapter 3 adapts and tests an inverse model to retrieve TTSW in a Mediterranean vineyard, assessing its accuracy at a local scale. Chapter 4 evaluates the proposed TTSW estimation method using a farmsourced dataset from a Southern French wine appellation, assessing the results' coherence on a regional scale. This research provides a comprehensive overview of IM applications in agriculture, offering a practical guide for selecting inversion methods. It introduces a novel transfer function that predicts temporally autocorrelated variables while accounting for prediction uncertainty, demonstrating a combination of predictive and inverse modeling techniques. The study proposes a cost-effective, vine-shoot-growth-based method for estimating TTSW in commercial vineyards, highlighting the importance of factors such as plant density. Despite the need for substantial data pre-processing, the findings illustrate the potential of farmsourced data in regional TTSW retrieval, notably their capability of revealing large-scale TTSW variation. Future research should aim to develop a more adaptable TTSW retrieval method that is less dependent on specific climatic conditions. Efforts should focus on generalizing the transfer function and incorporating additional sources of variability. Special consideration should be given to the unique challenges in viticulture for TTSW estimation, particularly regarding the influence of plant density on vine shoot growth. To maximize the potential of farmsourced data, strategies to guide data collectors to provide more complete information and to refine their data collection practices are necessary. Formalizing cooperation among data collectors, researchers, and other viticultural stakeholders will be crucial in improving sampling strategies for regional-scale data generation.

Jury compound:

  • Maria Paz Diago Santamaria, Université de la Rioja, Espagne (Rapporteure)
  • Stéphane Follain, l’Institut Agro Dijon, France (Rapporteur)
  • Cornelis Van Leeuwen, Bordeaux Science Agro, France (Examinateur)
  • Anne Pellegrino, l’Institut Agro Montpellier, France (Examinatrice)
  • Léo Pichon, l’Institut Agro Montpellier, France (Examinateur)
  • Bruno Tisseyre, l’Institut Agro Montpellier, France (Directeur de thèse)

Contact : yulinzhang9 [AT] gmail.com - Tél: +33658910155

Social networks: LinkedIn - ResearchGate

Communications / Papers:

  • (open access) Zhang, Y., Pichon, L., Pellegrino, A., Roux, S., Péruzzaro, C., and Tisseyre, B. (2024). Predicting predawn leaf water potential while accounting for uncertainty using vine shoot growth and weather data in Mediterranean rainfed vineyards. Agricultural Water Management, 302, 108998. https://doi.org/10.1016/j.agwat.2024.108998
     
  • (open access) Tarraf, B., Brun, F., Raynaud, L., Roux, S., Zhang, Y., Davadan, L., and Deudon, O. (2024). Assessing the impact of weather forecast uncertainties in crop water stress model predictions. Agricultural and Forest Meteorology, 349, 109934. https://doi.org/10.1016/j.agrformet.2024.109934
     
  • (open access) Zhang, Y., Pichon, L., Roux, S., Pellegrino, A., Simonneau, T., and Tisseyre, B. (2024). Why make inverse modeling and which methods to use in agriculture? A review. Computers and Electronics in Agriculture, 217, 108624. https://doi.org/10.1016/j.compag.2024.108624
     
  • (open access) Oger, B., Zhang, Y., Gras, J.-P., Valloo, Y., Faure, P., Brunel, G., and Tisseyre, B. (2023). High spatial resolution dataset of grapevine yield components at the within-field level. Data in Brief, 50, 109580. https://doi.org/10.1016/j.dib.2023.109580
     
  • (open access) Pichon, L., Brunel, G., Zhang, Y., and Tisseyre, B. (2023). Vers une cartographie régionale de l’état hydrique de la vigne basée sur des observations collaboratives: Article prenant sa source de l’article de recherche “Towards a regional mapping of vine water status based on crowdsourcing observations” (OENO One, 2022). Langue originale de l’article : français. IVES Technical Reviews, vine and wine. https://doi.org/10.20870/IVES-TR.2023.7499
     
  • (open access) Pichon, L., Brunel, G., Zhang, Y., and Tisseyre, B. (2022). Towards a regional mapping of vine water status based on crowdsourcing observations: This article is published in cooperation with Terclim 2022 (XIVth International Terroir Congress and 2nd ClimWine Symposium), 3-8 July 2022, Bordeaux, France. OENO One, 56(2), Article 2. https://doi.org/10.20870/oeno-one.2022.56.2.544
     
  • (open access) Zhang, Y., Pichon, L., Taylor, J. a., Oger, B., and Tisseyre, B. (2023). 82. Introducing Bayesian priors to semi-variogram parameter estimation using fewer observations. In Precision agriculture 2023 (pp. 651–658). Wageningen Academic Publishers. https://doi.org/10.3920/978-90-8686-947-3_82