PAGES Magazine articles

Kees Klein Goldewijk1, M.-J. Gaillard2, K. Morrison3, M. Madella4 and N. Whitehouse5

LandCover6k general meeting, Utrecht, The Netherlands, 14-16 June 2016

The LandCover6k working group brought together a group of 50 scientists from various disciplines (archaeology, history, geography, paleoecology, land-use and climate modeling) to discuss progress and future plans to improve historical land-use descriptions for Holocene climate modeling. It was the group’s second general workshop - the first being the launch meeting in October 2015 (Gaillard et al. 2015). Alongside this, a first workshop of the historians and archaeologists group was held in 2015 (Morrison et al. 2016) and several regional workshops have also been organized. The aim of this second general workshop was to update the group on the progress made during the working group’s first year of activities, and plan the next steps.

LandCover6k’s rationale is that adequate incorporation of land cover in global and regional climate models is not satisfactory, and the effect of anthropogenic activities on global climate via bio-geochemical and bio-geophysical processes in the past is not yet fully understood (pastglobalchanges.org/landcover6k).

Figure 1: Livelihood distribution map for western Africa, AD 600 (Kay et al., unpublished data).

An overview of modeled Holocene land-use changes was presented, based on a new version of the HYDE database (Klein Goldewijk et al. 2011) - updated with better data, irrigated and rain-fed cropland areas, and a distinction between managed pasture and extensive rangelands. The group reviewed patterns of HYDE, and regional comparisons were made with maps from historical geographers and archaeologists. One of the conclusions was that global allocation rules of HYDE do not perform well in the deep past due to a Eurocentric view of how people engaged in settlement and agriculture (expansion). Input from archaeologists and historians is clearly essential to this work and progress has already been made in southern and Central America, western Africa (see Fig. 1) and southern India. These regions are used as test cases of the land-use classification proposed by the LandCover6k working group (Morrison et al. 2016).

The group also reviewed progress made over the globe on regional pollen-based reconstructions of land cover) based on studies with the REVEALS model (Sugita 2007). REVEALS reconstructions for Europe except the Mediterranean and easternmost regions are already published (Trondman et al. 2015). REVEALS reconstructions will be available at the end of 2016 for parts of northern America, northern and temperate China, Siberia, and Cameroon.

Finally, current climate-modeling projects (e.g. within PMIP and CMIP) were reviewed and their needs in terms of past land-use and land-cover characteristics were discussed in the context of LandCover6k’s aims and goals.

Participants agreed that focus should be on testing the global scale land-use categorization established by the archaeologists involved in LandCover6k in southern America and southern India, and on the pollen-based reconstructions in regions where they are achievable within the relatively near future, i.e. northern America, southern and eastern Europe, Siberia, and India. Moreover, a case study is planned to incorporate empirical archaeological data from a single region into the HYDE system, in order to see whether this would produce a significant difference in modeled past agricultural patterns, and to evaluate whether that difference would significantly impact on climate change. It was also decided that Neotoma (www.neotomadb.org) will act as the major archive of all case studies from LandCover6k.

The LandCover6k working group will continue to be present at many international conferences, organizing sessions and presenting scientific results. Forthcoming events are the European Geosciences Union Assembly 2017, and PAGES’ 5th Open Science Meeting (OSM). The third general workshop of LandCover6k will be held in conjunction with the OSM on 16-17 May 2017 in Zaragoza, Spain. For more details, see: pastglobalchanges.org/calendar/127-pages/1637

affiliations

1Copernicus Institute of Sustainable Development, Utrecht University, The Netherlands
2Department of Biology and Environmental Science, Linnaeus University, Kalmar, Sweden
3Department of Anthropology, University of Pennsylvania and the Museum of Archaeology and Anthropology, Philadelphia, USA
4Department of Humanities and ICREA, Universitat Pompeu Fabra, Barcelona, Spain
5School of Geography, Earth and Environmental Sciences, Plymouth University, UK

contact

Kees Klein Goldewijk: c.g.m.kleingoldewijkuu.nl

references

Gaillard M-J et al. (2010) Clim Past 6: 483-499

Gaillard M-J et al. (2015) PAGES Mag 23: 81

Klein Goldewijk K et al. (2011) Glob Ecol Biogeogr 20: 73-86

Morrison K et al. (2016) PAGES Mag 24: 40

Sugita S (2007) Holocene 17: 229-241

Trondman A-K et al. (2015) Glob Chang Biol 21: 676-697

N’dji dit Jacques Dembele

Bamako, Mali, 26-30 September 2016

Since 2009, WAQUA (West African Quaternary Research Association) has been promoting Quaternary research in West Africa by organizing meetings for researchers involved in Quaternary science.

One of the main issues raised during these meetings is the absence of contributions from West African researchers to the understanding of paleoenvironmental conditions in the region, because of the lack of trained Quaternary scientists. To help address this issue, WAQUA organized a training session for 60 early-career scientists and students from the West Africa region. Attendees were mainly archaeologists, geographers, historians, and botanists, from Cote d’Ivoire, Benin, Mali, Niger and Nigeria.

The four invited speakers discussed Quaternary palynological methods (Chantal Kabonyi, Democratic Republic of Congo), analysis of palynological data (Monique Tossou, Benin), dendrology (Ntamwira Miranda, Democratic Republic of Congo) and quaternary sediments and geomorphological evidences (N’dji dit Jacques Dembele, Mali).

The workshop focused on seven topics that were addressed during the five days. Day one introduced the stratigraphic framework of the Quaternary and the geomorphological evidences of Quaternary paleoenvironments. These presentations were followed by debates and discussions.

Figure 1: Workshop participant Diarra Cheick Oumar sampling a lacustrine profile of the Niger River right bank, near Bamako, during the field excursion.

The second day was devoted to Quaternary sediments: the different types, their textures and structures; the methodology to analyze the textures and structures; the different environments of deposits and the methodology of their reconstruction; the provenance of the sediments using heavy-minerals analysis; and the use of Quaternary sediments to infer Quaternary paleoenvironmental change. While the morning was devoted to theory, participants spent the afternoon on a field trip to see Quaternary sediments of the Niger River at Bamako (Fig. 1).

Day three focused on Quaternary palynology. Methods were presented to collect, treat and extract pollen samples; to make pollen slides and analyze them using the microscope; to count and classify pollen using morphology and typology; and to build pollen diagrams. The afternoon was dedicated to practical courses. The participants learnt to extract and identify pollen, and build a pollen diagram. The used samples were from honey, as the preparation of sediment sample necessitates hazardous chemicals such as hydrofluoric acid.

The morning of the fourth day was dedicated to archaeology. Presenters discussed the types of archaeological sites in West Africa, the Paleolithic and Neolithic Periods of West Africa, and the archaeological excavation techniques in the West African context. The afternoon was spent discussing major issues concerning paleoenvironmental research in West Africa, such as the lack of quaternary science courses in West African universities; the fact that African researchers have to send their samples to Europe and Asia for analysis; the lack of communication between researchers from different regions; and the lack of funding.

The fifth and last day of the training workshop was used to define new research paths for possible paleoenvironmental research; to discuss publishing issues (language, high fees, and lack of paleo journals interested in West African case study results). The participants proposed to investigate the options for creating a new PAGES working group on paleoenvironmental dynamics of the Niger River during the Holocene. Also, following on from the day’s discussions, a group was mandated to assess the feasibility and scope for a new journal where Quaternary scientists from West Africa could publish their results more easily. It was also proposed that WAQUA should hold a workshop every year so that West African researchers can remain informed about new research methods and techniques.

In addition to the various daily agendas, a session titled “Meet the Experts” was held every day during the training workshop, to enabled early-career researchers and doctorate students to discuss their research issues with senior scientists. The participants felt inspired following this much-welcome training meeting and look forward to implementing their new skills and interact with the new colleagues they met.

acknowledgements

WAQUA members want to express their sincere gratitude to PAGES and the University of Social Sciences and Management of Bamako (USSGB) for their financial and material support, and for their encouragements.

affiliation

Faculty of History and Geography, University of Social Sciences and Management of Bamako, Mali

contact

N’dji dit Jacques Dembele: ndjijacquesicloud.com

Jing Li1 and Nico Caltabiano2

Qingdao, China, 18-25 September 2016

Figure 1: Plenary session during the CLIVAR Open Science Conference.

More than 600 scientists from 50 countries gathered in Qingdao, China, for the second CLIVAR Open Science Conference (OSC). The event was titled “Charting the course for climate and ocean research” and marked the 20th anniversary of CLIVAR. Over the course of five days, scientists showcased major advances in climate and ocean research. By design, more than one third of participants were early-career scientists, who presented their work through 234 posters and oral presentations, including plenary talks (Fig. 1).

Throughout the conference, the value of cross-disciplinary integration and international cooperation was emphasized. Critical issues addressed included the impacts of a warming ocean on future climate; regional variations in ocean and climate warming; the respective contributions of thermal expansion and melting ice to sea-level rise; the connections between the oceans and the global water and energy cycle; the changes taking place deep in the ocean; and the consequences of excess carbon on sea life. Different aspects concerning climate or proxy modeling that could bring about closer interactions between PAGES and CLIVAR were highlighted during the PAGES Town Hall at the CLIVAR OSC, with specific focus on: regional and high resolution modeling for paleoclimate studies; proxy modeling and assimilation of paleo observations in climate models; study on monsoon and changes in ENSO-monsoon teleconnections at different timescales, etc. Both parties could benefit from common model simulations, data syntheses and methodologies in future cooperation.

Closing a successful week of high-level talks and discussions, and new or strengthened collaborations, the final day of the conference was devoted to climate information for sustainable development and the future of climate and ocean science. Presenters and panelists concluded that CLIVAR of the future needs to help build a society resilient to environmental changes by expanding the understanding of uncertainty in climate risk and providing regional climate information and seamless predictions across timescales. All participants argued for a continued international CLIVAR coordination effort to improve our observational records (including paleo), models and process understanding, and our ability to communicate scientific discoveries translating knowledge into useful information. It was emphasized that further improvements in our understanding of climate processes are required, covering scales ranging from a centimeter to global and from hours to decades, and longer.

One of the important aims of the conference was to engage the future generation. Jointly with the OSC, CLIVAR successfully organized an Early Career Scientists Symposium (ECSS), hosted by China’s First Institute of Oceanography, State Oceanic Administration (SOA/FIO). More than 120 students and early-career scientists (some of them supported by PAGES) participated in the symposium. Participants enjoyed an informal atmosphere, while discussing in groups the key research challenges that the scientific climate community faces at the moment and highlighting the need for international science collaboration. They also discussed the best ways to engage potential stakeholders with their scientific information and suggested their vision for the future of CLIVAR.

The conference injected an amazing degree of enthusiasm about CLIVAR’s research and fortified the notion that CLIVAR is and will be critical to meet society’s needs for climate information. International cooperation of the type that CLIVAR fosters will continue to be indispensable to developing the human capacity and infrastructure that underpin all major scientific breakthroughs. The conference certainly helped in handing over the enthusiasm for CLIVAR and its science to the next generation, whose excitement is a promise for a very bright future.

The conference was hosted by the Qingdao National Laboratory for Marine Science and Technology - a new campus designed to be an open and innovative world-class marine research center. More details about the Open Science Conference can be found at www.clivar.org.

affiliations

1International CLIVAR Project Office, Qingdao, China

2International CLIVAR Project Office, Southampton, UK

contact

Nico Caltabiano: nico.caltabianoclivar.org

Emilie Capron1, A. Govin2 and E.J. Stone3

We compiled ice and marine records of high-latitude temperature changes and placed them on a common timescale. We also produced climatic time slices for 115, 120, 125, and 130 ka. They represent improved benchmarks to perform Last Interglacial model-data comparisons.

The Last Interglacial (LIG, ~129-116 thousand of years BP, ka) represents a test bed for climate model feedbacks under warmer-than-present conditions. A spatio-temporal picture of LIG temperature evolution is indispensable to perform robust model-data comparisons and examine the General Circulation Models (GCMs) used for future climate projections (e.g. Lunt et al. 2013; Bakker and Renssen 2014).

However, existing LIG data syntheses (e.g. Turney and Jones 2010; McKay et al. 2011) consist of a single time slice and introduced dating uncertainties of up to several thousands of years by using the paleoclimatic records on their original timescales. Nevertheless, there is evidence that LIG surface temperatures peaked asynchronously around the globe (e.g. Bauch and Erlenkeuser 2008; Govin et al. 2012).

Here, we overcome the difficulty of aligning temperature records from different paleoclimatic archives retrieved across the world by limiting a new data synthesis to the high-latitudes. Our compilation provides a dynamic representation of the LIG temperature evolution, allowing more robust model-data comparisons.

A new LIG data synthesis associated with a coherent temporal framework

We combined 47 surface air and sea surface temperature records across the LIG polewards of 40°N and 40°S, respectively (Capron et al. 2014). Surface air temperature records are deduced from water stable isotopic profiles of ice cores. Sea surface temperatures (SST) are reconstructed from foraminiferal Mg/Ca ratios, alkenone unsaturation ratios, and faunal assemblages in marine sediment cores.

We use as a reference chronology for both marine and ice records, the recent Antarctic ice core chronology AICC2012. It is the first integrated timescale over the LIG, based on a multi-site approach including both Greenland (NGRIP) and Antarctic ice cores (EDC, EDML, TALDICE, Vostok). The numerous new stratigraphic links significantly reduce the absolute dating uncertainty down to ±1.6 ka (1σ) during the studied time interval (Bazin et al. 2013) making it a particularly well-constrained age scale.

In a first step we have transferred the Dome F and NEEM ice cores onto AICC2012 since they were not initially included in the construction of AICC2012. In a second step, to align marine records onto AICC2012, we follow the strategy of Govin et al. (2012). It is based on the assumption that surface-water temperature changes in the sub-Antarctic zone of the Southern Ocean (respectively in the North Atlantic) occurred simultaneously with air temperature variations above Antarctica (respectively Greenland).

Figure 1: Temperature time slices across the Last Interglacial. NH air and sea surface temperature anomalies (A) associated with 2σ uncertainties (B). (C) and (D) are the same as (A) and (B) for the SH. The size of the dots (both for temperature anomaly and associated 2σ uncertainty) follows the temperature scale given in the box. For all panels, warming (cooling) compared with modern temperature is represented in red (purple). The temperature anomalies are relative to the World Ocean Atlas (WOA) 1998, 10 m-depth data for marine records and relative to present-day instrumental surface air temperature measurements for ice records.

The aligned time series have already been used for comparison with transient model simulations (as in Loutre et al. 2014). In addition, we constructed four time slices to assist snapshot simulation assessment. We calculated temperature anomalies relative to present-day conditions over 2 ka time windows centred on 115, 120, 125, and 130 ka.

The aligned surface temperature time series and the time slice reconstructions benefit from quantitative uncertainty estimates. We used a Monte-Carlo analysis, taking into account temperature reconstruction errors as well as the propagation of dating uncertainties (see Capron et al. 2014 for details). This results in a final uncertainty estimate of ± 2.6°C on average for SST records and ± 1.5°C for surface air temperature Antarctic records (Fig. 1).

LIG climatic features

Figure 2: Model-data comparison of time slice reconstructions for 130 ka (A and B) and 125 ka (C and D) with respective snapshot simulations using the CCSM3 and HadCM3 models. Summer SST temperature anomalies from the marine sediment data (dots) superimposed onto model July-August-September SSTs in the NH and model January-February-March SSTs in the Southern Ocean; Annual mean surface air temperature anomalies from ice core data (dots) superimposed onto the model annual surface air temperature.

Producing robust age models and SST reconstructions remains difficult in the Nordic Seas. Thus, we focus on the robust climatic patterns found in the North Atlantic and Southern Ocean. The time slices (Fig. 2) capture the major features characterizing the spatial sequence of events described in the time series (Capron et al. 2014). In particular, the comparison of the 130 ka time slices from the two polar regions illustrates that warming in the Southern Hemisphere (SH) preceded warming of the North Atlantic. This interhemispheric asynchrony, which has already previously been reported (e.g. Govin et al. 2012), is attributed to the “bipolar seesaw” mechanism, induced by changes in the intensity of the Atlantic Meridional Overturning Circulation (AMOC; Stocker and Johnsen 2003). The melting of northern ice sheets extended beyond the penultimate deglaciation into the early LIG and the associated release of meltwater into the North Atlantic was suggested to have delayed the full establishment of a vigorous AMOC, resulting in peak Antarctic temperatures while the North Atlantic was still cold (Govin et al. 2012).

The 130 ka time slice also reveals SSTs significantly cooler-than-present-day conditions, e.g. up to 7.5 ± 3°C cooler in the northern high latitudes, while temperatures were slightly warmer than present-day (1.7 ± 2.5°C on average) in most of the SH sites. Warmer-than-present-day climatic conditions are clearly visible in the 130, 125, and 120 ka time slices in the SH, while they are only observed locally and with a relatively high uncertainty in the 125 and 120 ka time slices of the Northern Hemisphere (NH). Warmer-than-present-day conditions hence lasted longer at southern high latitudes than at northern ones. Finally, the magnitude of temperature changes is larger at northern than at southern high latitudes during the LIG onset and demise.

Toward more robust model-data comparisons

To illustrate the potential of our new LIG data synthesis for model-data comparisons, we compared 125 and 130 ka snapshot simulations performed with two GCMs, CCSM3 (Otto-Bliesner et al. 2013; Collins et al. 2006), and HadCM3 (Gordon et al. 2000), to the corresponding data-based time slices (Fig. 2; Capron et al. 2014).

We observe that both models predict warmer-than-present-day North Atlantic conditions at 130 ka, i.e. earlier than in our data compilation. Furthermore, neither of the models reproduce the reconstructed early SH warming (Fig. 2A,B). This means that the 130 ka simulations do not reproduce the bipolar seesaw pattern identified from the data synthesis. This is possibly because they only simulate the climate response to the static orbital and greenhouse gas forcing at 130 ka without taking into account potential dynamic AMOC responses to persistent NH ice sheet melting. Indeed, simulations where freshwater forcing to the North Atlantic from the Laurentide and Eurasian ice sheets is introduced perform better in reproducing the late NH warming (Govin et al. 2012). Also both models run under modern polar ice sheet and vegetation configurations, not under 130 ka ones. Other simulations considering an additional feedback linked to the disintegration of the West Antarctic ice sheet resulted in an additional warming over Antarctica (Holden et al. 2010; Otto-Bliesner et al. 2013).

At 125 ka, both models produce warmer-than-present-day conditions in the North Atlantic region (Fig. 2C,D). In Antarctica, CCSM3 suggests a cooler climate at 125 ka compared to present-day while ice core data suggest warmer conditions (Fig. 2C). The observed dissimilarities between CCSM3 and HadCM3 simulations are likely related to their different sea ice sensitivities (Otto-Bliesner et al. 2013).

Summary and outlook

We produced the first data synthesis that documents the spatio-temporal evolution of high-latitude temperatures from the LIG onset to its demise, using one single reference chronology. We paid careful attention to estimate quantitatively the temperature errors, including the propagation of dating uncertainties. Our study reveals asynchronous temperature changes between the polar regions and highlights the importance of considering the LIG sequence of events rather than averaging climate conditions over the entire LIG period.

This work should encourage more in-depth model-data comparison exercises with both snapshots and transient model simulations. Future work should not only consider surface temperature, but include additional climatic parameters (e.g. deep ocean circulation changes, sea ice extent) to move towards a more complete picture of LIG climatic and environmental changes.

acknowledgements

This work has also received funding from the UK-NERC consortium iGlass (NE/I009906/1).

affiliations

1British Antarctic Survey, Cambridge, UK

2MARUM/Center for Marine Environmental Sciences, University of Bremen, Germany

3School of Geographical Sciences, University of Bristol, UK

contact

Emilie Capron: ecapbas.ac.uk

references

Bakker P, Renssen H (2014) Clim Past 10: 1633-1644

Bauch HA, Erlenkeuser H (2008) Polar Res 27: 135-151

Bazin L et al. (2013) Clim Past 9: 1715-1731

Capron E et al. (2014) Quat Sci Rev 103: 116-133

Collins WD et al. (2006) J Clim 19: 2122-2143

Gordon C et al. (2000) Clim Dyn 16: 147-168

Govin A et al. (2012) Clim Past 8: 483-507

Holden PB et al. (2010) Clim Past 6: 431-443

Loutre MF et al. (2014) Clim Past 10: 1541-1565

Lunt DJ et al. (2013) Clim Past 9: 699-717

McKay NP et al. (2011) Geophys Res Lett 38, doi:10.1029/2011GL048280

Otto-Bliesner B et al. (2013) Phil Trans R Soc London, Ser A, doi: 10.1098/rsta.2013.0097

Stocker TF, Johnsen SJ (2003) Paleoceanography 18, doi:10.1029/2003PA000920

Turney CSM, Jones RT (2010) J Quat Sci 25(6): 839–843

Veres D et al. (2013) Clim Past 9: 1733-1748

Laura Rodríguez-Sanz1 and P. Graham Mortyn2,3

During the last glacial termination, atmospheric and oceanic teleconnections promoted millennial-scale episodes of widespread salinification in the tropical and extratropical Pacific Ocean that coincided with the Younger Dryas and Heinrich Stadial 1 cold phases in the North Atlantic.

D.M. Roche1,2, H. Renssen1, C. Morrill3,4, H. Goosse5 and A. Mairesse5

A suite of model simulations performed within the Past4Future project have investigated the effect of freshwater fluxes and volcanic eruptions on global climate during interglacial climate conditions. Results for the Holocene evidence impacts that persist for centuries to millennia.

Reconstructions of the past climate unambiguously show that climate can shift abruptly on decadal timescales. Evidence of abrupt transitions comes from both the present “Holocene” Interglacial and the Last “Eemian” Interglacial (e.g. Alley et al. 1997; Pol et al. 2014). However, there is still considerable uncertainty concerning the role of the different climate forcings in causing abrupt shifts. Constraining the sensitivity of interglacial climates to each forcing is a pre-requisite for assessing the future risk of abrupt changes and the potential impacts on societies.

To better understand our capacity to reproduce the reconstructed past climate changes, the Past4Future project evaluated the results of large coordinated modeling experiments, including those that exposed interglacial climate to freshwater and volcanic forcing perturbations. We report here on the results regarding (i) the well-known 8.2 kyr cold event and (ii) the effect of volcanic forcing in Holocene transient simulations.

Freshwater forcing of the 8.2 kyr cold spell

The 8.2 kyr event is an abrupt cooling that lasted ~150 years and is clearly identified in water isotopic records from Greenland ice-core records (e.g. Johnsen et al. 1992). It is generally associated in models with a temporary weakening of the Atlantic Meridional Overturning Circulation (AMOC) as a consequence of freshwater forcing pulse; there are data evidences that this is also the case in reality (Kleiven et al. 2008). Though some uncertainties exist on the precise amount, duration, and sequence of the freshwater pulse(s) delivered to the North Atlantic at this time, the cause is generally regarded to be the final drainage of giant North American proglacial lakes that stored deglacial meltwater from the retreating Laurentide ice-sheet (Barber et al. 1999). The largest freshwater contribution into the North Atlantic Ocean likely came from proglacial Lake Agassiz-Ojibway.

The 8.2 kyr event is an easy target for model intercomparison studies because (i) of its short duration, (ii) it occurred at the very end of the deglaciation in a climate relatively similar to our current one, and (iii) of the relatively well-known characteristics of the freshwater pulse. As such, it is accessible for the most recent complex coupled climate models. Additionally, this event is of particular interest since previous modeling studies reported that the resulting oceanic circulation decrease is of comparable magnitude to that obtained in future climate projections by the end of the century (Cheng et al. 2013; LeGrande et al. 2006).

Morrill et al. (2013) analyzed the results of four simulations from three climate models obtained using the 8.2 kyr event simulation scenario proposed within Phase 3 of the Paleoclimate Modelling Intercomparison Project (PMIP3). The three models analyzed are of various complexities and resolution (see Table 1). They show reasonable skill in reproducing the present-day climate, and of particular interest here, the strength of the oceanic circulation. There are, however, notable differences between the models in the regions where deep oceanic waters are formed.

All studied models simulate an AMOC weakening following the imposed forcing anomaly where freshwater was added for one year at a rate of 2.5 Sv (1 Sv = 106 m3 s-1). Generally, the AMOC reduction is a consequence of surface ocean freshening and sea-ice expansion in the North Atlantic. However, the three models show differences in the duration and amount of the AMOC decrease. Two of them show a reduction of 10% of the AMOC and a return to pre-freshwater forcing values within a few decades. However, the third model shows a more substantial 25% AMOC reduction, with a longer lasting response of a century.

Figure 1: Comparison of mean annual temperature anomalies for the 8.2 kyr event from a multi-model ensemble (top) and proxy evidence (bottom). Stippling shows grid cells where at least three of four simulations agree on the sign of the temperature anomaly. Quantitative estimates of temperature anomalies from proxies are shown in degrees Celsius.

As a consequence, surface temperature changes are observed in all models (Fig. 1a). Overall, a significant cooling is observed in the Northern Hemisphere, with a reduction in mean annual temperature of less than 0.5°C over most continents and around 1°C over some regions of the North Atlantic. Models also show a consistent pattern of warming of about 0.3°C or less in parts of the Southern Ocean. These features are consistent with the spatial distribution of temperature anomalies from proxy data (Fig. 1b). However, the models generally underestimate the amount of cooling by a factor of two and obtain only a shorter temperature response of just a few decades instead of the observed 150 years.

Analysis of model simulation differences shows that other factors may also influence the model result and account for some of the data-model discrepancies (Morrill et al. 2013). For example, the models may be sensitive to the presence or absence of deep oceanic convection in the Labrador Sea and to the length, amplitude, and other details in the freshwater forcing scenario.

Volcanic forcing of Holocene climate variability

Large volcanic eruptions are, in addition to freshwater releases, another well-known driver of abrupt climate variations. During explosive eruptions, sulphate can be injected into the stratosphere. This yields an enhanced absorption of solar and surface radiation and provokes a net stratospheric heating that in turn results in a net surface cooling. This effect was clearly observed following the Pinatubo eruption in 1991, which cooled lower troposphere temperatures globally by approximately 0.5°C (Dutton and Christy 2012). This cool anomaly lasted up to several years, linked to the lifetime of sulphate aerosols in the stratosphere.

Analysis of historical archives and sulphate deposited on ice-sheets enables us to reconstruct the history of large explosive volcanic eruptions. Volcanic forcing scenarios covering the last 10 kyr were developed to be utilized in climate models, expressed here as radiative impact (Total Solar Irradiance, TSI). The Mt. Tambora eruption in 1815, the largest of the last millennium, had an estimated TSI equivalent impact of 12 W m-2 and large eruptions in the Holocene may have had a magnitude many times greater than that, although the precise number is still uncertain (Mairesse 2014).

Figure 2: Global annual mean temperature difference between the simulation with and without volcanism. Gray curve is the unfiltered simulation. Also shown are temperature filtered by a running mean of 101 (red), 251 (blue), 501 (green) and 1001 (pink) years. Note that the vertical scale does not show the whole range of extreme values. The strongest annual global cooling was 6°C (in 6500 BC).

Using a newly developed volcanic forcing scenario, Mairesse (2014) performed transient fully coupled climate simulations with the LOVECLIM model over the last 10 kyr, with and without including the radiative effect of volcanoes. This enables the impact of large eruptions on Earth's global mean temperature to be analyzed (Fig. 2). The results show a significant (> 0.1 °C) imprint of volcanic eruptions on the simulated global climate. Medium-sized eruptions (injections of ~50 Tg of sulfate) result in recurrent coolings of 0.1 to 0.2°C at the centennial scale. The maximum effect for individual years is much larger but the exact magnitude of the response is still uncertain as the radiative impact of very large eruptions does probably not scale linearly with the sulphate emissions. Without further information, it remains difficult for now to reach a consensus on how to improve the forcing.

Outlook

The reported model experiments suggest that both volcanism and freshwater forcing have a significant impact at the century scale on the Earth's climate during the Holocene. There is a need, however, to refine the relationship between the TSI forcing and the sulfate injections for very large eruptions, since there is no analogue in the instrumental era.

The feedback from human activities on the melting of the Greenland ice-sheet may yield increased freshwater fluxes to the ocean over the course of this century. The strength of the fluxes can in principle be influenced by mitigation measures. While the rate and occurrence of disruptive volcanic eruptions is beyond human control, a risk assessment of their magnitudes and impacts may be useful for planning disaster prevention.

affiliations

1Department of Earth Sciences, VU University of Amsterdam, The Netherlands

2Laboratoire des Sciences du Climat et de l’Environnement (LSCE), Gif-sur-Yvette Cedex, France

3Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, USA

4NOAA’s National Climatic Data Center, Boulder, USA

5Georges Lemaître Centre for Earth and Climate Research, Université catholique de Louvain, Louvain-la-Neuve, Belgium

contact

Didier Roche: didier.rochelsce.ipsl.fr

references

Alley RB et al. (1997) Geology 25: 483-486

Barber DC et al. (1999) Nature 400: 344-348

Cheng W et al. (2013) J Clim 26: 7187-7197

Dutton EG, Christy JR (2012) Geophys Res Lett 19: 2313-2316

Johnsen SJ et al. (1992) Nature 359: 311–313

Kleiven HKF et al. (2008) Science 319: 60-64

LeGrande AN et al. (2006) PNAS 103: 837-842

Mairesse A (2014) Analysis of the Holocene climate variability using a data assimilation method applied to the model LOVECLIM, PhD Thesis

Morrill et al. (2013) Clim Past 9: 955-968

Pol K et al. (2014) Geophys Res Lett 41: 4004–4012

Hubertus Fischer1, J. Schmitt1, S. Eggleston1, R. Schneider1, J. Elsig1, F. Joos1, M. Leuenberger1, T.F. Stocker1, P. Köhler2, V. Brovkin3 and J. Chappellaz4

High-precision ice core data on both atmospheric CO2 concentrations and their carbon isotopic composition (δ13Catm) provide improved constraints on the marine and terrestrial processes responsible for carbon cycle changes during the last two interglacials and the preceding glacial/interglacial transitions.

Natalie Kehrwald1, P. Zennaro1,2, S. Schüpbach3, T. Kirchgeorg1, J. R. McConnell4, R. Zangrando2, A. Gambaro1, and C. Barbante1,2,5

New data from Greenland ice cores reveal a major peak in boreal biomass burning during the 1600s AD, presumably related to major regional droughts in Central Asia. This climate-related peak in fire activity is greater even than post-industrial biomass burning.

Fires ignited by humans to manage the landscape may have started altering atmospheric greenhouse gas concentrations and warming the planet thousands of years ago (Ruddiman 2003). Today, biomass burning produces up to 50% as much carbon dioxide as fossil fuel burning (Bowman et al., 2009) while at the same time shrinking an important carbon sink.

Polar ice cores present a unique medium to quantitatively investigate past biomass burning at the global scale because they trap and archive fire-related aerosols before and after humans began influencing the global carbon cycle.

Figure 1: Distribution of the high northern latitude land cover and location of the NEEM ice core drilling site at 77.49°N, 51.2°W, 2480 m above sea level. Modified from the European Commission Global Land Cover 2000 database and based on the work of cartographer Hugo Alhenius UNEP/GRIP-Arendal.

Here, we synthesize information from three geochemical fire activity proxies, i.e. levoglucosan, black carbon, and ammonium measured in the NEEM and NEEM-2011-S1 ice cores from Greenland (Fig. 1) to investigate interactions between Northern Hemisphere fire activity, climate, and possible human influences in boreal regions over the last 2000 years.

Fire proxies in ice cores

Several tracers of past biomass burning can be measured in polar ice cores. Aerosols are transported in smoke plumes from Eurasia and North America over thousands of kilometers. Some of the aerosols are transported all the way to Greenland, deposited there and trapped within the ice sheet (Kehrwald et al. 2012; Zennaro et al. 2014). Ammonium is a tracer of biomass burning, especially in boreal regions (e.g. Legrand et al. 1992), but also enters the atmosphere through interactions with fertilizers, manure, and marine sources. Black carbon is a more specific proxy as it is produced by the incomplete combustion of both biomass and fossil fuels (McConnell et al. 2007). Both black carbon and ammonium can be determined using high-resolution continuous flow analysis techniques (CFA; Zennaro et al. 2014). Fire information is also inferred from the isotopic composition of methane (CH4). The δ13C of CH4 is enriched if it was produced by a pyrogenic process compared to biogenic sources including rice cultivation, ruminants or wetlands (Sapart et al., 2012) Note, however, that due to methane’s atmospheric lifetime of several decades, the δ13C of CH4 is a global mixture from sources with different isotopic signatures (Sapart et al. 2012). Finally, levoglucosan, a monosaccharide anhydride, has strong potential for tracing broad-scale fire activity. It is only produced by burning cellulose at temperatures of 300°C or higher (Simoneit 2002) and represents one of the principal constituents of smoke plumes. Levoglucosan is currently only determined in discrete samples, resulting in lower-resolution records than parameters analyzed by CFA techniques.

Our study acknowledges both the strength and the weaknesses inherent in most ice core fire proxies and remedies the latter by integrating the results from ammonium, black carbon, and levoglucosan records from the NEEM ice cores over the past 2000 years (Zennaro et al. 2014).

Boreal biomass burning and climate

Figure 2: Climatic and environmental records over the past 2000 years, presented as z-scores smoothed by incorporating the nearest 10% of data. (A) NEEM levoglucosan concentrations; (B) Boreal charcoal concentrations (Marlon et al. 2008); (C) NEEM black carbon versus levoglucosan concentrations; (D) Siberian forest fire reconstruction (Eichler et al. 2011); (E) Northern Hemisphere land temperature (Mann et al. 2008); (F) NEEM pyrogenic CH4 emissions (Sapart et al. 2012). Dashed blue lines in panels B-F are the levoglucosan record from A for reference. The yellow vertical bar marks the levoglucosan-based 17th century fire activity maximum highlighted in the text.

Fires are individual events lasting hours to weeks, yet droughts, increased temperatures, or land use changes can increase net fire activity over years to centuries. We compare our records with other biomass burning and climate reconstitutions from elsewhere to locate the geographical extent of the fire activity reconstructed from the NEEM ice cores and to investigate their driving factors. To all of the records we applied a locally-weighted scatterplot smoothing (LOWESS, details on the statistical method in Zennaro et al. 2014) and normalized all data as z-scores to ease the comparison between the different records (Fig. 2).

The NEEM levoglucosan record peaks over a century-long period, centered on 1640 AD (Fig. 2). Although less prominent, high values around that time also exist in the NEEM black carbon records (Fig. 2). A regional average record of northern high latitude (>55°N) charcoal data from the Global Charcoal Database of the Global Palaeofire Working Group (www.gpwg.org) shows similar trends to the levoglucosan record throughout most of the last two thousand years, with the interesting exception of low charcoal values during the 1640 AD levoglucosan peak (Fig. 2). However, the majority of the boreal charcoal data are located in North America, thus geographically weighting the resulting synopsis. This could explain the observed difference between the levoglucosan and hemispheric charcoal record syntheses during this time period.

Tree-ring, speleothem, and alpine ice core data (e.g. from the Belukha ice core in Mongolia) demonstrate widespread, decadal-scale droughts over most of Central Asia coincident with major fire peaks in the NEEM ice core (Zennaro et al. 2014). These Central Asian droughts, and associated fire activity peaks, do not correspond with either increased regional or Northern Hemisphere land temperatures (Fig. 2). Therefore, over decadal timescales, we propose that precipitation changes may be an important fire-controlling factor. Except during the 1640 AD fire peak, the resemblance observed between the NEEM levoglucosan, black carbon, and Northern Hemisphere temperature reconstructions suggests that temperature might be the major control of boreal biomass burning over centennial timescales.

Conclusions

Our multi-proxy (levoglucosan, black carbon, and ammonium) reconstruction of fire activity based on the NEEM ice cores suggests that temperature was the main control on boreal fire activity over the past two thousand years, while major droughts influenced biomass burning over decadal timescales. Our dataset suggests a peak in fire activity in the mid-1600s, which coincides with the most severe Central Asian droughts of the past two millennia. This fire activity peak is higher than any bio­mass burning recorded at NEEM after the Industrial Revolution, when land-clearing rates were the highest in recorded history. Therefore, this dominance of the 1640 AD fire peak suggests that climate affected boreal biomass burning more than human activity over the past two thousand years.

acknowledgements

The research leading to these results received funding from the European Union’s Seventh Framework programme (FP7/2007-2013) under grant agreement no. 243908, grant agreement no. 267696, and the U.S. National Science Foundation award no. 0909541.

affiliations

1Department of Environmental Science, Informatics and Statistics, Cà Foscari University of Venice, Italy

2IDPA-CNR, Venice, Italy

3Climate and Environmental Physics, Physics Institute and Oeschger Centre for Climate Change Research, University of Bern, Switzerland

4Desert Research Institute, Department of Hydrologic Sciences, Reno, USA

5Centro B. Segre, Accademia Nazionale dei Lincei, Rome, Italy

contact

Natalie Kehrwald: Kehrwaldunive.it

references

Bowman et al. (2009) Science 324: 481-484

Eichler A et al. (2011) Quat Sci Rev 30: 1027–1034

Holmes BJ, Petrucci GA (2007) J Atmos Chem 58: 151-166

Kehrwald N et al. (2012) Tellus B 64: 18196

Kirchgeorg T et al. (2014) Org Geochem 71: 1-6

Legrand M et al. (1992) Geophys Res Lett 19: 473-475

Mann et al. (2008) PNAS 105: 13252-13257

Marlon et al. (2008) Nat Geosci 1: 697–702

McConnell JR et al. (2007) Science 317: 1381-1384

Ruddiman WF (2003) Clim Change 61: 261-293

Sapart C et al. (2012) Nature 490: 85-88

Savarino J, Legrand M (1998) J Geophys Res Atmos 103: 8267–8279

Simoneit BRT (2002) Appl Geochem 17: 129-162

Zennaro P et al. (2014) Clim Past 10: 1905-1924