Jena, March, 1999

I. The Paleotrace Gas and Aerosol Challenge

Understanding the decadal to millennial-scale interactions between chemical components of the atmosphere (CO2, N2O, CH4 and other reactive gases, SO42- aerosol, mineral and organic aerosols) and climate remains a major and fundamental challenge to be addressed by global change scientists. Without a predictive understanding of the natural changes that have taken place over the past 150,000 years, we have little hope of accurately foreseeing the long-term consequences of human activities on global climate over the next decades to centuries. Paleodata from ice core, marine and terrestrial data sources provide information about former states of the ocean, terrestrial biosphere and physical climate. These diverse sources of paleodata will help us to develop and test predictive, multicomponent Earth system models, within a framework of known temporal and spatial constraints.

With this background, it was decided that IGBP should act to promote interdisciplinary and international co-operative action, by mounting a concerted attack on the paleotrace gas and aerosol problem using a combination of ice-core atmospheric composition data (the primary "target’ for explanation), state-of-the-art models, and the best available global paleodata sets to constrain models. Hence,t the IGBP launched its ‘Paleotrace Gas and Aerosol Challenge’; a joint initiative of the IGBP Global Analysis, Interpretation and Modelling (GAIM), Past Global Changes (PAGES) and International Global Atmospheric Chemistry (IGAC) programs.

The workshop, assigned the acronym TRACES (Trace Gas and Aerosol Cycles in the Earth System), brought together people and concepts from multiple research communities (terrestrial, marine, ice-core, atmospheric chemistry, carbon cycle, paleoclimate modelling) to discuss common research themes related to the regulation of atmospheric composition in the past. The aim of the workshop was twofold; first, to establish the current state of Earth system modelling and data acquisition relative to this problem, and second, to identify future high-priority needs which would require collaborative, international and multidisciplinary research.

At this workshop, TRACES participants were presented with the challenge of identifying outstanding research initiatives which satisfied the following two criteria: 1) that our understanding of the processes and mechanisms underlying Earth system feedbacks that control atmospheric composition, would be greatly enhanced by such activities, and 2) that these goals could realistically be met within a time frame of about five years. This report begins with a brief summary of the state of the science in paleodata collection, paleoclimate modelling and biogeochemical cycling, to serve as a background for the scientific research priorities established at the workshop.

II. Sources of Paleodata

(a) Ice cores

High resolution ice core records now provide global change scientists with a detailed record of atmospheric trace gases spanning the last four glacial-interglacial cycles. Not only can concentrations of relatively long-lived trace gases such as CH4, N2O and CO2 be measured in ice; also levels of gases with short residence-times such as CO and biogenic species including NH3, acetate and formate, can be traced through time. Biogenic trace gas and isotopic measurements in ice cores provide global constraints on ecosystem processes.

Methane and CO2 dynamics during glacial-interglacial cycles have become an important focus of global change research, because they are key components of the earth’s carbon cycle, and their concentrations show different dynamic behavior and relations to climate change. For example, during the Younger Dryas cooling event, CH4 decreased but CO2 continued rising. Atmospheric CO2 apparently responds to prolonged periods of cooling, thus providing a positive feedback in the climate system, but not to rapid climate oscillations such as Dansgaard-Oeschger events. We are still largely unable to explain the processes and mechanisms underlying these dynamics.

There are uncertainties associated with the measurement of trace gases in polar ice, as well as with the chronology of samples of oldest and deepest ice. Future research directed toward addressing the following issues was strongly recommended. 1) Does the CO2 concentration measured at the bottom of the Vostok ice core (at 400 ka) represent the highest interglacial CO2 level ever reached, or are there earlier periods with even greater CO2 concentration? 2) How important is the consideration of changes in atmospheric alkalinity for measurements of CO2 concentrations in ice? 3) What was the chemical status of the stratosphere at the time of the last glacial maximum, and how did it affect global climate processes? 4) What is the potential for tropical glaciers to contribute relevant data on atmospheric chemistry and trace gas concentrations through the Holocene?

(b) Marine Paleodata

Key characteristics of surface ocean waters are temperature, salinity, pCO2, nutrient utilization and alkalinity, and a variety of proxies have been developed to indicate these variables. Sea surface temperatures and the location and strength of coastal upwelling zones can be reconstructed for key time periods in the past (e.g. Last Glacial Maximum) from plankton assemblages using species-abundance transfer functions and modern-analog approaches, stable O-isotopes, and alkenone ratios. Sr/Ca ratios in corals have also been used for paleotemperature reconstructions. Proxies for subsurface conditions, including d13C and Cd/Ca ratios in benthic foraminifera, have been widely used to reconstruct paleocirculation changes. However, further work is needed, especially to clarify contradictions among different proxies for changes in marine productivity, which have an important bearing on the past marine sources of N2) and DMS and on the CO2 problem.

(c) Terrestrial Paleodata

Because of the increasing focus on data-model comparisons in paleoclimate and paleovegetation modelling, the creation of temporally and spatially resolved terrestrial databases is a high priority. Global terrestrial databases currently in use include: 1) the Global Lake Status Data Base which contains estimates of changes in lake level or area reconstructed from biostratigraphic or lithological data, which are a function of changes in precipitation minus evaporation (P-E); 2) Paleovegetation Mapping Project (known as BIOME6000), which contains paleovegetation data sets for 6000 years B.P and the last glacial maximum (LGM); 3) Dust Indicators and Records of Terrestrial and MArine Paleoenvironments data base (DIRTMAP), which provides data on dust accumulation and flux rates as recorded in ice cores, marine cores and terrestrial (loess) sections, over a period from the last glacial period to the present day; 4) the 21 ka tropical data synthesis project, which has used multiple proxies (pollen, plant macrofossils, noble gas thermometry, speleothem isotopic paleotemperatures, and lake-status changes) to quantify changes in glacial-age temperature and water balance parameters in low latitudes. Continued collection and synthesis of terrestrial paleo data is needed in light of the increasing need for data sets to both validate and calibrate newly developed Earth system models. Improvement to the chronologies of individual records is also necessary in order to use these data to evaluate model experiments focusing on rapid changes.

III. Paleoclimate Modelling

Climate model complexity has rapidly expanded from the introduction of the first Atmospheric General Circulation Model (AGCM) approach to the evaluation of paleoclimatic problems over two decades ago. Sensitivity studies have demonstrated that both land- and ocean-surface processes are important for accurate predictions of past climates. By testing the sensitivity of climate models to different coupling scenarios (atmosphere, biosphere, hydrosphere), provided sufficient experiments are conducted, the strength of interactions between climate components can be quantified, and the type of interaction classified as additive or positively or negatively synergistic. Fully coupled Atmosphere-Ocean-Vegetation climate models (AOVGCMs) are under active development.

Comparisons of the response of different climate models to identical external forcings can help to determine which results are model-dependent and which are robust. Differences between models can, in principle, be related to differences in model formulation. The Paleoclimate Model Intercomparison Project (PMIP; Joussaume & Taylor, 1995) is an international initiative to compare climate simulations from about 20 climate models, and to evaluate the ability of these models to reproduce both last glacial maximum and 6 ka climate features. The mid-Holocene (6 ka) was selected as a focus for PMIP simulations in order to study climate response to changes in insolation without the complications of changes in ice sheets (Joussaume et al., 1999). PMIP simulations all show enhancement of the Afro-Asian monsoons and high-latitude warming, however, none of the simulations adequately reproduce the magnitude of 6 ka monsoons and high latitude warming. This magnitude can only be approached when both vegetation and sea-surface conditions are allowed to change.

Two recommendations are made for improving paleoclimate modelling as a necessary foundation for TRACES research. 1) Fully coupled models (AOVGCMs) need to be developed, tested and applied to paleoclimate problems routinely. 2) Intercomparisons need to be carried out for AOVGCMs (or sub-components) in order to test the robustness of those simulations.

IV. Biogeochemical Cycles

(a) Carbon

An understanding of all of the major physical and biological processes modifying atmospheric CO2 levels is essential if we are to be able to predict the future of CO2 added to the atmosphere over time scales of a century or longer. On glacial-interglacial time scales, these processes must involve change in the oceans, which are able to greatly alter atmospheric CO2 concentration through changes in the balance between CO2 removal by deep water sequestration and CO2 release by CaCO3-compensation. Several hypotheses to explain the low CO2 concentrations during glacial periods invoke changes in externally supplied nutrients (especially eolian deposition of Fe), resulting in an enhanced biological pump. It seems that mechanisms, in addition to changed biological-pump activity, must at least partly be responsible for the drawdown of atmospheric CO2, because even when OGCM-simulated ocean productivity is unlimited by Fe, N or PO4, the magnitude of simulated CO2 decrease does not match that observed during the last, and earlier, glaciations.

As with ocean carbon cycling, many questions concerning the terrestrial component of the global carbon budget remain unanswered. Despite advances in all three of the methods used to reconstruct biospheric carbon (data-based reconstructions, global ocean d13C-budget, and climate-biosphere models), estimates of terrestrial carbon storage at the last glacial maximum have still to be determined within a reasonable margin of error. The role of aerosols in adding nutrients to terrestrial systems may be important. Continued research on the role of terrestrial ecosystems in influencing the global carbon cycle is needed even though it is clear that the main glacial-interglacial changes in CO2 cannot be explained by terrestrial processes alone.

Methane’s role in global carbon cycling has been relatively under-appreciated up to now; so much has yet to be investigated beyond the reconstruction of flux measurements from ice. Surprisingly, most CH4-source studies focus on northern temperate peatlands despite indications that tropical wetlands likely hold a key role in affecting global carbon cycling. Even with CH4-research focused on northern temperate regions, measurements of CH4 fluxes in the field are often inconsistent with predictions of current models. Future research needs to include work on the biological processes underlying CH4-cycling, and enhancement of available CH4 flux measurements in the tropics.

(b) Nitrogen

Nitrogen is fundamentally important component of the biogeochemical system, because of its role as a proximally limiting nutrient both on land and in the oceans, but also because of the ? ? N20 and NOx. The global ocean N budget may be unbalanced in both glacial and interglacial periods. N2O is a climatically active gas with major biogenic sources both on land and in the ocean. NOx, produced in soils, is important in determining the troposphereic content of 03, another climatically active gas. Reactive nitrogen on land and in the oceans is provided by biological N2-fixation and as such is influenced by the supply of other nutrient elements, especially P on land and Fe in the ocean.

c) Water

Modelling of the processes involved in transport of water from the continents via river systems and lakes to the ocean helps to provide us with a better understanding of the sensitivity of the hydrological system to changes in climate. High-resolution continental hydrological models are in their early stages of development and have not been coupled to other models. They can provide a realistic way of estimating freshwater discharge to the oceans, which in turn affects ocean circulation. Hydrological modelling predictions are strongly constrained by the representation of topography. They have been useful for estimating the impact of climate and land-use change on river discharge at the continental-scale, and as a regional to global model diagnostic. Model simulations under modern climate show that the Caspian Sea and north African and Indonesian basins have significantly greater capacity than their current water storage, indicating a potentially high sensitivity of these basins to past changes in precipitation. Simulations of the mid-Holocene hydrological cycle over northern Africa, driven by output from PMIP 6 ka simulations, have been employed to document how much these experiments underestimate observed paleohydrological changes. In the future, continental hydrology models should be linked interactively to the other climate systems model components in order to represent e.g., the effect of plant physiological responses to CO2 on water storages and flows and the impacts of changes in continental climates on freshwater supply to the oceans.

(d) Dust

Under current atmospheric conditions, mineral aerosols (i.e. desert dust) can have very large local effects on heat transfer. Areas which are likely to be sources of atmospheric dust are those with little vegetation and pronounced seasonal changes in moisture. Mineral aerosol abundance is estimated to have been considerably greater during the last glacial maximum, and therefore likely exerted an even stronger influence on radiation budgets than at present. Model simulations of dust accumulation during the last glacial maximum show that increases in wind strength and a decrease in the strength of the hydrological cycle are insufficient to explain the atmospheric dust loading at the LGM; however an increase in source area results in an increase in atmospheric loading by a factor of 2.5. This source area increase can be explained by a combination of low precipitation and the physiological effects of low CO2 on vegetation during glacial periods.

Several issues concerning paleodust cycling remain outstanding, and were recommended as targets for future research: 1) quantification of increased dust accumulation during the last glacial maximum, 2) mechanisms promoting increased aerosol production and deposition, and 3) implications for atmospheric loading. Although the significance of dust for climate forcing is well recognized, its role as either an indicator of altered climate, or primary driver of paleoclimate, must still be determined. The biogeochemical role of dust in the oceans also requires better quantification, taking into account especially the specific pathways of Fe uptake by marine ecosystems.

V. Short-term (5-year) Research Objectives

TRACES participants identified the following research tasks as being of great importance for the advancement of Earth System science. Table 1 provides a summary of the ranking of these objectives as either moderate (3), high (2) or very high (1) in relative priority.

(a) Paleoclimate Modelling

• Create an ocean biogeochemistry model which incorporates classification of marine biota as "marine functional types" (analogous to the characterisation of terrestrial vegetation as plant functional types).

• Develop fully-coupled, three-dimensional Atmosphere-Ocean-Vegetation GCMs (AOVGCMs), or Earth System Model (ESMs). These models should be validated using paleodata to assess their ability to capture radically different climates and environments from those of today. Model development should be accompanied by specific model-intercomparison efforts.

• LGM vegetation simulations should be conducted with process-based (mechanistic) biospheric models, so that plant-level physiological effects of low atmospheric CO2 can be properly addressed, scaled to the ecosystem-level, and allowed to interact with the physical climate system.

• Conduct continuous simulations across key climatic boundaries (e.g. start of Bšlling-Alleršd; Younger Dryas - Preboreal transition) for periods of several thousand years using ESMs of intermediate complexity, and for periods of less than one thousand years using 3D-ESMs.

• Conduct time-slice simulations, with a first focus on the LGM as compared to the pre-industrial period.

(b) Biogeochemical Cycles

• Perform in-depth intercomparisons (model versus observed) of marine productivity for several key time slices.

• Introduce a more refined treatment of weathering processes in global carbon models.

• Establish long-term, land-based flux measurements of atmospheric trace gases (CO2, CH4, N2O, CO, NMHCs, NOx, dust), in combination with experiments investigating the relationship between flux measurements and physical, chemical, and biological processes. A greater appreciation of the variability (seasonally, spatially, and temporally) inherent in trace gas emissions is also required.

• Progressively replace current correlative atmospheric gas flux relationships with process-based functions and parameterisations in ecosystem models. Paleoatmospheric chemistry (particularly oxidative capacity) can then be approached by combining ecosystem modelling and atmospheric chemistry transport modelling.

• Estimate peat accumulation rates based on field measurements and biological (ecophysiological) understanding (e.g. Sphagnum- versus sedge-dominated wetlands).

• Extend current research on the dating and identification of periglacial dust source regions (Alaska, Europe, Siberia, China) using mineralogical and isotopic tracer techniques.

• Attempt to create a map of soluble Fe input into the oceans at the last glacial maximum to better understand glacial-interglacial dust cycles and their relation to atmospheric pCO2 changes.

• Refine established proxies and perhaps develop new proxies for reconstructing ocean productivity from both marine paleosediments and ice cores.

(c) Paleodata

• Characterise oceanic paleosediments on the basis of marine functional types to facilitate direct model-data comparisons.

• Create distribution maps of modern wetlands (remote sensing) and paleowetland (14 C-dated peat records). Define target margins of error, keeping in mind that estimates of the present extent of wetlands vary by a factor of two or more. A standardized wetland classification must be adopted.

• Continue compilation and improvement of existing global paleodata bases.

• Improve of chronologies of terrestrial and marine records, specifically loess records.

• Improve isotopic tracer data relevant to the cycling of C, N, and O.

• Synthesise major river basin discharge data from marine paleosediments for validating hydrological models.

• Compile oceanic source data for N2O to help provide explanations for why N2O concentrations parallel trends in CH4 during periods of rapid climate change.

• Map distributions of (well-dated) loess source regions and accumulation rates of loess.

• Update DIRTMAP to contain information on loess grain size, Fe-content, micro-aggregates and mineralogy (hematite, Fe, kaolinite, chlorite).

• Create a global synthesis of paleoproxy data for ocean circulation and ocean alkalinity.

(d) Target Glacial-Interglacial Time Slices (in order of priority)

• pre-industrial

• last glacial maximum (21 ka)

• glacial initiation (115 ka)

• intervals of rapid change during the deglaciation start of Bolling-Allerod, 14.5 ka; end of Younger Dryas, 11.6 ka)

• Holocene (9 ka, 6 ka)