Effects of Human-Caused Greenhouse Gas Emissions on U.S. Climate, Health, and Welfare (2025)
Chapter: 4 Impacts of Greenhouse Gas Emissions on Future Climate
4
Impacts of Greenhouse Gas Emissions on Future Climate
4.1 KEY MESSAGES
EPA (2009a) provided projections of future changes in the climate system associated with human-caused greenhouse gas (GHG) emissions. Many of these projected changes have been observed since 2009, as described in Chapter 3, including increasing surface temperatures, higher sea levels, and regional variability across the United States in other physical and biological systems.
Continued emissions of greenhouse gases from human activities will lead to more climate changes in the United States, with the severity of expected change increasing with every ton of greenhouse gases emitted. Despite successful efforts in many parts of the world to reduce emissions, total global GHG emissions have continued to increase, and additional warming is certain.
Models have proven skillful and are effective at simulating a fingerprint of human influence on the changing climate that is now observed. Climate models, which simulate the Earth system, have been used since the 1960s to examine the role of different climate forcings in driving climate variability. Models have simulated certain “fingerprints” of the climate response to human-caused GHG emissions that have since been observed, including the vertical structure of temperature changes and enhanced warming over land relative to oceans.
All climate models—regardless of assumptions about future emissions scenarios or estimates of climate sensitivity—consistently project continued warming in response to future atmospheric greenhouse gas increases. Projections of future change draw primarily on physically based climate models, which have advanced in spatial resolution, process representation, and evaluation since 2009, improving confidence in understanding of the implications of future emissions. Applying fundamental physics of the Earth system leads to the same conclusion about future warming as projected by climate models.
Continued changes in the climate increase the likelihood of passing thresholds in Earth systems that could trigger tipping points or other high impact climate surprises. These surprises are difficult to predict, can occur abruptly, and, in some cases, would be irreversible.
4.2 CLIMATE MODELS
Climate models are numerical simulations of the Earth system, including the atmosphere, ocean, land, freshwater systems, and sea ice, and the coupling among these components. These models are based on the underlying physics that govern these systems and evolved from numerical weather prediction models, first developed in the
1960s (e.g., Manabe and Wetherald, 1967). Aided by improvements in their process1 representation and improvements in supercomputers, these models have increased in complexity. These models are now often referred to as “Earth System Models” and can incorporate many additional components and processes, including terrestrial and marine ecosystems, atmospheric chemistry, land ice, and glacial dynamics.
Since the 1960s, climate models have been used to examine climate variations and the role of different climate forcers, such as GHGs and volcanic emissions, in driving that variability. They provide projections of future climate conditions subject to scenarios of future emissions of GHGs and aerosols, as well as land cover changes. They are also an important tool for attribution in that they allow for controlled experiments, for example by isolating the climate response to increasing GHGs versus other climate drivers (e.g., Gillett et al., 2016). Notably, these models have improved at simulating changes in the climate that have already been observed.
However, the climate system is complex, and models are imperfect tools. Climate projections have uncertainty due to internal variability in the climate system, uncertainty in future emissions of GHGs and aerosols, and structural uncertainty in the models themselves. Comparisons across models, which include different details of process representation and numerical implementation, are useful for quantifying structural model uncertainty (e.g., Hawkins and Sutton, 2009). A common metric used for this comparison is the equilibrium climate sensitivity (for a recent review, see Jeevanjee et al., 2025), which is the equilibrated global mean surface warming for a doubling of CO2. A subset of climate models has equilibrium climate sensitivity that is higher than the likely range as assessed by the Intergovernmental Panel on Climate Change (e.g., Zelinka et al., 2020). Understanding why these models have a higher equilibrium climate sensitivity is an area of active research. That said, while this metric is useful for understanding climate response and model uncertainty, it has somewhat limited relevance for projected change in the near term because the climate system is not equilibrated and factors such as changing aerosol emissions play a role (Jeevanjee et al., 2025).
Since EPA (2009a), updates to climate models have occurred with the newest models available under the Coupled Model Intercomparison Project 6 (Eyring et al., 2016). As a group, these models have improvements to process representation and include additional capabilities. For example, advances have been made since 2009 in the ability for model ensembles to quantify the influence of internal climate variability on projections (Kay et al., 2015) and in model resolution. These improvements have enabled the use of models for new applications, such as multiyear prediction of flood frequency (Zhang et al., 2025) and extending the time horizon of tropical cyclones forecast to seasons (Murakami et al., 2025).
The availability of longer observational time series since EPA (2009a) has also allowed for improved validation of model-simulated trends in the historical record and improved understanding of model successes and challenges that are still present (Simpson et al., 2025). The models have allowed the detection of a “fingerprint” of human influence (see Section 2.4) across many observed changes in the Earth system (Eyring et al., 2021), including the vertical (Santer et al., 1996) and regional (Hegerl et al., 1996) structure of temperature changes, seasonal cycle changes for tropospheric (Santer et al., 2022) and sea surface (Shi et al., 2024) temperatures, and daily precipitation variability (Ham et al., 2023), among others.
While models are not perfect, they are useful and skillful tools for attribution of anthropogenic signals in the changing climate and understanding of future climate changes in response to GHG emissions. All climate models consistently project continued warming in response to future GHG increases, regardless of climate sensitivity levels or future emission scenarios. Notably, they are just one line of evidence of human influence on historical climate change. When combined with observational evidence, paleoclimate information, and theoretical understanding, it is unequivocal that many climate changes underway can be attributed in large part to rising GHG emissions from human activity. This evidence also indicates that every additional quantity of emissions will strengthen, and in some cases accelerate, those changes for the future.
___________________
1 In climate and Earth system models, a “process” refers to a physical, chemical, or biological phenomenon—such as cloud formation, ocean circulation, or carbon uptake by plants—that influences the climate system and is described mathematically in the model.
4.3 EXPECTED CHANGES IN U.S. CLIMATE
The long lifetime of energy, transportation, industrial, and other built infrastructure creates challenges in rapidly reducing CO2 emissions (NASEM, 2021b, 2024a). Hence, despite mitigation efforts in many parts of the world and the achievement of downward CO2 emission trends in many advanced economies, total global emissions have continued to grow and are expected to remain near current levels over the coming decade.
Given the persistence of CO2 and other long-lived GHGs in the atmosphere, past emissions have increased atmospheric concentrations of these gases, which will sustain Earth’s energy imbalance (see Section 2.3). Warming will continue until net CO2 plus N2O emissions reach (and remain at) ~0 and emissions of CH4 are constant or decreasing. As long as global emissions of CO2 stay above zero, concentrations and radiative forcing will continue to increase and global temperature will increase roughly in proportion to cumulative CO2 (i.e., each additional ton emitted adds an increment more to temperature increase) with small contributions from other long-lived gases including N2O and F-gases. As global emissions of GHGs are spread across all nations, a collective effort at reducing emissions is required to limit future warming.
Analysis of policies in place in 2023 showed that CO2e emissions would stay roughly constant over 2025–2035 (UNEP, 2024), driving continued warming that would lead to a projected peak global mean warming in 2100 of about 4.9°F (2.7°C) (4.1 to 5.4°F, or 2.3 to 3°C range) (Hausfather, 2025). Were all countries to fully implement their 2023 policies and their unconditional pledges to the United Nations Framework Convention on Climate Change, global CO2e emissions would drop by ~5%, leading to a projected peak 21st-century warming of about 4.4°F (2.4°C), or only about 0.5°F (0.3°C) less than under current policies (UNEP, 2024). The likelihood of exceeding global mean 3.6°F (2°C) warming relative to preindustrial temperatures under these two cases is estimated at 97% with current policies continuing and 94% for unconditional pledges continuing (UNEP, 2024). Warming beyond 3.6°F (2°C) is expected to have many negative consequences for human health and welfare in the United States (see Chapters 5 and 6).
Emission scenarios used in simulations of projected climate (O’Neill et al., 2016; Riahi et al., 2017) have been updated since EPA (2009a). These scenarios encompass a range of possible futures, including high emissions scenarios, which assume significant “regional rivalry,” and “sustainability” scenarios, which assume deep and sustained reductions in emissions, with net negative CO2 emissions by 2100. Using this range of scenarios and large numbers of model simulations, projected impacts of future warming on the United States can be assessed probabilistically. Risks of future impacts for some quantities—including heat, sea level rise, and some extreme events—can be assessed with relatively high confidence, while risks for other expected impacts, including regional droughts and hurricane intensities, continue to have large uncertainties in their quantification. Climate-related damages increase with every quantity of GHGs emitted and the damages per ton of emissions also rise as the Earth continues to warm owing to non-linearities in impacts (Cissé et al., 2022; EPA, 2023). In simulations with sustained GHG emission reductions, the increase in atmospheric CO2 concentrations slows after 5–10 years, and global warming slows after several decades (Lee et al., 2021). This is consistent with studies that examine the climate response to a complete cessation of CO2 emissions, which indicate that temperatures would stabilize or even decrease over time (Jones et al., 2019; Matthews and Weaver, 2010).
With each increment of continued GHG emissions and warming, surface and near-surface air temperatures (and thus heat exposure to humans and, for example, crops, animals, and ecosystems) increase; extreme heat becomes more frequent and extreme precipitation events increase across some regions, while aridification and drought persist in others—patterns that often scale approximately linearly with global temperature, though not uniformly across all metrics or places (USGCRP, 2023). The oceans continue taking up heat and CO2, driving higher ocean heat content, rising sea levels from thermal expansion and land-ice loss, and decline in ocean pH; these changes persist for decades to centuries even if temperatures stabilize (Lee et al., 2021). These findings are independent of the equilibrium climate sensitivity level in any specific model.
4.4 ABRUPT CLIMATE CHANGE
In response to increasing GHGs, many climate changes have been observed and are anticipated for the future. In many cases, these changes exhibit a roughly linear relationship to changed forcing, as with global temperature, discussed above. However, the climate system can also exhibit an abrupt response to climate warming when certain thresholds (sometimes referred to as “tipping points”) are passed. Paleoclimate data indicate that abrupt shifts have occurred in the past (e.g., Capron et al., 2021). With continued and accelerating climate warming, the likelihood for surpassing thresholds grows with the potential for rapid and dramatic disruption to human systems (NRC, 2013).
EPA (2009a) states, “Climate warming may increase the possibility of large, abrupt regional or global climatic events” (p. ES-4). This was and remains accurate, supported by more evidence on additional possible “tipping elements” that could undergo abrupt change. EPA (2009a) discussed several specific elements where abrupt changes were possible, including megadroughts, disintegration of the Greenland Ice Sheet, collapse of the West Antarctic Ice Sheet, catastrophic release of CH4 from sea floor CH4-hydrates and/or permafrost soils, and slowing down of the Atlantic Meridional Overturning Circulation (AMOC)—which is a major component of Atlantic Ocean circulation. For many of these elements, abrupt change in the 21st century was considered to be low probability but high impact. There remains uncertainty in whether or when tipping points might be reached in these elements. For example, studies on AMOC simulate a range of responses to changing buoyancy (heat and freshwater) forcing (e.g., Jackson et al., 2023) and observation-based early warning systems may suggest a higher likelihood of collapse than seen within climate models (e.g., Ditlevsen and Ditlevsen, 2023). Paleoclimate evidence strongly suggests instances of AMOC collapse during the Younger Dryas, Heinrich events, and potentially during Dansgaard–Oeschger2 events in the Last Glacial Period (e.g., Lynch-Stieglitz, 2017).
Since 2009, evidence has emerged for some abrupt changes underway. For example, numerous rapid changes in the Antarctic environment have occurred (Abram et al., 2025), including rapid reductions in sea ice (Purich and Doddridge, 2023), regime shifts in biological systems (e.g., Fretwell et al., 2023), and increasing ice sheet mass loss (Rignot et al., 2019), with consequences for sea level rise. Research has also highlighted additional potential “tipping elements,” including rapid changes in numerous terrestrial and marine ecosystems, expansion of oxygen minimum zones, the potential collapse of the Antarctic Overturning circulation (Abram et al., 2025), and loss of alpine glaciers, among others (e.g., NRC, 2013). Assessments have quantified an increasing likelihood of passing multiple climate tipping points with increasing warming (e.g., Armstrong McKay et al., 2022), many of which could be irreversible. Work has also highlighted that tipping elements can interact and often do so in destabilizing ways, thus setting up the possibility of “tipping cascades” (Wunderling et al., 2024).
___________________
2 Dansgaard–Oeschger (D-O) events are periods of rapid warming (over a few decades) followed by a slow cooling (over a few hundred years). During the Last Glacial Period, there were 25 recorded D-O events that occurred every few thousand years.
