Will more data rescue us from the climate and nature emergencies?
This article considers that more data is not needed to tell us about the climate and nature crises. It argues that we need data to help us figure out the solutions and actions. How do we transition the data discussion away from the problem and focus on how data can support environmental solutions?
Digital resources have the potential to transform our knowledge and understanding of the environment, nature and their relationships with people. However, to extract the embedded intelligence, we need to improve discoverability and become better at connecting quite disparate data sources. The development of new technology also means that more and more data are becoming immediately available (at least potentially). Data are an immensely valuable asset, but only if we use them well, there is also the potential for misinformation. This article outlines the current digital landscape and reflects on the near future.
Provocation
Statement 1: We need more data to tell us about the climate and nature crises.
Statement 2: We do not need more data to tell us that we are in crisis. We need data to help us figure out the solutions and actions.
The provocation is intended to re-pivot our thinking about environmental data, and to consider a slightly different question: “how do we transition the data discussion away from the problem and focus on how data can support environmental solutions?”
Introduction
The rationale for much environmental data is often on setting a baseline, reporting against a target for condition or status, and tracking changes – typically describing a trend (improving, stable or deteriorating) through time. Data gives clear sight of the trajectory, looking backwards (retrospective), but also looking forward (prospective) and evidencing the effect of an intervention. The latter use case needs careful design in the collection of data, indeed careful design is important in all situations as is being clear what the purpose for the data collection is. The precision (or uncertainty) needed to address the purpose is equally important, but often neglected, and is fundamental to ‘why we might need more data’.
Who sets the agenda for data, and defines what is needed and from where?
From a research perspective, this is determined by the investigator team and the scientific questions being addressed in the research project. There would not always be a recognition of the bigger picture or of the benefits of designing something which provide data that dovetail with other campaigns.
From a regulatory perspective, this is driven by policy, linked to any targets which are set, and by reporting requirements. It is perhaps worth reflecting that in many of the EU Directives, which dominated water regulation (freshwater, bathing water, drinking water, groundwater) in the UK for decades, a minimum requirement for the monitoring was defined. Member States did however have some freedom in the detail. Again, historically there was no push to think about the bigger environmental picture (a very simple example could be the situation that biology and water quality monitoring are infrequently co-located).
Regardless of the perspective, it is important to recognise that the nature of the question being addressed is critical in determining the data which will be needed. It is also true to say that there are monitoring and network designs with known statistical properties which have been used by regulatory agencies (e.g. Generalized Random Tessellation Stratified (GRTS) used by the US EPA and also more recently by the EA).
“Longer term, we are investing in ways to improve our river monitoring and give a fuller, more accurate picture of the problems in England’s water bodies. For example, our new long-term ‘river surveillance network’ monitoring programme (based on GRTS) for rivers will give a national overview of the state of England’s rivers. This network will provide a statistically robust assessment of the health of rivers and how this changes over time at a national scale and over the long term”.
But the hypothetical, perfect scheme also needs to be practicable and feasible given budget, which means changes are often made to deliver hybrid schemes.
In the UK, there are many monitoring schemes delivering environmental data on water, air, biodiversity etc. Some have been in operation for many decades, all are likely to have changed over that time as the techniques and technologies available have changed.
Today, we live in a time when more and more environmental data are being generated using new technologies, whether these be new sensors, drones, new satellite missions or through citizen science.
The collection of more data has, in that sense, become easier, and ‘data lakes’, or ‘data deluges’ are terms that are sometimes used to describe our current situation. But the question is, has this revolution or evolution in data generation delivered any change in our understanding of what is happening and why; something I will return to later in this piece.
What is a digital earth?
The data deluge, has led to a vision of a digital earth, a phrase which is growing in use – but what does it mean? Within the EU (Nativi et al., 2021), the “Digital Earth is a concept describing an interactive digital replica of the entire planet that can facilitate a shared understanding of the multiple relationships between the physical and natural environments and society.” A digital earth enables us to think about the earth system, where “Earth system science (ESS) is a rapidly emerging transdisciplinary endeavour aimed at understanding the structure and functioning of the Earth as a complex, adaptive system. The grand challenge for ESS is to achieve a deep integration of biophysical processes and human dynamics to build a truly unified understanding of the Earth System” (Steffen et al., 2020). So we could argue, the deluge offers us a potential route to discovering solutions.
In this sense, a simple answer to the question why do we need (more) data is that from our data grows increasing understanding and better intelligence about our planet and nature. This enables us to think about systems, and become aware of sometime hidden interconnections. Armed with this enhanced knowledge, we can hope to make better decisions.
The provocation could end there, with this simple explanation. However, there are challenges when we have these growing bodies of data, collected in different ways and by different organisations and individuals, and using new technology. These are at least four-fold: locating them; being sure of their provenance; understanding their sampling frame and hence their properties; and being able to link them together. Increasingly in the landscape of ‘big data’ there are also overarching concerns around governance, privacy and security, and ethics in terms of data acquisition and use. There are a number of governmental, NGO and expert bodies settings standards and defining strategies around open and geospatial data (geospatial simply put means that the data has a geographic location), use of AI, and open data.
As one example, the Geospatial Commission is an expert committee responsible for setting the UK’s geospatial strategy and coordinating public sector geospatial activity. The UK Geospatial Strategy 2030 presents the government’s plan “to unlock the power of location data and technologies across the UK economy. The strategy proposes harnessing technologies such as AI, satellite imaging and real-time data to boost location-powered innovation”.
Closer to home, the Scottish Science Advisory Committee (SSAC) produced a report in 2021 “Future Landscapes: Report on Geospatial Knowledge” and identified a key set of recommendations and gaps including:
- a central data directory for all spatially referenced data with links to existing cross-sector data portals
- Government-held or funded data catalogues should provide clearer user guidance, and follow GEMINI/ INSPIRE standards; Gemini is the UK standard for spatial metadata, while INSPIRE legislation requires most UK public bodies (central government, local authorities and all other public authorities as defined in the Environmental Information Regulations 2004) to publish certain datasets.
- A Scotland-wide geospatial framework should be established to combine expertise and datasets across policy areas, starting with the environment,
- Long-term data collection projects should be given adequate priority by public funders, with in-built flexibility in the collection process, to sustain continued dynamic data collection.
Principles around data have also been developed, with perhaps the best known being the FAIR 2016 principles – Findable, Accessible, Interoperable and Re-usable.
Findable addresses the issue of there being many holders/curators of data (sometimes we talk about data siloes), how do we identify where the relevant data are for any particular question?
Accessible tackles the challenge, that not all data are open and freely available, there may be licence conditions, paywalls etc.
Interoperable and re-usable are linked, the ideal situation is that the many different data streams ‘can talk with each other’, this means common language (ontology), similar frames of reference (geographically and temporally) and ultimately that what is collected for one purpose can be re-used/re-purposed. Data have many attributes, they will have been gathered for a primary purpose, but the ambition is that they can be used for other purposes (some of which may be unexpected or unforeseen).
Open data has been a strategy of the Scottish Government since 2015. Open data is defined as non-personal and non-commercially sensitive data which are findable, accessible, and with open licensing, encouraging reuse.
In the context of AI (artificial intelligence), which is also potentially transforming how we use data, there are similarly fast moving developments, for example the UK Government White Paper on AI regulation in 2023 and the EU Artificial Intelligence Act which came into force in August 2024 to foster responsible artificial intelligence development and deployment in the EU. The ethical and potential bias implications in the use of AI are also topics of considerable concern.
Such a plethora of activity reflects the ambition in the digital space, and emphasises that developments are fast moving. In Scotland (and the UK more generally), our environmental and ecological data are often held in diverse repositories (NERC Environmental Data Centres, in regulatory agencies, city and regional councils, in commercial organisations and in universities and research institutes). This means for any investigation, it can be challenging to identify and use the data.
To give some context to this statement, I offer two examples. In the first, a national evaluation of freshwater quality and the pressures it faces. This is a research question being addressed through use of existing monitoring data funded by UKRI NERC MOT4Rivers. The starting point for the research would be water quality data collected and held by regulators (SEPA, EA, NRW). At the same time, to understand the environmental/ecological context, information about river flow, from the National Rivers Flow Archive (NRFA) would be useful, which could then be supplemented with data about soils, and land use (the most likely source for these being UK CEH). This example illustrates some of the complexity to address what seems like a simple, focussed enquiry. This has resulted in collation of data on a few hundred chemicals in freshwater from thousands of sites across the UK. Ideally the data are comparable (consistency of laboratory methods, and in reporting), but we have encountered a number of inconsistencies, emphasising that data quality assurance is important but often undervalued, and that before any serious modelling is undertaken, checks need to be carried out.
In the second example, the focus is the global crisis of biodiversity loss. The science question is what is the status of biodiversity globally, and to address this question, we can ask what is measured and by whom, where is it measured and where can the data be found. Biodiversity studies are undertaken by many organisations, so this is a crowded and complex landscape, even within the UK. New technology (e.g. acoustic studies, eDNA) are generating new data streams which can be challenging to integrate with more traditional monitoring; this is one space where citizen science has a long pedigree. Prioritising which species and in which habitats also make the data story more challenging, since our regulatory agencies do not have the resources to measure everything, everywhere. So we tend to be selective understandably, we focus on sites of special interest, or headline species or aggregate them (e.g. all farmland birds).
Citizen science is an area where there has been considerable growth in generating new data, traditionally this is used in biodiversity, but there are moves to see more widespread use of such schemes, including through social media. Popular citizen science projects have included annual surveys, for example run by BTO and RSPB for birds; increasingly these have been supplemented by the use of apps for the identification of species, for example bats and plants. Citizen science can present some challenges in its integration with other data streams, not least related to some of the potential sampling biases that can exist (e.g. if recording insects at water bodies, the public may choose to visit when the weather is good, more people will visit sites close to home etc). This could be described more broadly in terms of judging how representative the data are. I have mentioned the word bias, and this is a word that is also discussed in the context of AI (AI learns or is trained on existing data, if these data are biased and unrepresentative, then the model and any inference is biased,). Social media data would be a prime example of a potentially biased source of data since the researcher or regulator has no control (and potentially limited knowledge) about the basis of the sampling – who are returning results, what are their motives, are they pursuing an individual agenda?
A glimpse of the future-digital twins and earth system models
Overall the complexity of data continues to grow, supplemented by data from social media, and data of many different forms, such as images, sound tracks and DNA. There is more urgency to find solutions – especially in adaptation to the intensification of climate extremes, the drive to net zero, reversing biodiversity loss and ensuring just transitions. Increasingly the linkage between environment and human health and wellbeing is being recognised, the sustainability agenda has high profile (at various scales: settlement, city, region, country…), so more data, and better connected data are essential. As well, there needs to be the development of tools to make better use of the existing data.
What does the data lake allow us to do differently? It allows us to think about systems rather than siloes, which leads naturally to thinking about the ESS models (which today are often also described as digital twins).
“A digital twin is a set of virtual information constructs that mimics the structure, context, and behaviour of a natural, engineered, or social system (or system-of-systems), is dynamically updated with data from its physical twin, has a predictive capability, and informs decisions that realize value.” (NAS, 2024)
Digital twins depend on large (real time/near real time) data sets to allow the physical world to be mimicked in the virtual world, which will allow us to test scenarios and interventions, to optimise management and to reflect the complexity of the systems we must deal with. This requires developments in the national digital infrastructure, otherwise we will not be able to take advantage of the opportunities offered.
New analytic/modelling approaches, including AI, are also essential. There are many developments in AI including the use of specific AI/ML tools such as large language models, neural networks to mention only a few of the main algorithmic approaches.
What does the future look like?
As a major supporter of environmental data generation, the Natural Environment Research Council (NERC) launched a 10 year digital strategy in 2020. “Digital approaches will be at the heart of NERC’s drive to deliver world-class environmental science. Advances in computational power and advanced simulations (such as digital twinning) will further enhance our ability to model the environment. Through improvements in sensing technology combined with artificial intelligence (AI) and machine learning we can provide environmental data in greater detail than ever before, and develop the tools needed to analyse high volumes of information” (NERC, 2020).
Delivering this vision, and in particular the development of the system approach, requires many actors, interconnected data and models, new infrastructure and tools to improve our environmental intelligence. There needs to be an alignment of many moving parts. Digital twins need accurate and reliable data, infrastructure delivering data storage, management, and accessibility. Governance policies are critical in providing standards and guidelines. Data flow is essential for the success of digital twin technology, and there needs to be a willingness to share. Data sharing is often held up over issues around data ownership and intellectual property
Discovering and then reusing the multiple data streams including citizen science, historical data and real time data from sensors can be challenging.
What are some of the environmental challenges that remain only partly addressed, where the data we have access to are inadequate?
I can identify a few areas where our data driven knowledge partially addresses an environmental challenge.
Natural Capital and Ecosystem Assessment (NCEA) is a programme covering land and water environments in England with data synthesis at its core. It is designed to collect data on the extent, condition and change over time of ecosystems and natural capital, and the benefits to society. Objectives include to:
- increase the spatial scale of some data (for example, species distribution, biodiversity),
- provide reliable data to help account for carbon stocks within England’s soils, peatlands, trees, and marine habitats,
- assessment of ecosystems in inland and coastal waters, marine environment, and on land including soil ecosystems.
The Office for National Statistics recently published their National Capital Accounts for 2024. The policy landscape within which these data are relevant and their future use is also evolving, with this growing emphasis on natural capital and ecosystem services. But on a day to day basis we are often confronted by the lack of data that exists for a specific site development, or a strategy (e.g. tree planting) or nature based flood alleviation often because the effects take a number of years to manifest so there remains a tension, and a data gap. It is an area where there are likely to be substantial data developments in the next few years, as we attempt to tension classical environmental data and environmental indicators with social and economic values and semi-quantitative or qualitative data to inform ecosystem service provision.
Other examples include understanding of effects of complex chemical mixtures. You might imagine that this has been studied for many years, but the focus has tended to be one chemical at a time in a water body. Limited data (as well as the need for new approaches to environmental quality standards) make it difficult to disentangle effects.
There are some instances where they may be contradictory scientific evidence, for example on the effects of muirburn on carbon (sink or source?) and the answer may be dependent on the timescale over which we observe the system.
Many of the data gaps are on interconnections and networks (so back to systems), the rates and flows in complex inter-connected systems.
In conclusion, there are many digital challenges that remain, but I am optimistic that more data will become available, along with new analytic tools, and that our ability to offer and evaluate solutions at systems level to our climate, biodiversity and pollution crises will improve. So in answer to my question, do we need more data, my response is yes, more data still has great potential not solely to illuminate the crises we face but to offer evaluations of the solutions we consider.
References
FAIR. 2016 https://www.go-fair.org/fair-principles
Nativi, S.; Mazzetti, P.; Craglia, M. 2021. Digital Ecosystems for Developing Digital Twins of the Earth: The Destination Earth Case. Remote Sens. 13, 2119. https://doi.org/10.3390/rs13112119
National Academies of Sciences, Engineering, and Medicine. 2024. Foundational Research Gaps and Future Directions for Digital Twins. Washington, DC: The National Academies Press. https://doi.org/10.17226/26894
NERC (2020) Digitally enabled environmental science 2021-2030
MOT4Rivers Monitoring, Modelling And Mitigating Pollution Impacts In A Changing World: Science And Tools For Tomorrow’s Rivers, NE/X01620X/1, 2022-2026
Steffen, W., Richardson, K., Rockström, J. et al. 2020. The emergence and evolution of Earth System Science. Nat Rev Earth Environ 1, 54–63. https://doi.org/10.1038/s43017-019-0005-6
- This article is part of a series from the RSE’s Environment Community of Interest. The aim is to stimulate informed public debate on the many aspects and elements of this complex and vexed subject. We’d love to hear your thoughts about this piece, email us on publicpolicy@therse.org.uk.
- To be notified when the next article in this series is live, sign up to the RSE’s monthly policy advice newsletter.
- The RSE’s blog series offers personal views on a variety of issues. These views are not those of the RSE and are intended to offer different perspectives on a range of current issues.
