Droughts and heat waves jeopardize terrestrial ecosystem carbon sequestration and hinder EU's goal of being climate-neutral by 2050. Developing an open digital twin of the soil-plant system can help monitor and predict the impact of extreme events on ecosystem functioning. We illustrate how our recently developed STEMMUS-SCOPE model, via linking comprehensive soil-plant processes to novel satellite observables (e.g. solar-induced chlorophyll fluorescence), contributes to building such a digital twin. This approach allows a mechanistic window for tracking above- and below-ground ecophysiological processes with remote sensing techniques. Following Open Science and FAIR principles for data and research software, we present the soil-plant digital twin's building blocks that include three pillars: process-based soil-plant model, physics-informed machine learning, and the assimilation of Earth Observation data.
Applying the soil-plant digital twin to simulate the ecosystem's water-energy-carbon fluxes facilitates a swirled evolving loop between the digital twin and the soil-plant physical twin, in terms of enhancing the digital representation of physical system. Such swirled evolving process pushes the frontiers of process-based model developments, for example, to include dynamic vegetation growth, integrated unsaturated-saturated processes, and explicit plant hydraulic pathways into the STEMMUS-SCOPE model. However, it also leads to a major bottleneck of applying such advanced process-based model at regional to global scale, due to the expensive computational cost. The machine learning algorithm helps enable a computationally effective yet physically consistent technique to approximate the original model with a surrogate model to bypass such computational burden. This study emphasizes the importance of FAIR-enabling digital technologies, which translate research needs and developments into reproducible and reusable software, data and knowledge.