Sessions held on the third day delved into the regional aspects of climate through two lenses that have direct relevance to human timescales. Building on earlier discussions on feedbacks, the first session focused on paradigms and applications of paleo hydrology, which is critical for understanding the impacts of future climate change on habitability. The second session focused on coupled modes of ocean-atmosphere variability, which is important to contextualize current changes in temperature and rainfall across the Earth.
The distribution of water is a critical expression of climate change in terrestrial systems. Naomi Levin, University of Michigan, posed several fundamental research questions to understand hydrologic responses to future climate change: How does water availability vary across topographic and ecological gradients through the Cenozoic? How can the influence of climate change be separated from other processes that control the distribution of water on landscapes? How can insights from hydrologic variability in the past be used to understand responses to future climate change? In order to answer these questions, integrated, multi-proxy datasets that capture spatial variability in hydroclimate would be required, as well as leveraging new tools for quantitative constraints by combining thermodynamically based data with higher resolution biologically based data, Levin said.
Matt Huber, Purdue University, discussed the importance of monsoons as the main source of water for more than 50 percent of the world’s population, motivating a need to understand capabilities to predict monsoons in the future and use the past to constrain and inform those predictions. The complexity of monsoons can be thought of as a microcosm for the functioning of the whole Earth system, and paleoclimate can provide insights into monsoon behavior and its drivers, for both the global monsoon and its regional expressions. Specifically, he said, paleoclimate science may be uniquely positioned to answer questions of how monsoon dynamics change in a warming world. Using past climates as a test bed, one open research questions stands out: Are there tipping points in the monsoon or are there smooth transitions, and what are those transitions a function of? As one strategy, Huber suggested looking to
past climates that were warmer than the present day and selecting different periods of time that can test individual theories of monsoon drivers using a factorization approach.
The biosphere affects the atmosphere through exchanges of energy, through carbon, water, and aerosols. Paleoclimate science has historically had an important role in advancing understanding of biosphere-atmosphere interactions, explained Christopher Skinner, University of Massachusetts Lowell, for example through the study of paleo monsoons. Research in recent years has demonstrated that vegetation changes are not only a key driver of local climate, but also have connections to the remote and global climate (Pausata et al., 2020) (Figure 18). The impact of incorporating biosphere-atmosphere interactions in models is apparent when the inclusion of prescribed or simulated changes in vegetation brings climate model output and proxy estimates of climate into better agreement. Huber added that it is important to include interactions with the biosphere, soil, and aerosols in the modeling of monsoons. Skinner explained that one gap in current understanding is how robust recently identified biosphere-atmosphere impacts on climate are across climate models (Otto-Bliesner et al., 2020). In order to overcome the limited coverage of vegetation boundary conditions needed to run paleo-vegetation experiments, he said, modeling centers could translate simplified maps of vegetation into useful forms for land surface models, and greater vegetation proxy coverage could also help to constrain vegetation models. In order to move beyond prescribed vegetation as a boundary condition in models, Earth system models would need to be able to simulate vegetation processes in dynamic vegetation modules. Skinner suggested that to improve the ability of Earth system models to simulate past vegetation changes, there is a need to examine soil vegetation processes, particularly soil characteristics, vegetation moisture stress parameterizations, and carbon pools in models.
Juan Lora, Yale University, drew attention to gaps in understanding of extreme events, which have major impacts from the perspective of regional hydroclimate and habitability, and as other participants have noted, extreme hydroclimate events have particular policy relevance. Lora argued understanding extreme events and their global and regional drivers in warm and cold past climates, particularly beyond the Holocene, is a necessity. Hydroclimate processes interact strongly on different timescales, and attention to hydroclimate processes on synoptic, seasonal, multidecadal, and forcing timescales in past climates may be essential. He noted that specific gaps in understanding include the responses and drivers of regional hydroclimate
processes in different climates beyond annual and seasonal timescales, and the identification of what proxies are sensitive to what timescales.
Workshop participants and panelists discussed how the range of hydroclimate archives can be used to answer outstanding research questions regarding modes of ocean-atmosphere variability. In the breakout sessions, discussions focused on applications of the following paleoclimate archives for this purpose: ocean, lake, and continental sediment cores; terrestrial outcrops, paleosols, and speleothems; ice cores; and tree rings.
Sediment records are the predominant climate archive for much of Earth’s history, formed by rainfall and runoff eroding rocks exposed on the continents and transporting eroded debris in streams and rivers in particulate and chemical (dissolved) forms. Sediments are then deposited in lower-energy environments, such as lakes or shallow, inland seas, and on continental shelves. Most sediment is eventually transported to the ocean, where it may persist for tens of millions of years until removed by tectonic processes. As a result, the
oldest marine sediment records extend back to ~170 million years; records of older periods are derived from the continents. There, wind and ice also drive erosion and transport, creating unique sedimentary signatures indicating Earth system dynamics in the past. Marine sediments in the deep sea were deposited in a relatively calm, less destructive environment, producing many continuous sediment records of tens of millions of years of Earth’s history.
Ice cores, corals, and tree rings provide relatively recent records of Earth’s history. Formed through the annual deposition of snow, ice deposits from mountain glaciers and massive polar ice sheets several kilometers thick contain multiple proxies for past climate variability. While cores of mountain glaciers record up to 10,000 years of climate change, cores recovered from Antarctic ice sheets contain records dating back to ~800,000 years before present.
Additionally, in regions of sufficient rainfall, groundwater percolating through soil and rock can dissolve and redeposit limestone in caves, containing records of hydroclimate intervals extending to several hundred thousand years. Tree-ring climate records, in contrast, offer a hydroclimate archive for the interval of the past few tens, hundreds, or thousands of years; however, they can provide annual resolution. The development of annual growth bands, manifest as rings in cross-sections of the tree trunk, depends on hydrographic conditions; therefore, descriptions of the rings can provide a powerful, high-resolution insight into regional patterns of climate conditions, such as precipitation or temperature.
Ocean, Lake, and Continental Coring and Drilling
Participants in the continental drilling and lake coring breakout moderated by James Russell, Brown University, and Sherilyn Fritz, University of Nebraska-Lincoln, discussed how sedimentary records from continental drilling and coring remain the premier tool to obtain multi-proxy records of continental climate and ecosystems. However, dense, long continental records, particularly pre-Holocene, would still likely be needed in order to unravel how continental climates respond to forcings and perturbations. Participants noted one key science question: What is the role of greenhouse gas forcing in past changes in continental temperature, ice, rainfall, and ecosystems, particularly during intervals warmer than the present? An important strategy to better understand spatial dynamics going forward would be sampling transects across important climatic, geomorphic, and ecological gradients.
Participants in the ocean coring and drilling breakout session, moderated by Yair Rosenthal, Rutgers University, and Melissa Berke, University of Notre Dame, identified a range of research questions that can be addressed by ocean coring and drilling tools, including reconstructing large-scale temperature patterns, ocean heat and carbon uptake processes, biogeochemical cycling, atmospheric CO2 variability, reconstructing changes in the hydrologic cycle, changes in ocean circulation, Arctic and Antarctic ice-sheet history, and the
variation of submarine volcanism and seafloor spreading rates over time. The breakout participants discussed an integrated surface to seafloor approach that would couple the whole water column marine biogeochemistry with paleo reconstructions.
Outcrops, Paleosols, and Speleothems
Two overarching questions discussed in the outcrops and paleosols breakout session, moderated by Katie Snell, University of Colorado, and Jeremy Caves Rugenstein, Colorado State University, were how to best leverage the inherent variability in outcrops and terrestrial sections, and how to best design studies that address paleoclimate questions using records that archive different parameters. Participants in the speleothems breakout session moderated by Kathleen Johnson, University of California, Irvine, and Jessica Oster, Vanderbilt University, discussed how speleothems can address climate questions from the tropics to the polar regions, including moving beyond Intertropical Convergence Zone paradigms and exploring applications for constraining zonal and regional hydroclimate, paleo-ENSO (El Niño Southern Oscillation), and extreme events, as well as permafrost timing and extent. Speleothems, which have the potential to be used for spatial hydroclimate reconstructions through new proxies and syntheses, could allow for evaluation of seasonality and seasonal bias and contribute to long-term cave monitoring. Participants discussed the importance of replication and reproducibility in the speleothem field that also needs to be balanced with cave conservation.
Participants in the ice breakout session moderated by Brad Markle, University of Colorado Boulder, and Mathias Vuille, State University of New York at Albany, discussed how new ice-core records could help to address outstanding questions about ice-sheet stability and ice-sheet extent. Developing proxies more related to ice-sheet stability would help to better understand surface melt, ice-sheet elevation, and accumulation. Ice-core records could also be used to access what the climate system looked like over the last million or several million years using the unique proxies ice cores can provide, such as paleo greenhouse gas concentrations. Participants also suggested combining ice-core work with other archive work, such as drilling rock cores at the bottom of ice sheets.
The Arctic is warming twice as fast as the rest of the world, with strong coupling and feedbacks between the cryosphere, precipitation, plants, and
carbon. Major knowledge gaps remain with respect to hydrological feedbacks in the Arctic region. Elizabeth Thomas, State University of New York at Buffalo, identified the need to fill spatial gaps in the terrestrial and marine Arctic to understand rapid warming intervals and to fill temporal gaps on orbital timescales, where terrestrial archives in the Arctic and marine archives that preserve terrestrial proxies can be used. Participants in the breakout discussion noted that more spatial coverage of ice cores would help to assess spatial variability and understand regional dynamics in the past, and increasing the temporal resolution of ice-core records would help examine the evolution of seasonal cycles. Thomas called for isotope-enabled climate models that would serve as good priors for data assimilation, better modern observations that can be gridded for the high Arctic latitudes, and a large increase in observation networks for water isotopes, precipitation, soil, lakes, and water vapor.
Participants in the tree rings breakout session, moderated by Valerie Trouet, University of Arizona, and Rosanne D’Arrigo, Lamont-Doherty Earth Observatory, discussed factors that could be uniquely resolved by tree rings, including spatiotemporal patterns of anomalies and interannual variability including seasonality. Tree rings could be used to address gaps in understanding of hydroclimate mechanisms and processes, including ENSO, monsoons, and other modes of variability. Some participants suggested a new application for tree rings could be identifying tipping points and thresholds and their causes in past climates. There may be an opportunity to focus on the linkages between the tropics and higher latitudes, but there is a need for better temporal coverage of tree-ring data for Africa, Asia, and the tropics.
Connie Woodhouse, University of Arizona, advocated for using tree rings to look more closely at the temperature component of hydroclimate—specifically, to better understand how temperature interacts with precipitation over time—drawing attention to the lack of chronologies with independent temperature information available for the mid-latitudes, particularly North America. Woodhouse called for independent proxy records for temperature and precipitation to understand the role of temperature as part of moisture-driven metrics and to answer research questions, such as: How has drought varied over time, what was the role of temperature, is that role changing under current climate, and what is that change in the context of the paleo record? In addition to temperature information, Woodhouse repeated calls made throughout the workshop to look more closely at seasonality in order to develop a more holistic understanding of past climate and, specifically, to gain understanding of moisture delivery mechanisms and interactions with environmental conditions. Similarly, there is a need for independent proxies for seasonable variables and an opportunity to employ underutilized or emerging technologies, such as inter-annual ring measurements, wood anatomy measurements, blue light-intensity measurements, isotopic analysis, and process modeling.
As mentioned previously, climate changes occur on multiple spatial and temporal scales. Similarly, responses to forcings can vary regionally. Natural variability in the climate system can result in relatively rapid changes, or fluctuations on the order of years to decades in duration, both of which affect particular regions. Examples of known multidecadal climate modes recorded in
proxy records or observed directly include Atlantic Multidecadal Variability, Dansgaard–Oeschger Events, ENSO, Heinrich Events, North Atlantic Oscillation, Pacific Decadal Oscillation, Quasi-biennial Oscillation, and Southern Ocean Centennial Variability (von der Heydt et al., 2021) (Figure 19). Panelists discussed both data and model needs and opportunities for paleoclimate research. Discussions during this session also highlighted the outstanding challenges of chronology, inter-site, and proxy-model comparisons for reconstructing the various modes of climate variability on both long and short (human) timescales.
Data Needs and Opportunities
While there are no perfect geologic analogs for future climate changes, a mechanistic understanding of climate feedbacks could improve predictions of future climate. Modes of climate variability—defined by preferred spatial patterns and temporal variability (Figure 19)—change as a response to changes in background states, providing an opportunity to increase predictive power by understanding how the nature and strength of climate feedbacks change, explained Christina Ravelo, University of California, Santa Cruz. Studying climate modes involves long time series that capture many cycles, as well as records that span a range of background states, including warm periods. Ravelo explained how ocean sediments fulfill both of these requirements and can be used, for example, to reconstruct ENSO variability using individual foraminifera analysis (Rustic et al., 2020; White et al., 2018) and to reconstruct mean background state and orbital variability (Peterson et al., 2020). Ravelo recommended future studies that constrain the potential state dependency of regionally specific and global feedbacks using any combination of data collection, data analysis, or data-model integration approaches.
Sam Stevenson, University of California, Santa Barbara, used ENSO as an illustrative example to discuss outstanding questions in the context of modes of paleoclimate variability. Recent work has shown there is a large internally driven, unforced variability in ENSO from proxies and models (e.g., Grothe et al., 2020; Wittenberg et al., 2009), and the tendency for El Niño events to follow strong volcanic eruptions based on large ensembles of climate models (Stevenson et al., 2016). However, proxy records suggest that forced responses shown by models may not agree with paleoclimate observations (Dee et al., 2020), but it remains unclear whether this indicates a problem with the simulated response to eruptions or a signal-to-noise problem with observations (e.g., there are not enough data to detect a significant response). Proxy system
models, which translate between climate changes and signals recorded in the proxy, are also uncertain, complicating interpretations for model-proxy disagreement. Stevenson called for more high-resolution marine and terrestrial proxy records; long-term in situ monitoring of proxy sites to understand how climate signals are recorded at those sites; an emphasis on water isotope observations both at proxy sites and in a longer-term monitoring context more broadly; and facilitating the ease of use of proxy data for modelers, including working toward the inclusion of isotope-enabled models in standard model versions. Kim Cobb, Georgia Institute of Technology, drew attention to approaches that would bolster the utility of existing paleoclimate archives. Cobb echoed workshop discussions about the need to direct resources towards modern-day process studies to inform proxy system modeling, reproducibility of high-value records, long-term monitoring of paleo-relevant climate variables (e.g., water isotopes), and approaches for paleoclimate data-model intercomparison, including paleoclimate data assimilation.
Modeling Needs and Opportunities
To advance paleoclimate research capabilities and understand future multicentury changes in the Earth system, Pedro DiNezio, University of Colorado Boulder, discussed the need for a fully configurable Earth system model with expanded functionalities that would include more components of the Earth system, including sediments, soils, the carbon cycle, ice sheets, and vegetation. He noted that it may be essential for such a model to be capable of managing complexity (Held, 2005; Polvani et al., 2017), and be community driven to engage the entire community. Community-driven modeling, following the example of the Community Earth System Model (CESM; Danabasoglu et al. 2020), could add, maintain, and tune newly developed community components, have regional or grid refinement modeling capabilities, and preserve key features of models (e.g., a realistic El Niño, climate sensitivity consistent with past changes). Considering computational efficiency of parametrizations may be important, particularly when upgrading model
components and parameterizations. In order to improve modeling capabilities, DiNezio repeated the importance of optimally increasing paleoclimate observations, particularly for modes of variability. Additionally, many different models or perturbation experiments within one model may be needed to combat structural uncertainties in paleoclimate simulations.
In order to connect noncontinuous observations, Hali Kilbourne, University of Maryland, offered a vision for adapting modern data assimilation techniques for paleoclimate proxy data. Recently, climatologists and meteorologists have used reanalysis to take sparse data and assimilate it in a model to fill in gaps in space and time, producing a continuous dataset of climate system variables that can be used to investigate climate processes (Hersbach et al., 2020). While paleoclimate data assimilation is still in its infancy, and there are challenges to adapting modern data assimilation techniques for the sparse, low-temporal resolution of proxy data, data assimilation has the promise of answering questions about climate variability, such as: What are the mechanisms behind multidecadal variability in the Atlantic? A focus on obtaining high-quality annual or better resolution records from recent centuries in the region could enable better testing of model-based hypotheses. Kilbourne posed another key question: Did the AMOC slow down in recent centuries? In order to answer this question, modern observations could be put in historical context over the geologic record, which would involve building evidence from available variables that record information about circulation. Building a widely used data assimilation product would require improving proxy networks with the infrastructure to build and maintain machine-readable metadata-rich databases, better characterizing physical and proxy variables, and improving models.
Outstanding questions about the climate system could be answered by linking multiple localities, which would require major advances in chronology. This is also needed to attain new high-resolution records, explained participants in the Cenozoic and Cretaceous breakout session, moderated by Celli Hull, Yale University, and Lisa Tauxe, Scripps Institution of Oceanography. Participants also discussed the need for targeted data collection and modeling during key intervals (similar to current efforts in the Pliocene [PlioMIP], Miocene [Mio-MIP], and Eocene [DeepMIP]) from the Cretaceous to the modern day. Using models with a range of complexities,
including biogeochemical and ecosystem modeling, and constraining boundary conditions with proxy data, may be central to leveraging paleo archives for understanding fundamental dynamics in the climate system. While participants noted that there are more opportunities to extract value from existing archives, participants in the late Holocene breakout session, moderated by Julie Cole, University of Michigan, and Sloan Coats, University of Hawaii, discussed that new data from tropical Africa, South America, the tropics, and the southern hemisphere is also needed to address outstanding questions about climate modes in the late Holocene. Research questions moving forward could include the following: How can climate modes in the past be distinguished from the background climate state or external forces? What is the variability in the late Holocene? What is the relationship between variability and external forces?
In the breakout session on glacial climates and abrupt change, moderated by David McGee, Massachusetts Institute of Technology, and Bryan Shuman, University of Wyoming, participants posed two key research questions: What are the dynamics driving abrupt changes, how do they depend upon the system state, and how can data-model integration be used to constrain these dynamics? What are the key responses, particularly rates and feedbacks, of
abrupt changes across the cryosphere, ecosystems, and hydroclimate at local and global scales that are relevant to future transitions and tipping points? Several participants noted a need for chronologies beyond radiocarbon to look farther back in time, using tools for synchronization (e.g., paleomagnetism, tephras) and absolute dating. In the deep time breakout session, moderated by Francis Macdonald, University of California, Santa Barbara, and Matthew Clapham, University of California, Santa Cruz, participants discussed the opportunity to look at the full range of extreme events over long timescales and use events that differ in rates, magnitude, and initial conditions to better understand processes. Participants also discussed the opportunity to look at how climate change is linked to extinction events, ocean anoxia, and the carbon cycle on different timescales.