1. Introduction
Climate change is a threat to human well-being and the health of the planet. Past and current emissions have likely already committed society to further climate disruption to food, energy, and water systems, with the potential for cascading impacts including increased competition for scarce resources, civil unrest, violent conflict, and mass migrations increasing for higher future emissions (IPCC 2023a,b).
The need to build resilience to committed changes (adaptation) and reduce the risk of further change (mitigation) is “urgent but it is not too late” (Vallance and Belcher 2021). Achieving this imperative requires (i) improved understanding of vulnerabilities and risks to humans, infrastructure, and environmental systems; and (ii) improved ability to identify optimal adaptation and mitigation options across complex, interconnected systems. However, taking action requires decision-makers to balance many competing factors and uncertainties that are currently hard to quantify, particularly using information that is often noisy, incomplete, or difficult to access. In addition, best practice for climate-related decision-making currently lags behind other applications. This can lead to decision paralysis or reliance on heuristics that may be biased (Tversky and Kahneman 1974) resulting in maladaptations (e.g., Nalau et al. 2021), and underutilization of weather, climate, and other environmental information for decision-making (Findlater et al. 2021).
Moving beyond heuristics and making these critical decisions more tractable will require information and tools for accessing it that are suited to the scale and complexity of the challenges. Digital twins are ideally suited to this task because they provide dynamic, virtual representations of physical systems, making intelligent use of multidisciplinary data, knowledge, and high-fidelity simulations (Rasheed et al. 2020; Semeraro et al. 2021). While digital twins are established in some industrial sectors (e.g., manufacturing), they are an emerging concept in the environmental sciences, and practical demonstrations are currently limited. Exploratory steps range from planning “ambitious digital twins of planet Earth” (Voosen 2020) and scoping requirements for representing the world’s forests (Mõttus et al. 2021), to addressing specific challenges such as flood prediction (Huang et al. 2022), urban air quality (Topping et al. 2021), and improving efficiency of the agrifood sector (Tzachor et al. 2022). However, significant scientific, technical, and data challenges of representing complex environmental systems remain a barrier (Fuller et al. 2020; Juarez et al. 2021).
Identifying shared challenges and synergies across industry and environmental applications will help unlock the potential of digital twins. However, this requires concerted efforts to bridge the current scientific and technological gap between digital twins for industrial sectors and digital twins of the environment. To achieve this, we identify the need for “environment aware” digital twins (EA-DTs) that comprise weather, climate, and environmental information systems interacting with existing digital twins of environmentally sensitive systems, such as cities, ports, flood barriers, energy grids, transport networks, and food–energy–water networks EA-DTs maximizing the use of DTs to manage environmentally sensitive assets. Bringing together the expertise of environmental scientists with those developing digital twins for industrial sectors provides a pathway for maximizing the benefits of the technology. In particular, through codeveloping a shared practical understanding of information management, modeling capabilities, and theoretical understanding in a user-centric framework, environment-aware digital twins will help to accelerate and improve adaption and mitigation decisions. To support the wider development of environment-aware digital twins and facilitate discussions about best practice we propose a new ontology that describes the functionality required for different decision types and levels of uncertainty, and outline potential applications across sectors and time scales.
2. Moving toward EA-DTs
Digital twin technology was developed in the early 2000s in the manufacturing sector and defined as a “virtual, digital equivalent to a physical product” (Grieves 2015), and its use spread across industry (Colombo et al. 2017; Jiang et al. 2021). Since then, there have been numerous definitions (Barricelli et al. 2019; Jones et al. 2020; Semeraro et al. 2021) to help differentiate between digital models, shadows, and twins (Fig. 1; Table 1). Three common elements of digital twins are (i) a virtual model of a physical system that (ii) exchanges live/new data with the system, coupled with (iii) a user interface that supports decision-making and enables action. The second and third elements together distinguish digital twins from models and shadows.
Schematic of digital models, shadows, and twins following Department for Business Energy and Industrial Strategy (2022).
Citation: Artificial Intelligence for the Earth Systems 2, 4; 10.1175/AIES-D-23-0023.1
Description of digital models, shadows, and digital twins.
More recently digital twin technology has received increasing attention because of its potential to deliver a step change in our understanding of Earth and environmental science (Blair 2021), and provide corresponding breakthroughs in our ability to address climate change risks. This ambition is reflected in projects such as Destination Earth that aims to develop a highly accurate digital model of Earth to monitor and predict environmental change and human impact to support sustainable development (European Commission 2021), and is designed to offer “a digital modeling platform to visualize, monitor and forecast natural and human activity on the planet” (Nativi et al. 2021). However, realizing this potential at scale will require two major breakthroughs in today’s information systems: (i) for information completeness and quality, and (ii) for information access, intervention, and exchange (Bauer et al. 2021). Overcoming these challenges may take time and will require targeted scientific and technological development. In the meantime, there are opportunities to move beyond traditional modeling approaches (digital models and shadows) for informing adaptation and mitigation decisions by developing digital twins of environmentally sensitive assets and systems.
3. Building environmental information into digital twins for decision-making
As weather extremes become more frequent and severe, there will be increasing socioeconomic benefits from building cross-time-scale weather, climate, and environmental information into digital twins of these critical systems such as energy grids, manufacturing systems, transport networks, and food–energy–water networks. Developing these environment-aware digital twins will require partnerships across industry and the environmental sciences to build more complete, accurate and accessible representations of the real world, supporting a wider range of practical applications for decision-making. Integrating digital twin technology with environmental data will also stimulate necessary advances in uncertainty quantification and the decision sciences that are needed to fully utilize the information. This will help to ensure that critical assets, systems, and processes are safer, more efficient, sustainable, and resilient to the challenges of today’s weather and tomorrow’s climate.
EA-DTs could be achieved by linking existing digital twins with information systems that generate and transfer the weather, climate, and environmental information. An EA-DT for a physical asset would offer significant benefits for decision-making in comparison with offline modeling or heuristic approaches by more fully quantifying risks and uncertainties in the modeling and data, thereby improving the asset’s operational performance and extending its lifetime. Achieving these benefits will require high-performance interoperable scientific, technology, and software systems that allow federation of very different modeling approaches, alongside integration, exchange, and transmission of huge quantities of data with varying origin, type, and quality. These are significant transdisciplinary challenges, and overcoming them in practical and sustainable ways will require significant advances in our current understanding, including guidelines and standards for the governance of infrastructure, software, computing and data resources, and appropriate regulation/legislation.
Figure 2 illustrates a high-level representation of an EA-DT for a physical asset, for example, an energy grid. In this example, the state of the physical asset is a function of design, usage, maintenance history, weather, and environmental hazards. Noisy sensors measure the condition of the asset, and observable inputs to the system, including environmental conditions. The digital asset model assimilates these measurements in near–real time to compute a more complete representation of the asset’s current state, including quantities that cannot be directly measured such as predicted asset lifetime. In addition, the digital asset model can be stress tested by ingesting weather, climate, and environmental scenarios that could occur during the asset’s lifetime. Each model and data component will have associated uncertainties and limitations, highlighting the need for effective data quality and information management frameworks. Errors and uncertainties in the digital asset model, environmental data sources, and other system inputs can be evaluated and updated regularly by comparing predicted and observed states of each asset component and model input. The feedback permits user knowledge, machine learning, and artificial intelligence to continuously refine the digital asset model and improve its performance across time scales.
Example schematic showing the main components of an environment-aware digital twin.
Citation: Artificial Intelligence for the Earth Systems 2, 4; 10.1175/AIES-D-23-0023.1
The EA-DT interface links the different components and allows users to select data sources including environmental hazard scenarios (e.g., using weather forecasts to multidecadal climate projections); quantify risks and uncertainties and explore the asset’s vulnerabilities; and test and implement decisions, incorporating high-quality information generated by the digital model, external considerations, and subjective expertise about additional uncertainties and important ethical considerations. Identifying decisions and effective solutions will require significant advances in the decision sciences, particularly our knowledge of complex optimization and predictive control problems under risk and uncertainty on longer time scales. This will require new theoretical and practical insights that make better use of the huge amounts of information available, drawing heavily on explainable machine learning and artificial intelligence approaches to build trust.
To support the use of digital twins for managing environmental risks to physical systems, we propose a new ontology for EA-DTs, focusing on applications of weather and climate information.
4. A new digital twin typology to support EA-DTs
Adopting a new technology often requires a period of exploration to establish user-relevant definitions of terms. This has been the case for digital twins, where conflating digital models, shadows, and twins, alongside varying levels of technological maturity, runs the risk of ambiguity or hyperbole, which could alienate potential users (Wright and Davidson 2020). In addition, while there is growing interest in digital twins for weather and climate applications, there is little clarity on which digital twin characteristics lend themselves to applications across different time scales and, perhaps more importantly, decision types. Without this clarity, the environmental science community risks failing to capitalize on technological developments that have the potential to help translate science into action.
We propose that digital twin functionality exists across a spectrum from digital twins for industry to digital twins of the environment, with environment-aware digital twins bridging the wide gap between them. The broad range of possible EA-DT applications lends itself to categories that describe both the application and utility of the digital twin for decision-support. Using categories to describe stages of technology application has been successfully applied to other transformative technologies (e.g., Czech et al. 2018) and represents a critical first step toward realizing the opportunity of digital twins in supporting decision-making.
Table 2 outlines three broad categories of EA-DTs linked to different decision characteristics and uncertainties: (i) human-in-the-loop digital twin, (ii) semiautomated digital twin, and (iii) fully automated digital twin. In addition to improved decision-making, consistently applying this ontology will help manage expectations about what the technology can currently deliver, prioritize future developments, match tools to challenges, and avoid overpromising or costly, overengineered solutions.
New ontology for categories of environment-aware digital twin (EA-DT).
5. Summary
Digital twins can be part of a transdisciplinary approach that makes adaptation of physical systems and climate change mitigation decisions easier to identify and implement at scale—integrating data, models, and theoretical understanding in a user-centric framework. However, while the diverse applications of digital twins are increasingly recognized in many sectors (Rasheed et al. 2020), we suggest that their utility has so far been underestimated within the environmental sciences, and their development held back partly because of the inherent transdisciplinary challenges of modeling complex environmental systems and associated information management requirements. For example, digital twins of the environment must address sustainable requirements for high-performance hardware and software (information technology infrastructure, including options for provision of on-demand infrastructure at cost through the “cloud”), architecture, data collection, quality control, and optimization so that the highest quality data are fed into the artificial intelligence algorithms underpinning digital twin technology. These concerns, and others, are well addressed elsewhere (Blair et al. 2019; Blair 2021; Fuller et al. 2020; Juarez et al. 2021; Rasheed et al. 2020).
As a route forward for bridging the scientific and technological gap between digital twins for industrial sectors and digital twins of the environment, we identify the need for “environment aware” digital twins that federate weather, climate, and environmental information with existing digital twins of environmentally sensitive systems. Alongside this we present a new ontology to assist thinking about best practice, design, and applications of EA-DTs. With this new ontology in place, and a strong commitment to make better use of continued advances in weather, climate, and environmental sciences, digital twins can transform how we address some of the most serious environmental challenges facing society. We call on the research and user communities to take the next step in developing EA-DTs to urgently address climate challenges.
Acknowledgments.
We are grateful to Kirsty Lewis for discussions and a review of the final draft. Authors Kirstine I. Dale, Edward C. D. Pope, Aaron R. Hopkinson, Theo McCaie, and Jason A. Lowe were supported by the Met Office Hadley Centre Climate Programme funded by BEIS and Defra.
Data availability statement.
Data sharing is not applicable (e.g., for review articles or theory-based articles) because no datasets were generated or analyzed during the current study.
REFERENCES
Barricelli, B. R., E. Casiraghi, and D. Fogli, 2019: A survey on digital twin: Definitions, characteristics, applications, and design implications. IEEE Access, 7, 167 653–167 671, https://doi.org/10.1109/ACCESS.2019.2953499.
Bauer, P., B. Stevens, and W. Hazeleger, 2021: A digital twin of Earth for the green transition. Nat. Climate Change, 11, 80–83, https://doi.org/10.1038/s41558-021-00986-y.
Blair, G. S., 2021: Digital twins of the natural environment. Patterns, 2, 100359, https://doi.org/10.1016/j.patter.2021.100359.
Blair, G. S., P. Henrys, A. Leeson, J. Watkins, E. Eastoe, S. Jarvis, and P. J. Young, 2019: Data science of the natural environment: A research roadmap. Front. Environ. Sci., 7, 121, https://doi.org/10.3389/fenvs.2019.00121.
Colombo, A. W., S. Karnouskos, O. Kaynak, Y. Shi, and S. Yin, 2017: Industrial cyberphysical systems: A backbone of the fourth industrial revolution. IEEE Ind. Electron. Mag., 11, 6–16, https://doi.org/10.1109/MIE.2017.2648857.
Czech, P., K. Turoń, and J. Barcik, 2018: Autonomous vehicles: Basic issues. Zesz. Nauk. Transp. Politech. Śląska, 100, 15–22, https://doi.org/10.20858/sjsutst.2018.100.2.
Department for Business, Energy and Industrial Strategy, 2022: Enabling a national cyber-physical infrastructure to catalyse innovation: Understanding the opportunities for connected digital twins and other advanced cyber-physical systems. U.K. Government, 58 pp., https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1071905/enabling-national-cyber-physical-infrastructure.pdf.
European Commission, 2021: Destination Earth. Accessed 8 May 2023, https://digital-strategy.ec.europa.eu/en/library/destination-earth.
Findlater, K., S. Webber, M. Kandlikar, and S. Donner, 2021: Climate services promise better decisions but mainly focus on better data. Nat. Climate Change, 11, 731–737, https://doi.org/10.1038/s41558-021-01125-3.
Fuller, A., Z. Fan, C. Day, and C. Barlow, 2020: Digital twin: Enabling technologies, challenges and open research. IEEE Access, 8, 108 952–108 971, https://doi.org/10.1109/ACCESS.2020.2998358.
Grieves, M., 2015: Digital twin: Manufacturing excellence through virtual factory replication. Digital Twin White Paper, 8 pp., https://www.3ds.com/fileadmin/PRODUCTS-SERVICES/DELMIA/PDF/Whitepaper/DELMIA-APRISO-Digital-Twin-Whitepaper.pdf.
Huang, T., and Coauthors, 2022: An Earth system digital twin for flood prediction and analysis. IGARSS 2022–2022 IEEE Int. Geoscience and Remote Sensing Symp., Kuala Lumpur, Malaysia, Institute of Electrical and Electronics Engineers, 4735–4738, https://doi.org/10.1109/IGARSS46834.2022.9884830.
IPCC, 2023a: Climate Change 2022: Impacts, Adaptation and Vulnerability. Cambridge University Press, 3056 pp., https://doi.org/10.1017/9781009325844.
IPCC, 2023b: Climate Change 2022: Mitigation of Climate Change. Cambridge University Press, 2030 pp., https://doi.org/10.1017/9781009157926.
Jiang, Y., S. Yin, K. Li, H. Luo, and O. Kaynak, 2021: Industrial applications of digital twins. Philos. Trans. Roy. Soc., A379, 20200360, https://doi.org/10.1098/rsta.2020.0360.
Jones, D., C. Snider, A. Nassehi, J. Yon, and B. Hicks, 2020: Characterising the digital twin: A systematic literature review. CIRP J. Manuf. Sci. Technol., 29, 36–52, https://doi.org/10.1016/j.cirpj.2020.02.002.
Juarez, M. G., V. J. Botti, and A. S. Giret, 2021: Digital twins: Review and challenges. J. Comput. Inf. Sci. Eng., 21, 030802, https://doi.org/10.1115/1.4050244.
Mõttus, M., and Coauthors, 2021: A methodology for implementing a digital twin of the Earth’s forests to match the requirements of different user groups. GI Forum, 9, 130–136, https://doi.org/10.1553/giscience2021_01_s130.
Nalau, J., E. Torabi, N. Edwards, M. Howes, and E. Morgan, 2021: A critical exploration of adaptation heuristics. Climate Risk Manage., 32, 100292, https://doi.org/10.1016/j.crm.2021.100292.
Nativi, S., P. Mazzetti, and M. Craglia, 2021: Digital ecosystems for developing digital twins of the Earth: The Destination Earth case. Remote Sens., 13, 2119, https://doi.org/10.3390/rs13112119.
Rasheed, A., O. San, and T. Kvamsdal, 2020: Digital twin: Values, challenges and enablers from a modeling perspective. IEEE Access, 8, 21 980–22 012, https://doi.org/10.1109/ACCESS.2020.2970143.
Semeraro, C., M. Lezoche, H. Panetto, and M. Dassisti, 2021: Digital twin paradigm: A systematic literature review. Comput. Ind., 130, 103469, https://doi.org/10.1016/j.compind.2021.103469.
Topping, D., T. Bannan, H. Coe, J. Evans, C. E. Jay, E. Murabito, and N. Robinson, 2021: Digital twins of urban air quality: Opportunities and challenges. Front. Sustainable Cities, 3, 786563, https://doi.org/10.3389/frsc.2021.786563.
Tversky, A., and D. Kahneman, 1974: Judgment under uncertainty: Heuristics and biases. Science, 185, 1124–1131, https://doi.org/10.1126/science.185.4157.1124.
Tzachor, A., C. E. Richards, and S. Jeen, 2022: Transforming agrifood production systems and supply chains with digital twins. npj Sci. Food, 6, 47, https://doi.org/10.1038/s41538-022-00162-2.
Vallance, P., and S. Belcher, 2021: The need for action on climate change is urgent. U.K. Government, accessed 20 September 2022, https://www.gov.uk/government/speeches/the-need-for-action-on-climate-change-is-urgent.
Voosen, P., 2020: Europe is building a ‘digital twin’ of Earth to revolutionize climate forecasts. Science, 1 October, https://doi.org/10.1126/science.abf0687.
Wright, L., and S. Davidson, 2020: How to tell the difference between a model and a digital twin. Adv. Model. Simul. Eng. Sci., 7, 13, https://doi.org/10.1186/s40323-020-00147-4.