Department of Energy - Critical Review of Impacts of GHG Emissions on the US Climate July 2025

Geophysical Research Letters 38, L14803, doi:10.1029/2011GL047711 McKitrick, R., and J.R. Christy (2019). Assessing changes in US regional precipitation on multiple time scales. Journal of Hydrology, 578, 124074. https://doi.org/10.1016/j.jhydrol.2019.124074 McKitrick, R. and J.R. Christy (2025), “Data and Code for CWG2025 Report”, Mendeley Data, V1, doi: 10.17632/by76s7rxfb.1 McNider, R.T., G.J. Steeneveld, A.A.M. Holtslag, R.A. Pielke Sr., S. Mackaro, A. Pour-Biazar, J. Walters, U. Nair, and J.R. Christy, 2012: Response and sensitivity of the nocturnal boundary layer over land to added longwave radiative forcing. Journal of Geophysical Research., 117, D14106, doi:10.1029/2012JD017578. Null, J. and J. Hulbert, “California Washed Away: The Great Flood of 1862,” Weatherwise 60, no. 1 (2007): 26–30, https://doi.org/10.3200/wewi.60.1.26-30 National Forest Foundation. (2022). Blazing battles: The 1910 fire and its legacy. https://www.nationalforests.org/our-forests/your-national-forests-magazine/blazing-battles-the-1910fire-and-its-legacy 72

National Interagency Fire Center. (2024). Wildfire statistics. https://www.nifc.gov/fireinformation/statistics/wildfires NCA4. (2017). Climate science special report: Fourth National Climate Assessment, Volume I (D. J. Wuebbles, D. W. Fahey, and K. A. Hibbard, Eds.). U.S. Global Change Research Program. https://doi.org/10.7930/J0J964J6 NCA5. (2023). Fifth National Climate Assessment (A. R. Crimmins et al., Eds.). U.S. Global Change Research Program. https://doi.org/10.7930/NCA5.2023.CH1 (accessed May 22, 2025) NOAA Hurricane Research Division. (2025a). All U.S. hurricanes (1851–2019). https://www.aoml.noaa.gov/hrd/hurdat/All_U.S._Hurricanes.html NOAA Hurricane Research Division. (2025b). Most intense (3, 4, 5) continental United States hurricanes: 1851–1970, and 1983–2023. https://www.aoml.noaa.gov/hrd/hurdat/most_intense.html Overland, J. E. (2021). Causes of the record-breaking Pacific Northwest heatwave, late June 2021. Atmosphere, 12(11), 1434. https://doi.org/10.3390/atmos12111434 Pan, M., M. Lu and U. Lall (2024). Diversity of cross-Pacific atmospheric river main routes. Communications Earth and Environment, 5, 378. https://www.nature.com/articles/s43247-02401552-y Parks, Sean, Christopher Guiterman, Ellis Margolis et al. (2025) A fire deficit persists across diverse North American forests despite recent increases in area burned.

Nature Communications 16 https://www.nature.com/articles/s41467-025-56333-8 Pielke, R. A., et al. (2011). Land use/land cover changes and climate: Modeling analysis and observational evidence. Wiley Interdisciplinary Reviews: Climate Change, 2, 828–850. https://doi.org/10.1002/wcc.144 Porter, K., et al. (2011). Overview of the ARkStorm scenario. U.S. Geological Survey. Quinlan, F. T., T. R. Karl, and C.N. Williams Jr. (1987). United States Historical Climatology Network (USHCN) serial temperature and precipitation data (NDP-019). Oak Ridge National Laboratory. Runnalls, K.E. and T.R. Oke, 2006: Technique to detect microclimate inhomogeneities in historical records of screen-level air temperature. Journal of Climate, 19, 959-978. https://doi.org/10.1175/JCLI3663.1 Samborska, V., and H. Ritchie (2024). Wildfires. Our World in Data. https://ourworldindata.org/wildfires Seneviratne, S. et al. (2021) “Chapter 11: Weather and Climate Extreme Events in a Changing Climate.” In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, edited by V. MassonDelmotte, P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, et al., 1513–1766. Cambridge: Cambridge University Press, 2021. https://doi.org/10.1017/9781009157896.013. Sommer, L. (2016, November 7).

Let it burn: The Forest Service wants to stop putting out some fires. KQED. https://www.kqed.org/science/1134217/let-it-burn-the-forest-service-wants-to-stop-puttingout-some-fires Stephens, Scott et al., (2021) “Fire, Water, and Biodiversity in the Sierra Nevada: A Possible Triple Win,” Environmental Research Communications 3, no. 8 (August 6, 2021): p. 081004, https://doi.org/10.1088/2515-7620/ac17e2. Tyukavina Alexandra , Peter Potapov, Matthew Hansen et al. (2022) Global Trends of Forest Loss Due to Fire From 2001 to 2019, Frontiers in Remote Sensing. Volume 3 https://www.frontiersin.org/journals/remote-sensing/articles/10.3389/frsen.2022.825190/full U.S. Forest Service. (2022). Managing fire. https://www.fs.usda.gov/science-technology/fire USGCRP. (2023). USGCRP indicators platform: Heat waves. https://www.globalchange.gov/indicators/heat-waves Vecchi, Gabriel A. and Thomas R. Knutson (2011) “Estimating Annual Numbers of Atlantic Hurricanes Missing from the HURDAT Database (1878–1965) Using Ship Track Density,” Journal of Climate 24, no. 6 (2011): 1736–1746, https://doi.org/10.1175/2010JCLI3810.1.

73

Villarini, Gabriele, Gabriel A. Vecchi, and James A. Smith, “U.S. Landfalling and North Atlantic Hurricanes: Statistical Modeling of Their Frequencies and Ratios,” Monthly Weather Review 140, no. 1 (January 2012): 44–65, https://doi.org/10.1175/MWR-D-11-00063.1. Web Archive. (n.d.). Archive of NIFC fire data. https://web.archive.org/web/20200212033452/https://www.nifc.gov/fireInfo/fireInfo_stats_totalFires. html Yang, L., Y. Yang, Y. Shen, et al. (2024). Urban development pattern’s influence on extreme rainfall occurrences. Nature Communications, 15, 3997. https://doi.org/10.1038/s41467-024-48533-5 Zhang, W., G. Villarini, G.A. Vecchi, and J.A. Smith, J. A. (2018). Urbanization exacerbated the rainfall and flooding caused by Hurricane Harvey in Houston. Nature, 563, 384–388. https://doi.org/10.1038/s41586-018-0676-z

74

7 CHANGES IN SEA LEVEL Chapter Summary Since 1900, global average sea level has risen by about 8 inches. Sea level change along U.S. coasts is highly variable, associated with local variations in processes that contribute to sinking and also with ocean circulation patterns. The largest sea level increases along U.S. coasts are Galveston, New Orleans, and the Chesapeake Bay regions - each of these locations is associated with substantial local land sinking (subsidence) unrelated to climate change. Extreme projections of global sea level rise are associated with an implausible extreme emissions scenario and inclusion of poorly understood processes associated with hypothetical ice sheet instabilities. In evaluatingAR6 projections to 2050 (with reference to the baseline period 1995-2014), almost half of the interval has elapsed by 2025, with sea level rising at a lower rate than predicted. U.S. tide gauge measurements reveal no obvious acceleration beyond the historical average rate of sea level rise.

7.1

Global sea level rise

Global sea level rise is arguably the most important climate impact driver that is unambiguously associated with increasing temperatures. At the global level, warming raises sea level through thermal expansion of sea water and through melting of glaciers and ice sheets. Variations in land water storage are another important factor. At the regional scale, sea level change is influenced by large-scale ocean circulation patterns, and geologic processes and deformation from the redistribution of ice and water. Locally, vertical land motion from geologic processes, ground water withdrawal, and fossil fuel extraction are also important. AR6 estimates that global mean sea level increased by 7.9 (5.9–9.8) inches between 1901 and 2018, with the rate of sea level rise accelerating in recent decades. At the ocean basin scale, sea levels have risen fastest in the Western Pacific and slowest in the Eastern Pacific over the period 1993–2018 (Fox-Kemper et al., 2021). The rate of global sea level rise is estimated to be 0.12 inches/year, about the height of two stacked pennies (NASA, 2020). The observing systems for global sea level rise have advanced significantly in the satellite era, particularly with the advent of satellite altimeters in 1993. Local tide gauges have provided useful data for the past century, and even longer for a few locations.

Following the end of the Little Ice Age in the midnineteenth century, tide gauges show that the global mean sea level began rising during the period 1820– 1860, well before most anthropogenic greenhouse gas emissions.

7.2

U.S. sea level rise

Observed and predicted rates of mean global sea level rise might have little scientific relevance for specific locations, owing to local processes (NOAA, 2025). Figure 7.1 shows that in Canada and Alaska (and also northern Washington), sea level is decreasing, owing to uplift from glacial rebound. Most of the Pacific coast tide gauges show low rates of sea level rise, while largest U.S. rates are on the Gulf coast (Louisiana and Texas) and in the mid-Atlantic states (Chesapeake Bay region).

75

Figure 7.1. Map of rates of relative sea level rise along the U.S. coast (NOAA, https://tidesandcurrents.noaa.gov/sltrends/). For reference, 3 mm = 0.12 in.

Measurements of relative sea level rise from tide gauges conflate the climatologically relevant increase in the volume of seawater with local vertical land motion. The latter, which varies from place to place, is best measured using a Global Positioning System (GPS) station located near the tide gauge. It is driven by a range of processes that can be comparable to the climate signal and can locally exacerbate (subsidence/sinking) or mitigate (uplift) the risk of sea level rise (Wöppelmann and Marcos, 2016). Human activities triggering subsidence are often significant. They include soil drainage (e.g. for urban development) and subsurface extraction of groundwater or hydrocarbons. Table 7.1 shows absolute sea-level rise (ASLR) for selected locations, determined from the sum of uncorrected relative sea-level rise (RSLR) as estimated from tide gauge time series (NOAA, 2025) and the vertical land motion (VLM) measurements (NAS, 2012; Letetrel et al., 2015; Karegar et al., 2016). The absolute sea level rise for each of these locations is significantly smaller than the measured relative sea level rise owing to local subsidence. More than half of the measured relative sea level rise is attributed to land sinking for these locations: San Francisco, Galveston, Grand Isle. For reference, the global average rate of absolute sea level rise is estimated to be 0.12 inches/year.

76

Location San Francisco, CA Galveston, TX Grand Isle, LA St Petersburg, FL New York City, NY

RSLR +0.08 +0.26 +0.36 +0.12 +0.11

VLM -0.06 -0.19 -0.28 -0.02 -0.05

ASLR +0.02 +0.07 +0.08 +0.10 +0.06

Table 7.1 Absolute sea level rise (inches/year) consisting of Relative Sea Level Rise (RSLR) plus Vertical Land Motion (VLM)

San Francisco Bay Over the past 100 years, relative sea level in the San Francisco Bay area has risen by 7.8 inches, at an average rate of 0.08 inches/year (Figure 7.2). As shown in Table 7.1, San Francisco’s vertical land motion is -0.06 inches/year (sinking), producing a recent absolute rate of +0.02 inches/year. Portions of Treasure Island, San Francisco International Airport, and Foster City are sinking as fast as 0.4 inches/year (Shirzaei and Bürgmann, 2018). Problems in the San Francisco Bay area, including threats to the airport, are caused primarily by soil compaction in landfill zones that were formerly wetlands, not by the slow creep of global sea level rise.

Figure 7.2. Tide gauge measurements at San Francisco, California, obtained from NOAA https://tidesandcurrents.noaa.gov/sltrends/sltrends_station.shtml?id=9414290 (downloaded 4/22/25).

Galveston - Houston Long-term tide measurements at Galveston Pier 21 show sea level rise of 2.18 feet in the past century, or a rate of 0.26 inches/year (Figure 7.3). Vertical land motion (subsidence) at Galveston is estimated at -0.19 inches/year, yielding an absolute rate of rise of +0.07 inches/year (Table 7.1). The U.S. Geologic Survey found that most of the land-surface subsidence in the Houston-Galveston region is a direct result of groundwater withdrawals (Kasmarek and Ramage 2017), which caused compaction of the aquifer sediments, mostly in the fine-grained silt and clay layers. By 1979, as much as 10 feet of subsidence had occurred in Houston. 77

Figure 7.3. Tide gauge measurements Galveston Pier, TX, obtained https://tidesandcurrents.noaa.gov/sltrends/sltrends_station.shtml?id=8771450 4/22/2025).

from NOAA (downloaded

New Orleans and the Mississippi delta Long-term tide gauge measurements at Grand Isle, Louisiana, show that sea level has risen by slightly more than 3 feet over the last 100 years at an average rate of 0.36 inches/year (Figure 7.4). Vertical land motion (subsidence) is estimated -0.28 inches/year. Table 7.1 gives the absolute sea level rise as +0.08 inches/year.

Figure 7.4. Tide gauge measurements at Grand Isle, LA, obtained from NOAA https://tidesandcurrents.noaa.gov/sltrends/sltrends_station.shtml?id=8761724 (downloaded 4/22/25).

78

The issues of sea level rise and land loss in the Mississippi Delta region are complex, with geological subsidence and the decline in sediment transported by the Mississippi River being the dominant drivers. The construction of dams in the basin since the 1950s has decreased the Mississippi’s suspended sediment load by ~50 percent (Maloney 2018). A new subsidence map of coastal Louisiana finds the coastal region to be sinking at about one third of an inch per year (Neinhuis et al. 2017), associated with groundwater and resource withdrawal. As the city’s elevation averages one to two feet below sea level, sea level rise from anthropogenic warming is hardly the dominant driver of New Orleans’ problems. New York City New York City is particularly vulnerable to the effects of sea level rise because it is built primarily on islands and has 520 miles of coastline. Measurements by a tide gauge at the southern tip of Manhattan (The Battery) show that relative sea level has risen over 11 inches over the past century, at an average rate of 0.11 inches/year (Figure 7.5). But vertical land motion in the New York City area is -0.05 inches/year (roughly 5 inches per century), so that the absolute rate of sea level rise at The Battery is 0.06 inches/year, or about 55 percent of the measured relative sea level rise.

Figure 7.5. Tide gauge measurements at The Battery, New York, obtained from NOAA https://tidesandcurrents.noaa.gov/sltrends/sltrends_station.shtml?id=8518750 (downloaded 4/22/25).

7.3

Projected sea level rise

The concern over sea level rise is not about the roughly eight inches of global rise since 1900. Rather, it is about projections of accelerated rise based upon simulations of a warming climate through the 21st century. AR6 finds high agreement across published global mean sea level projections for 2050 with little sensitivity to emissions scenarios. Considering only projections incorporating ice-sheet processes in whose quantification there is at least medium confidence, the global sea level projections for 2050, across all emissions scenarios, fall between 3.94 and 15.75 inches (5th–95th percentile very likely range) relative to the 1995– 2014 baseline period (Fox-Kemper et al., 2021). 79

Conversely, AR6 states there is low agreement across published global mean sea level projections for 2100, particularly for higher emissions scenarios. Considering only projections representing ice sheet processes in whose quantification there is at least medium confidence, the AR6 global mean sea level projections for 2100 lie between 7.9 and 39.4 inches (5th–95th percentile very likely range) under the medium emissions scenario SSP2–4.5 (Fox-Kemper et al., 2021). There is deep uncertainty surrounding projections of sea level rise to 2100 owing to uncertainties in ice sheet instabilities, particularly for the higher emissions scenarios. In February 2022, NOAA issued its projections of sea level rise for various sites along the U.S. coast (Sweet et al., 2022). They claim that by 2050, the sea will have risen one foot at The Battery in Manhattan (relative to 2020). A one-foot rise in thirty years would be more than twice the current rate and about three times the average rate over the past century. In that historical context, NOAA’s projection is remarkable— as shown in Figure 7.6, it would require a dramatic acceleration beyond anything observed since the early 20th century. But even more noteworthy is that Sweet et al. (2022) say this rise is “locked in”—it will happen no matter what future emissions are. We should know in a decade or so whether that prediction has legs.

Figure 7.6 Rate of sea level rise at the Battery in Manhattan. Shown is the historical thirty-year trailing trend, together with the allegedly “locked in” NOAA predicted trend for 2050. Historical data: NOAA Tides and Current.

References Fox-Kemper, B., Hewitt, H. T., Xiao, C., Adalgeirsdóttir, G., Drijfhout, S. S., Edwards, T. L., ... & Yu, Y. (2021). Ocean, cryosphere and sea level change. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (pp. 1211–1362). Geneva, Switzerland: Intergovernmental Panel on Climate Change. Karegar, M. A., Dixon, T. H., & Engelhart, S. E. (2016). Subsidence along the Atlantic Coast of North America: Insights from GPS and late Holocene relative sea level data. Geophysical Research Letters, 43(7), 3126–3133. https://doi.org/10.1002/2016GL068015 80

Kasmarek, M. C., & Ramage, J. K. (2017). Water-level altitudes 2017 and water-level changes in the Chicot, Evangeline, and Jasper aquifers and compaction 1973–2016 in the Chicot and Evangeline aquifers, Houston-Galveston Region, Texas (Scientific Investigations Report 2017–5080). U.S. Geological Survey. https://doi.org/10.3133/sir20175080 Letetrel, C., Becker, M., Llovel, W., & Cazenave, A. (2015). Estimation of vertical land movement rates along the coasts of the Gulf of Mexico over the past decades. Continental Shelf Research, 111(Part A), 42–51. https://doi.org/10.1016/j.csr.2015.10.018 Maloney, J. M., Bentley, S. J., Xu, K., Obelcz, J., Georgiou, I. Y., & Miner, M. D. (2018). Mississippi River subaqueous delta is entering a stage of retrogradation. Marine Geology, 400, 12–23. https://doi.org/10.1016/j.margeo.2018.03.001 National Academy of Science (NAS, 2012). Sea-level rise for the coasts of California, Oregon and Washington: Past, present, and future. Appendix A: Vertical land motion and sea-level data along the West Coast of the United States. National Academies Press. https://nap.nationalacademies.org/catalog/13389/sea-level-rise-for-the-coasts-of-california-oregonand-washington NASA. (2020). NASA sea level rise portal: 2020 edition.

https://www.nasa.gov/specials/sea-level-rise2020/ National Oceanic and Atmospheric Administration (NOAA) Center for Operational Oceanographic Products and Services. (n.d.). Sea level trends. https://tidesandcurrents.noaa.gov/sltrends/ NOAA Tides & Currents. (n.d.). Relative sea level trends. https://tidesandcurrents.noaa.gov/sltrends/ Nienhuis, J. H., Törnqvist, T. E., Esposito, C. R., Liang, M., & Ma, H. (2017). A new subsidence map for coastal Louisiana. GSA Today, 27(6), 28–29. https://doi.org/10.1130/GSATG337GW.1 Shirzaei, M., & Bürgmann, R. (2018). Global climate change and local land subsidence exacerbate inundation risk to the San Francisco Bay Area. Science Advances, 4(3), eaap9234. https://doi.org/10.1126/sciadv.aap9234 Sweet, W. V., Hamlington, B. D., Kopp, R. E., et al. (2022). Global and regional sea level rise scenarios for the United States (USGS Report 70229139). United States Geological Survey. https://www.usgs.gov/publications/global-and-regional-sea-level-rise-scenarios-united-states Wöppelmann, G., & Marcos, M. (2016). Vertical land motion as a key to understanding sea level change and variability. Reviews of Geophysics, 54, 64–92. https://doi.org/10.1002/2015RG000502

81

8 UNCERTAINTIES IN CLIMATE CHANGE ATTRIBUTION Chapter summary “Attribution” refers to identifying the cause of some aspect of climate change, specifically with reference to anthropogenic activity. There is an ongoing scientific debate around attribution methods, particularly regarding extreme weather events. Attribution is made difficult by high natural variability, the relatively small expected anthropogenic signal, lack of high-quality data, and reliance on deficient climate models. The IPCC has long cautioned that methods to establish causality in climate science are inherently uncertain and ultimately depend on expert judgement. Substantive criticism of the main IPCC assessments of the role of CO2 in recent warming focus on inadequate assessment of natural climate variability, uncertainties in measurement of solar variability and in aerosol forcing, and problems in the statistical methods used for attribution. The IPCC does not make attribution claims for most climate impact drivers related to extreme events. Statements related to statistics of global extremes (e.g. event probability or return times, magnitude and frequency) are not generally considered accurate owing to data limitations and are made with low confidence. Attribution of individual extreme weather events is challenging due to their rarity.

Conflicting claims about the causes of the 2021 Western North America Heatwave illustrate the perils of hasty attribution claims about individual extreme events.

8.1

Introduction

The Intergovernmental Panel on Climate Change (IPCC) distinguishes between detection of climate change and attribution of its cause. As defined by the AR6 WGI Glossary (IPCC, 2025): Detection: Detection of change is defined as the process of demonstrating that climate or a system affected by climate has changed in some defined statistical sense, without providing a reason for that change. An identified change is detected in observations if its likelihood of occurrence by chance due to internal variability alone is determined to be small, for example, <10%. Attribution: Attribution is defined as the process of evaluating the relative contributions of multiple causal factors to a change or event with a formal assessment of confidence. Both detection and attribution rely on statistical analysis. Detection focuses on whether changes are significant enough to stand out from random variation regardless of cause. Attribution involves comparison of observed events to model-generated counterfactuals. Since experimentation on the climate is not possible, attribution requires statistical inference to assess how much human activities (such as GHG emissions) versus natural factors (like volcanic eruptions) contribute to observed changes. Attribution methods assume all external and internal drivers of the system are known and represented.

There are ongoing scientific debates around attribution methods, especially those for attributing extreme weather events to climate change. The IPCC has long cautioned that methods to establish causality in climate science are inherently uncertain and ultimately depend on expert judgement. AR4 Working Group I (Hegerl et al., 2007) explained it as follows:

82

‘Attribution’ of causes of climate change is the process of establishing the most likely causes for the detected change with some defined level of confidence... unequivocal attribution would require controlled experimentation with the climate system. Since that is not possible, in practice attribution of anthropogenic climate change is understood to mean demonstration that a detected change is ‘consistent with the estimated responses to the given combination of anthropogenic and natural forcing’ and ‘not consistent with alternative, physically plausible explanations of recent climate change that exclude important elements of the given combination of forcings’ (IPCC, 2001)… The approaches used in detection and attribution research described above cannot fully account for all uncertainties, and thus ultimately expert judgement is required to give a calibrated assessment of whether a specific cause is responsible for a given climate change. AR5 Working Group II (Cramer et al., 2014) makes the following statement: Two broad challenges to the detection and attribution of climate change impacts relate to observations and process understanding. On the observational side, high-quality, long-term data relating to natural and human systems and the multiple factors affecting them are rare.

In addition, the detection and attribution of climate change impacts requires an understanding of the processes by which climate change, in conjunction with other factors, may affect the system in question. Because of the complexity of the causal chains in the climate system, investigation of these relationships is exceptionally challenging.

8.2

Attribution methods

The IPCC employs several attribution methods to assess the causes of observed climate changes, distinguishing between natural and human-induced factors. Below is a concise description of the key IPCC attribution methods. Optimal Fingerprinting uses linear regression to explain variations in observed climate data as a weighted sum of climate model simulations run with and without anthropogenic forcings. The data used in the regression model is weighted to try to minimize the influence of random noise and the estimation method is chosen to account for climate model error. Time Series Analysis exploits statistical differences between anthropogenic forcing and natural variability to see which dominates observed temperatures and also uses variations in the timing of changes to determine if causal dependence across data series can be inferred. Process-Based Attribution focuses on understanding the physical mechanisms behind observed changes. This approach combines observations, climate models, and theoretical understanding to attribute changes in specific processes to forcings. This approach is often used for regional climate phenomena, such as monsoon changes or polar amplification Extreme Event Attribution assesses the role of human influence in the likelihood of occurrence or intensity of extreme weather events (e.g. heat waves or droughts). Methods include: •

Probabilistic Event Attribution uses large ensembles of climate model simulations to compare observed events to model-generated counterfactuals

•

Storyline Approach examines the physical processes driving an event and evaluates how anthropogenic forcings might have modified those processes 83

8.3

Attribution of global warming Attribution statements for global warming in the three most recent IPCC Assessment reports are: AR4: Most of the observed increase in global average temperatures since the mid-20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations. (IPCC 2007) AR5: It is extremely likely that more than half of the observed increase in global average surface temperature from 1951 to 2010 was caused by the anthropogenic increase in greenhouse gas concentrations and other anthropogenic forcings together. The best estimate of the human- induced contribution to warming is similar to the observed warming over this period. (IPCC 2013) AR6: The likely range of total human-caused global surface temperature increase from 1850–1900 to 2010–2019 is 0.8°C to 1.3°C, with a best estimate of 1.07°C. It is likely that well-mixed GHGs contributed a warming of 1.0°C to 2.0°C, other human drivers (principally aerosols) contributed a cooling of 0.0°C to 0.8°C, natural drivers changed global surface temperature by –0.1°C to +0.1°C, and internal variability changed it by –0.2°C to +0.2°C. It is very likely that well-mixed GHGs were the main driver of tropospheric warming since 1979. (IPCC 2021)

The AR4 and AR5 attribution statements reference the warming since the mid-20th century, the period when greenhouse gas emissions began increasing rapidly. The words “most” and “more than half” are deliberately imprecise, potentially ranging from 51 to 99 percent of the warming – presumably this imprecision is to account for structural uncertainties such as natural internal variability. The confidence level increases from very likely to extremely likely from AR4 to AR5. The structure of the attribution statement in the AR6 is fundamentally different, referencing the warming to the period 1850-1900. The AR6 attribution statement is more precise numerically, but with a lower level of confidence at “likely” – AR6 attributes essentially all the warming to increases in greenhouse gases. The most confident statement from AR6 relates to the tropospheric warming since 1979, using the words “main driver” and “very likely.” There are three areas of substantive criticism of the IPCC’s assessment of the causes of the recent warming: inadequate assessment of natural climate variability, inappropriate statistical methods, and substantial discrepancies between models and observations. The last is discussed in Chapter 5, while this chapter discusses the first two factors.

All of these criticisms are relevant to the IPCC’s attribution of the recent warming, which also underpins extreme event attribution 8.3.1 Natural climate variability AR6 states that natural external drivers since 1850-1900 have changed global surface temperature by – 0.1°C to +0.1°C, and internal variability has changed it by –0.2°C to +0.2°C – on average having essentially no net impact on the warming since 1850-1900. As discussed below, this minimal contribution of natural variability has been disputed by several publications that question the magnitudes of solar variability and internal variability from large-scale ocean circulations. Solar variability AR5 concluded that the best estimate of radiative forcing due to Total Solar Irradiance (TSI) changes over the period 1750–2011 was very small (0.05 W/m2, Myrhe et al., 2014). AR6 acknowledges substantially higher values and a much larger range of estimates of changes in TSI over the last several centuries, stating that the TSI between the Maunder Minimum (1645–1715) and the second half of the twentieth century increased by 0.7–2.7 W/m2, a range that includes both low and high variability TSI data 84

sets (Gulev, 2021). However, the recommended forcing dataset for the CMIP6 climate model simulations used in AR6 for attribution studies averages two data sets with low solar variability (Matthes, 2017). While AR6 shows a substantially greater solar impact than does AR5, the overall impact of solar forcing on the climate was still assessed to be small compared to anthropogenic forcing. However, the impact of solar variations on the climate is uncertain and subject to substantial debate (Lockwood, 2012; Connolly et al., 2021) - something that is not evident in the IPCC assessment reports. The variations of TSI over time remains a challenging problem. Since 1978, there have been direct measurements of TSI from satellites. However, the data exhibits non-negligible inconsistencies, and interpreting any multi-decadal trends in TSI requires comparisons of observations from overlapping satellites. There are several rival composite TSI datasets that disagree as to whether TSI increased or decreased during the period 1986–96 (the ACRIM gap; see Chapter 4). Further, the satellite record of TSI is used to calibrate proxy models that infer past solar variations from sunspots and cosmogenic isotope measurements (Velasco Herrera et al., 2015).

There is substantial evidence for high solar activity in the second half of the 20 th century (starting in 1959) and extending into the 1990’s, before a decline in the early 21st century; this period is often termed the “Modern Maximum.” (Chatzistergos et al., 2023; Solanki et al., 2004; Usoskin et al., 2007). However, some scientists have concluded that it is not possible to be confident of any multi-decadal trend in TSI (Schmutz, 2021). This uncertainty causes some reconstructions of TSI from 1750 to have low variability (implying a very low impact of solar variations on global mean surface temperature) whereas datasets with high TSI variability can explain more than 70 percent of the temperature variability since preindustrial times (Scafetta, 2013; Stefani, 2021). The choice of TSI satellite record used in an analysis can therefore substantially influence how much climate change is attributed to human versus natural forcings. There is growing evidence that other aspects of solar variability, which are referred to as solar indirect effects, either amplify TSI forcing or are independent of TSI forcing. Scafetta et al. (2023) suggests that ~80 percent of solar influence on climate might stem from non-TSI mechanisms.

There are numerous candidate processes, including solar ultraviolet changes; energetic particle precipitation; atmosphericelectric-field effect on cloud cover; cloud changes produced by solar-modulated galactic cosmic rays; large relative changes in the magnetic field; and the strength of the solar wind. Such solar indirect effects are not included in climate models, although indirect methods of estimating their impacts suggest they are significant. However, claims of non-TSI mechanisms influencing climate are uncertain and debated. Natural variability of large-scale ocean circulations Variations in global mean surface temperature are linked to recurrent large-scale variations in ocean circulation patterns, including the Atlantic Multidecadal Oscillation (AMO), the Pacific Decadal Oscillation (PDO) and the El Nino-Southern Oscillation (ENSO). These circulations influence ocean heat uptake and heat distribution and also influence atmospheric circulation patterns and cloud distributions. There is some debate as to whether these variations are strictly internal to the climate system, or whether this variability can have a solar/astronomical origin or can be influenced by large volcanic eruptions.

While climate models simulate the large-scale ocean circulations and internal climate variability, most models have too little amplitude compared to observations at multi-decadal frequencies and phasing outof-sync with the observed climate (Kravtsov et al. 2024). Averaging multiple simulations effectively averages out the internal variations, leaving only the forced climate variability (e.g. CO2 forcing). With most of the modern warming beginning in the late 1970’s, the recent 50-year warming is on the same time scale as the multi-decadal oscillations of the AMO and PDO. Here are summary statements from the AR5 and AR6 reports: 85

AR5: Decadal variability in the Pacific, associated with the PDO or IPO [Interdecadal Pacific Oscillation], contributes significantly to regional and global temperature trends, but the relative contributions of internal variability and external forcing are difficult to disentangle in CMIP5 simulations. AR6: Since AR5, there has been increased understanding of the role of internal variability, such as ENSO, PDO, and AMO, in modulating regional climate trends. However, limitations in simulating the exact timing and amplitude of these modes in CMIP6 models contribute to uncertainties in attributing observed changes to anthropogenic forcing (high confidence). The amplitude of the peak-to-trough impact of the multi-decadal oscillations on global temperatures has been assessed by the IPCC AR6: “The likely range of change due to internal variability is -0.2°C to +0.2°C (IPCC, 2021).” This implies a trough-to-peak change of 0.4°C. Over many centuries, any global temperature changes from the troughs and peaks will cancel out, with little to no net impact. However, with a nominal timescale of 60-80 years for oscillations such as the AMO and PDO, the timing of the peaks and troughs can be confused with the secular trend. This becomes very relevant for attributing the warming for the past 50 years, when most of the recent warming has occurred.

Even if climate models have the correct amplitude of the multidecadal oscillations, the timing of the peaks and troughs is not adequately simulated, with each model and individual ensemble simulating a different phasing. Once the simulations of the ensemble members and different models are averaged, the natural internal variability contribution is averaged out owing to differences in phasing, effectively cancelling the role of natural internal variability in the attribution process. The time series of global surface temperature anomalies since 1850 (Figure 8.1) shows irregular variations of significant amplitude against the background of an overall warming trend and, after 1976, a pronounced difference in trends between the Northern Hemisphere and the Southern Hemisphere. The period between 1905-1945 shows a strong trend of warming. The following 30 years from 1945-1976 showed slightly declining temperatures. The most recent warming period started in 1977.

86

Figure 8.1 Global average surface temperature anomalies 1850—2025. Top: Northern hemisphere. Middle: Southern hemisphere. Bottom: Global average. Source: UK Hadley Centre https://crudata.uea.ac.uk/cru/data/temperature/HadCRUT5.0Analysis.pdf

The causes of the early 20th century warming are discussed by Hegerl et al. (2017; 2019). The atmospheric CO2 concentration increased from 298 ppm in 1905 to 310 ppm in 1941, implying that CO2 had little impact. Volcanic activity during this period was very low, and solar forcing is uncertain. Yet Hegerl et al. (2017) somehow inferred that 40-54 percent of this warming could be attributed to external forcing, with the rest associated with internal variability. Bronniman et al. (2024) focused on the causes of the cooling in the first decade of the 20th century in the Southern Hemisphere. They found that the cooling was related to a La-Niña-like pattern in the Pacific, a cold tropical and subtropical South Atlantic, a cold extratropical South Pacific, and cool southern midlatitude land areas. The Southern Annular Mode was positive, with a strengthened Amundsen–Bellingshausen seas low, although the spread of the data products is considerable. The warming in the 1930’s and subsequent cooling during mid-century was particularly pronounced in the Arctic.

Bokuchava and Semenov (2021) find that these variations were most likely caused by a combined effect of long-term natural climate variations in the North Atlantic and North Pacific with a contribution of the natural radiative forcing related to the reduced volcanic activity and variations of solar activity, as well as growing greenhouse gases concentrations. Tokinaga et al. (2017) showed that the combined effect of internally generated Pacific and Atlantic interdecadal variability intensified Arctic warming in the early 20th century. The synchronized Pacific-Atlantic warming drastically alters planetaryscale circulations over the Northern Hemisphere; these same circulation patterns have a global influence. 87

The cooling period between 1945 and 1976 has been referred to as the “grand hiatus.” Numerous causes have been hypothesized: natural internal variability from fluctuations associated with the Pacific Decadal Oscillation and Atlantic Multidecadal Oscillation; cooling from increased emissions of aerosols from industrial activities; increased heat uptake in the Atlantic, Southern and Equatorial Pacific Oceans. Thompson et al. (2010) find that that the hemispheric differences in temperature trends in the middle of the twentieth century stem largely from a rapid drop in Northern Hemisphere sea surface temperatures of about 0.3 °C between about 1968 and 1972. The Great Pacific Climate Shift of 1976-1977 was a notable climatic event characterized by an abrupt change in the North Pacific Ocean's atmosphere-ocean system that interacted with global climate patterns. This shift is closely associated with the Pacific Decadal Oscillation (PDO) that oscillates between warm and cool phases over decades. The 1976 shift marked a transition from a predominantly negative (cool) PDO phase (1947–1976) to a positive (warm) phase (1977 through 2000), with significant implications for global and regional climate patterns, including the frequencies of El Nino and La Nina events. The Great Pacific Climate Shift coincided with the beginning of a period of accelerated global warming.

When the Great Pacific Climate Shift is accounted for in climate attribution analyses since 1950, 40 percent or more of the warming in the second half of the 20th century is attributed to natural internal variability (McLean et al., 2009; Tung and Zhou, 2013; Chylek et al., 2016; Scafetta, 2021).

8.3.2 Optimal fingerprinting Optimal fingerprinting is a statistical technique introduced by Allen and Tett (1999) that compares observed climate data to climate model simulations to identify patterns (or "fingerprints") associated with human or natural forcing. It involves taking a vector of observed climate changes, such as warming rates in locations around the world, and decomposing them into a weighted sum of “signals”, which are analogues to the observations generated by climate models with different types of forcings. The weights are chosen using a regression method called Total Least Squares (TLS). The analysis commonly uses just two signals, one generated by models using only anthropogenic forcings and one using only natural forcings. If the estimated weighting coefficient attached to a signal is significantly different from zero, then that signal is said to be “detected”. If it is close to 1.0, then the model generating the signal is said to be consistent with observations. If it is less than 1.0, then the model signal for that forcing is too strong and needs to be scaled down, and vice-versa. Both the observed data and the signals are weighted using a model-generated estimate of patterns of randomness in the climate system so as, in principle, to put maximum weight on regions where natural variability is minimized.

Optimal fingerprinting is the primary tool for attribution in the research literature. While it has been widely used and prominently featured by the IPCC since 2001, there is very little literature examining the statistical properties of the results it generates. One of its inherent weaknesses is that results depend on assumptions about the accuracy of climate models, especially regarding their representation of natural variability (IPCC AR5 Ch 10.2.3). A series of papers by McKitrick (McKitrick 2021, 2022, 2023, 2025) has challenged the optimal fingerprinting method, arguing that it is inherently unreliable and that practices adopted from econometrics can provide more valid results. Statistical theory requires that for regression models to yield unbiased coefficients, a set of assumptions called the Gauss-Markov conditions must hold. McKitrick (2021) argued that the Allen and Tett (1999) methodology violates the Gauss-Markov conditions, leading to potentially biased and inconsistent fingerprinting coefficients, undermining the reliability of the method. Chen et al. (2023) confirmed this analysis, although they argued that the method could yield valid results under very stringent assumptions.

McKitrick (2022) and (2023) raised a further concern that climate scientists—virtually alone among scientific disciplines—have used TLS to estimate anthropogenic greenhouse gas signal coefficients, despite its tendency to be unstable unless some strong assumptions hold that in practice are unlikely to be true. 88

Under conditions that easily arise in optimal fingerprinting, TLS estimates have a large positive bias. Thus, any study that used TLS for optimal fingerprinting without verifying its applicability in the specific data context has likely overstated the attribution. McKitrick (2025) presented an empirical example comparing the results of conventional optimal fingerprinting against methods drawn from mainstream econometrics that are known to be valid for the specific application of signal detection. While the IPCC optimal fingerprinting method yields an anthropogenic signal coefficient close to 1.0 on a global temperature data set spanning 1900 to 2010, the consistent method yields a coefficient around 0.4, which rises to about 0.65 on data spanning 1980 to 2010, implying the model response to greenhouse gases needs to be scaled down by about half to optimally match observations. The natural forcing signal coefficient, by contrast, is between 2.0 and 4.0, implying the climate model signals of natural forcing need to be scaled up two-to four-fold to match observed climate change.

The fingerprinting coefficients estimated in McKitrick (2025), when used to scale the average sensitivity of the climate models used to generate the forcing signals in his data set, imply a Transient Climate Sensitivity of 1.4°C, which is consistent with the estimate by Lewis (2023) using a different estimation method and multiple independent data sets. These findings indicate that the basis on which the optimal fingerprinting method has long been viewed as reliable is not valid. Re-examining previous results individually would be required to determine which findings are statistically robust.

8.3.3 Time series methods The IPCC (AR5 WGI 10.2.2) drew attention to alternative approaches of assessing causality that emerged from the time series econometrics literature. These have the advantage of not depending on assumptions about the accuracy of climate models, but the disadvantage that they depend on difficult-totest assumptions about the data generating process underlying climate and forcing data. The methods, now referred to as climate econometrics, use the tools of unit root testing, Granger causality and cointegration analysis, all of which are familiar in economics and finance and which are slowly being adopted in climate science. Time series analysis methods hold out the possibility of determining whether anthropogenic or natural forcings are the primary drivers of climate change without requiring the use of climate models, although key questions remain unsettled (e.g. Dergiades et al., 2016, Balcombe et al., 2019, Dagsvik et al., 2020, Razzak, 2022, Dagsvik and Moen, 2023). “Granger causality” modeling is a tool for determining directions of influence in data series that move together. It is a statistical, not a physical, concept. If knowing the current value of one variable significantly improves the forecast accuracy of another variable, we can infer a causal connection exists; this is called Granger causality.

The modeling tools, called vector autoregression, measure direct impulse-response patterns and feedbacks. An application to the well-known Vostok ice core data revealed an error in Al Gore’s documentary An Inconvenient Truth. Gore showed the data and drew attention to the coherence of temperature and CO2 changes over a 440,000 year span, which he asserted was due to CO2 driving temperature changes. But temperature changes can also affect atmospheric CO2 levels. Davidson et al. (2015) examined the series and found that temperature Granger causes CO2 but not the reverse. In other words on the time scales represented in the Vostok data, the coherence in the series is primarily due to the influence of temperature on CO2 levels, not the feedback of CO2 levels on temperature. In summary, the primary statistical methods for attributing causation in climate data are optimal fingerprinting and time series analysis. Applications of the Allen and Tett (1999) optimal fingerprinting method dominate the attribution literature and have underpinned past IPCC conclusions, but results depend on the accuracy of climate models and the method has recently been criticized as inherently biased. Time 89

series methods do not depend on climate models but require assumptions of their own and have generated results that have thus far not converged on a consensus.

8.4

Declining planetary albedo and recent record warmth

A sharp recent increase in global average temperatures has raised the question of short-term drivers of climate. One such candidate is the fraction of absorbed solar radiation which has also increased abruptly in recent years. The question is whether the change is an internal feedback to warming caused by greenhouse gases, or whether something else increased the fraction of absorbed radiation which then caused the recent warming. The planetary albedo is the fraction of incoming solar radiation that is reflected back into space rather than being absorbed by the planet. Highly reflective surfaces like cloud tops and snow and ice are most important in this regard. The Earth's albedo is approximately 30 percent, meaning almost a third of the sunlight that reaches Earth is directly reflected back to space. A lower albedo implies more solar energy is absorbed by the planet to be then re-radiated as heat. Hence, other things being equal, a decline in planetary albedo is associated with a warming of the Earth. Arguably the most striking change in the Earth’s climate system during the 21st century is a significant reduction in planetary albedo since 2015, which has coincided with at least two years of record global warmth. Figure 8.2 shows the planetary albedo variations since 2000, when there are good satellite observations.

The 0.5 percent reduction in planetary albedo since 2015 corresponds to an increase of 1.7 W/m2 in absorbed solar radiation averaged over the planet (Hansen and Karecha, 2025). For comparison, Forster et al. (2024) estimate the current forcing from the increase in atmospheric CO2compared to preindustrial times to be 2.33 W/m2 . Looking back prior to 2000 with less adequate data, Goessling et al. (2025) assessed that the planetary albedo was relatively low around the 1940’s and 50’s before rising industrial aerosol precursor emissions increased the albedo until the 1980’s. The strongest planetary albedo excursions were high-albedo episodes caused by volcanic eruptions, such as after the Mount Pinatubo eruption in 1991. Although uncertain, the 2023 planetary albedo minimum might have been the lowest since at least 1940. Changes in surface characteristics cannot explain this decrease in planetary albedo since 2015: • • • •

Arctic sea ice extent has declined by about 5 percent since 1980 (https://nsidc.org/data/seaice_index/images/s_plot_hires.png), although following 2007 there has been a pause in the Arctic sea ice decline (England et al., 2025) Regarding Antarctic sea ice, the IPCC AR6 concludes that “There has been no significant trend in Antarctic sea ice area from 1979 to 2020 due to regionally opposing trends and large internal variability.” (Summary for Policymakers, A.1.5) Northern hemispheric annual snow cover has been slowly declining since 1967, with barely significant trends. The data show the Northern Hemisphere has snowier winters, accompanied by more rapid melt in spring and summer, see http://climate.rutgers.edu/snowcover/ and Section 5.6 Global greening (Chapter 2) is contributing to the decrease in planetary albedo, as forests have a lower albedo than open lands or snow. However, there is some evidence that forests increase cloud cover (high reflectivity), which counteracts the direct albedo decrease associated with increasing forested area.

90