To be policy-relevant, future climate research must include the noncontiguous United States

The research community devotes a lot of time and resources toward monitoring historic climate changes, tracking current impacts, and modeling future risks in the United States. But even as climate science has made significant advances over the last several decades, insights on how climate change has and will affect the nation as a whole are incomplete. That’s because much of the available information needed to understand climate risks and inform our response to the climate crisis typically omits key areas outside of the contiguous United States (OCONUS), i.e., Alaska, Hawai’i, US-Affiliated Pacific Islands, and the US Caribbean. The science community, including research funders, project managers, and local experts, needs to work together to close this gap, particularly if researchers want their work to remain relevant to climate mitigation and adaptation planning and policymaking.

Much of the available information about climate risks omits key areas outside of the contiguous United States such as Maui (pictured here) and the other Hawaiian islands. We need to close this gap, particularly if researchers’ work is to remain relevant to climate mitigation and adaptation planning and policymaking. Image credit: Shutterstock/Zane Vergara.

Failing to include all parts of the United States in climate research results in inaccurate national estimates of climate risks. It also leaves local governments, practitioners, and communities without the information they need to support timely, climate-related policy decisions. At the national level, efforts to quantify the benefits, costs, and equity implications of climate policies and investments are excluding a large number of citizens from the equation. At local levels, climate information is important for mitigation and adaptation planning in OCONUS states and territories, which often face unique or especially extreme climate hazards. For example, Alaska and the rest of the Arctic are warming faster than the global average change in temperature (1). Island communities across Puerto Rico and the US Virgin Islands face more intense and slower-moving hurricanes combined with sea-level rise that threaten coastal infrastructure (2). And in Hawai’i and the US-Affiliated Pacific Islands, saltwater intrusion puts local agriculture and ecosystems at risk (3). As the country moves forward on climate action, it’s important that the entire nation be represented in the evidence that underpins our collective decisions.

Why Data Gaps Matter

One example of how data gaps affect broader efforts to support decision-making through science is the National Climate Assessments (NCAs), the US Government’s premier assessments on climate change impacts, risks, and response. As a key source of climate information for all regions of the United States (Fig. 1), it is critical that these reports assess and consider the entire nation. However, the existing literature and available downscaled and derived datasets that NCA authors can assess are often limited to the contiguous United States (CONUS).

Fig. 1.

Three of the 10 regions used in the Fifth National Climate Assessment are outside of the contiguous United States, including Alaska, Hawai'i, the US-Affiliated Pacific Islands (which include the Commonwealth of the Northern Mariana Islands; the unincorporated territories of American Sāmoa, Guam, and the Pacific Remote Islands; and the Freely Associated States: the Federated States of Micronesia, the Republic of the Marshall Islands, and the Republic of Palau), and the US Caribbean (which includes Puerto Rico and the US Virgin Islands). Image credit: Data from ref. 4, USGCRP, 2023.

The most recent NCA, the Fifth National Climate Assessment (NCA5), was released in November 2023 (4) and offers a case in point. Building on past NCA efforts to improve geographic representation, NCA5 leadership provided further guidance to assessment authors and set a new requirement for recognizing all 10 NCA regions in spatial figures that report national data. This requirement outlined a preferential approach to handling varying types of data gaps.

In an ideal situation, where consistent datasets were available for the entire nation (i.e., including Alaska, Hawai’i, US-Affiliated Pacific Islands, and the US Caribbean), authors were instructed to include this information in figures or findings. Where some OCONUS data were available, but not for the same time period or not using the same methods or source as CONUS data, the information was included in national figures, with those differences in the datasets described in the caption and metadata. If a lack of data for a part of the United States meant that a region could not be represented visually in a national-scale figure, general information about observed or expected trends in those regions was added to the caption. If even general information on trends in those regions was not available, authors were asked to explicitly acknowledge that data gap.

Authors improved geographic representation in the final report compared to drafts developed prior to this guidance; yet, the assessment also highlights the significant data gaps that remain (see Boxes 23.2 and 30.1 in refs. 2 and 3). Table 1 breaks down how OCONUS data were incorporated into NCA5 national maps. A lack of data meant that more than half of the national figures in the assessment took the least-preferred approach: noting in the caption that observed or projected climate-relevant OCONUS data were unavailable.

Table 1.

Visualization of OCONUS data in the Fifth National Climate Assessment

Preferred approach (ranked)	Method for displaying regional data in NCA5 figures representing national climate information	Percent of figures (%)
1	Displays full US extent with the same dataset	22
2	Similar or related OCONUS data identified and included in visual display even if time-series, sources, or methods vary from CONUS data	4
3	Impacts or trends in OCONUS are known and described in text and/or caption, but datasets are unavailable for inclusion in the visual display	21
4	Datasets with full US coverage are unavailable; data gaps are acknowledged in the figure caption	53

Assessment authors were constrained by the availability of empirical data and model outputs that covered all 10 NCA regions. Of the 388 total figures in NCA5, 58 figures were maps designed to convey national representations of climate-relevant information (e.g., projected increase in average surface temperatures across the US under different global warming levels). Authors were required to include data for all 10 NCA regions in every national graphic or caption, or to explicitly acknowledge the lack of available datasets in the figure caption. Data are from ref. 4, USGCRP, 2023.

Lack of data across the entire United States could limit the applicability of some assessment findings to only those areas where data exist to support the conclusions. The paucity of timely and usable data in OCONUS makes the jobs of NCA authors—and the decision-makers who use the NCA—much more difficult. To truly represent national climate information, all maps in the report would show all US regions, and the percent of graphics noting OCONUS gaps would be zero.

Regional data gaps further undermine timely policy decisions by limiting the applicability of climate services: usable products and sources of information that help governments make decisions and take actions relevant to climate change risks, impacts, and responses. Someone trying, for example, to determine how to update building codes to harden infrastructure in locations exposed to increasing hazards might look to a climate services product to make an informed decision based on scientific data (5, 6). Local, state, Tribal, territorial, regional, and national decision-makers developing mitigation and adaptation plans rely on government products, such as atlases (7), geographic information systems (GIS) tools, and comprehensive multisectoral reports and assessments (8). When OCONUS data are not available or are left out, those products are less robust and less useful.

The implications of excluding OCONUS from climate-relevant datasets are clear. But actually filling these gaps remains difficult. There are a multitude of unique challenges associated with research in OCONUS, ranging from empirical data collection to climate modeling.

Closing the Data Gaps

Observations and measurements can be hard to obtain in remote, hard-to-access areas or locations with limited internet access. Projecting climate impacts in these areas can also be difficult. Although global climate models have historically lacked the fine resolution needed to inform mitigation and adaptation planning for small islands, scientific advancements and increases in computing power have improved the geographic resolution of climate projections (9). Yet, modeling future climate in places like Hawai’i, US-Affiliated Pacific Islands, and the US Caribbean is still complicated because islands have more land–sea interactions than many contiguous states with coastlines, and they may have complex terrain or experience unique climate phenomena (10, 11). The physical landscape of Alaska has also proved challenging, with its tall mountains, seasonally varying sea ice, and large seasonal swings in temperature (12).

As a result of challenges in projecting climate impacts in such locations, high-resolution regional climate models that capture the effects of local terrain and phenomena may have more scientific fidelity in OCONUS areas than downscaling methods that calculate future statistics from past climate observations. For instance, regional modeling includes simplified representations, known as parameterizations, for coastlines, hydrology, and processes affected by land use and land cover, which may not be represented in more general statistical methods (13). Additionally, locally sourced information may be more salient and credible to the decision-makers in these respective communities. Where it is possible to integrate National Weather Service station data or community knowledge into regional climate products, the results could also be more tailored to local information needs (8).

A good example of how to address OCONUS data limitations comes from the hydrometeorology community, which has found ways to improve the quality of gridded observational data and subsequent localized climate projections in the face of missing or sparse data. This is important because observed precipitation extremes in global gridded datasets vary widely across different climates and geographies (14), and it can be difficult to determine whether differences are attributable to specific methodological choices. In addition to improvements in gridded observational data, computer algorithms have also been used on global climate model outputs to compensate for station sparsity and mountainous terrain, before being input to hydrologic model simulations for Hawai’i and Alaska (11, 15, 16). These techniques allow scientists to develop trend projections in areas where it was not previously possible. Having such detailed analyses allows researchers to better evaluate datasets and can inform choices by project funders and local decision-makers, such as city planners and water utility managers.

Examples like these demonstrate that the research community has the capability to solve the scientific problems that have previously excluded OCONUS from national climate datasets. Particularly as researchers continue to overcome the technical reasons for the lack of data, it is important to explore other creative approaches to dismantle institutional barriers to data collection and development in these areas.

Taking Down Barriers

One institutional barrier stems from the long history of these geographies being underserved or underresourced, resulting in a paucity of observational data. Much of these missing data represent exclusion in data-collection efforts and perpetuate historical social injustices. For example, the location of rain gauges throughout the national Community Collaborative Rain, Hail and Snow Network shows a statistical relationship with race and class: the probability of at least one rain gauge in a US Census Tract increases with the percentage of White residents and with median household income (17). The US Caribbean is populated by groups classified by the US government as minorities; the US Virgin Islands has a majority Black population, Puerto Rico has a predominantly Hispanic or Latino Spanish-speaking population, and both areas face socioeconomic challenges. There is also a relatively high representation of Tribal and Indigenous communities in OCONUS, including Native Hawaiians and Alaska Natives. Improving climate data coverage in these parts of the United States will improve access to critical climate information for underserved communities.

In addition to sparse monitoring and exclusion from downscaled climate projections, researchers from the Pacific islands also report an absence of hazard and exposure mapping for floods and sea-level rise, insufficient data on water supply, limited information on ecosystem responses to climate impacts, and gaps in socioeconomic and health data important to understanding climate impacts and risks (3). Researchers in the US Caribbean note similar gaps, as well as missing data on carbon storage and population projections (2). Granting organizations and contract officers could begin to fill these gaps and shift institutional norms by adding requirements that funded monitoring and modeling research must include all states and territories of the United States.

Improving climate data coverage in these parts of the United States will improve access to critical climate information for underserved communities.

Furthermore, sparse datasets can be supplemented by incorporating local and traditional climate information, ground-truthing modeling data with improved monitoring and observations, and building partnerships with the communities that have historically been underrepresented and underserved (17). Recent guidance to Federal agencies on use of Indigenous Knowledge (18) highlights examples of successful coordination and shows commitment to further efforts to consult and collaborate with Tribal Nations and Indigenous Peoples in Federal decision making (see, for example, ref. 19).

When informed by or coproduced with communities in OCONUS, more complete and higher-resolution data may result in more usable climate information, particularly where information is provided in accessible formats, with place-based context and decision-making timescales salient to the communities living in these locations (13, 20, 21). For example, Arctic studies where Indigenous community members were engaged from the start of the research process and where Indigenous and Western scientific knowledge bases were integrated scored higher on indicators of community engagement. These indicators included whether the findings informed action (e.g., development of adaptation plans or capacity-building) and/or directly benefited community members (e.g., development of services such as maps or curricula) (22).

Prioritizing research that includes full geographic coverage of the entire nation enhances climate science, improves the relevance and uptake of research outputs to all communities, and better positions the United States to take informed mitigation and adaptation actions. To meet the climate crisis and build a climate-resilient nation, people living across the entire country need to see themselves represented in climate research to better serve their communities and make informed choices.

That means all researchers who want their data to be useful, usable, and used to support future policy-making must address this gap between CONUS and OCONUS data availability. Future national assessments, such as the National Nature Assessment (23), are likely to continue the trend of prioritizing assessment of data sources inclusive of the entire nation (24). The next cycle of development for the NCA has already begun; research efforts that consider full national representation must begin now if they are to be published in time to be included in NCA6.