Understanding Past Climate Forcings and Sensitivity

UNDERSTANDING DRIVERS OF PAST CLIMATE CHANGE

Climate forcings drive changes in the Earth system, causing its components to respond and interact in many ways, often on different timescales. Tectonic processes, for example, generated by mantle dynamics, altered the paleogeography of the Earth’s surface, transforming mountains and ocean basins, prompting volcanic eruptions, and altering the way that mass and energy are transferred and distributed across the planet over both short and long timescales. Changes in the Earth’s orbit around the sun alter climate by changing the amount of solar radiation by season and by latitude; they occur over tens to hundreds of thousands of years. Changes in the strength of the sun also affect the amount of solar radiation that reaches the Earth. While the strength of the sun has slowly increased over the past 4.5 billion years (Sagan and Mullen, 1972), short-term variations in solar irradiance can occur over decades or longer.

During this workshop session, participants were asked to consider the drivers of past climate change with a focus on the knowns and the unknowns of climate forcings. The session featured panelists who spoke to various forcings of climate variability, including greenhouse gas, volcanic and aerosol, paleogeographic, and orbital forcings. In the subsequent breakout discussions, participants considered gaps and strategies related to understanding drivers of past climate change.

Greenhouse Gas Forcing

Bärbel Hönisch, Columbia University, highlighted current uncertainties in CO₂ forcing during the Cenozoic. Using a research coordination network that has compiled published paleo-CO₂ data into one database (Figure 4), various proxy records of atmospheric CO₂ can be compared, and radiative forcing can be calculated from the CO₂ data. While the data can show major climate shifts from warmer to cooler periods in the past, there is disagreement in CO₂ concentrations derived from different terrestrial and marine proxies. Additionally, the associated uncertainties are not well constrained. Moving forward, Hönisch suggested that it is critical to understand proxy

disagreements, developing a more uniform proxy theory in some cases, and utilizing detailed databases like paleo-CO₂ to conduct sensitivity studies and reconstructions from the same locations to compare different proxy records. There are also large data gaps in the Cenozoic record for CO₂, and it could be useful to develop data from multiple proxies during the long-term transitions between major climate shifts. During the breakout discussion on greenhouse gas proxies moderated by Yige Zhang, Texas A&M University, and Gordon Inglis, University of Southampton, participants suggested improving frameworks (process-based, proxy-model, and proxy-proxy) to integrate information in a way that does not rely on a single proxy or record. In order to better understand proxy records and their uncertainties, participants pointed to a need to understand the underlying processes that govern each proxy (e.g., biological, ecological, oceanographic, diagenetic) and to understand whether the modern processes that underpin these proxies are the same as they have been in the past.

While reconstructions of atmospheric CO₂ are important and can allow for understanding of radiative forcing, Hönisch emphasized that it is also important to look at reservoirs of CO₂ beyond the atmosphere. This work could include reconstructing the extent of vegetation on land, carbon storage in the ocean, weathering of the solid earth, and volcanism. Advances have been made in recent years mapping paleo-ocean carbon storage over long periods of time using isotope ratios of the calcite shells of benthic foraminifera (e.g., Farmer et al., 2019; Yu et al., 2020). While most of this work has focused on the

Pleistocene, these techniques could be used to map carbon storage farther back in time. Better understanding between different carbon reservoirs and the atmosphere would allow for greater confidence in reconstructions of atmospheric CO₂.

Ed Brook, Oregon State University, discussed climate forcings from other well-mixed greenhouse gases, specifically methane and nitrous oxide (N₂O). On Quaternary timescales, methane and N₂O can be thought of more as feedbacks than forcings in the climate system in the sense that researchers are interested in the mechanisms that cause variability in these greenhouse gases. Methane is linked to processes in the carbon cycle, including emissions from wetlands with implications for paleohydrology, and N₂O is linked to changes in the global nitrogen budget, including in soils and in the ocean. From ice cores, the concentration history is known reasonably well back to 800,000 years; however, constraints on the drivers of methane and N₂O variability are limited. He noted that the concentration history before 800,000 years is largely unknown, and there is an international effort to extend the ice core record back to 1.5 million years. Prior to the ice core record, finding proxies for these gases is difficult. One challenge in using the geologic record to constrain concentrations of these gases is that their sources are highly localized, making it difficult to understand their importance as a global climate forcing.

Participants in the breakout session discussed developing new proxies for the non-CO₂ greenhouse gases to fill in data gaps both during a particular time period globally, as well as farther back in time, and suggested the need to go deep in time (e.g., pre-Quaternary Cenozoic) to find analogs for the greenhouse gas system. As an alternative approach to ice cores and proxies, Brook discussed using atmospheric chemistry models, constrained by data, to try to understand variations in sources and sinks of greenhouse gases on longer timescales. There is also work to be done understanding climate system “wild cards” like methane hydrates and permafrost, he said, and whether they cause large-scale abrupt climate shifts or more chronic, positive feedbacks.

Volcanic and Aerosol Forcing

Aerosol radiative forcing is an important factor in estimating climate sensitivity. For example, uncertainty in aerosol radiative forcing in the preindustrial period drives uncertainty from the preindustrial period to today, explained Natalie Mahowald, Cornell University, accounting for the largest source of uncertainty in anthropogenic radiative forcing (IPCC, 2014). Tropospheric aerosols have a short atmospheric lifetime (~2 weeks) making it challenging to characterize their global distribution over time using proxy records. Similarly, for past climates, knowing the aerosol radiative forcing is

important. In the breakout discussion on volcanic and aerosol forcing moderated by Sarah Aarons, Scripps Institution of Oceanography, and Clay Tabor, University of Connecticut, participants discussed the importance of acknowledging that understanding of aerosol forcing in the present is still limited, so quantifying uncertainties and performing sensitivity studies may be necessary to constrain aerosols in the past. One approach that could be leveraged is data-data comparisons, for example, comparing different proxy data for wildfires at different resolutions.

For the proxy records that do exist, Mahowald noted, terrestrial dust records from the Last Glacial Maximum (LGM), for example, are limited because they do not show variability in dust during that time period. Furthermore, they cannot provide a complete picture of the aerosol record because dust is only one of a number of aerosols important for understanding the associated climate forcing. Other aerosols that could contribute to past forcing have been neglected in paleoclimate studies. Natural aerosols (e.g., dimethyl sulfide [DMS], volcanic, terpenes, dust, and fires) can contribute 0-1 W m^-2 each, equivalent to the radiative forcing from anthropogenic aerosols (Rap et al., 2013) (Figure 5). Contributions from wild and open fires remain a large source of uncertainty in the aerosol forcing between the preindustrial period and the present (Hamilton et al., 2018; Wan et al., 2021).

Mahowald argued there is a need to use ice core, marine sediment, and lake sediment records to examine charcoal archives that can qualitatively indicate whether there was fire in the past. New proxies that have high resolution temporally and spatially could be useful to better understand distributions of aerosols over both ocean and land. Advancements in the paleoceanography community have contributed to well-calibrated proxies for aerosols and dust that have integrated proxies with records of grain size pattern. Workshop participants discussed the importance of particle size distribution and composition for quantifying aerosol forcing, and suggested integrating modern observations, modeling, and assimilation of proxy data. Participants in the breakout session described how better constraints on the quantity and composition of aerosols emitted to the atmosphere during volcanic eruptions, as well as constraints on the timing and forcing of marine and terrestrial eruption events, could be useful. Participants of the breakout discussion on temperature response to volcanic forcing in the later session agreed that improving knowledge and reducing uncertainty in the eruptions themselves (e.g., location, timing, magnitude, severity, patterns of plumes) would be necessary to reconstruct the climate response to volcanic forcing. Workshop participants discussed interest in extending the paleo-aerosol record beyond the Pleistocene and into the Pliocene, and synthesizing these proxies into global datasets as has been done for the LGM.

Ilya Bindeman, University of Oregon, demonstrated the use of volcanic tephra to understand climate forcing in the past. Volcanic eruptions, a climate forcing, can release large amounts of sulphur dioxide into the atmosphere, and in the upper atmosphere, oxidation can produce sulfuric acid aerosols that can cool the climate for several years. Oxygen isotopes in volcanic tephra preserved in the geologic record can be used to investigate the influence of super volcanic eruptions on the ozone layer (Martin and Bindeman, 2009), as tephra deposited by major super eruptions have a unique isotopic signature derived from ozone and the products of its decomposition in the atmosphere. Furthermore, he said, during past snowball glaciations, volcanoes interacted with the hydrosphere and cryosphere, leading to caldera collapses (e.g., Crater Lake) and the formation of unique geological features (e.g., pinnacles). Evaluating the interaction of caldera-forming eruptions with ice in the past using stable isotopic signatures of ancient rocks can provide evidence of glacial climate states in deep time.

Orbital Forcing

In the orbital forcing breakout discussion moderated by Lorraine Lisiecki, University of California, Santa Barbara, and Zhengyu Liu, Ohio State University, participants explained how orbital forcing can be leveraged to better understand climate feedbacks and identify thresholds and tipping points because the forcing is known, eliminating one source of uncertainty. Remaining research questions include understanding the signatures of orbital forcing, both the global and regional impacts, and the seasonal and spatial variations. Better spatial coverage of proxies, especially at high latitudes and in the Indian Ocean, and seasonally resolved proxies could help to address these questions. Looking at past warm periods to understand the rate and natural variability during those periods would help to apply the paleoclimate record to current and future climate forcings. Participants also noted that tipping points have important implications for the present, and the gradual known changes from orbital forcing provide some leverage to identify tipping points in the past.

Paleogeographic and Tectonic Forcing

Paleogeography is the study of Earth’s geographical features in the past, including the location and elevation of continents, orographic features, and the opening and closing of ocean basins and gateways. Chris Poulsen, University of Michigan, explained that paleogeography is a forcing that can affect every component of the Earth system, including the atmosphere, hydrological cycle, ocean, poles, glacial and ice-sheet formation, surface vegetation, weathering, and biogeochemical cycles. While paleogeography can complicate the argument that paleoclimate can serve as an analog for future climate change, it

is critical to quantify its contribution to both proxy records and model simulations and to distinguish paleogeography from other climate forcings. Poulsen stated that there is broad agreement based on paleoclimate modeling that paleogeography can have large local and regional effects and should be considered when interpreting individual proxy records. However, there is less consensus on whether paleogeographic forcing is important on a global scale (e.g., Lowry et al., 2014; Lunt et al., 2016; Zhu et al., 2020). Changes in geographical features over time may play a greater role in long-term climate change than has been previously recognized, and that raises questions about the robustness of Earth system models (Box 4). There are also large uncertainties in the evolution of paleogeography through time, particularly past continental elevations and ocean depths.

Moving forward, Poulsen suggested there is an opportunity to collaborate with the geodynamic and tectonics community to refine knowledge of paleogeography and the uncertainties associated with paleogeographic reconstructions, including making proprietary data publicly available. Participants in the paleogeographic and tectonic forcing breakout session, moderated by Jung-Eun Lee, Brown University, and Kate Huntington, University of Washington, also pointed to the need for diverse and collaborative teams of researchers with expertise in data and modeling to identify regions of interest and better assess uncertainties in paleogeography. Earth system models (Box 4) could also be used to promote systematic studies of paleogeographic forcing and the response to long-term paleogeographic change, using both single-model studies and model intercomparisons for specific intervals. The participants also discussed how a hierarchy of models—including regional and global, high and low resolution, and simple to intermediate complexity—could be used to separate the effect of paleogeography from other forcings. Because there is large variability in the model response to paleogeography, Poulsen argued studying past climates when paleogeography was substantially different from the present can inform understanding of processes and the robustness of models.

RECONSTRUCTING GLOBAL CLIMATE CHANGE AND CLIMATE SENSITIVITY

This session shifted focus from understanding climate forcings to assessing the impacts of climate change on a global scale. The session included a panel and breakout discussions on the knowns and unknowns of reconstructing global climate change and climate sensitivity, including connections to policy. Paleoclimate science provides an independent estimate of the climate sensitivity of the planet and key evidence for assessing and evaluating the accuracy of climate models. Earth’s sensitivity to the forcing of increased greenhouse gas concentrations can be quantified as the global average change in surface temperature caused by a doubling of CO₂ concentrations beyond the preindustrial value in a global model. Using doubled CO₂ concentrations as an initial boundary condition, a climate model can be run until the simulated temperature comes into (near) equilibrium with the higher CO₂ level (IPCC, 2014). The global average increase in simulated temperature indicates the climate sensitivity, or the sensitivity of the global model—a key index for comparing and evaluating models used to project future climate changes.

Climate Sensitivity and Policy Relevance

Motivating the importance of understanding and identifying processes that control climate sensitivity and other parts of the climate system, Gabe Bowen, University of Utah, noted that the climate system is already in or headed into a state that the Earth has not seen or experienced for millions of years. Thus, he noted that it is necessary to document climate states in the past and look for patterns to evaluate models in order to understand the Earth system under a range of conditions that are possible in the future.

Given that policy is future facing, Gavin Schmidt, National Aeronautics and Space Administration (NASA), argued that in order for paleoclimate science to influence policy, it has to affect future projections. As examples, predictions of sea-level rise and global temperatures are being assessed by models that have been constrained by past climate. Paleoclimate can provide external constraints that reduce the spread of uncertainty across models and can directly impact future projections and future policy through work on climate sensitivities. Climate sensitivity is not just one number, but a function of time-scale and processes (Figure 6). As models have become more complex, Schmidt said, they have extended the ability of paleoclimate researchers to characterize different variations in climate sensitivity (e.g., Sherwood et al., 2020). For example, ice sheets are just now being included in models, but how the inclusion of these processes affects the ability of models to calculate climate sensitivity is not yet well understood.

In order to contribute to predictability and projections that are policy relevant, Bowen argued that more can be gleaned from existing records. Additionally, strong frameworks for integrating records and different types of data may be needed to do the kind of synthesis work deeper in time that has been done for the shallower time record in order to improve mechanistic and systematic understanding.

Data Needs to Constrain Temperature and Climate Sensitivity

With respect to the distribution of proxy networks from the LGM to present, Shaun Marcott, University of Wisconsin–Madison, explained that there are relatively good constraints both temporally and spatially (Kaufman et al., 2020; Osman et al., in press). However, in order to understand climate sensitivity, forcings, and feedbacks, there is an inadequately dense proxy network for other glacial-interglacial periods that is needed to understand each unique interglacial period in the past (Cheng et al., 2016). Spatially, there are important data gaps hindering understanding of the Pacific (Walczak et al., 2020), Marcott said, as well as throughout the water column (Rosenthal et al., 2013), with implications for understanding ocean temperatures from ice cores. In order to resolve the spatial and temporal aspects of the temperature response to volcanic forcing and account for biases and uncertainty, participants in the breakout session moderated by Kevin Anchukaitis, University of Arizona, and Christina Karamperidou, University of Hawaii, discussed the need for high resolution proxy networks, particularly in under-sampled regions, including the southern hemisphere and tropics. In addition, development of proxies relevant for the long-term influence of volcanism beyond the transient response of the climate system to a particular event could be useful. Marcott explained that while there are networks of paleoproxies and glacial chronologies (Brosius et al., 2021; Dyke et al., 2014), the records are lower in quality going farther back in time. There is a lack of land sediment records that would serve as counterparts to marine records in order to understand climate sensitivity and land-sea contrasts.

Discussions in the climate sensitivity breakout session, moderated by Cristi Proistosescu, University of Illinois, and Dan Lunt, University of Bristol, focused on a need to quantify proxy uncertainties when translating localized records to global quantities used to characterize climate sensitivity, and participants suggested data assimilation as one approach forward. In order to better understand surface temperature patterns, participants in the terrestrial surface temperature patterns breakout session, moderated by Katherine Glover, University of Maine, and Ben Laabs, North Dakota State University, discussed the need for proxy calibration and understanding—for example, identifying

proxies that reflect specific seasonal temperatures in order to sort out seasonal biases. Participants echoed approaches discussed during the session, including proxy system models (Box 4), multi-proxy approaches to develop records, modern calibration work to enhance the utility of current proxies, and the use of existing but underutilized records (e.g., pollen, geomagnetic data). In the ocean temperatures breakout session, moderated by Aradhna Tripati, University of California, Los Angeles, participants discussed how multi-proxy approaches and proxy intercomparisons can be used to identify a robust temperature signal, necessary to understand thermal history and equilibrium climate sensitivity. Paired with improved understanding of such variables as the timescale, non-thermal effects, and carbonate chemistry recorded by a proxy, multi-proxy approaches could go beyond ocean temperature reconstructions to get at seasonality, hydroclimate, and biogeochemical feedbacks.

Modeling Approaches to Constrain Temperature and Climate Sensitivity

In order to reconstruct global temperatures in the past, Jess Tierney, University of Arizona, explained that proxy system models (Box 4) are needed to translate between a proxy network and temperature. Modern process studies aimed at improving a proxy system have a role to play, along with better collaboration between proxy experts and modelers who are using proxy system models to do reconstructions. Strides have been made in the past decade to describe mathematically how different proxies respond to climate forcing (e.g., Tierney et al., 2019). Bowen echoed Tierney’s emphasis on the importance of proxy system models to critically examine what is and is not known about proxies, the overlap between proxies, and how that information can be used to assess uncertainty in a more comprehensive way. With the heterogeneous distribution of proxies in space and time, discussed earlier in

the workshop, statistical techniques are used to turn a sparsely distributed proxy network into spatiotemporal information. Tierney explained how proxy information and climate model information can be leveraged together using data assimilation to develop a statistically robust global reconstruction of temperature (e.g., Osman et al., in press). Bowen agreed that statistical frameworks that integrate data and models to test ideas are necessary, and suggested another method, Bayesian hierarchical modeling, that can test different models of how the climate system works using proxy system models and a diversity of data.

While much of this work has been used to understand the climate system during the past millennium, there is an opportunity to use these tools to look father back in time. Participants in the climate sensitivity breakout session discussed how more simulations before the historical record can be used to constrain climate models and climate sensitivities. Such simulations could include more time periods and with models of different resolutions, parameter sensitivities, and perturbed physics ensembles, as well as the use of different model hierarchies. The development of a suite of model simulations for other time periods beyond the Deep-Time Model Intercomparison Project (DeepMIP)⁴ could advance work on deeper time. From Tierney’s perspective, the inclusion of terrestrial proxies in deep time data assimilation poses a challenge because of their unknown associated elevation. She noted that improving constraints on climate sensitivity in deep geologic time also requires knowledge of changes in temperature and CO₂; however, temperature and CO₂ proxies are interdependent, particularly through the carbonate system, and more work needs to be done on refining the global temperature and CO₂ curves (Figure 7).

Jiang Zhu, National Center for Atmospheric Research (NCAR), discussed cloud feedbacks, which remain the most uncertain of the forcing feedback processes but may hold key information about past and future warming. While the feedback is likely positive (Sherwood et al., 2020), Zhu argued that one of the biggest unknowns hampering the ability to use past information to inform the future is how and why the cloud feedback depends on background states (Caballero and Huber, 2013; Schneider et al., 2019; Zhu et al., 2019). To understand the state dependence, more work with state-of-the-art modeling and analysis techniques may be needed. More accurate reconstructions of past sea-surface temperature patterns could contribute to understanding of the coupling between cloud feedbacks and large-scale processes across a range of past climates (Dong et al., 2019; Erfani and Burls, 2019; Zhou et al., 2017). Participants in the climate sensitivity breakout session also discussed the

___________________

⁴ See https://www.deepmip.org.

importance of state dependence to constrain feedbacks and climate sensitivity. Zhu added that cloud-aerosol interactions are likely very important (e.g., Kiehl and Shields, 2013; Kump and Pollard, 2008), and while models are becoming more sophisticated, modeling studies are still lacking in Zhu’s opinion, and reconstructions of past aerosol distributions need to be refined, as discussed above.

This page intentionally left blank.