EcoExtreML: Accelerating Process Understanding for Ecosystem Functioning under Extreme Climates with Physics-Aware Machine Learning
Droughts and heatwaves impact ecosystem water, energy and carbon fluxes, and jeopardize terrestrial ecosystem carbon sequestration. Remote sensing of fluorescence and plant-hydraulics-based vegetation models are state-of-the-art approaches to monitor and predict drought responses of ecosystem functioning. However, the disciplinary disconnect between the two approaches has hazed the full potential of synergizing them. This project will couple the vegetation photosynthesis model (SCOPE) with the soil moisture model (STEMMUS, considering dynamic root growth), synergized with Earth-Observation data, to understand how water-carbon dynamics of ecosystem vary with variable environmental and climate stress.
The bottleneck in applying STEMMUS-SCOPE globally is its expensive computational cost. As the first step, the coupled STEMMUS-SCOPE model will be exposed to Basic-Model-Interface, serving as the first level acceleration. Second, a physics-aware machine learning emulator, based on a limited number of STEMMUS-SCOPE runs, will be developed. Furthermore, to address the âdata-gapâ issue of satellite reflectance products (i.e., revisit-time (5â27days) and cloudy condition), OpenDA will be deployed to assimilate multiscale/multi-sensor data to generate spatiotemporally continuous information on ecosystem functioning. This project will develop an open digital twin of soil-plant system (see Figure 1) and provide a variety of new opportunities for Earth-Observation for retrieving higher-level products like root-zone-soil-moisture and belowground-carbon-allocation, besides land-atmosphere gas exchanges.
Figure 1 Three main components of a soil-plant digital twin.
2023-05-24:Â Good Practices for SUSTAINABLE Development OF STEMMUS-SCOPE
Together with NLeScience colleagues, we enjoyed a fruitful workshop to follow the good practice in Sustainable Software Development, in terms of the collaborative workflow, programming styles, and many more.
- Â 10:00 â 10:30 Â Introduction and collaborative workflow
- 10:30 â 11:00 Â Demo
- 11:00 â 11:15 Â Coffee break
- 11:15 - 12:00 Â Modular code development
- 12:00 â 13:00 Â Lunch
- 13:00 â 13:30 Â Code styles and documentation
- 13:30 â 14:15 Â Exercise â Setup, fix an issue, and submit a pull request
- 14:15 â 14:30 Â Coffee break
- 14:30 â 15:30Â Exercise - Review a pull request and merge
- 15:30 â 15:45 Â Coffee break
- 15:45 â 16:30 Â Wrap-up and questions
Please find the slides here and the collaborative document here.
2023-04-26: EGU Presentation on Soil-Plant Digital Twin
I presented "Towards a Digital Twin of Soil-Plant System" at EGU.https://lnkd.in/ezREGWVm
For a more detailed explanation of the soil-plant digital twin, please check the recording at the OpenGeoHub meeting:Â https://lnkd.in/ecgWqhPk
2023-03-07: SOIL-PLANT DIGITAL TWIN
2023-02-15: UT-NEWS - INTERNATIONAL SOIL MODELLING CONSORTIUM (ISMC) ELECTS YIJIAN ZENG AS CO-CHAIR
UT researcher Yijian Zeng, from the Department of Water Resources at the ITC Faculty, has been elected as the Co-Chair of the International Soil Modelling Consortium (ISMC). For the coming three years, he will work together with top soil scientists towards an open âSoil Digital Twinâ, contributing to EU Soil Strategy for 2030 that aims to achieve healthy soils, which is at the heart of Europeâs twin green and digital transition.
2023-02-01: EcoExtreML in LSM Workshop Bonn
Towards a Digital Twin of Soil-Plant System (Importance of Leaf Water Potential)
2023-01-17: EcoExtreML in CRIB-OpenGeoHub workshop
Towards a Digital Twin of Soil-Plant System
2022-11-17: EcoExtreML in CRIB-eScience Center Minisymposium
EcoExtreML for a Soil-Plant Digital Twin
2021-12-03: UToday - Predicting Vegetation Health
UT scientist Yijian Zeng and his team, from the Faculty of Geo-information Science and Earth Observation (ITC), combined two existing computer models into a powerful tool to monitor the impact of drought and heat waves on vegetation. They received a three year grant from Netherlands eScience Center to further develop and refine their new model.
Link:Â https://www.utoday.nl/science/70731/predicting-vegetation-health
2021-11-17: ECOEXTREML KICK OFF
EcoExtreML kicked off successfully online.Â
2021-09-23: UT-News - Monitoring And Predicting The Effect Of Climate Extremes On Ecosystems
UT researchers Dr. Yijian Zeng, Prof. Dr. Bob Su, Dr. Christiaan van der Tol, Prof. Dr. Raul Zurita-Milla and Dr. Michael Ying Yang (all from ITC Faculty) have been awarded a grant from Netherlands eScience Centerâs ASDI 2020 call. In their research project, they will combine two computer models to understand how water-carbon dynamics of ecosystems vary with climate extremes, such as droughts and heatwaves.
Link:Â https://bit.ly/EcoExtreML-EN
There are three core components (see Figure1): i) The soil-plant model for a digital representation of the soil-plant system; ii) Physics-aware machine learning algorithms to approximate the soil-plant model; and iii) Data assimilation framework to digest Earth Observation data to update the states of soil-plant system. To address these digitalization challenges, the conceptual workflow for a soil-plant digital twin engine is presented in Figure 2, which consists of three pillars: âforward modellingâ, âphysics-aware machine learningâ, and âEarth Observationsâ.
Figure 2 Conceptual workflow of the soil-plant digital twin engine.
The coupled STEMMUS-SCOPE model is a digital replica of soil-plant system (see Figure 3). The coupled STEMMUS-SCOPE model integrates the SIF remote sensing with the plant-hydraulics-based SPAC model to advance our mechanism understanding of the complex soil-water-plant-energy interaction. The SIF remote sensing can acquire explicit information about photosynthetic light responses and steady-state behaviors in vegetation to evaluate photosynthesis and water-stress effects, across a range of biological, spatial and temporal scales. The plant-hydraulics-based SPAC model links mechanistically tissue-level stress to ecosystem-level water and carbon fluxes, via a resistor-based manner with the tissue-level hydraulic traits (of roots, stems and leaves) and stomatal optimality theory (i.e., photosynthetic gain vs. hydraulic risk).
Figure 3 Physically-based process model, STEMMUS-SCOPE, as a digital replica of soil-plant system (Wang et al. 2020 GMD)
The data assimilation technique (e.g., OpenDA) will be applied here to facilitate the data fusion, and this is actually creating the engine for the soil-plant digital twin. This digital twin engine will combine medium-resolution (10m â 1km) Copernicus satellite data with high-resolution (cm) unmanned aerial system (UAS) data, in situ observations, analytical- and physically-based models, as well as machine learning algorithms, for understanding ecophysiological dynamics of ecosystems and their responses to climate extremes across the globe. The OpenDA is an open interface standard for (and free implementation of) a set of tools to quickly implement data-assimilation and calibration for any dynamic models, and will be used to build the soil-plant digital twin engine.
