Tension #1: Diversity of Sources, Actors, Purposes, and Products

First, the diversity of (new) climate data sources and of key actors and stakeholders involved in the climate data chain, each bringing their own priorities and values to the discussion, creates difficulties and tensions in the data ecology.

Box 1 indicates some of the diversity, listing a sample of data sources, actors, purposes, and products involved in climate data. The diversity is only likely to increase with the continued expansion of the Internet of Things, accompanied by a plethora of sensors and other new data collection methods. In addition, climate data, once primarily collected and used by scientists, is now making its way into a variety of domains—policymaking, the private sector, disaster response, etc.—which leads to a potential divergence of methods and priorities among an ever-widening group of stakeholders (including statistical agencies, scientists, citizen scientists, policymakers, civil society, the private sector, and more). Advances in machine learning and AI are likely to further complicate the picture, leading to unpredictable uses of data and equally unpredictable outcomes.

The cornucopia of interests and stakeholders—marked both by plenty and increasing divergence—calls for new approaches to governance. In particular, there is a need for greater multi-stakeholder governance that could align interests, sensitivities, and requirements at all levels of decision-making.Footnote ⁹ Data stewards have a natural role to play in such processes, and a number of initiatives (e.g., the Climate Data Store, Climate Montreal, and the Pacific Climate Impacts Consortium) have provided hubs to bring together stakeholders and help ensure legitimacy and trust in the climate data ecosystem. These are essential ingredients in a field that often processes PII; in addition, multi-stakeholder governance can also help protect independence and rigor for scientists and other researchers, a vital concern in the often contested field of climate data.

Tension #2 Competing concerns and lack of common principles

Among the divergence of sources and interests marking the climate data field, few are as pronounced as those separating climate change and climate justice actors on the one hand and private sector stakeholders on the other. The former category (which includes policymakers, researchers, and activists) calls for more data related to the environment, and more openness (embodied by the “Right to Know” movement), and seeks to increase data justice while limiting data extraction and data colonialism. Corporations and private actors, on the other hand, are often motivated by a desire to maintain competitive advantage. In addition, the private sector is itself marked by competing interests, depending on whether companies primarily produce, process, or reuse data.

In short, the field is marked by an absence of common principles for how data should be (re)used, with different stakeholders upholding different principles to advance their respective agendas and priorities. There is an urgent need for a common normative and ethical framework that could guide the collection, processing and (re)use of climate data. In particular, we need to move from a concept of data ownership, which exacerbates asymmetriesFootnote ¹⁰, to data stewardship. Data stewards could increase accessibility and transparency while accommodating individual and collective concerns and rights, making room for a variety of stakeholders.

Tension #3 Power imbalance: who decides what to measure? And what to collect? And what to share?

Decisions concerning what data to collect, analyze, and use—and how to use it—underlie another tension at the heart of climate data. The divergence of stakeholders means similar divergence in approaches to measuring, collecting, and sharing data. The potentially deleterious impact of these divergences is heightened by power asymmetries within the digital ecology, particularly as they relate to data access and agenda-setting. An agenda for data-driven collaboration can “inform the strategic allocation of resources for new research projects, indicators for regular monitoring, and the formation of cross-sectoral data-sharing collaborations.”Footnote ¹¹ Despite clear potential benefits and widespread agreement on the general principle of using climate data, in other words, there remain important open questions about what specific forms of data are used, what types of conclusions are derived, and who benefits from data-driven decision-making.

Solutions do exist, but it is important to understand that the choice—and application—of solutions is itself not value-neutral. The notion of purpose specification, for instance, which may help limit data abuses, is an inherently political choice: defining a purpose will inherently narrow the scope of who benefits. To help mitigate such risks, it is essential that equity and participation be considered as overarching principles throughout the data value chain and the broader ecology of climate data governance. The quality of participation can also be enhanced by more effective communication and awareness building with civil society groups and the public at large. Several climate initiatives are now advancing new frontiers in data stewardship by making data useful not only for researchers and scientists, but by pioneering new approaches to visualization and storytelling (e.g., the WORLDLING initiative at MIT, or the Climate Impact Lab).

Tension #4 Extraction through the collection. Proportionality and collective rights. Data ownership

Datafication is taking place in a world marked by long-standing hierarchies, inequalities, and socio-economic divisions. This is perhaps especially true of climate data, which is being collected, processed and used across national and cultural boundaries, raising important questions about relations between and the relative rights of communities. Concerns exist about data sovereigntyFootnote ¹² for indigenous populations, feminist and anti-colonial movements, and the rights of populations subject to so-called “helicopter research”Footnote ¹³ and extractive data practices. Such concerns are magnified by a general distrust of data collection practices between populations with unequal rights or powerFootnote ¹⁴, and growing awareness regarding flawed data consent provisions, which call for new “social licenses” to govern how data is collected and used.

To ensure that the field of climate data mitigates rather than exacerbates existing divisions, such concerns should be acknowledged by and embedded within emerging governance frameworks. Community-based participatory research and collection can help minimize extractive data practices. Representation should also be taken into consideration when selecting data stewards, whether they be individuals or bodies and institutions (which can be composed of a variety of community voices so as to ensure various stakeholder interests are taken into account).

In addition, it is essential to uphold proportionality as a core principle, for example by tailoring data collection efforts to the minimal necessary for meaningful insights, ensuring that the benefits of data use are equitably distributed among all stakeholders, and implementing safeguards that prevent the over-surveillance of vulnerable communities. Underlying these methods and approaches is a recognition of data as a public—rather than private—good, and a commitment to upholding the principle of digital self-determinationFootnote ¹⁵ throughout the data ecology.

Tension # 5: Quality, provenance, and standards

As the volume of available data grows, so do concerns over data quality and integrity. In addition, it is essential to ensure—and acknowledge—the “situatedness” of data; so-called “thick data”Footnote ¹⁶ helps ensure that data is processed, analyzed and used in a contextually relevant and sensitive way. Attention to data quality and thickness must be accompanied by—and can help encourage—the development of data standards to promote interoperability and responsible reusability. Responsible reuse is especially important for climate data, which is inherently global and cross-national in scope.

Data quality is in part a matter of ensuring that decisions involving climate technology are supported by appropriate quality-assured engineering standards and processes. But, as with all the tensions discussed in this paper, resolving these tensions is about far more than just technology. Governance and policy steps are also essential. These can include frameworks that help operationalize and standardize quality assurance, for instance, by establishing clear metrics for data accuracy, consistency, and completeness across different stages of data collection, analysis, and usage. In addition, data quality (and relevance) can be enhanced if institutions have policies that track what decisions they make about data throughout the data lifecycle.

Data stewards have a valuable role to play in this process, for instance by leveraging their ability to bring together and coordinate multiple stakeholders. A good example can be found in the Datzilla error reporting and tracking system, which offers a web-based portal to identify and correct data discrepancies in climate-related data sets from the National Oceanic and Atmospheric Administration (NOAA). One key to the tool’s effectiveness is its multistakeholder nature: hosted by Texas A&M University, it allows academic researchers, government bodies, and civil society organizations to coordinate to enhance the quality of climate data.

Tension #6 Timeliness, continuity, and sustainability

Ensuring the timely collection and release of climate data is critical for effective risk management and the development of innovative approaches to advance climate mitigation and climate justice. However, persistent concerns regarding the financial (and other forms of) sustainability raise questions about timeliness, continuity, and the long-term viability of climate data and efforts at climate justice. Such concerns are heightened by the political context of the global climate debate, which poses challenges to policy continuity and financial sustainability. In particular, a risk exists that bigger stakeholders could walk away from the conversation, thus imperiling the functioning of the larger ecosystem dedicated to climate mitigation.

As a result of these tensions, it is clear that any efforts at sustainability need to be at the core of an effective governance framework for climate data. This will require fit-for-purpose incentives for investment and institutionalization so that all stakeholders are aligned. In addition, governance will need to ensure the timeliness of climate data, to ensure its relevance and effectiveness. Some mechanisms that can help achieve these objectives include public-private data collaborativesFootnote ^{17 ,} Footnote ¹⁸ (e.g., the Net Zero Public Data UtilityFootnote ¹⁹), standards for data quality and timeliness, and international agreements on climate data exchange and reporting. In addition, fostering community-based monitoring programs and leveraging technological innovations, such as artificial intelligence for data analysis, can further enhance the governance framework’s effectiveness.

Tension # 7 Access, openness, and transparency

Our era’s challenges necessitate nuanced solutions beyond merely “opening” data—a narrow approach that may inadvertently serve corporate agendas. Mere access to raw data does not guarantee transparency, leaving room for manipulation. To navigate these complexities, fostering data collaboratives, empowering data stewards, and enhancing data-sharing practices are essential. However, for these strategies to be genuinely effective, it is crucial to establish and clarify incentives for private sector data holders (“the business case” for data collaborationFootnote ²⁰) and to improve the harmonization of cross-border data sharing efforts.

As a broad approach, embedding FAIR data principlesFootnote ²¹ into the climate data conversation may provide an overarching framework to reduce access asymmetries and achieve meaningful openness. These principles (which embody findability, accessibility, interoperability, and reusability) were first established in the context of academic data. In addition, data collaboratives have proven particularly useful vehicles for bridging gaps between the private and public sectors, and more generally for promoting greater data sharing. Finally, the field of climate data could benefit from a commitment by relevant stakeholders to share data in response to specific—and specific types or scales—of crises. This can at least ensure that the right data will be available to serve the public good when it is most needed.

Tension # 8: Bias, capture, and whitewashing

Tension #3, above, highlights the risks of power imbalances within the data ecology. There are various ways these imbalances can manifest. In addition to the problems posed by unequal access, bias in the underlying data and algorithms is emerging as a serious concern—particularly given the growing prominence of artificial intelligence and large language models (LLMs). These risks are accompanied by concerns over “whitewashing” (in which stakeholders may cover up or obfuscate data collection or disclosure), and concerns regarding academic captureFootnote ²² by the private (and public) sectors. All of these affect the rigor and credibility of data and data efforts; more generally, they undermine trust in the ecosystem.

An adequate framework to address these concerns must encompass both technological solutions and policy interventions. Technological solutions might involve the development of automated systems capable of auditing and tracing data flows within the lifecycle, thereby ensuring accountability through alerts on potential discrepancies or misuse. However, these technological measures need to be underpinned by a robust governance framework that not only operationalizes ethical principles but also extends the notion of responsibility in the use of climate data.

This broader governance framework should include comprehensive approaches to protect individual and community rights. Some existing approaches to data stewardship already include tools to increase local participation—for example, the Protected Areas Database of the United States (PAD-US) initiative, a multi-stakeholder project that facilitates local review of protected lands and areas in the United States. But beyond formal participation mechanisms and tools, it is essential more broadly to establish a “social license to operateFootnote ²³,” which entails gaining and maintaining public trust by demonstrating that data collection and use are conducted in ways that align with societal values and expectations. This concept goes beyond legal compliance to include ethical considerations, transparency, and community engagement, ensuring that data practices are perceived as legitimate and beneficial by the wider community.

Tension # 9: Local vs global: subsidiarity and cultural difference

Climate change is a global problem, and climate data is therefore also global in scope. But within this broad context, there exists little consensus on what types of questionsFootnote ^{24 ,} Footnote ²⁵—or solutions—should be devolved to the local level, and how to account for cultural and structural differences across jurisdictions. As a general principle, governance at the local level is more conducive to citizen participation, and rapid feedback loops to iterate and improve upon data initiatives. In addition, local governance allows for more efficient management of climate resources and challenges. At the same time, it is important to recognize that many localities may lack the type of data expertise available at national and international levels. An overly local focus may likewise lack the global perspective required to mitigate broader challenges.

Several pathways are available to resolve these tensions. To begin, it is essential to embed notions of subsidiarity within data governance design; these can help establish the principle that, where possible and advantageous, climate data should be governed at the local level. Highlighting subsidiarity also upholds the importance of local participation and cultural and context sensitivity. Alongside subsidiarity, however, it is equally important to identify issues and challenges that are better addressed at the global (or supra-local) level—for example, data related to the management of common resources or the upholding of shared values and rights. Ultimately, an effective governance framework will coordinate the local and the global, and harmonize the need for contextual sensitivity and community participation with the wider scope of the problem and of the data itself.

Tension #10: Disputes, accountability and use

While it is increasingly clear that data can help mitigate the climate crisis, there exists little agreement on how to implement it, and what processes or mechanisms exist to harmonize values and priorities, avoid misuse and harm, and ensure accountability. Disputes are inevitable and require agile and independent processes to be resolved in a productive manner. In addition, as with all uses of data, accountability is essential. Climate justice requires not just judicious use of data but also clear lines of responsibility and accountability.

Existing processes from other data verticals may provide guidance. For example, there now exist well-established procedures and mechanisms (technological and otherwise) to establish decision provenance (i.e., transparency about who is responsible and accountable for the use of climate data). Auditing tools and frameworks can also be repurposed, and designed to ensure independence and agility within dispute resolution processes. While repurposing such established steps, it is also important to keep in mind specific needs and variations that may be necessary in the climate context—e.g., the tensions between the global and the local or the urgent need for real-time data to inform immediate climate action, the necessity for integrating indigenous and traditional knowledge systems, and the importance of addressing data sensitivity and security concerns related to vulnerable ecosystems and communities.