The new problematic is the elaboration of coupled models across spatio-temporal scales, therefore involving the use high spatial resolution and dense time series from multiple EO missions and diverse non-traditional data sources including seismic infrasound records. This is the new challenge of Big EO Data. The tutorial introduces and explains solution based on AI for the DT design. An DT is the convergence of the sensing physical mechanisms tightly connected, communicating and continuously learning, from and with mathematical models, data analytics, simulations and user interaction.
The presentation covers the major developments, of hybrid, physics aware AI paradigms, at the convergence of forward modelling, inverse problem and machine learning, to discover causalities and make prediction for maximization of the information extracted from EO and related non-EO data. The majority of EO applications or services require the complementary EO multi-sensor and non-EO data, i.e., sensor fusion and multitemporal observations. The tutorial explains how to automatize the entire chain from multi-sensor EO and non-EO data, to physical parameters, required in applications by filling the gaps and generating a relevant, understandable layers of information.
The DT are technologies looking o the evolution of EO at least for the horizons of next two decades. The explosive present advance of AI methods was obtained thanks to mainly two factors, the advancement of theoretical bases and performance evolution of the IT, i.e. computation, storage and communication.
The methods and techniques presented will cover topics as:
- Counterfactual explanations in EO applications
- Explainable AI for adaptation at climate changes effects at scales of human actvity
- Physics-aware AI for EO and in-situ diverse sensor data
- Analyzing the possible effects of improper explanations
As a practical illustration, we will present a first case study of an extreme meteorlogical event that impacted Romania’s Black Sea coast, setting new records for precipitation and emphasizing the growing need for improved monitoring of extreme weather events. In this example, we demonstrate how diverse non-traditional data sources—including seismic infrasound, and satellite observations can be integrated to track the full lifecycle of a storm. Infrasound microseismic bands (0.1–1 Hz) captured the distinct phases of the storm and showed the coupled atmosphere-lithosphere interactions. This case study demonstrates how integrating heterogeneous, physics-rich datasets can significantly improve the detection, characterization, and early warning of extreme weather events, offering a powerful example of the potential of Digital Twins for real-world climate adaptation strategies.
A second case study will be extended to bio-, chemical, physical parameters extraction from Sentinel-1 and Sentinel-2 products, and the interactive analysis of their spatio-temporal evolution in relation to meteorological parameters, as temperature, precipitations or wind parameters. A new paradigm of Virtual Sensing is used to predict un-observed signatures, as training a DNN model with Lidar measurements and applying it to Sentinel-2 for canopy height prediction. The demonstration is focused to produce adaptation knowledge as requested for actions aiming to protect biodiversity and ecosystems. Further the eutrophication of sweet-water lakes is addressed analyzing time series of cyanobacteria, turbidity or water indexes in spatio-temporal context to detected causalities. EO support for OneHealth applications are also envisaged to anticipate fight vector-borne emerging animal and zoonotic infectious diseases, implementing an integrated approach considering the links between human health, animal health, and environmental health. The coupled DTs components will support and promote a geographical diversity approach, involving various regions and communities, following a systemic approach converging several cross-modalities themes and areas of innovation, implemented as an inclusive methodology to bring together public administrations, private sector, civil society, and finally the citizens in person.