Digital Twins

Overview

Digital Twins are digital models of physical systems, processes or objects, aiming to accurately reflect the current state of the subject. They allow large volumes of data to be coalesced into singular models for highly complex simulation and prediction of the subject’s current and future states in response to a variety of factors. The notion of a digital twin originated in response to the Apollo 13 oxygen tank explosion, where a ‘living model’ of the Apollo mission was created in an attempt to evaluate the failure. This model consisted of a physical replica of the vehicle as well as digital components allowing analysis of the failure and partial simulation of potential solutions. As seen in this example, complete replicas, digital or otherwise, are invaluable in testing products, processes and scenarios. This can apply across a variety of fields, from agriculture, water use and disaster monitoring to manufacturing, urban planning and healthcare.

One specific use of this emerging technology has been in digital twins of the Earth. These models are partial digital representations of the Earth and its systems. They enable monitoring, forecasting and assessment across a range of fields, including climate phenomena, hydrological processes and ecological systems. Digital twins are often constructed as field-specific tools, modelling singular phenomena on a global scale, aiding simplicity and data requirements. Examples of these specific digital twins are the Weather-Induced Extremes Digital Twin, which is designed to support responses to meteorological, hydrological and air quality extremes, and the Climate Change Adaptation Digital Twin, which aims to model the impacts of anthropogenic climate change over a multidecadal timescale. Both of these models are a part of the EU Destination Earth (DestinE) initiative, which aims to create a general digital twin of the entire Earth system. 4) 9) 13) 18)

The Earth System Digital Twins (ESDT) project from NASA defines three main components of a functional digital twin:

• The digital replica of an Earth system: constructed by continuous, near real time observations of relevant variables, defined by the chosen system being modelled. These observations undergo a process of data assimilation and fusion, providing an accurate representation of the current system state.
• Dynamic forecasting models: using historical data and current observations to predict future states of the desired system.
• Impact assessment capabilities: incorporating aspects of both the digital replica and the forecasting models through advanced computational and visualisation capabilities to quickly run a large number of simulated outcomes across a range of spatial and temporal scales, allowing the investigation of multiple hypothetical future scenarios. 1) 2) 4) 8) 10) 14) 18)

Importance and Applications

Digital twins have immense potential for revolutionising both the understanding of Earth systems and the predictive ability of current forecasting models. Through advancements in remote sensing applications and the increasing volume of data collected, it has become possible to extract valuable insights and information, including more accurate and timely predictions of Earth system events.

​

A key example of this is climate modelling, as shown above, particularly with regards to anthropogenic climate change. A digital twin of the Earth’s climate and atmosphere would allow predictions of the Earth’s state under different emissions scenarios, potentially providing clearer targets for limiting climate change, and furthering an understanding of the true threat of a changing climate. Additionally, digital twins created for systems such as Earth’s hydrosphere could aid in managing the evolving challenges surrounding sea level rises and water use by predicting droughts or floods, as well as monitoring freshwater availability and ensuring sustainable distribution in water scarce areas. Digital twins of different biomes and ecosystems can similarly aid in modelling deforestation and biodiversity loss, as well as providing valuable visualisations for public education and environmental stewardship messaging. Another valuable aspect of accurate digital twins is their role in disaster management. During natural disasters such as hurricanes, cyclones and other extreme weather events, digital twins of climate or hydrologic systems can help to more accurately predict the impact and course of such disasters, aiding relief and evacuation efforts. 1) 4) 8) 13) 14)

The creation of successful digital twins is thus a significant step in developing a complete understanding of the Earth and its processes. Through the wealth of Earth observations now available, digital twins have the potential to revolutionise modelling and forecasting, driving more reliable projections of Earth’s future state. 4) 14)

Examples

Destination Earth (DestinE)

DestinE is the European project  aiming to create a full digital replica of the Earth by 2030. Initiated in 2022, the project is managed under a partnership between the European Space Agency (ESA), the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) and the European Centre for Medium-Range Weather Forecasts (ECMWF). DestinE consists of three main components:

• A public platform, providing access to modelling, real time variable monitoring and impact assessment capabilities;
• A ‘data lake’, the pooled Earth observation data from ESA, EUMETSAT and ECMWF, as well as data from Copernicus and a range of other sources; and
• The digital twin itself. Under the program, a variety of different models covering various Earth systems and regions will be created, contributing to the ‘full’ digital replica. 3) 8) 10) 13) 15) 17)

As part of DestinE, several of these narrower twins have already been created or are being developed:

Two other digital twins are being developed by ECMWF, the Weather Induced Extremes Digital Twin (Extremes DT) and the Climate Change Adaptation Digital Twin (Climate DT). Extremes DT aims to provide kilometre level spatial resolution simulations, with 5 day forecasting being run daily at a spatial resolution of 4.4 km on the ECMWF Atos supercomputer since August 2023. Climate DT aims to provide annual climate simulations on a multi-decade scale. These projections will be made with local granularity, incorporating sector specific information to relevant fields, such as monitoring of renewable energy resources, the Earth’s hydrological processes and land use for applications such as urban planning and agriculture. 1) 2) 3) 5) 8) 9) 10) 17)

​

Earth System Digital Twin (ESDT)

The Earth System Digital Twin (ESDT) effort is an Advanced Information Systems Technology (AIST) initiative that aims to:

1. Develop information system frameworks to provide continuous and accurate representations of systems as they change over time;
2. Mirror various Earth Science systems and utilize the combination of Data Analytics, Artificial Intelligence, Digital Thread, and state-of-the-art models to help predict the Earth’s response to various phenomena;
3. Provide the tools to conduct "what if" investigations that can result in actionable predictions.

The end goal of the AIST ESDT project is the creation of digital twins that integrate relevant Earth system models and simulations, other relevant models, continuous observations, long term records, and analytics and artificial intelligence tools. Such a model will allow both real time monitoring and modelling of future hypothetical scenarios. To achieve this goal, AIST has identified three areas of investment:

• Novel Observing Strategies (NOS): coordination and operation of various observing assets, such as space based Earth observation, on the ground weather and climatological measurement or other sources of relevant data.
• Digital Twin Infrastructure: tools used to assess uncertainties and causality in observed data, often through machine learning, as well as the processes of data visualisation and leveraging insights gained from the creation of digital twins.
• Analytic Collaborative Frameworks (ACF): simplifying access to diverse and large quantities of data, as well as the necessary analytics and modelling tools to meaningfully interpret such data. 6) 7) 11) 12)

To target these areas, AIST has selected 14 projects across the fields of ESDT infrastructure, AI-Surrogate Modelling for ESDT, Analytics Frameworks Development towards ESDT and ESDT prototypes. The details for these selected projects are shown below.

Nvidia Earth-2

Earth-2 is a climate focussed digital twin developed by technology company Nvidia. It forecasts global climate change scenarios, simulates climate processes at unprecedented scales and aims to provide insights into the mitigation of climate risks. The project is a digital twin cloud platform combining artificial intelligence, simulation and modelling methods, and computer graphics tools, allowing visualisation and forecasting of weather and climate processes. 19) 20)

​

Earth-2 has four core features: high-resolution simulations, AI-driven analysis, integration of big data, and scalability. The model provides ultra-high resolution climate models that capture processes on a global scale. Higher resolution modelling and monitoring not only provides higher accuracy, capturing smaller-scale processes, but also enhances the predictive capabilities of the model as a whole, by reducing uncertainty in key variables, such as wind speed or precipitation patterns. Earth-2 will also utilise data from a variety of sources as inputs, using Earth observation satellite data, historical records and on the ground meteorological information from sources such as weather stations or buoys, to produce its real time twin of the Earth’s climate. The combination of these disparate data sources into a unified framework demands significant computing power, so the model will make use of Nvidia’s Deep GPU Xceleration (DGX) deep learning system. These supercomputers are optimised for parallel processing, and can therefore handle the computational requirements of global climate simulations. 19) 20) 21)

Earth-2 aims to be a valuable policy tool for decision makers to better understand, model and predict a changing climate and the impacts it has had and will have. It aims to provide early warning systems and management strategies for natural disasters, predict global and regional climate trends such as heat waves or droughts, and track ecosystem changes, such as deforestation or polar ice cap trends. 19) 21)

References

1) “Building a highly accurate digital twin of the Earth.” Destination Earth, URL: https://destination-earth.eu/

2) “Climate Change Adaptation Digital Twin: a window to the future of our Planet.” ECMWF, 30 April 2024, URL: https://destine.ecmwf.int/news/climate-change-adaptation-digital-twin-a-window-to-the-future-of-our-planet/

3) “DestinE Uses.” ECMWF, URL: https://destine.ecmwf.int/destine-uses/

4) “Digital Twin Earth: the next-generation Earth Information System.” Frontiers, 5 March 2024, URL: https://www.frontiersin.org/journals/science/articles/10.3389/fsci.2024.1383659/full

4) “Digital Twin Engine | Destination Earth.” Destination Earth, URL: https://destine.ecmwf.int/digital-twin-engine/

6) “Digital Twins and Living Models at NASA.” NASA Technical Reports Server, 1 Novemeber 2021, URL: https://ntrs.nasa.gov/citations/20210023699

7) “Digital Twins and Living Models at NASA.” NASA Technical Reports Server, 3 November 2021, URL: https://ntrs.nasa.gov/api/citations/20210023699/downloads/ASME%20Digital%20Twin%20Summit%20Keynote_final.pdf

8) “The Digital Twins | Destination Earth.” Destination Earth, URL: https://destine.ecmwf.int/digital-twins/

9) “A digital twin to sharpen our vision of extreme weather.” ECMWF, 4 March 2024, URL: https://destine.ecmwf.int/news/a-digital-twin-to-sharpen-our-vision-of-extreme-weather/

10) “Earth System Digital Twin.” Earth Science Technology Office, URL: https://esto.nasa.gov/earth-system-digital-twin/

11) “Earth System Digital Twins (ESDT) Technology for NASA Earth Science.” NASA Technical Reports Server, 17 May 2022, URL:  https://ntrs.nasa.gov/citations/20220007620

12) “Earth Systems Digital Twins (ESDT).” NASA Technical Reports Server, June 2023, URL:  https://ntrs.nasa.gov/api/citations/20230010074/downloads/2023-06-15_Keynote-STC-DT-Webinars_Final.pdf

13) “ESA - Digital Twin Earth.” European Space Agency, URL: https://www.esa.int/ESA_Multimedia/Images/2020/09/Digital_Twin_Earth

14) Hazeleger, W., et al. “Digital twins of the Earth with and for humans.” Communications Earth and Environment, 27 August 2024, URL: https://www.nature.com/articles/s43247-024-01626-x

15) “Phase Two of Destination Earth Confirmed.” ECMWF, 13 December 2023, URL: https://stories.ecmwf.int/phase-2/index.html

16) “Project Selections for AIST-21 - NASA Earth Science and Technology Office.” NASA ESTO, 4 May 2022, URL: https://esto.nasa.gov/project-selections-for-aist-21/

17) “Understanding DestinE's DIGITAL TWINS.” ECMWF, URL: https://stories.ecmwf.int/explainer-digitaltwins/index.html

18) “Working towards a Digital Twin of Earth.” European Space Agency, 14 October 2021, URL: https://www.esa.int/Applications/Observing_the_Earth/Working_towards_a_Digital_Twin_of_Earth

19) “Nvidia Earth-2.” Climate Science Fair, URL: https://climatesciencefair.emersoncollective.com/nvidiaearth-2

20) “Earth-2.” Nvidia, URL: https://www.nvidia.com/en-au/high-performance-computing/earth-2/

21) “NVIDIA Announces Earth Climate Digital Twin.” NVIDIA Newsroom, 18 March 2024, URL: https://nvidianews.nvidia.com/news/nvidia-announces-earth-climate-digital-twin