PAGES - Past Global Changes Magazine

PAGES - Past Global Changes Magazine

Material Information

PAGES - Past Global Changes Magazine
Series Title:
Alternate Title:
formerly PAGES news
International Geosphere-Biosphere Programme
International Geosphere-Biosphere Programme
Publication Date:


Subjects / Keywords:
Education ( local )
Environment ( local )
serial ( sobekcm )


General Note:
PAGES (Past Global Changes) supports research aimed at understanding the Earth's past environment in order to make predictions for the future. We encourage international and interdisciplinary collaborations and seek to promote the involvement of scientists from developing countries in the global paleo-community discourse. PAGES scope of interest includes the physical climate system, biogeochemical cycles, ecosystem processes, biodiversity, and human dimensions, on different time scales - Pleistocene, Holocene, last millennium and the recent past. Founded in 1991, PAGES is a core project of the International Geosphere-Biosphere Programme (IGBP) and is funded by the U.S. and Swiss National Science Foundations, and the National Oceanic and Atmospheric Administration (NOAA). It is overseen by a Scientific Steering Committee (SSC) comprised of members chosen to be representative of the major techniques, disciplines and geographic regions that contribute to paleoscience.
Open Access
Original Version:
Vol. 21, no. 2 (2013)
General Note:
See Extended description for more information.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
K26-03240 ( USFLDC DOI )
k26.3240 ( USFLDC Handle )
17163 ( karstportal - original NodeID )

USFLDC Membership

Added automatically
Karst Information Portal

Postcard Information



This item has the following downloads:

Full Text

PAGE 1 El Nio-Southern Oscillation Vol 21 No 2 August 2013 Editors: Pascale Braconnot, Chris Brierley, Sandy P. Harrison, Lucien von Gunten and Thorsten Kiefer ne w s Beyond just beautiful Tridacna spp. are faithfully recording paleoENSO variability. These giant, long-lived bivalves are reef dwelling organisms characteristic of the Indo-Pacic region. The presence of symbionts provides the multitude of colors commonly observed in Tridacna spp Mini section on data assimilation p. 72-79


46 Announcements Inside PAGES Celebrating 20 years In this issue we celebrate 20 years of PAGES news. You can read our paleobibliographic analysis on the next page, and take a trip back in time and enjoy the rst ever copy of PAGES news from 1993. A reprint is includ ed in this issue as an insert (it was much smaller then). Communications update PAGES has now joined the Twitter-sphere. Follow us to get all the latest news: @PAGES_IPO. You can also subscribe to our new YouTube channel: Past Global Changes, and of course, weve still got the Facebook fan page: PAGES Past Global Changes. Sta update Welcome to our two new sta mem bers: Nicole Wegmller (Finance and Oce Manager) and Leonie Goodwin (Communications and Project Ocer). PAGES OSM and YSM The Open Science and Young Scientists Meetings continue to resonate. We re cently uploaded videos of seven plenary talks with the accompanying PowerPoint slides and some short video montages to our YouTube channel: Past Global Changes. You can also read reports on the YSM activi ties written by the participants, starting on page 89. PAGES umbrella programs The transformation of the landscape of Global Environmental Change programs is taking shape. The new Future Earth program has now established a Scientic Committee and an Interim Secretariat with an Interim Director. You can check the names of the personnel on the Future Earth website ( and keep informed via their media, including a newsletter and a blog. Our current um brella program IGBP has decided to sunset by the end of 2015. The PAGES SSC has de cided that over the next two years PAGES should join the Future Earth network, while at the same time continuing to collaborate with other organizations. IGBP Scientic Committee in Bern In April, PAGES, together with ProClim and the Oeschger Centre, hosted the IGBP Scientic Committee Meeting in Bern. IGBP SC members came together from around the globe to discuss potential synthesis projects, IGBPs legacy and how best to transition into the new integrated Future Earth super-program in the coming years. Support for meetings During its meeting in June, the PAGES SSC granted support for a total of ten scientic and educational meetings. The next dead line for applying for PAGES meeting sup port is 20 September 2013. Support can be sought for workshop-style meetings relevant to PAGES Foci and Cross-Cutting Themes. The three eligible categories are PAGES Working Group meetings, work shops with a training or education focus, and an open call for other workshops that are relevant to PAGES science. Application guidelines and forms can be found on the PAGES website > My PAGES > Meeting Support. Guest scientists We are pleased to welcome two guest scientists to the PAGES oce this sum mer: Gisela Winckler from the LamontDoherty Earth Observatory, US and Bernd Zolitschka from the University of Bremen, Germany. Among other things, they are working alongside PAGES sta to edit up coming editions of PAGES news. You can learn more about our Guest Scientists and the work they are doing on the PAGES web site > People > Guest Scientists. Introducing the new mini section We hope you enjoy our rst Science Highlights mini section; in this issue it fo cuses on Data Assimilation. The new mini section format will feature 4-5 articles focusing on a specic topic, and might appear more regularly in future issues de pending on demand. Upcoming newsletters The next two issues of PAGES news will fo cus on dust and on annual recorders of the past. While the dust issue is already closed, suitable articles for the annual record ers issue are still welcome. Contributions should explore the question of how natu ral archives with annual resolution are ap proaching the temporal resolution of in strumental records. Submissions should be discussed with Bernd Zolitschka ( and be submitted before the 15 th of September. As always, you are invit ed to submit Science Highlights, Program News, and Workshop Reports for the Open Section of PAGES news. Author guidelines can be found on the PAGES website > My PAGES > Newsletter. New on the PAGES Bookshelf Tracking Environmental Change Using Lake Sediments, Volume 5 Data Handling and Numerical Techniques Editors: H. John B. Birks, Andr F. Lotter, Steve Juggins, John P. Smol Springer, 2012 Another volume in the Development in Paleoenvironmental Research Series, this book is the rst of its type to cover the full range of modern data-analytical and statistical techniques used in paleolimnology and paleoecology. It features numerical and statistical techniques, such as exploratory data analysis, error estimation, clustering, ordination and modern statistical learning techniques. It also includes case studies on human impact, lake development and climate change.


47 47 Annoucements 20 years of PAGES news T he four-page insert included in this issue is a copy of the rst ever PAGES news created 20 years ago in Spring 1993. We thought this milestone provided an opportunity to reect on how far PAGES news has come in that time, but also how many of the issues discussed in our rst newsletter still remain pertinent today. A quantitative paleobibliographic analysis reveals the steady growth of PAGES news in the last 20 years. We found evidence of ve distinct stylistic eras and observed that the number of pages in each issue has increased 12-fold from four in 1993 to around 50 in current issues (Fig. 1). Likewise there is a strong correlation between the number of pages and the number of science highlights articles (r 2 = 0.92; P<0.01), with early issues usually con taining one article and more recent issues 15. If this trend continues over the next 20 years, PAGES news will be a daunting 100 pages thick and feature 35 articles by the year 2033! Hans Oeschger, the rst chair of PAGES, observed in his introduction to the rst issue that global change research is advancing at a remarkable pace, and many of the issues he highlighted back in 1993 are still relevant today. An improved un derstanding of past global change is cru cial for evaluating present environmental conditions and for creating predictive cli matic models. Paleoclimatic and paleoen vironmental data have the unique ability to provide detailed insights into ecosys tem responses to climate change at dier ent time scales, which can inform future policymaking. We hope you enjoy a little trip back in time via our rst issue. And for anyone thinking about replying to the Call for Title Suggestions advertisement on the last page: Although we have happily stuck to the unimaginative title PAGES news for 20 years, we are, of course, still open to your suggestions for improvement. PAGES Calendar Paleore data synthesis using R 02 06 Oct 2013 Besanon, France Holocene Circum-Arctic Peatland Carbon Dynamics 12 16 Oct 2013 Bethlehem, USA PALSEA 2013 Workshop 21 24 Oct 2013 Rome, Italy Joint PAGES 2k & PAST2k-PMIP Workshop 04 06 Nov 2013 Madrid, Spain Ramsar Wetlands: Ecological Character 06 08 Nov 2013 Queenscli, Australia PMIP Ocean Workshop 2013 04 06 Dec 2013 Corvallis, USA Age Models, Chronologies, and Databases 13 16 Jan 2014 Belfast, Northern Ireland PAGES Focus 4 workshop 03 07 Feb 2014 Leuven, Belgium Figure 1: Paleobibliographic analysis of PAGES news.


48 PAGES news Vol 21 No 2 August 2013 Editorial: El Nio-Southern Oscillation observations and modeling PASCALE BRACONNOT 1 C. BRIERLEY 2 AND S.P. HARRISON 3,4 1 Laboratoire des Sciences du Climat et de lEnvironnement, unit mixte CEA-CNRS-UVSQ, Saclay, France; 2 Department of Geography, University College London, UK; 3 Department of Biological Sciences, Macquarie University, North Ryde, Australia; 4 Centre for Past Climate Change and School of Archaeology, Geography and Environmental Sciences, University of Reading, UK T he El Nio-Southern Oscillation (ENSO) is one of the major climate phenomena affecting our global en vironment (Figs. 1 and 2), but current understanding of ENSO is limited. Instrumental records are too short to allow us to document its spectrum of variability and there is little knowledge of how variability alters with changes in the climate mean state, for example with anticipated global warming. In the PAGES news issue Paired perspectives on global climate change (2012), Amy Clement highlighted some important issues about the decadal variability, predictability and modeling of ENSO with global warming. As point ed out by Julien Emile-Geay in the same issue, paleoclimate data are a valuable resource to improve our understand ing of these issues, even though the past does not provide direct analogs for the future. Analyses of high-reso lution paleoclimate indicators and the possibility of long simulations with the same climate models used for future climate projections offer new opportu nities for improving our understanding of ENSO. One of the foci of the work ing group on climate variability in the 3 rd phase of the Paleoclimate Modeling Intercomparison Project (PMIP3; Braconnot et al. 2012a) is to foster the synthesis of high-resolution data and the development of new data-model comparison methodologies to exam ine changes in ENSO. The future of ENSO research The suite of papers presented in this newsletter is one outcome of the PMIP workshop on Tropical climate variability with a focus on last mil lennium, mid-Holocene and Last Glacial Maximum which took place in the south of France in September 2011 and was co-sponsored by PAGES (Braconnot et al. 2012b). The articles provide an overview of high-resolution records and modeling studies that can be regarded as a basis for future ENSO research. Individual contributions address how one can extract ENSOrelevant climate information from tree rings, corals, giant bivalves, or shells, and how modeling can be used to test explicit hypotheses to improve the in terpretation of paleoclimate records as well as to infer the mechanisms of climate change. The articles also il lustrate some fundamental questions that need to be addressed in preparing multi-proxy reconstructions of past ENSO behavior and in analyzing paleo climate modeling results. We consider these studies to be the first step to wards a comprehensive evaluation of whether climate models incorporate the right physics and feedbacks to re produce the diversity of ENSO events. Pulling the pieces together An exciting development, showcased in this special issue, is that many dif ferent types of records exist that Figure 1: La Nia event of 1999. Left: Positive (purple) and negative (blue) sea surface temperature anomalies. Right: Positive (blue) and negative (brown) rainfall anomalies (mm). The four El Nio regions referred to in the following articles are depicted in the right panel. Images from NASA. Science Highlights: Editorial


49 PAGES news Vol 21 No 2 August 2013 document short term climate variabil ity (see Elliot et al. p. 54 and Carr et al. p. 56) and that pooling information from multiple types of records pro vides a pan-tropical picture of ENSO. The contributions here and in other recent papers (e.g. Cobb et al. 2013), demonstrate that the process-level un derstanding of individual records has improved over recent years (see Elliot et al. p. 54 or McGregor et al. p. 52). This includes the recognition that it is rare for paleoclimate records to reflect a single climate variable (e.g. temper ature or salinity) or aspect of climate variability (e.g. seasonal phasing or magnitude changes in mean frequency or in extremes). The challenge now is to find ways to synthesize the dispa rate aspects of the records and to take advantage of the wider range of infor mation available to provide an over view of ENSO variability (see Russon et al. p. 62 or Thompson et al. p. 60). Modeling improvements A second important development is the improvement in modeling capacity and analytical approaches. Coupling between different components of the climate system is notoriously difficult and this poses problems for simulat ing aspects of the climate system, such as ENSO, that are primarily driven by the coupling between the ocean and atmosphere (see Capotondi et al. p. 58 or Lazareth et al. p. 66). Nevertheless, with improved model physics, in creased spatial resolution, more rou tine incorporation of tracers (e.g. isotopes) and biogeochemistry, and the ability to run much longer simula tions, state-of-the-art models are in creasingly yielding important insights into ENSO. Model simulations show that ENSO displays non-stationary behavior in space and time, includ ing the strength of ENSO teleconnec tions (Capotondi et al. p. 58, Merkel et al. p. 68). The simulations imply that the ENSO phenomenon has to be de coupled into different components, such as mean state, hydrology and teleconnections that change indepen dently depending on the forcing (see Braconnot et al. p. 64). The lessons learned from models and their integra tion with observations pose ambitious challenges to the interpretation of the ENSO-relevant signal in paleoclimate records (Russon et al p. 62, Brierley p. 70, Lazareth et al. p. 66). Towards data and model integration It is a pivotal time for the ENSO com munity. We see a real need for new ac tivities. First and foremost, there needs to be a more integrated approach to combining data and modeling for hypothesis testing. To facilitate this, the community needs to synthesize more primary data across regions and realms. This will allow us to capitalize on the improvements in process un derstanding that are emerging from both data analyses and modeling ex periments. The other requirement is developing statistical and analytical tools that combine the two sources of information (data and model out puts) to disentangle the evolution of mean state, seasonality, and interan nual to decadal variability. To develop these approaches and tools, we see a need for a dedicated activity within the PAGES and PMIP communities that brings together the full range of ex pertise. This is the only way to answer the fundamental questions about the relationships between the background climate state, the seasonal cycle and ENSO variability. Selected references Full reference list online under: Braconnot P et al. (2012a) Nature Climate Change 2: 417-424 Braconnot P, Harrison SP, Tudhope S, Michaut C (2012b) PAGES news 20(1): 51-52 Cobb KM et al. (2013) Science 339: 67-70 Science Highlights: Editorial Figure 2: El Nio event of 1997. Left: sea surface temperature anomalies. Blue: negative, purple: positive. Right: Rainfall anomalies (mm). Brown: negative, blue: positive. Images from NASA.


50 Myanmar monsoon drought variability inferred by tree rings over the past 300 years: linkages to ENSO ROSANNE D ARRIGO 1 J. PALMER 2 C. UMMENHOFER 3 N.N. KYAW 4 AND P. KR USIC 5 1 Lamont-Doherty Earth Observatory (LDEO), Palisades, USA; 2 Gondwana Tree-Ring Laboratory, Canterbury, New Zealand; 3 Woods Hole Oceanographic Institution, Woods Hole, USA; 4 Forest Department, Naypyidaw, Myanmar; 5 Bert Bolin Centre for Climate Research, Sweden University, Stockholm, Sweden A new tree-ring record of teak from Myanmar yields information about past tropical monsoon rainfall variability, including sustained drought conditions, and climatic eects of the El Nio-Southern Oscillation (ENSO), dating back nearly four centuries. E xtreme climatic conditions linked to the Asian monsoon and modulated by the ENSO system, such as droughts and oods, severely impact human populations. In 2008, Cyclone Nargis caused >138,000 fa talities and more than an estimated 10 bil lion US$ in economic damage. However, our knowledge of monsoon climate and ENSO-related impacts remains limited, not the least due to sparse paleoclimatic information. Such information is especially scarce for Myanmar, which is directly im pacted by the eects of the monsoon but where even instrumental climate records and related studies of monsoon climate dynamics are rare. Much of the country experiences pronounced wet and dry seasons linked to the seasonality of the monsoon. Myanmar rainfall correlates sig nicantly with indices of ENSO and Indian monsoon rainfall (e.g. All India Summer Monsoon Rainfall and Core Indian Rainfall), although these relationships are spatially and seasonally variable and not very well understood. Myanmars location in the transition zone between the South Asian and East Asian monsoon systems results in a particularly complex spatial pattern of precipitation variability. Asian Monsoon dendroclimatology Gleaning additional information on the long-term climate variability of Myanmar and greater monsoon Asia relies on records from natural or historical paleoclimatic ar chives. Tree rings are often ideal as they can yield precisely dated, annual time series, also referred to as chronologies, dating back centuries to millennia. Studying tropi cal tree rings poses considerable diculties because of the lack of pronounced seasons at low latitudes where monsoonal rainfall occurs, and the associated obstacles in identifying tree species that can be dated for dendrochronology. Nevertheless, much progress in chronology development has been made in recent years that now al lows us to reconstruct past climate in tropical Asia from tree rings (e.g. Cook et al. 2010, Monsoon Asia Drought Atlas MADA; Buckley et al. 2007, 2010; Ummenhofer et al. 2013). The MADA tree-ring dataset was analyzed for past changes in the Palmer Drought Severity Index (PDSI) and thus pro vided evidence for major droughts and wet events in Southeast Asia (e.g. Indonesia, Thailand, and Vietnam). However, in the case of Myanmar, tree-ring data coverage is still exceedingly sparse and represents a sizeable gap in paleoclimate information of monsoon Asia. Here we present a new re cord for Myanmar that can be incorporated into the MADA network. Myanmar teak Teak ( Tectona grandis ) is one of the tree spe cies that has been used most successfully Figure 1: Map of Myanmar and adjacent southern monsoon Asia. Pink dot is Maingtha Forest Reserve (MFR) teak tree-ring site north of Mandalay. Background shows monthly Global Precipitation Climatology Centre rainfall data for the period 1951-2007, averaged for MayOctober when the majority of Myanmar annual rainfall occurs. The gure shows the lower mean rainfall in central Myanmar (modied from DArrigo et al. 2011).


51 for tropical dendroclimatology. It is endem ic to Myanmar, but due to its great com mercial value it is now rapidly disappear ing across the country. Our new ring width chronology of teak for the Maingtha Forest Reserve is one of the rst high-resolution proxy records that were developed for Myanmar (Figs. 1 and 2; DArrigo et al. 2012). It is based on 38 individual series from 20 living trees and spans the years 1613-2009 AD. Teak growth at the site is positively correlated with rainfall and PDSI variabil ity over Myanmar, during and prior to the May-September wet monsoon season. Accordingly, the Maingtha Forest Reserve record reveals past variability of Myanmar hydroclimate. Monthly correlations between the Myanmar teak ring-width chronology (Fig. 2) and instrumental PDSI (1950-2003 AD) were computed for the region overlap ping our teak site (15-25N, 95-105E), for the period, from the year prior to the cur rent year of radial growth to the current year. Statistically signicant (95% level), positive correlations were found for the months concurrent and just prior to the wet monsoon season, from around April to August. Correlations with PDSI are stron gest when averaged over May and June (r = 0.32, n = 54), reecting the importance of moisture availability for teak growth in central Myanmar around the time of mon soon onset. Comparison with local monthly station rainfall records (e.g. for Mandalay and Yangon) further conrms that teak growth in central Myanmar is controlled by moisture availability. This teak record also correlates sig nicantly with larger-scale climate indices, including those for core Indian rainfall and ENSO. With the IITM All-India monsoon rain fall index, correlation for May-Jun is r = 0.36 (p < 0.01, n = 57). Correlations are stronger using the core northeast Indian monsoon rainfall index, which represents the region of India opposite the Bay of Bengal from Myanmar (23N, 76E). Averaged over JanJun (1950-2006), the correlation is r = 0.46 (p < 0.001). For East Uttar Pradesh, a subdi vision of this core region, the correlation is 0.47 since 1950 (Jan-Jul, p < 0.001, n = 57); and r = 0.29 over the maximum common period (p < 0.001, n = 136; 26N, 82E). The Myanmar teak chronology corre lates negatively with Nio-3 SST (Jul-Aug, r = -0.27, p < 0.05, n = 60). This is consistent with the tendency for positive Nio-3 SSTs, as during El Nio warm events, to be linked to drought conditions and decreased teak growth over southeast Asia, due to the eastward migration of the Walker circula tion. Notably, Myanmar teak growth in the year 1999, following the so-called El Nio of the Century of 1997-1998, is the low est of any year on record (Fig. 2). The year 1998 also shows the lowest annual rainfall value for central Myanmar. Myanmar teak growth is also below average during the Strange Parallels drought (Fig. 2) found across southeast Asia in the 1700s. The late Victorian Great Drought was associated with a major ENSO warm episode (1876 to 1878) and is perhaps the most spatially pervasive and severe drought in the MADA tree-ring data set for monsoon Asia (Cook et al. 2012; Buckley et al. 2007, 2010). The potential of the new Myanmar record Our new teak tree-ring width chronol ogy from central Myanmar, one of the rst paleoclimate records published for this country, is signicantly and positively cor related with local and regional precipita tion as well as with larger scale indices of Indian monsoon core rainfall, particularly over northeastern India. The Myanmar teak records also correlate negatively with Nio-3 SST, consistent with the tenden cy for El Nio warm events to be linked to drought over Southeast Asia, as they were following the pronounced 19971998 warm event. Thus, this teak record reects not only local conditions, but also the large-scale strength of the circulation of the Asian monsoon-ENSO system. The timing of the inferred drought conditions coincides with megadroughts identied elsewhere in southeastern Asia, including Thailand and Vietnam (Buckley et al. 2007, 2010), and which are attributed to vari ability in the tropical Indo-Pacic climate system and the Intertropical Convergence Zone. Overall, our results indicate much potential for generating reconstructions of monsoon climate dynamics for Myanmar and the wider region from tree-ring data. Note MADA dataset available online at the NOAA Paleoclimatology website, International TreeRing Databank (ITRDB). References Buckley B et al. (2010) PNAS 107: 6748-6752 Buckley B et al. (2007) Climate Dynamics 29: 63-71 Cook E et al. (2010) Science 328: 486-489 DArrigo R et al. (2011) Geophysical Research Letters 38, doi:10.1029/2011GL049927 Ummenhofer C et al. (2013) Climate Dynamics 40: 1319-1334 Figure 2: Tree-ring width chronology of teak (in dimensionless indices or values) for central Myanmar, spanning 1613-2009 AD, with sample depth (number of individual tree samples). Sample depth is near its peak over much of the chronology, gradually declining prior to the middle 1700s. Plot labels time of Strange Parallels (SP) drought in the 1700s (Buckley et al. 2007, 2010), the late Victorian Great Drought (VGD), and narrow ring at time of 1997-98 El Nio of the Century (modied from DArrigo et al. 2011).


52 Coral microatoll reconstructions of El Nio-Southern Oscillation: New windows on seasonal and interannual processes HELEN V. MCGREGOR 1 C.D WOODROFFE 1 M. FISCHER 2 M.K. GAGAN 3 AND D FINK 2 1 School of Earth and Environmental Sciences, University of Wollongong, Australia; 2 Institute for Environmental Research, Australian Nuclear Science and Technology Organisation, Lucas Heights, Australia; 3 Research School of Earth Sciences, The Australian National University, Canberra, Australia Porites coral microatolls show 18 O signal reproducibility and delity comparable to more conventional coral growth forms. Longer-lived and fossil microatolls, which grow in suitably ushed environments, contain 18 O signals that can signicantly extend instrumental records of the El Nio-Southern Oscillation. P orites corals are the most commonly used genus for reconstructing El NioSouthern Oscillation (ENSO). This herma typic coral is found in all tropical reef en vironments (Veron 2000) with a variety of growth forms. Climate reconstructions of a century or more have been obtained from the most common, dome-shaped Porites growth form, whereby the colonies, be ginning from the substrate, grow outward and upward towards the ocean surface (Knutson et al. 1972). Domed structures, however, are not the only Porites growth form. Coral microatolls Porites coral microatolls are found on shallow reefs where reef topography en ables individual colonies to grow up to the average spring low tide level. Further upward growth is limited due to expo sure of the upper coral surface at low tide (Stoddart and Scon 1979). At this point, the coral then grows laterally, resulting in a at-topped discoid growth morphology termed microatoll (Fig. 1). Coral microatolls can live for decades to many centuries (McGregor et al. 2011a), are distributed broadly across the IndoPacic region (Scon and Stoddart 1978), and their preservation potential is particu larly high due to the possibility for rapid burial beneath sand and coral rubble through storm ridge or beach deposition. Microatolls provide information about past water levels, from which sea level, climatic, or tectonic histories have been derived (Natawidjaja et al. 2004; Sieh et al. 2008; Smithers and Woodroe 2001; Taylor et al. 2008, 1987; Woodroe and McLean 1990; Woodroe et al. 2012; Zachariasen et al. 1999). Microatolls also have the advantage of sampling a narrow depth range over long periods of time, which is desirable when reconstructing depth-de pendent, ENSO-related variables, such as sea surface temperature (SST) and sea sur face salinity (SSS), together with changes in ocean dynamic height, in the tropical Pacic. Studies of domed Porites show that there can be signicant dierences in skeletal 18 O on the sides and tops of the corals and this is equally a concern for laterally-growing microatolls (e.g. Cohen and Hart 1997; McConnaughey 1989). However, testing of 18 O variability within and between Porites sp. microatolls living on reef ats around Kiritimati (Christmas) Island in the central Pacic ocean, demon strates no signicant dierences between 18 O records from dierent growth orien tations within a single microatoll, or be tween records from microatolls in dier ent reef settings (McGregor et al. 2011b). Moreover, 18 O records from microatolls and from conventional domed Porites from elsewhere on the atoll (Evans et al. 1998b; Nurhati et al. 2009) also show simi lar patterns and magnitude of variability. Together, the results show that Porites mi croatolls can be used interchangeably with dome-shaped corals to reconstruct tropical climate variability. ENSO and 18 O in modern microatolls at Kiritimati Island Kiritimati Island is optimally located (Evans et al. 1998a) for reconstructions of ENSO. The island lies within the dry equatorial zone of the central Pacic, and in the NINO3.4 index region where SST variations dene ENSO events (Bjerknes 1969; Ropelewski and Halpert 1987). In this region El Nio events result in marked positive SST anomalies of up to 3C in the boreal winter, whereas La Nia events produce negative SST anomalies of 1-2C (Wyrtki 1975; Fig. 2a). Rainfall also shows a dominant ENSO signal with higher annual precipitation during El Nio years. Porites microatolls from Kiritimati reg ister these climatic variations. Variations in a composite (stacked) monthly microat oll 18 O record spanning the years 19782007 show a strong inverse correlation of r = -0.71 with SST and records major El Nio events (McGregor et al. 2011b; Fig. 2a). This is similar to ndings for domedPorites from Kiritimati where 70% of the variance is shared with SST (Evans et al. 1999). The stacked microatoll 18 O record Figure 1: Porites coral microatoll image and X-radiograph. A) Porites coral microatolls on a reef at at low tide. B) Positive X-radiograph crosssection through a Porites microatoll. Dark and light bands are the high and low-density bands, respectively, that form as the coral grows. Starting from the center, the coral grows upwards until further upward growth is constrained by exposure during the minimum low water level (in this case, 1997/1998). Lateral growth then ensues resulting in a discoid microatoll structure. The location of the living surface in 2007 when the coral was collected is indicated. Dashed lines indicate the outline of coral pieces not X-rayed.


53 also corresponds with anomalously high rainfall years (Fig. 2a); annual mean coral 18 O values of less than -5.3 are found only in years when total annual rainfall is above 1800 mm. In addition to local SST and rainfall, the microatoll 18 O signal is also negatively correlated with both SST and precipitation amount over a broad area of the equatorial Pacic (McGregor et al. 2011b). Since most of the covariance between 18 O, SST and precipitation is due to ENSO, the spatial correlations reect the characteristic ENSO pattern. ENSO and seasonal cycle variance patterns Understanding ENSO annual and interan nual cycle variance and interactions can provide important information on ENSO processes (Guilyardi et al. 2009). ENSO variance is recorded in the stacked mi croatoll 18 O record. The record (Fig. 2b,c) tracks SST variability at interannual (ENSO; 53% of the 18 O variance) and annual timescale (14%), consistent with existing analyses of instrumental tropical Pacic SSTs (Chiu and Newell 1983). Changes at annual and interannual scales at Kiritimati are reminiscent of the climate signal of the eastern equatorial Pacic (Chen et al. 1994; Mitchell and Wallace 1992). ENSO events occur irregularly every 2-8 years, yet individual events show a distinctive SST pattern tied (or phaselocked) to the seasonal cycle, such that El Nio SST anomalies peak during the boreal winter (DJF). The interannual com ponent of SST and rainfall records for Kiritimati Island show maxima in February, as does microatoll 18 O (Fig. 2d). At the annual scale, the microatoll 18 O, which peaks in July-August, varies predomi nantly in-phase with SST, rather than with rainfall. The annual maximum in Kiritimati rainfall occurs in March-April, due to the position of the Intertropical Convergence Zone (ITCZ) (An and Choi 2010; Horel 1982; Mechoso et al. 1995; Waliser and Gautier 1993). That the microatoll 18 O tracks SST at the annual and interannual scale is important; SST variations in the NINO3.4 Index region are used to dene ENSO variations. Accordingly, microatoll 18 O from the NINO3.4 region, such as Kiritimati Island, can be used to recon struct past ENSO variations at multiple timescales. ENSO signal in fossil microatoll 18 O Fossil Porites microatolls, which were growing in well-ushed environments, of fer opportunities to reconstruct tropical SST and ENSO variability beyond the limits of the instrumental record. Initial studies conrm reduced ENSO variability during the middle Holocene (Woodroe et al. 2003). Individual ENSO events in the late Holocene (Fig. 2e) however, appear at least as intense as those experienced in the past two decades (Woodroe et al. 2003). One particular El Nio event from 1740 yr BP (Fig. 2e) shows a negative 18 O excursion to ~5.6, which suggests substantial ad dition of 18 O-depleted rainfall (Woodroe and Gagan 2000). The stronger ENSO in the late Holocene may represent tighter coupling in the Pacic between the more southerly ITCZ, the east Pacic cold tongue and the Southern Oscillation, which could amplify ENSO precipitation variability and associated teleconnections. Such a sce nario is consistent with terrestrial paleocli mate records indicating a marked increase in El Nio activity from ~3000 yr BP. We are undertaking further analysis of fossil coral microatolls and their annual and interan nual variability to test this scenario. Note Data are archived at WDC-paleoclimatology p=519:1:4336223539645946::::P1_STUDY_ ID:12278 Selected references Full reference list online under: Evans MN, Kaplan A, Cane MA (1998a) Paleoceanography 13: 502-516 Mitchell TP, Wallace JM (1992) Journal of Climate 5: 1140-1156 McGregor HV et al. (2011b) Geochimica et Cosmochimica Acta 75: 3930-3944 Woodroe CD, Gagan MK (2000) Geophysical Research Letters 27(10): 1511-1514 Woodroe CD, Beech MR, Gagan MK (2003) Geophysical Research Letters 30, doi: 10.1029/2002GL015868 Figure 2: Variability of Kiritimati Island records. A) Comparison of the stacked microatoll 18 O record (red line) with Kiritimati SST (blue line) and monthly rainfall (green line). The stacked 18 O is a composite of three microatoll records from Kiritimati. The microatoll 18 O record is strongly anti-correlated with Kiritimati SSTs (r = -0.71) and is sensitive to El Nio events (yellow bars). B) Annual-scale and (C) interannual-scale wavelets for the stacked microatoll 18 O (red, y-axis inverted), SST (blue), rainfall (green). D) The circular phase plots show the calendar month (and 95% condence intervals) when, on average, the wavelets in B) and C) reach their maximum value. In general the microatoll 18 O record tracks SST variations. The rainfall is in phase with the SST and coral 18 O data at the interannual scale, but is out of phase at the annual scale. E) Prole of a 18 O microatoll from Kiritimati dated at ~1740 yr BP, placed on a new oating age scale (unpublished data) that shows El Nio events (yellow bars) similar to those of the past few decades. Black horizontal line is the mean 18 O. Figure modied after McGregor et al. (2011b), and Woodroe and Gagan (2000). SST and rainfall data from ERSSTv3b (Smith et al. 2008) and GPCPv2 (Adler et al. 2003), respectively.


54 Giant clam recorders of ENSO variability MARY ELLIOT 1 K. WELSH 2 AND R. DRISCOLL 3 1 Laboratoire de Planetologie et Geodynamique, Universit de Nantes, France; 2 School of Earth Science, University of Queensland, St. Lucia, Australia; 3 School of GeoSciences, University of Edinburgh, UK Giant clam stable isotope proles from Papua New Guinea faithfully record all the major El Nio events between 1986 and 2003, thus illustrating the usefulness of this archive to reconstruct past ENSO variability. C onsiderable uncertainty remains about the response of the El NioSouthern Oscillation (ENSO) to future climate scenarios (Merryfield 2006). Reconstructions of past changes in seasonality and ENSO from natural ar chives have a key role in providing in formation for understanding both the full range of variability and the sensi tivity of ENSO to changes in climate boundary conditions. Geochemical time series extracted from skeletons of annually banded reef-building corals and mollusks constitute powerful re cords in this regard. A number of excit ing recent studies have illustrated how clams (i.e. bivalves) can be used in pa leoenvironmental studies (e.g. Sano et al. 2012; Wanamaker et al. 2012). Here we specifically illustrate the use fulness of one bivalve species, Tridacna gigas (Fig. 1) as a natural archive for paleo-ENSO. Massive Porites spp. cor als and Tridacna spp. clams are both reef-dwelling, aragonite secreting or ganisms. Their annual bands can be subsampled and analyzed to derive profiles of oxygen isotope ratios ( 18 O) which have been shown to reflect the combined effects of regional sea sur face temperature (SST) and sea water 18 O from which they precipitate their aragonite structures (Tudhope et al. 1995; Welsh et al. 2011). Time series of 18 O in modern and fossil corals col lected in northern Papua New Guinea in the heart of the Western Pacific Warm Pool have been used to reconstruct ENSO variability for short windows of time over the past 130 ka (Tudhope et al. 2001). These records are however extremely rare because of the ten dency for the porous Porites skeleton to undergo diagenetic alteration dur ing periods of subaerial exposure. An advantage of T. gigas is that they have relatively impervious and finely layered shells that inhibit infiltration of ground waters that would lead to the diagenet ic processes of dissolution, recrystal lization and precipitation of secondary calcite. Finally, while coral 18 O show an isotopic disequilibrium, Tridacna spp. precipitate their shells in isotopic equi librium. This provides the possibility to more accurately quantify past changes in absolute SST and see water 18 O. Figure 1: Photo of a live Tridacna gigas from Heron Island. T. gigas are reef dwelling mollusks which have symbiotic algae living within their mantle. Valves are 50 cm from end to end. Photo K. Welsh.


55 Calibration To illustrate the potential of T. gigas as paleo-ENSO recorders, we obtained a high-resolution 18 O profile from a modern specimen that we compared to modern Porites coral 18 O profiles and an ENSO index. Samples were col lected from three localities along the Huon Peninsula in northern Papua New Guinea. Profiles of 18 O were obtained by subsampling the annual growth bands using high precision microdrill ing devices. The age of the coral and bivalve 18 O profiles were obtained in dependently (i.e. they were not tuned to one another) by counting the annual growth bands when visible and using the 18 O maxima and minima to posi tion the warmest and coolest months. The T. gigas 18 O profile covers the pe riod 1986-2002 and the Porites 18 O re cords cover the period 1987-2001 (Fig. 2). Average SSTs at the Huon Peninsula are around 29C with an annual range of 0.5-1.5C in monthly means. The pre dicted equilibrium skeletal annual aver age 18 O is -1.6. Therefore, our results confirm that T. gigas precipitate their shell close to isotopic equilibrium as has been shown previously (e.g. Aharon et al. 1991). Comparison of bivalve and coral profiles A striking feature is the high degree of resemblance between the coral and bi valve records despite their geographic separation of approximately 30 km and their average 18 O offset of ~4 (Fig. 2). Profiles correlate in detail on the seasonal and on the interannual levels. This correlation is particularly interest ing given that paleoclimate archives ob tained from coastal areas characterized by strong SST and salinity gradients can potentially be significantly influenced by the local micro-environmental hy drography. Our results clearly show that corals and clams record large-scale regional patterns. Furthermore, the good correlation between 18 O coral and bivalve profiles remains constant although measurements have been obtained from different carbonate se creting organisms with fundamentally different biological controls on car bonate formation and different growth rates. Giant Clams as recorders of ENSO events In northern Papua New Guinea precipi tation and temperatures are coupled on seasonal and interannual timescales. El Nio periods are associated with lower than average SST and drier conditions, whereas La Nia periods are associated with higher than average SST and wet ter conditions. The associated changes in see water 18 O and SST will thus have cumulative effects on shell 18 O, which will become more positive during El Nio and more negative during La Nia phases. The comparison of the ENSO index with the T. gigas and Porites 18 O records shows that each El Nio event is recorded in the shell and coral pro file by isotopic shifts of around 1.0 to 1.2 toward more positive values (Fig. 2) reflecting the combined influence of lower temperatures and decreased rainfall. During the El Nio phase of the Southern Oscillation, the region expe riences relative drought and slightly reduced SSTs (~-0.2 to -0.5C anomaly, see Fig. 2). These factors combine to drive skeletal 18 O to heavy values, with SST explaining about 30-50% of the skeletal 18 O range. Take away message We show that shells of T. gigas can be used to produce multi-decadal climatic records, hence providing a valuable re source for investigating changes to the frequency and strength of ENSO events in the past. The excellent reproduc ibility of clam and coral 18 O profiles illustrates the strength of using these archives to reconstruct large-scale hy drographic changes. References Elliot M et al. (2009) Palaeogeography, Palaeoclimatology, Palaeoecology 280: 132-142 Sano Y et al. (2012) Nature Communications 3: 761, doi:10.1038/ncom ms1763 Tudhope AW et al. (2001) Science 291: 1511-1517 Welsh K et al. (2011) Earth and Planetary Science Letters 307: 266-270 Figure 2: Comparison of T. gigas 18 O prole with ENSO index, local temperature and rainfall data. A) NINO3.4 index, (B) 3pt smoothed monthly rainfall anomaly (mm day -1 NASA/GPCPV2) for 146.25E, 6.25S, (C) T. gigas 18 O record, (D) Porites 18 O proles and (E) 3pt smoothed monthly SST anomaly (from IGOSS) for the same grid box as the rainfall data. Y-axes of the 18 O are inverted. The shaded bands indicate El Nio events.


56 Reconstructing ENSO in the Eastern Tropical Pacic from short-lived marine mollusks MATTHIEU CARR 1 S. PURCA 2 AND J.P. SACHS 3 1 UM2-CNRS-IRD, Institut des Sciences de l'Evolution de Montpellier, France; 2 Instituto del Mar del Peru, Callao, Peru; 3 School of Oceanography, University of Washington, Seattle, USA Shells of mollusks from oshore Peru were analyzed to reconstruct variability of the seasonal cycle and ENSO. The data provide insights into past changes in ENSO-related interannual variability in the Eastern Tropical Pacic and in the spatial structure of ENSO. T he focus of much paleoclimate work on ENSO has been on records span ning multiple decades, such as those de rived from corals. Such long records are, however, relatively rare, especially in the Eastern Tropical Pacic. This important limitation for studying past changes in the spatial pattern of ENSO activity can be compensated for by obtaining re cords from a larger number of shorterlived organisms. In this case, information about past climate is not available as a continuous record, but is compiled to ex tract climate statistics. Carr et al. (2013) recently presented a technique that responds to the critical need for quantitative estimates of tropi cal marine interannual variability. This technique uses the shells of marine mol lusks that live for 1-4 years, and thus al low us to reconstruct the seasonal range of sea surface temperature (SST). These data can then be compared to coral re cords and GCM outputs. The technique of using marine mollusk shells shares similarities with the approach of coral studies in that it produces oating win dows of climate record at a very high, often monthly, resolution. It also shares similarities with the approach of analyz ing many foraminifera shells individually from the same sediment layer (Koutavas et al. 2006; Leduc et al. 2009) in that pa leoclimate statistics are estimated from a random sample. Isotopic records in mol lusks enable independent reconstruc tions of the seasonal cycle. This approach is valid for any coastal mollusk species that faithfully records at least one annual SST cycle, and therefore opens up new opportunities for direct, quantitative pa leo-ENSO reconstructions in the Eastern Tropical Pacic, using either archeologi cal shell middens or uplifted fossil shell banks from Peru. ENSO characterization in the Nio1+2 region SST variability on the Peruvian coast is largely dominated by ENSO-related interannual variability. The amplitude of the annual cycle (T) on the coast is also clearly related to the Nio1+2 index, with larger amplitudes during El Nio and smaller ones during La Nia. As a result, the variance of the annual cycle amplitude, Var(T), on the Peruvian coast over any period of time is an indicator of ENSO variance generally in the Nio1+2 region (Fig. 1a). Oxygen stable isotopes ( 18 O) in marine shells from central and southern Peru primarily reect SST variability since seawater 18 O is not signicantly aect ed by freshwater input and evaporation (Carr et al. 2013). Mesodesma donacium shells grow continuously throughout the year and record the full annual cycle of coastal SST in Peru over a window of 1 to 4 years at a monthly resolution (Fig. 1b). Modern shells (n=13) collected during the late 20 th century provide a random sample cumulatively providing about 25 years of statistics that faithfully repro duces the skewed distribution of ENSO Figure 1: A) Annual mean SST and precipitation for the 1950-2010 period. Location of the Nio1+2 region in the Eastern Tropical Pacic and the Peruvian localities mentioned. B) Polished section of a M. donacium shell, and the associated 18 O plotted on a time scale. The shell was continuously sampled so that every data point integrates about one month. The chronology was determined using shell growth lines. Triangles indicate seasonal extrema used for the calculation of seasonal amplitudes T.


57 anomalies of this period in Callao, Peru and in the Nio1+2 region (Fig. 2a). It should be noted that the extreme warm events of 1982-83 and 1997-98 are not recorded in the data as the exceptionally warm conditions induced mass mortality in the mollusks. However, such events are so extraordinary even at the millennial scale (Rein 2007) that they are arguably not representative of ENSO (Takahashi et al. 2011). The mollusk shells are comple mentary to rainfall-related archives as they are sensitive to variability in the ma rine manifestation of ENSO while rainfallrelated archives are more sensitive to oc casional catastrophic events induced by the atmospheric anomalies of extreme ENSO events (Rein 2007). Estimating the reconstruction uncertainties Estimating reconstruction uncertain ties is a necessary challenge in paleocli mate studies if meaningful comparisons with GCM simulations are to be made. To estimate the reconstruction error, we used an in situ instrumental time series and simulated the reconstruction pro cess (forward proxy model). This was done by randomly extracting short time samples and adding dierent types of noise to them representing uncertainty sources (such as analytical error, random growth breaks, or temperature toler ance). Iterating this process thousands of times provides an estimate of systematic biases (the mean value of the error pop ulation) and of the standard error (the standard deviation of the error popula tion) (Fig. 2b). This shell sample yields thus an estimate of mean annual tem perature with a precision of .4C, of ENSO variance with a precision of %, and of ENSO skewness with a precision of .3. This procedure can also be used to evaluate the relative contributions of error sources, and improve our under standing of the proxy (Carr et al. 2012). This sample will be used as a modern reference to normalize past reconstruc tions, minimize systematic biases due to the archive or to local eects, and allow meaningful comparisons with coral re cords and climate simulations. Discriminating Central from Eastern Pacic modes in the past Capotondi et al. (2013, this issue) present some recent developments in our under standing of ENSO diversity. Two types of El Nio events, Central Pacic events and Eastern Pacic events, have been dened by the location of the maximum SST anomaly (Ashok et al. 2007). Because of their signicantly dierent impacts, the question of the evolution of ENSO in both the past and future, should now also address the variable contribution of Central Pacic and Eastern Pacic ENSO events. In the Nio1+2 region, in the far Eastern Pacic, Eastern Pacic and Central Pacic modes can be distinguished by the shape of their SST distributions. The Eastern Pacic mode is characterized by positively skewed SST distributions while the Central Pacic mode produces a sym metric distribution. Such a change in ENSO asymmetry has been noted since the 1990s (Yeh et al. 2009; Boucharel et al. 2011; Dewitte et al. 2012). La Nia cold events generate negatively skewed SSTs in the Nio1+2 region. The ENSO anomaly distribution obtained from the modern Peruvian shell sample faithfully reproduces the positive skewness of the late 20th cen tury ENSO, even without recording the most extreme ENSO events (Fig. 2a). Fossil mollusk shell samples from Peru could therefore potentially be used to track past changes in the frequency of Eastern Pacic and Central Pacic events, and provide novel insights into the rela tionships between ENSO modes and the mean climate. Selected references Full reference list online under: Carr M et al. (2013) Palaeogeography, Palaeoclimatology, Palaeoecology 371: 45-53 Carr M, Sachs JP, Wallace JM, Favier C (2012) Climate of the Past 8: 433-450 Rein B (2007) Quaternary International 161: 56-66 Takahashi K, Montecinos A, Goubanova K, Dewitte B (2011) Geophysical Research Letters 38, doi:10.1029/2011GL047364 Yeh S-W et al. (2009) Nature 461: 511-514 Figure 2: A) Distributions of annual Nio1+2 index from 1950 to 2002, seasonal T values in Callao, Peru from 1950 to 2002 excluding extreme El Nio events in 1982-83 and 1997-98, and T values calculated from a modern sample of M. donacium shells from Ica, Peru. B) Results of Monte Carlo simulations undertaken to estimate the uncertain ties of mollusk-derived paleoclimate reconstructions. Plots show the true value (green), the ensemble average value (black line), and the standard error (blue area) vs. the sample size, for the reconstruction of the mean and variance of T.


58 Challenges in understanding and modeling ENSO ANTONIETTA CAPOTONDI 1 E GUILYARDI 2 AND B. KIRTMAN 3 1 CIRES, University of Colorado, Boulder, USA; 2 LOCEAN, Institut Pierre Simon Laplace, Paris, France; 3 RSMAS/MPO, University of Miami, USA Some new exciting directions in ENSO research explore inter-event differences in spatial patterns, teleconnections and impacts, asymmetries between warm and cold phases, and the role of extra-tropical regions in triggering ENSO events. However, large uncertainties remain regarding ENSO projections. T he El NioSouthern Oscillation (ENSO) is a naturally occurring fluc tuation that originates in the tropical Pacific region and affects ecosystems, agriculture, freshwater supplies, hurri canes and other severe weather events worldwide. Over the last thirty years significant progress has been made in improving our understanding of the dynamic processes underlying ENSO, including the ocean-atmosphere feed backs that are essential to this coupled phenomenon. The oscillatory nature of ENSO, al ternating between El Nio and La Nia events, can be described in terms of the recharge and discharge of warm water to and from the equatorial thermocline (recharge oscillator; Jin 1997) or in terms of thermocline depth changes associated with wave propagation (de layed oscillator, e.g. Suarez and Schopf 1988). These simple paradigms of ENSO as a linear oscillator capture basic dy namical processes; however, they fail to explain differences among events and asymmetries between warm and cold episodes. Moreover, they ignore the im portant role of stochastic atmospheric phenomena (e.g. westerly wind bursts) and other non-linear effects. Understanding and predicting the diverse characteristics of El Nio and La Nia events is important since their re gional climatic impact can vary heavily depending on the longitudinal location of the SST anomalies. Also, understand ing how teleconnections vary depend ing on the event type is crucial when proxy records are used to reconstruct past ENSO. Hence, exciting new research developments have emerged to address this observed ENSO diversity. Understanding ENSO dynamics The first development is a renewed in terest in inter-event differences and the related El Nio Modoki debate. Based on a statistical analysis of SST in the tropical Pacific, Ashok et al. (2007) sug gested the existence of another type of El Nio, called Central Pacific El Nio (or Date Line El Nio or El Nio Modoki by various authors). They argued that this type of El Nio is not the same as the ca nonical El Nio because its center of ac tion is in the central Pacific instead of the eastern Pacific, as illustrated in Figure 1. It was also suggested that Central Pacific El Nios have become more frequent in recent decades, and their frequency may increase further with global warm ing (Yeh et al. 2009). Subsequent ob servational and modeling studies have tried to define the Central Pacific El Nio more precisely or differently (Kug et al. 2009; Kao and Yu 2009). However, as yet no agreement has been reached on the best way to characterize the new Central Pacific-type of El Nio. Some studies have tried to distinguish the central Pacific and eastern Pacific (canonical) warm events based on their underlying dynamical processes, and their relation ship with the oceanic mean state (e.g. Choi et al. 2011; McPhaden et al. 2011). A number of other studies dispute the statistical significance of the distinc tion between the two El Nio types or at least of the increasing occurrence of the Central Pacific variety. They argue either that the reliable observational record is too short to detect such a dis tinction (Nicholls 2008; McPhaden et al. 2011), or that they have found no trend using other approaches (Giese and Ray 2011; Newman et al. 2011; Yeh et al. 2011). Other authors alternatively suggest to distinguish between other types of El Nio, such as standard and extreme El Nios (Lengaigne and Vecchi 2010; Takahashi et al. 2011). Due to the asymmetric nature of the warm and cold phases of ENSO, Kug and Ham (2011) could not identify analogous distinc tions for La Nia, neither in observations nor in the simulations of the Climate Figure 1: Composite spatial pattern of SST anomalies for the canonical (top) and Central Pacic (bottom) El Nio types (SODA 2.0.2/3) from 1958 to 2007 computed with the approach of Kug et al. (2009). Canonical El Nios are characterized by a boreal winter (DJF) Nio3 index larger than 0.5C and larger than the Nio4 index (red and blue dashed boxes, respectively), and vice versa for the Central Pacic El Nio (from Capotondi, in press). Observed El Nio events can be described as blends of these two end-member types.


59 Model Intercomparison Project version 3 (CMIP3). Due to the large societal rel evance of the impacts of ENSO, it is im portant to predict not only whether an El Nio (or La Nia) event is expected, but if possible which expression the anomaly will take. Fueled by these early studies, new questions are now emerg ing asking, for instance, if discrete classes of ENSO events emerge from observations, paleoclimate records and model simulations, or if ENSO diversity is better described as a continuum with a few characteristic extremes (e.g. Wu and Kirtman 2005). Other new lines of research in ENSO diversity include revisiting the relative roles of the ocean and the atmosphere in shaping ENSO (Kitoh et al. 1999; Guilyardi et al. 2004; Dommenget 2010; Clement et al 2011; Lloyd et al. 2011) and exploring the role of regions outside the tropical Pacific in triggering ENSO events (Vimont et al. 2003; Izumo et al. 2010; Terray 2011; Wang et al. 2011). An exam ple of remote influence is the seasonal footprinting mechanism (Vimont et al. 2003): Atmospheric variability originat ing in the North Pacific can impact the subtropical ocean during winter, and the resulting springtime SST anomalies alter the atmosphere-ocean system in the tropics during the following summer, fall and winter. The diversity of geographical sources and mechanisms proposed may explain the diversity of El Nio events, both in observations and in models. ENSO in climate models and future projections Most of our understanding of the rep resentation of ENSO in climate models has been derived from the analysis of the model simulations of the Climate Model Intercomparison Project versions 3 (CMIP3) and 5 (CMIP5). While the mod els appear to reproduce some of the ba sic processes and feedbacks associated with ENSO, the details of the SST anom aly patterns as well as the temporal evo lution of the anomalies often differ from the observed, and reflect model biases or erroneous atmosphere-ocean inter actions (Capotondi et al. 2006, Guilyardi et al. 2009; Guilyardi et al. 2012a). For example, in most of the CMIP3 mod els, the largest anomalies are located further west along the equator than in observations. Furthermore, in many models ENSO events tend to occur more frequently and regularly than in the real world. While the models keep improving in their simulation of ENSO, no quan tum leap was seen in CMIP5 compared against CMIP3 (Guilyardi et al. 2012b). Over the past few years, new prom ising methods have emerged, which could improve ENSO simulations, for example by bridging ENSO theoreti cal frameworks and climate modeling. Resulting innovations include the devel opment of indices that can be used to assess the stability of ENSO in Coupled General Circulation Models (CGCMs), and intermediate models that can be used to predict ENSO characteristics from aspects of the mean state. By focus ing on the key processes affecting ENSO dynamics (e.g. the thermocline feed backs or the wind stress response to SST anomalies), these new approaches have much potential to accelerate progress and improve the representation of ENSO in complex climate models. Not only can these new methods help address the question of whether the characteristics of ENSO are changing in a changing cli mate, but potentially they can also im prove the reliability of centennial-scale climate projections and predictions on seasonal time scales. Looking forward At present, we dont know enough about how ENSO has changed in the past (the detection problem) and what caused the changes i.e. the contribution from external forcing vs. that due to internal variability (the attribution problem). Given the much too short reliable obser vational record (both for ENSO and for the external forcing fields, Wittenberg 2009), the complexity and diversity of the paradigms and processes involved, and the shortcomings of current stateof-the-art models, understanding the causes of ENSO property changes, both in the past and in the future, remains a considerable challenge. For instance Collins et al. (2010) concluded that it is not yet possible to say whether ENSO activity will be enhanced or damped in future climate scenarios, or if the fre quency of events will change (Fig. 2). Paleoclimatic and paleoceanographic reconstructions have the potential to initiate the next quantum leap. Selected references Full reference list online under: Ashok K et al. (2007) Journal of Geophysical Research 112, doi:10.1029/2006JC003798 Capotondi A., Wittenberg A, Masina S (2006) Ocean Modeling 15: 274-298. Collins M et al. (2010) Nature Geoscience 3 (6): 391-397 Guilyardi E et al. (2009) Bulletin of the American Meteorological Society 90: 325-340 Guilyardi E et al. (2012a) Bulletin of the American Meteorological Society 93: 235-238 Figure 2: Standard deviation of Nio3 SST anomalies for thirteen CMIP5 model experiments. Blue bars, pre-industrial control experiments; orange bars, years 90-140 from the 1% year -1 CO 2 increase experiments; red bars, years 50-150 after an abrupt four-fold CO 2 increase. Model names are given on the x-axis. Error bars indicate the standard deviations over 50-year windows of Nio3 anomalies in the multi-century control experiments. Thus, when the Nio3 standard deviation in one of the CO 2 runs falls outside the error bar, the changes are deemed signicant (modied from Guilyardi et al. 2012b). As in CMIP3, this new set of model simulations does not provide a clear trend for ENSO strength in a warming climate.


60 Coral-model comparison highlighting the role of salinity in long-term trends DIANE M. THOMPSON 1 T .R. AULT 1 M.N. EV ANS 2,1 J.E COLE 1,3 J. EMILEGEAY 4 A.N. LEGRANDE 5 1 Department of Geosciences, University of Arizona, Tucson, USA; 2 Department of Geology and Earth System Science Interdisciplinary Center, University of Maryland, College Park, USA; 3 Department of Atmo spheric Sciences, University of Arizona, Tucson; USA; 4 Department of Earth Sciences, University of Southern California, Los Angeles, USA; 5 NASA Goddard Institute for Space Studies and Center for Climate Systems Research, Columbia University, New York, USA We use a simple proxy model to compare climate model simulations and coral records over the 20 th century. While models and observations agree that the tropical oceans have warmed, they disagree on the extent and origin of freshening. T he response of the tropical Pacic Ocean to anthropogenic climate change re mains uncertain, in part because we do not fully understand how the region has re sponded to anthropogenic change over the 20 th century. Analysis of 20 th century temper ature and salinity trends is hindered by lim ited historical data, lack of long-term in situ measurements, and disagreement among coupled general circulation model (CGCM) hindcasts. High-resolution paleoclimate records, particularly the large network of tropical Pacic coral oxygen isotope records, are an alternate means of assessing tropi cal climate trends. However, these natural archives of past climate are biased by their limited spatial and temporal distribution and their biologically mediated response to climate. By converting native climate vari ables (e.g. temperature and net freshwater ux) into synthetic (pseudo) proxy records via an explicit proxy system model (forward model), we can directly compare historical climate data and climate model simulations with coral records, and assess biases and un certainties associated with each. Pseudocoral modeling The stable oxygen isotope ratio ( 18 O) of cor al aragonite is a function of the temperature and the oxygen isotopic ratio of seawater ( 18 O sw ) at the time of growth; the latter is in turn strongly related to sea-surface salinity (SSS). As direct measurements of 18 O sw are scarce, we model the expected 18 O anoma lies of corals ( 18 O pseudocoral ) as a function of sea-surface temperature (SST) and salinity anomalies: We dene coecient a 1 as -0.22 C -1 based on the relationship between tem perature and the isotopic composition of the skeleton in well-studied coral genera (e.g. Evans et al. 2000). Coecient a 2 is esti mated from basin-scale regression analysis of available observations of 18 O sw on SSS (LeGrande and Schmidt 2006; LeGrande and Schmidt 2011). Uncertainty in the ap plication of the resulting bivariate model arises from the assumed independence and linearization of a 1 and a 2 and substitution of the second term for a direct dependence on 18 O sw We apply this simple forward model of 18 O pseudocoral to generate synthetic coral (pseudocoral) records from historical ob servations and CGCM simulations of tem perature and salinity (Thompson et al. 2011). When driven with historical climate data, we found that this simple model was able to capture the spatial and temporal pattern of ENSO and the linear trend observed in corals from 1958 to 1990. Modeling pseudocorals with temperature and salinity individually also demonstrated that although warming accounts for the majority of the observed 18 O coral trend (60% of trend variance), salin ity also plays an important role (40% of trend variance). The addition of the SSS term im proved agreement between modeled pseu docoral and observed coral 18 O trends over pseudocoral trends modeled from SST only (Thompson et al. 2011). 20 th century trends When driven with the output from 20 th century simulations of a subset of CGCMs from the third phase of the Coupled Model Intercomparison Project (CMIP3) sampled at the coral locations, none of the pseudo coral networks reproduced the magnitude of the secular trend, the change in mean state, or the change in ENSO-related vari ance observed in the actual coral network from 1890 to 1990 (Thompson et al. 2011). Applying this same approach to the newer (CMIP5) suite of historical climate simula tions, we nd that large discrepancies still remain in the magnitude (Fig. 1), spatial pattern and ENSO-related variance of the simulated and observed trends. Dierences between observed and simulated 18 O coral trends may stem from the simplicity of our forward model of 18 O coral biological bias in the coral records, or model-inherent bias in the CGCM SST and SSS elds. Although we cannot yet completely rule out a non-climatic origin for the ampli tude of the observed 18 O coral trend, previous work highlights biases in simulated salinity elds as a potential source of the observedsimulated trend discrepancy (Thompson et al. 2011). We found that the suite of CMIP3 and CMIP5 CGCMs simulate weak and heterogeneous salinity trends that are indistinguishable in magnitude from that of unforced control runs (Fig. 1). Further, the pseudocoral simulations (Fig. 1) illus trate that the magnitude of the simulated 18 O coral trend can be less than the sum of Figure 1: Magnitude of the trend slope ( per decade), computed by linear regression through the trend principal component (PC) in corals (far left) and pseudocorals modeled from Simple Ocean Data Assimilation (SODA) 20 th century reanalysis (Carton and Giese 2008; Compo et al. 2011), a 500-year control run from the CGCM version CM2.1 of the Geophysical Fluid Dynamics Laboratory (GFDL cm2.1) (Wittenberg et al. 2009), and all CMIP-3 and CMIP-5 model ensembles (average of all models from each modeling group). In each case, 18 O coral was modeled from SST and SSS (1), SST only (2), and SSS only (3). Error bars depict 1 standard deviation of the regression estimate. 18Opseudocoral= a1SST+ a2SSS


61 the individual trends arising from tempera ture and salinity when the temperature and salinity trends are confounding at the coral sites (as observed for SODA modeled pseu docorals). However, given the limited num ber of historical SSS observations, much uncertainty remains in the sign and magni tude of the 20 th century salinity trend. When forward-modeling pseudocorals with data from two recent versions of an extended reanalysis (SODA v2.2.4 and v2.2.6; Ray and Giese 2012), we found that even the rela tive contribution of temperature and salinity to the observed pseudocoral trend diers (Fig.1); this discrepancy likely arises from the choice of wind eld used in the reanalyses (G. Compo, personal communication). These results suggest a need for improved simula tion of moisture transport and additional proxy reconstructions of salinity and 18 O sw to better understand their relationship and the sign and magnitude of their change. 18 O sw vs SSS: insights from isotope enabled simulations In substituting the 18 O sw -SSS relationship calculated from the limited observational dataset for 18 O sw our simple forward model assumes that this relationship is not only a valid approximation for 18 O sw but also that this relationship does not vary signicantly through time or within regions. Although this assumption does not likely hold at the millennial timescale (e.g. LeGrande and Schmidt 2011), it may be appropriate for simulating tropical variability during the past century. Here we assess the stability of the 18 O sw -SSS relationship through space and time on monthly to decadal timescales using a control simulation of an isotopeenabled version of the Goddard Institute for Space Studies model (GISS ModelE2, pro vided by A. LeGrande). In these simulations, the relationship between 18 O sw and SSS was generally regionally consistent over month ly to decadal timescales (Table 1), suggest ing that the substitution of SSS for 18 O sw is unlikely to impose low-frequency variability on the modeled pseudocorals. However, we nd that the slope of the 18 O sw -SSS relation ship and its sensitivity to timescale varies within the broad regions of Table 1, par ticularly between the eastern and western Pacic (Fig. 2). Similar regional variability in the slope of the 18 O sw -SSS relationship was observed in an isotope enabled ver sion of the UK Met Oce model (HadCM3; Russon et al. 2013). Additionally, the slopes of the 18 O sw -SSS relationship simulated for the tropical regions in the GISS model were generally higher and more spatially consis tent than those calculated from the limited observations (LeGrande and Schmidt 2006; Table 1). If we analyze only model output corresponding to the location and time of observations, the data-model discrepancy is reduced but not eliminated (Table 1). These discrepancies likely arise from the scarcity of paired 18 O sw and SSS observations as well as from the modeling of precipitation process es, and will be reduced by a combination of continued seawater sampling and model development. If the current observational dataset underestimates the true magnitude of the 18 O sw -SSS slope, our simple forward model will underestimate the magnitude of the true 18 O coral trend when a signicant fresh ening is observed. Estimates of uncertainty in the 18 O sw -SSS slope should be incorpo rated in future work simulating pseudo coral trends. Nonetheless, the salinity trend in CMIP3 and CMIP5 models is weak, and near zero, suggesting that the uncertainty in the 18 O sw -SSS relationship is not likely the source of the dierence in the coral and pseudocoral trend magnitude. The presence of a signicant freshening in historical ob servations suggests that this discrepancy is more likely due to an underestimation of the 20 th century freshening in the CGCMs. Acknowledgements We thank A. Wittenberg, B. Giese and G. Compo for contributing data and analysis products, and all those responsible for the PMIP3/CMIP5 simulation archives. Supported by NOAAs Climate Change Data and Detection Program (NA10OAR4310115 to JEG, MNE, DMT; NA08OAR4310682 to JEC) and P.E.O International. Selected references Full reference list online under: Evans MN, Kaplan A, Cane MA (2000) Paleoceanography 15(5): 551-563 LeGrande A, Schmidt G (2011), Paleoceanography 26, doi:10.1029/2010PA002043 LeGrande AN, Schmidt GA (2006) Geophysical Research Letters 33, doi:10.1029/2006GL026011 Russon, T, Tudhope AW, Hegerl GC, Collins M, and Tindall J (2013) Climate of the Past Discussions 9, 741-773 Thompson DM et al. (2011) Geophysical Research Letters doi:10.1029/2011GL048224 Figure 2: Slope of the GISS ModelE2 simulated 18 O sw vs. salinity relationship at each gridbox on monthly (top) and decadal (bottom) timescales. Decadal series were calculated by averaging the monthly data at 10-year intervals. Region Observed slope Monthly ModelE2 Annual ModelE2 Decadal ModelE2 ModelE2 at observations Tropical Pacic 0.27 0.35 0.35 0.36 0.32 South Pacic 0.45 0.33 0.33 0.37 0.30 Indian Ocean 0.16 0.33 0.35 0.35 0.27 Table 1: Slope of the regional 18 Osw-SSS relationship calculated from observations (LeGrande and Schmidt 2006) and the GISS ModelE2 control simulation. Annual and decadal series were calculated by averaging the monthly data at yearly and 10-year intervals.


62 The response of pseudo-corals to ENSO in an isotopeenabled climate model TOM RUSSON 1 A.W TUDHOPE 1 G.C. HEGERL 1 M. COLLINS 2 AND J. TINDALL 3 1 School of GeoSciences, University of Edinburgh, UK; 2 College of Engineering, Mathematics and Physical Sciences, University of Exeter, UK; 3 School of Earth and Environment, University of Leeds, UK Coral stable isotope records provide information on past ENSO variability. However, separating the contributions from variability in ocean temperature and the hydrological cycle to such records remains challenging. Model simulations using water isotope-enabled climate models provide powerful tools to explore this. T he stable oxygen isotopic composi tion of the aragonite of reef-dwelling corals ( 18 O coral ) relates to both the tem perature, taken here as being the sea sur face temperature (SST), and the isotopic composition of the seawater ( 18 O sw ) in which calcication occurred. The relation ship between 18 O coral and SST, derived from modern calibrations, generally has a slope that is close to the value of -0.2 K -1 found for inorganically precipitated carbonates (e.g. Gagan et al. 2000; Zhou and Zheng 2003). Some long-lived corals generate suciently high growth rates as to allow measurement of 18 O coral at subannual resolution over multiple decades (e.g. Carr et al., this issue). These proper ties provide a strong basis for using fossil corals 18 O coral to reconstruct SST variabil ity associated with the El-Nio Southern Oscillation (ENSO) over the Holocene and LGM (e.g. Tudhope et al. 2001; Cobb et al. 2003). However, 18 O coral also depends di rectly on 18 O sw which is in turn inu enced by a range of factors. Some of these factors may be closely coupled to ENSO, such as the local precipitationevaporation balance, but others relate in stead to the integrated hydrological his tory of the precipitation. In regions with a very active hydrological cycle, where the 18 O sw contribution is thought to domi nate the overall 18 O coral signal, records have been used to infer past changes in precipitation, rather than SST (Cole and Fairbanks 1990). Fully quantifying the spatial pattern of relative contributions from SST and 18 O sw to 18 O coral remains a challenge for interpreting these records. Limitations of the instrumental record Instrumental records of 18 O sw are not available for most coral bearing loca tions and those that do exist are typi cally too short to allow robust quanti cation of inter-annual changes in 18 O sw (Schmidt 1999; LeGrande and Schmidt 2006). However, the 18 O sw contribution can be estimated empirically from an instrumental SST record, provided that (1) the ENSO-related 18 O sw uctuations relate linearly to those in SST and (2) this relationship remains stationary through out the period of interest. An example of a case in which the rst assumption may be compromised is if the source re gion for precipitation changes with the magnitude of ENSO events. The second assumption may be compromised if the dominant spatial modes of ENSO vari ability change through time (Yeh et al. 2009; Capotondi et al., this issue). Climate model realizations of the response of 18 O sw to ENSO uctuations have the po tential to better constrain the validity of such assumptions. Representing pseudo-corals in an isotope-enabled climate model Only a few coupled Ocean/Atmosphere General Circulation Models (GCMs) in clude the additional hydrological cycle processes required to directly simulate water isotope variables such as 18 O sw Modeling pseudo-coral records based on the use of 18 O sw proxy variables such as salinity, provide a strategy to avoid this limitation (Thompson et al. 2011, this Figure 1: The percentage of modeled interannual 18 O coral variance accounted for by 18 O sw assuming a SST 18 O coral slope of -0.2 K -1 The 10% and 50% levels are contoured in red and green respectively and the location of the NINO4 region is highlighted in gray.


63 issue). However, work with the isotopeenabled Goddard Institute for Space Studies ModelE-R shows that the slopes of the 18 O sw -salinity relationships may dier when calculated over temporal and spatial patterns of variability (LeGrande and Schmidt 2009). The results presented here are based on a 750-year long preindustrial control simulation of another isotope-enabled coupled GCM, the UK Met Oces HadCM3 (Russon et al. 2013; Tindall et al. 2009). The inter-annual vari ability of the tropical climate in HadCM3 is known to be dominated by processes exhibiting spatial and temporal patterns resembling, albeit with signicant biases, those of the observed ENSO phenom enon (Collins et al. 2001; Guilyardi et al. 2006). For this study, the water isotope regimes were brought to equilibrium by rst running the model for an additional 300 years from an assumed initialization state. The pseudo-coral 18 O coral eld is then calculated directly by inputting the monthly-mean SST and 18 O sw data for the ocean grid resolution of 1.25 by 1.25 over the tropical Pacic (30S-30N and 120E-80W) into a linear formulation of the standard isotope paleo-temperature equation, with an assumed 18 O coral to SST slope of -0.2 K -1 Quantifying the 18 O sw contribution Modeled inter-annual uctuations in 18 O sw vary inversely with those in SST across almost the entire tropical Pacic region such that they combine positively. The fraction of the inter-annual variance of pseudo-coral 18 O coral that could be accounted for by the inter-annual vari ance of modeled 18 O sw is less than 10% (red contour in Fig. 1) across much of the subtropical eastern and equatorial Pacic, but higher in the Warm Pool, South Pacic Convergence Zone, and central American coastal regions. This arms that the 18 O sw contribution is indeed im portant in regions of high precipitation variability (Tudhope et al. 2001; Cole and Fairbanks 1990). Consequently, whilst eastern Pacic pseudo-coral 18 O coral could be reasonably used as a proxy of SST uctuations, this is not the case for all locations. However, only in very limited regions does the 18 O sw contribution ex ceed 50% (green contour in Fig. 1). Even within the high precipitation regions, there are no locations where one would expect the SST contribution to be neg ligible. Therefore, interpreting western Pacic corals as solely (or even predomi nantly) dependent on either temperature or precipitation appears misguided for many locations in the model. Non-linearity between SST and 18 O sw The regional relationships between mod eled SST and 18 O sw are not always simple. For example, in the western equatorial Pacic NINO4 region (grey rectangle in Fig. 1), little relationship is seen between modeled SST and 18 O sw during La-Nia (blue crosses), neutral (grey crosses) and even moderate El-Nio (red crosses) re gimes (Fig. 2a). This results in pseudocoral 18 O coral values that lie close to the imposed 18 O coral -SST slope (Fig. 2b). In such situations, the 18 O sw variability eectively adds (a relatively small degree of) noise to the 18 O coral record. However, during larger El-Nio events a weak an ti-correlation between SST and 18 O sw becomes evident (lower right quadrant of Fig. 2a), such that for SST anomalies exceeding ~1.5K, a deviation from the imposed slope of the 18 O coral -SST rela tionship becomes noticeable (lower right quadrant of Fig. 2b). For large El-Nio events, estimating NINO4 SST directly from 18 O coral would result in a relative overestimation of the true SST anomaly by over 20%. This eect would compli cate attempts to accurately infer the relative magnitudes of the SST anomalies during El-Nio events of dierent magni tude from proxy records of 18 O coral alone. Acknowledgements This work was funded by NERC grant NE/ H009957/1 Note The model data presented here are available upon request from the corresponding author. Selected references Full reference list online under: LeGrande AN, Schmidt GA (2006) Geophysical Research Letters 33, doi:10.1029/2006GL026011 LeGrande AN, Schmidt GA (2009) Climate of the Past 5(3): 441-455 Russon T et al. (2013) Climate of the Past 9: 1543-1557 Thompson DM et al. (2011) Geophysical Research Letters 38, doi: 10.1029/2011GL048224 Tindall JC, Valdes PJ and Sime LC (2009) Journal of Geophysical Research 114, doi:10.1029/2008JD010825 Figure 2: Scatter plots of modeled monthly inter-annual anomaly data within the NINO4 box. A) 18 O sw plotted against SST. B) 18 O coral plotted against SST, with the assumed slope of -0.2 K -1 used to calculate 18 O coral shown as a dashed line. Points are color coded according to their SST anomaly values, such that those lying in the upper and lower standard deviations of the SST data are highlighted red and blue, and are associated with El-Nio and La-Nia events respectively.


64 ENSO and changes in the mean state in Holocene simulations P. BRACONNOT 1 AND Y LU AN 1,2 1 Laboratoire des Sciences du Climat et de lEnvironnement, Gif-sur-Yvette, France; 2 State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Beijing, China Simulations suggest that Pacic interannual changes in sea surface temperature (SST) are smaller than SST seasonality, whereas the opposite is modeled for precipitation. Nonstationarity in ENSO patterns may aect the interpretation of past variability changes from climate records. H igh-resolution paleoclimate indica tors provide a unique opportunity to reconstruct and understand how sea sonal climate variability and the El NioSouthern Oscillation (ENSO) have evolved in the past. However, most climate inter pretations of paleorecords assume sta tionarity, i.e. that the modern relationship between a given climate sensor and ENSO was the same in the past, which might not be true. Therefore, one of the diculties is to infer how changes in mean state and variability have shaped the evolu tion of SST and other environmental factors, such as precipitation and wind, in the past. Using a suite of simulations performed with the climate model of the Institut Pierre Simon LaPlace (IPSL-CM4; Marti et al. 2010), we discuss the relative eects of dierent Holocene forcings on seasonality and interannual variability between the early Holocene and the preindustrial period. Sensitivity experiments Previous simulations and model data comparison have established that longterm changes in insolation aected ENSO variability in the Holocene (Clement et al. 2000; Moy et al. 2002). However, the presence of melting remnant ice sheets in the northern hemisphere during the Early Holocene, may have partially oset the impact of the insolation forcing. In this study, simulations of the Early Holocene and the Mid-Holocene are used to infer how the slow variation of the Earths or bital parameters aected ENSO variability (Luan et al. 2012). We also conducted a set of sensitivity experiments to exam ine how meltwater release and the pres ence of remnant northern hemisphere ice sheets may have impacted ENSO charac teristics (Braconnot et al. 2012; Marzin et al. in press). The dierent model years in the simulations were classied either as El Nio, La Nia, or normal years based on the December-January-February SST anomalies in the Nio3 region (150W-90W, 5S-5N). Anomalies were only classied as El Nio or La Nia events when they crossed a SST threshold of 1.2 times the standard deviation derived from the pre-industrial SST time series. ENSO is the dominant mode of SST and precipitation variability in all of the simulations. However, both the pre-indus trial and past El Nio event simulations are aected by biases common to most climate models (Zheng et al. 2008). For example, the cold tongue (i.e. the equa torial region in the Pacic with cold SSTs) of normal years penetrates too far west along the equator (not shown), as does the equatorial warming associated with El Nio events (Fig. 1a,c). As a result the horseshoe structure of El Nios SST and precipitation patterns seen today in the western Pacic is not well pronounced. This discrepancy can lead to misinterpre tation when comparing simulated chang es with high-resolution coral records from the central equatorial Pacic region (Brown et al. 2008). In addition, simula tions of ENSO variability are also damped in the eastern Pacic when compared with modern observations. Spatial variability of ENSO patterns Our simulations indicate that over time changes in forcing inuence the loca tion and intensity of the maximum SST and precipitation anomalies. In the Early Holocene for example, insolation forcing slightly damps the strength of the peak of the event (Fig. 1) compared with preindustrial control simulations. In these Early Holocene examples a major reduc tion is simulated west of the maximum Figure 1: Composite sea surface temperature (SST) anomaly and precipitation anomaly maps during El Nio events as simulated by the IPSL-CM4 climate model for preindustrial and Early Holocene (9.5 ka BP) climates. The anomalies describe the departures from normal years in each simulation at the peak of the event in December-January. Isolines are plotted every 0.5C for SST with a renement of 0.25C around the 0 line, and every 1 mm d -1 for precipitation with a renement of 0.5 mm d -1 around the 0 line. The red and blue dashed boxes in the lower left panel show the Nio3 and Nio4 regions, respectively.


65 SST anomaly found in the pre-industrial control simulation, and another impor tant reduction occurs on the SouthAmerican coast. Figure 1 therefore illus trates that the teleconnections between the dierent parts of the Pacic basin, as well as between the Pacic and the Indian Ocean, vary depending on the mean cli mate state. This further suggests that the climatic relationships between regions to day are not stationary in time. Changing seasonality and interannual variability accross time Insolation forcing also aects the season ality of SST and precipitation. Changes in the magnitude of the seasonal cycle of precipitation mirror the changes in the seasonal cycle of SSTs. Figure 2 shows the Early Holocene insolation only simulations (EHnF), and pre-industrial control runs (CTRL) for two regions; the West (Nio4 region; 160E-150W, 5S-5N), and East (Nio3 region; 150W-90W, 5S-5N). Compared with the pre-industrial control run, the seasonal cycle of precipitation (seas in Fig. 2) was increased in the West (Nio4 region) and decreased in the East (Nio3 region) in the Early Holocene. These pre cipitation changes follow SST changes (not shown). It indicates that the seasonal variability in insolation was in phase with the SST seasonal cycle in the West and out of phase, in the East. The SST seasonal cycle is further amplied by the east-west asymmetry of cloud feedback and the dynamic response of SST to anomalous westerly winds in the eastern equato rial Pacic (Luan et al. 2012). As a conse quence, seasonality had a larger eect on SSTs than do changes in interannual ENSO variability, even in the east Pacic (Braconnot et al. 2012). With regards to precipitation, we note a reduction of larger absolute mag nitude associated with El Nio and La Nia events in both the Nio3 and Nio4 regions when compared with the preindustrial control runs (Fig. 2). Thus, both seasonality and interannual variability damp the SST and precipitation variations in the eastern Pacic, whereas seasonality enhance and variability damp precipita tion in the central Pacic. Our analyses have implications when interpreting records of past ENSO variability across dierent regions. In the eastern Pacic, Early Holocene seasonal ity and interannual SST and precipita tion variability are reduced compared with SST and precipitation variability in the pre-industrial run. Therefore climate reconstructions, which are primarily de termined by the relative sensitivity of climate sensors to seasonality and inter annual variability, must take this into ac count in their calibrations in order to de rive robust reconstructions. In the central to western part of the basin, changes in seasonality and interan nual variability act in opposite directions, and our results suggest that only the nat ural archives that are sensitive to precipi tation would register large ENSO changes in the Early Holocene. Role of additional forcings The addition of a fresh water ux in the North Atlantic in the Early Holocene simulations leads to increased interan nual variability and a slight increase in seasonality (Fig. 2; EHwF) compared with the simulation in which only insolation is changed (Fig. 2; EHnF). This suggests the fresh water ux partially osets the changes due to insolation compared with the pre-industrial simulation (Fig. 2). This result is similar to the results of fresh water ux experiments under mod ern (Timmermann et al. 2007) or glacial (Merkel et al. 2010; see also Merkel et al., this issue) conditions. The presence of ice sheets (Fig. 2; EHIS) in the simulations, leads to a strong reduction in seasonality and a further damping of interannual pre cipitation variability compared with the insolation-only simulation. Particularly in the western Pacic, the results suggest that the remnant ice-sheets in the Early Holocene may have oset the amplica tion of precipitation seasonality. Towards a better understanding Our results show that the pattern of ENSO anomalies between the east and west Pacic is aected dierently by forc ings, but that SST variations during the Holocene were predominantly inuenced by changes in seasonality driven by the Earths orbital parameters such as insola tion. Linking the development of an El Nio event with changes in the seasonal evolution of the thermocline depth is a key factor explaining the damping of the simulated ENSO in the IPSL models (Luan et al. 2012). Our sensitivity experiments show that fresh water uxes partially counteract the insolation-driven season al phasing and the melting remnant ice sheets strongly aect the mean thermo cline depth and east-west gradient in SSTs (Luan et al. submitted). However, it seems that precipitation and SSTs do not neces sarily vary with the same relative strength on seasonal and interannual timescales when subject to the same sensitivity ex periments. These ndings suggest that a better understanding of the controls and timescales of variability is necessary to interpret paleo-records of past ENSO vari ability correctly, and that paleo-records should be used with caution to test how well models reproduce changes in ENSO variability. Selected references Full reference list online under: Braconnot P, Luan Y, Brewer S, Zheng W (2012) Climate Dynamics 38: 1081-1092 Clement AC, Seager R, Cane MA (2000) Paleoceanography 15(6): 731-737 Merkel U, Prange M, Schulz M (2010) Quaternary Science Reviews 29(1-2): 86-100 Luan Y et al. (2012) Climate of the Past 8: 1093-1108 Zheng W et al. (2008) Climate Dynamics 30: 745-762 Figure 2: Sensitivity experiments: Diagrams of precipitation (mm d -1 ) averaged over the Nio3 and Nio4 regions in the Pacic Ocean. Groups of columns show the magnitude of the seasonal cycle and the peak (December) mag nitude of El Nio and La Nia events for each of the simulations discussed in the text: pre-industrial (CTRL), Early Holocene (9.5 ka BP, EHnF), Early Holocene with a fresh water ux mimicking ice sheet melting in the North Atlantic (EHwF) and Early Holocene with the presence of remnant Laurentide and Fennoscandian ice sheets (EHIS). See Braconnot et al. (2012) for details on the methodology and statistical signicance. The magnitude of the seasonal cycle is computed as the dierence between maximum and minimum monthly values at the annual time scale. El Nio and La Nia anomalies correspond to the value at the peak of the event in December.


66 A controversial insight into Southwest Pacic midHolocene seasonality: from corals to models CLAIRE E LAZARETH 1 M.G. BUSTAMANTE ROSELL 1 N. DUPREY 1 N. PUJOL 2 G. CABIOCH 1, S. CAQUINEA U 1 T CORRGE 2 F LE CORNEC 1 C. MAES 3 M. MANDENGYOGO 1 AND B. TURCQ 1 1 IPSL/LOCEAN, UPMC/CNRS/IRD/MNHN, Institut de recherche pour le dveloppement France Nord, Bondy, France; 2 University of Bordeaux, Environnements et Paloenvironnements Ocaniques et Continentaux, Talence, France; 3 Laboratoire dEtudes en Go physique et Ocanographie Spatiales, Institut de Recherche pour le Dveloppement, Toulouse, France; To Guy, in memoriam Seasonal amplitude was higher in the Southwest Pacic during the mid-Holocene, say the corals. It was not, reply the models. More work is needed, agree the researchers I n the Southwest (SW) Pacic, the sea sonal changes in seawater surface char acteristics, such as temperature (SST) and salinity (SSS), are governed mainly by the position and intensity of the South Pacic Convergence Zone (SPCZ) and by the occurrences of El Nio and La Nia events. The SPCZ is a southeast narrow cloudy belt that inuences the wind and rain conditions from Papua New Guinea to French Polynesia (Trenberth 1976). During the austral summer, the SPCZ moves southwest whereas dur ing the austral winter it moves north. Consequently, waters in the SW Pacic are generally warmer and less saline dur ing the austral summer, and colder and saltier in winter. At the interannual time scale, the main changes in SST and SSS in the SW Pacic relate to El Nio-Southern Oscillation (ENSO) dynamics, with a peri odicity of ca. 2-7 years (Trenberth 1976). During an El Nio event, warm waters of the equatorial West Pacic (the Warm Pool) and connected precipitation mi grate toward the central and eastern Pacic. Consequently, the SW Pacic SST decreases slightly while precipitation de creases strongly, leading to higher SSS (e.g. Delcroix 1998). The opposite occurs during La Nia events. Therefore, knowl edge of past water characteristics at a seasonal resolution has the potential to provide information on the position and intensity of the SPCZ as well as on the oc currence of ENSO events. Mid-Holocene (6 ka BP) hydrographic proxy data from the SW Pacic are rare. However, this is a key period, characterized by a change in ENSO amplitude that is unfortunately not yet entirely understood. Corals and numerical models: tools to look back in time Past seasonal data on SST and SSS can be obtained from archives such as massive coral skeletons. The aragonitic skeleton of coral is secreted by polyps over several decades or centuries, at a rate of around 1 cm year -1 for massive forms. The chemi cal composition of skeletal aragonite re ects the properties of the water in which the coral has lived. In tropical areas, stud ies focus on the massive Porites sp. cor als. The Strontium/Calcium (Sr/Ca) ratio (Corrge 2006) is a robust proxy for SST in these corals. The stable oxygen isotopic ratio ( 18 O) is used, combined with Sr/Ca, to reconstruct the isotopic composition of surface seawater ( 18 O sw ), which is in turn closely related to SSS (via the evapo ration vs. precipitation budget). Climate models, based on current knowledge of the various compartments of the Earth system, help understand modern climate variability and predict future changes. Climate models with ex ternal forcings dierent to current ones (e.g. dierent orbital forcings, ice sheets, greenhouse gas concentrations) can be used to simulate past climatic changes. Many model simulations exist for the mid-Holocene (e.g. 19 in the Paleoclimate Modelling Intercomparison Project [PMIP2] database; http://pmip2.lsce.ipsl. fr/; Braconnot et al. 2007). Their main forcing is a millennial-scale change in the seasonality of insolation. For the SW Pacic region, the insolation for January to March was lower than at present while it was higher for August to October. To gain an insight into SW Pacic mid-Holocene mean climate at sea sonal resolution and hence into ENSO characteristics, we studied corals from New Caledonia and Vanuatu which have been dated to 5.5 ka BP and 6.7-6.5 ka Figure 1: Comparison between coral (blue) and model run (orange) results in terms of sea surface temperature (SST; C) characteristics in the New Caledonia region for the present (light colors) and the mid-Holocene (dark colors) situations. A) Monthly coral time series from New Caledonia, at present (light blue, Stephans et al. 2004; Stephans et al. 2005) and from a coral fossil from the mid-Holocene (5.5 ka BP, dark blue; Lazareth et al. 2013). B) Monthly time series from the CCSM PMIP2-models for the New Caledonia region (160-164E 20-24S), at present (light orange) and at the mid-Holocene (6 ka BP; dark orange; Otto-Bliesner et al. 2006). C, D) Monthly SST histograms and calculated seasonal amplitude.


67 BP respectively. A ~1 cm-thick slab, cut along the axis of maximum growth of the coral colonies, was X-rayed to reveal the annual growth bands. To ensure the skeleton was well preserved, pieces were collected and analyzed for their mineral ogy using X-ray diraction, and for their microstructure using scanning electronic microscopy. If the skeleton preserva tion was found to be satisfactory, the slab was continuously sampled at 1-mm steps, providing on average one sample per month of coral growth. The samples were then dissolved and analyzed to de termine their chemical composition. The New Caledonia coral was too short (~20 years) to investigate ENSO dynamics and as the 18 O proved to be partly altered, only the SST reconstructed with Sr/Ca will be discussed here. The 6 ka BP PMIP2 model simulations show a cooling of the tropical Pacic SST (Zheng et al. 2008) and a decrease in sea sonal SST amplitude (south of 10S and between 160-260E for the South Pacic) related to the insolation change. The re sults obtained on the New Caledonia coral were compared with the six simula tions for which monthly data were avail able and which correctly reproduced the current SST cycle in the New Caledonia region on four 2 by 2 grid points (164168E, 20-24S). The SPCZ at 6 ka BP: Where was it located? In the New Caledonia lagoon, the midHolocene SST seasonal amplitude as seen by the corals was higher than nowadays (Fig. 1a). This dierence is mainly due to colder mid-Holocene winters. In the New Caledonia region, the histogram of both the modern and mid-Holocene coral-reconstructed SST monthly values (Fig. 1c), and the 0 ka BP model outputs (Fig. 1d) show two modes, correspond ing to the positions of the SPCZ in win ter (July-August-September, JAS) and in summer (December-January-February, DJF). In the 6 ka BP coral results however, colder temperatures prevail in the winter mode, with a wider distribution. We in terpret this as reecting a more variable position and/or a weakening of the SPCZ during mid-Holocene winters. None of the PMIP2 models for the New Caledonia region reproduce an increase in seasonal SST amplitude in the mid-Holocene and the bi-modality is maintained (Fig. 1d). The models generally follow the insola tion change: a warmer winter and a cold er summer, leading to reduced seasonal ity (Fig. 1b). For the Vanuatu corals, Sr/Ca and 18 O were measured as proxies for SST and SSS, respectively, and data from the longest colony were used to highlight ENSO characteristics during the early mid-Holocene. The precipitation regime in Vanuatu at 6.7-6.5 ka BP was dier ent from the current one. Fossil corals indicate peaking SSS in summer (DJF) whereas low SSS would have been ex pected from the reduced summer insola tion at that time. Indeed, summer today is accompanied by strong precipitation brought by the southward displacement of the SPCZ. The reconstructed high mid-Holocene SSS suggests that in the SW Pacic at 6.7-6.5 ka BP the SPCZ was located further north than nowadays (Duprey et al. 2012). At that time, ENSO variability was reduced by 20-30% com pared to the modern ENSO. It remains uncertain, however, whether the reduced ENSO variability reects a real trend in ENSO dynamics or if it resulted from a weaker coupling between the precipita tion regime and the SPCZ. Are proxy data and models compatible for 6 ka BP? While coral records suggest a reduced SPCZ inuence in the SW Pacic, possibly from 6.7-6.5 to 5.5 ka BP, the PMIP2 model maps of precipitation reveal only a small shift of the SPCZ towards the northeast and a decrease in associated precipita tion during the winter months (Fig. 2b). These small changes, although in the right direction, are, however, not su cient to simulate an increase in the SST seasonality in the SW Pacic region at 6 ka BP. The modeled insolation-driven hemispheric change in seasonality is not reected in the SW Pacic proxy data. This suggests the models have diculty in reproducing mid-Holocene changes in coupled ocean-atmosphere circulation in this region. This could be due to known model biases in representing the current, and thus also the 6 ka BP, SPCZ, and to the models large grid size to which the SPCZ seasonal displacement is sensitive. On the other hand, corals are shallow wa ter organisms and the SST and SSS they record may not be valid for open oceans. Clearly, more data and new model runs are needed to understand the amplitude and geographical pattern of western Pacic mid-Holocene changes. Selected references Full reference list online under: Braconnot P et al. (2007) Climate of the Past 3: 261-277 Corrge T (2006) Palaeogeography Palaeoclimatology Palaeoecology 232: 408-428 Delcroix T (1998) Bulletin de l'Institut Franais d'Etudes Andines 27: 475-483 Duprey N et al. (2012) Paleoceanography 27, doi:10.1029/2012PA002350 Trenberth KE (1976) Quarterly Journal of the Royal Meteorological Society 102: 639-653 Figure 2: Multi-model mean maps for the winter season (July-August-September, JAS). Six PMIP2 simulations having similar resolution and reproducing correctly the mean seasonal cycle in the New Caledonia region have been used. The arrow points to New Caledonia. A) SST multi-model maps at 0 ka (left) and dierence (6 ka 0 ka; right). B) Multi-model mean precipitation eld at 0 ka (left) and 6 ka (right). Modied from Lazareth et al. (2013).


68 Interannual variability in the tropical Pacic and associated atmospheric teleconnections during the last glacial period UTE MERKEL, M. PRANGE AND M. SCHULZ MARUM Center for Marine Environmental Sciences, University of Bremen, Germany; Simulations of the climate of Marine Isotope Stages 2 and 3 suggest pronounced ENSO variability during the Heinrich Stadial 1 period when the Atlantic overturning circulation was weaker. Our model results also highlight the nonstationarity of ENSO teleconnections through time. A well-known example of coupled ocean-atmosphere interaction on interannual timescales is the El NioSouthern Oscillation (ENSO) phenom enon in the tropical Pacific. Consensus is still lacking about how ENSO will behave under future climate conditions, even in the latest generation of comprehensive climate models (Guilyardi et al. 2012). Major goals of paleoclimatic re search are to provide constraints on the possible range of changes in response to modified boundary conditions and to identify possible feedback and ampli fication mechanisms in the climate sys tem. In this context, climate models are valuable tools for investigating different climate scenarios that have occurred in the past. Most ENSO studies, however, are limited to the two priority periods specified by the Paleoclimate Modeling Intercomparison Project (PMIP): the MidHolocene, and the Last Glacial Maximum (LGM; e.g. Zheng et al. 2008). However, proxy data from the tropical Pacific (e.g. Stott et al. 2002; Leduc et al. 2009; Dubois et al. 2011) suggest that ENSO also changed on millennial timescales, e.g. in association with pronounced abrupt climate changes related to the Dansgaard-Oeschger stadials and inter stadials during Marine Isotope Stage 3 (MIS3, 59-29 ka BP). Modeling ENSO for different glacial climate states The first modeling studies that ad dressed MIS3 were limited to intermedi ate complexity models (e.g. Ganopolski and Rahmstorf 2001; van Meerbeeck et al. 2009; Ganopolski et al. 2010), a re gional model approach (e.g. Barron and Pollard 2002), or an atmosphere-only setup (Sima et al. 2009), unsuitable to capture the complexity of ENSO dynam ics. Recently, the first simulations of MIS3 climate in a comprehensive coupled cli mate model were accomplished with the US National Center of Atmospheric Research's CCSM3 model (Merkel et al. 2010). The study used a timeslice approach with a focus on a period of relatively regular Dansgaard-Oeschger variability around 35 ka BP. When 35 ka BP boundary conditions (greenhouse gas concentrations, orbital parameters, continental ice sheet distributions) are prescribed, the model simulates a very weak Atlantic meridional overturning circulation (AMOC) of about 7 Sv, which is much weaker than the preindustrial value of 12 Sv, but also weaker than the ~10 Sv simulated for the LGM. Therefore, we consider the simulated 35 ka BP cli mate as a stadial climate state. The coun terpart of an interstadial climate state is induced in the model by a 0.1 Sv fresh water extraction from the North Atlantic over ~300 model years, thereby forcing a resumption of the AMOC to ~14 Sv. Our set of experiments also includes a simulation of a Heinrich Stadial 1 sce nario. This is set up by imposing a fresh water perturbation of about 0.2 Sv to a simulated LGM ocean state over 360 model years. This is motivated by ear lier studies which mimic past Heinrich events by hosing freshwater into the modern ocean and thereby demonstrate that a slowdown of the AMOC may have a pronounced impact on the tropical Pacific (e.g. Timmermann et al. 2007). One of our major findings was that interannual (about 1.5-8 years) variabil ity in sea surface temperatures (SST) of the eastern tropical Pacific was distinctly increased in our Heinrich Stadial 1 simu lation compared to pre-industrial times, whereas variability in our LGM and MIS3 simulations was systematically reduced, albeit only weakly (Fig. 1). Modern ENSO dynamics studies show that stronger ENSO variability is dynamically linked to a weaker annual cycle of SST and to a weaker meridional asymmetry of SST across the equator in the eastern tropi cal Pacific (Guilyardi 2006; Xie 1994). Our model results show that these relation ships also hold for the different simu lated glacial climate states. In particular, our Heinrich Stadial 1 simulation exhib its a much weaker north-south contrast in eastern tropical Pacific SST than under Figure 1: ENSO variability of sea surface temperature in the eastern tropical Pacic (Nio3 region: 150W-90W, 5S-5N) for dierent simulated climatic states.


69 modern conditions. This is attributed to an atmospheric signal communication from the strongly cooled North Atlantic into the tropical Pacific. Model-data comparison Further insights into tropical Pacific vari ability can be achieved through modeldata intercomparison. Felis et al. (2012) present findings from a fossil coral re trieved during IODP Expedition 310 near Tahiti. The coral has been dated to Heinrich Stadial 1. Its fast growth rate allows sampling at monthly resolution and provides a unique opportunity to investigate interannual SST variability in the southwestern tropical Pacific dur ing that period. The coral record exhib its pronounced variability at interan nual ENSO frequencies during Heinrich Stadial 1, consistent with the basin wide increase in ENSO variability in our Heinrich Stadial 1-analogue simulation. At the Tahiti location, the coral and the model are also quantitatively consis tent, as both suggest a strengthening of interannual SST variability by 20-30% compared to modern conditions. Modern and past ENSO teleconnections Modern ENSO is well known for its atmo spheric teleconnections of near-global extent. Understanding how teleconnec tions operate, both in the atmosphere and the ocean, is particularly relevant for the validity of paleoclimatic reconstruc tions, as they generally assume that at mospheric teleconnection patterns are stable. This may be particularly critical in the interpretation of proxy records not stemming from the core ENSO region. A composite analysis of atmospheric patterns (e.g. of sea level pressure) dur ing all El Nio and La Nia events in the different simulations revealed obvious deviations from the modern spatial dis tribution of anomalies (Fig. 2). In par ticular, the teleconnections to the North American continent and the North Atlantic region seem to be strongly al tered in terms of amplitude and spatial structure in the LGM and MIS3 simula tions. This difference is probably caused by the presence of the glacial continen tal ice sheets and the glacial cooling of the North Atlantic, which both affect the position of the upper-tropospheric jetstream and atmospheric storm tracks, and thus the tropical-extratropical sig nal propagation. The typical intensifi cation of the Aleutian low forced by El Nio (Fig. 2a) seems to be present dur ing the LGM but is less pronounced, and the atmospheric bridge to the North Atlantic region seems to be interrupted, as no clear large-scale pattern is simulat ed there (Fig. 2b). The MIS3 stadial con ditions (Fig. 2c) bear more resemblance to the control simulation over the North Pacific, whereas over the North Atlantic, the ENSO influence is clearly reduced, similar to the LGM situation. This points to a complex interplay of atmospheric dynamics with the various forcings in the different climatic states. The need to learn more about glacial climatic states In summary, our modeling study con firms that ENSO variability responds to various glacial climatic states. However, ENSO variability does not appear to be linearly linked to the strength of the AMOC. This calls for more detailed analyses, for instance in the form of glacial hosing studies in a multi-model approach (Kageyama et al. 2013). The different roles of the AMOC and the vari ous glacial boundary conditions with respect to their impact on ENSO need to be further disentangled. Likewise, we emphasize that the concept of station ary teleconnections should only be ap plied to past climatic states with caution as they may be altered by different past boundary conditions and forcings inter nal and external to the climate system. Selected references Full reference list online under: Felis T et al. (2012) Nature Communications 3, doi:10.1038/ncomms1973 Guilyardi E et al. (2012) CLIVAR Exchanges 58(17): 29-32 Kageyama M et al. (2013) Climate of the Past 9: 935-953 Merkel U, Prange M, Schulz M (2010) Quaternary Science Reviews 29: 86-100 Zheng W et al. (2008) Climate Dynamics 30: 745-762 Figure 2: ENSO teleconnections during boreal winter (Dec.-Feb.): El Nio minus La Nia composites of sea level pressure [hPa] for (A) pre-industrial control climate, (B) Last Glacial Maximum (LGM), and (C) Marine Isotope Stage 3 (MIS3) stadial climate. Figure modied from Merkel et al. 2010.


70 ENSO behavior before the Pleistocene CHRIS BRIERLEY Department of Geography, University College London, UK; The Pliocene was characterized by a weak equatorial sea surface temperature gradient in the Pacic, confusingly reminiscent of that seen eetingly during an El Nio. Data also show interannual variability in the Pliocene, raising questions about ENSOs dependence on the mean climate state. T he behavior of the El NioSouthern Oscillation (ENSO) in pre-Pleistocene climates is highly uncertain. This uncer tainty is rooted in a fundamental lack of evidence. However, several, recent studies focusing on past warm climates are begin ning to address this issue. These studies were motivated by suggestions that the climate of the early Pliocene was a per manent El Nio a term that has led to much confusion. After looking at the new evidence for ENSO, I will discuss the histo ry of the term permanent El Nio, before suggesting that it should be consigned to history as well. Pre-Pliocene ENSO Detection of interannual variability re quires paleoclimate indicators that moni tor changes over short timescales such as the thickness of varved sediments and isotope ratios in long-lived fossil mollusks or corals. Once a record spanning sucient years has been recovered, its power spectra can be analyzed for frequencies represen tative of ENSO. ENSO has been detected during the warm intervals of the Miocene (5.96-5.32 Ma; Galeotti et al. 2010), Eocene (45-48 Ma; Huber and Caballero 2003; Lenz et al. 2010) and Cretaceous (70 Ma; Davies et al. 2011, 2012) from layered deposits or varved sediments showing a strong peak in the 3-5 years range. It has also been seen in the Eocene (50 Ma; Ivany et al. 2011) from analysis of power spectra of carbon iso topes in fossil driftwood and bivalves. All of these analyses have used records gath ered in locations far away from the tropical Pacic, such as Antarctica. However, ENSO teleconnections depend on the mean cli mate state (Merkel et al., this issue) and may therefore have been dierent in these early warm periods. The plausibility of as sumed teleconnections of that time can be conrmed with climate model simulations of the period (Galeotti et al. 2010; Huber & Caballero 2003; Ivany et al. 2011). Pliocene ENSO Two studies have found evidence of ENSOstyle periodicity during the Pliocene (Fig. 1) from locations in the Tropical Pacic. Oxygen isotope records from fossil cor als (MacGregor et al., this issue) from the Philippines show power spectra similar to recent corals, hence likely representing ENSO variability (Watanabe et al. 2011). Analyses of individual foraminifera from the Eastern Equatorial Pacic (Scroxton et al. 2011) nd several instances of isotopic compositions outside the range predicted for the present-day seasonal cycle. This has been interpreted as showing an active ENSO cycle. Unfortunately, a foraminifer does not live through an annual cycle (un like mollusks; e.g. Carr et al., this issue), so changes in the seasonal cycle (Braconnot and Luan, this issue) are a potential source of uncertainty. Despite the complications associ ated with each individual study, a picture is emerging in which ENSO is a pervasive feature of past climate. However, a sys tematic eort will be needed to provide quantitative information from these prePliostocene intervals that could qualify for data-model comparisons. Figure 1: Variation in the Sea Surface Temperature in the Equatorial Pacic over the past ve million years. The estimates come from two ocean cores in the West (ODP 806 at 150E) and the East (ODP 847 at 95W). Two dierent types of record are used to reconstruct the temperatures: Mg/Ca (Wara et al. 2005) and alkenones (Dekens et al. 2007; Pagani et al. 2010). The period described as lacking ENSO variability (i.e. the period of the permanent El Nio) from a mistaken interpretation of Fedorov et al. (2006) is shown in orange. Times with observed ENSO variability, as found by Scroxton et al. (2011) and Watanabe et al. (2011), are marked in green. The time-slab used by PlioMIP and its precursors is marked as mPWP. Figure modied after Fedorov et al. (2013).


71 The permanent El Nio of the Early Pliocene One of the factors fuelling the hunt for ENSO in pre-Pleistocene warm climates was the idea that the early Pliocene was in state of permanent El Nio. Mg/Ca records (a proxy for sea-surface tempera tures, SST) from the Western and Eastern Equatorial Pacic (Wara et al. 2005) sug gest that no temperature gradient existed along the equatorial Pacic around four million years ago (4 Ma; Fig. 1). Subsequent work shows similar results for the equa torial SST gradient in the early Pliocene. Reconstructions of the SST gradient during older periods need further work, but pre liminary data suggest that a reduced SST gradient is not solely a feature of the early Pliocene (LaRiviere et al. 2012) although it may not be a ubiquitous feature of all warm climates (Nathan and Leckie 2009). An El Nio event (the warm phase of the ENSO oscillation) is characterized by a lack of SST gradient along the equatorial Pacic. Although Wara et al. (2005) empha sized that their Mg/Ca records reected the mean climatic state, i.e. a change in the long-term average climate rather than a change in interannual variability, they de scribed their observation using the short hand of a permanent El Nio. The term was propagated by Fedorov et al. (2006), who used model simulations to examine how such a state could be maintained. They emphasized the similarity between the long-term average state and the condi tions seen during recent El Nios. This sim ile has been read as an assertion that there was no ENSO variability before 3 Ma, al though this is not what the authors intend ed (Alexey Fedorov, personal communica tion) and is certainly not what is shown by the more recent studies described above. Equatorial SST gradient and ENSO in models The mid-Pliocene warm period (3.3-3.0 Ma; marked mPWP in Fig. 1) has been the focus of sustained eort by the data and modeling communities, most re cently under the auspices of the Pliocene Model Intercomparison Project (PlioMIP; Dolan et al. 2012). This period is one mil lion years later than the minimal SST gradi ent identied by Wara et al. (2005), but is thought to share similar climate forcings (Fig. 1). Haywood et al. (2007) found ENSO variability in both mid-Pliocene and mod ern simulations. However, the equatorial temperature gradient of the mid-Pliocene simulation was hardly smaller than in the modern simulation. Subsequent simula tions performed with updated boundary conditions (Dowsett et al. 2010), similarly show a lack of strong reductions in the equatorial temperature gradient between the mPWP and the modern day (Haywood et al. 2013) in comparison with the halv ing seen in the paleo-observations (Fig. 1). Attempts have been made to force coupled models to replicate a mean state with a weak SST gradient in the equato rial Pacic. One approach has been to in crease the background ocean vertical dif fusivity (Brierley et al. 2009), potentially to represent a changed tropical cyclone distribution (Fedorov et al. 2010). These simulations (Fig. 2) appear to show a rela tionship between equatorial SST gradient and the amplitude of ENSO, but not its pe riod (Fedorov et al. 2010). This result could easily be model dependent, but oers a scenario for a weak ENSO around 4.2 Ma, the time for which proxy data suggest that the SST gradient was very small (open or ange box in Fig. 1). Conclusion The introduction of the term permanent El-Nio in the literature has caused confu sion, but it has also motivated paleoclima tologists to look for (and nd) evidence of interannual ENSO variability in deep time. The assertion that there was no ENSO vari ability before 3 Ma, wrongly attributed to Fedorov et al. (2006), is not true. However, the relationship between a Pacic mean state with a minimal equatorial SST gradi ent and related ENSO properties merits further investigation. We have made prog ress towards uncovering ENSO behavior on geologic timescales, but there is still a long way to go. Selected references Full reference list online under: Fedorov A, Brierley C, Emanuel K (2010) Nature 463: 1066-1070 Fedorov A et al. (2006) Science 312: 1485-1489 Scroxton N et al. (2011) Paleoceanography 26(2), doi: 10.1029/2010PA002097 Wara MW, Ravelo AC, Delaney ML (2005) Science 309: 758-761 Watanabe T et al. (2011) Nature 471: 209-211 Figure 2: Nio 3.4 SST anomalies in two model simulations; a control (green) and one with an equatorial SST gradient that is approximately halved (red). The shaded area represents four standard deviations from a 30-year running window (Fedorov et al. 2010).


72 72 PAGES news Vol 21 No 2 August 2013 Overview of data assimilation methods GREGORY J. HAKIM 1 J. ANNAN 2 S. BRNNIMANN 3 M. CR UCIFIX 4 T EDWARDS 5 H. GOOSSE 4 A. PA UL 6 G. V AN DER SCHRIER 7 AND M. WIDMANN 8 1 Department of Atmospheric Sciences, University of Washington, Seattle, USA; 2 Research Institute for Global Change, JAMSTEC, Yokohama Institute for Earth Sciences, Japan; 3 Institute of Geography and Oeschger Centre for Climate Change Research, University of Bern, Switzerland; 4 Earth and Life Institute and Georges Lematre Centre for Earth and Climate Research, Universit catholique de Louvain, Belgium; 5 Department of Geographical Sciences, University of Bristol, UK; 6 MARUM and Department of Geosci ences, University of Bremen, Germany; 7 KNMI, De Bilt, The Netherlands; 8 School of Geography, Earth and Environmental Sciences, University of Birmingham, UK We present the data assimilation approach, which provides a framework for combining observations and model simulations of the climate system, and has led to a new eld of applications for paleoclimatology. The three subsequent articles explore specic applications in more detail. D ata assimilation involves the combi nation of information from observa tions and numerical models. It has played a central role in the improvement of weather forecasts and, through reanalysis, provides gridded datasets for use in cli mate research. There is growing interest in applying data assimilation to problems in paleoclimate research. Our goal here is to provide an overview of the methods and the potential implications of their applica tion. Understanding of past climate vari ability provides a crucial benchmark ref erence for current and predicted climate change. Primary resources for deriving past understanding include paleo-proxy data and numerical models, and studies using these resources are typically per formed independently. Data assimilation provides a mathematical framework that combines these resources to improve the insight derivable from either resource in dependently. The three articles that fol low describe the current activity in this emerging eld of study: transient state estimation (Brnnimann et al., this issue), equilibrium state estimation (Edwards et al., this issue), and paleo data assimilation for parameter estimation (Annan et al., this issue). Here we provide an overview of these methods and how they relate to existing practices in the paleoclimate community. In weather prediction, data assimilation uses observations to initialize a forecast (Lorenc 1986; Kalnay 2003; Wunsch 2006; Wikle and Berliner 2007). Since the shortterm forecast typically starts from an accu rate analysis at an earlier time, called the prior estimate, the model provides rela tively accurate estimates of the weather observations. Data assimilation involves optimizing the use of these independent estimates to arrive at an analysis (i.e. esti mate of the weather or climate state) with a smaller error than the model short-time forecast or the observations. For Gaussian distributed errors, the result for a single scalar variable (singlydimensioned variable of one size), x given prior estimate of the analysis value, x p and observation y is (1) where x a is the analysis value. The inno vation xxxxxxxxx, represents the infor mation from the observation that diers from the prior estimate. This comparison requires a conversion of the prior to the observation, which is accomplished by xx. For example, in a paleoclimate applica tion, xxxxx may estimate tree-ring width derived temperature data from a climate model (Fig. 1). The weight applied to the innovation is determined by the Kalman gain, K (2) where cov represents a covariance. The error variances associated with the ob servation and the prior estimate of the observation are given by p and y re spectively. Equation (1) represents a linear regression of the prior on the innovation Figure 1: Schematic illustration of how the innovation is determined in data assimilation for a tree-ring example. Proxy measurements are illustrated on the left, and model estimates of the proxy on the right. The observation operator provides the map from gridded model data, such as temperature, to tree-ring width, which is used to compute the innovation. Images credit: Wikipedia. Editors: H. Goosse and A. Paul


73 73 PAGES news Vol 21 No 2 August 2013 (the denominator of K is the innovation variance). Equivalently, the Kalman gain weights the innovation against the prior, resulting in an analysis probability density function with less variance, and higher density, than either the observation or the prior (Fig. 2, red solid line, dashed green line and dashed blue line respectively). Generalizing (1) and (2) to more than one variable is straightforward, with scalars becoming vectors and variances becom ing covariance matrices (for details see Brnnimann et al., this issue). These cova riance matrices provide the information that spreads the innovation in space and to all variables through a Kalman gain ma trix. Application of data assimilation to the paleoclimate reconstruction problem involves determining the state of the cli mate system on the basis of sparse and noisy proxy data, and a prior estimate from a numerical model (Widmann et al. 2010). These data are weighted according to their error statistics and may also be used to calibrate parameters in a climate model (Annan et al. 2005). Relationship to established methods While there are similarities between the application of data assimilation to weather and paleoclimate, there are also important dierences. In weather predic tion, observations are assimilated every 6 hours, which is a short time period com pared to the roughly 10-day predictabil ity limit of the model. However, transient state estimation in paleoclimatology in volves proxy data having timescales of years to centuries or longer, which gener ally exceeds the predictability of climate models, which are on the order of a de cade. Consequently, relative errors in the model estimate of the proxy are usually much larger in paleoclimate applications. Hoever, data assimilation reconstruction may still be performed, at great cost sav ings, since the model no longer requires integration and each assimilation time may be considered independently (Bhend et al. 2012). Paleoclimate data assimilation at tempts to improve upon climate eld reconstructions that use purely statisti cal methods. One well-known statistical approach for climate eld reconstruction (Mann et al. 1998; Mann et al. 2008) in volves limiting eld variability to a small set of spatial patterns that are related to proxy data during a calibration period. Data assimilation, on the other hand, retains the spatial correlations for loca tions near proxies, which may be lost in a small set of spatial patterns, and also spreads information from observations in time through the dynamics of the climate model. Another distinction between data assimilation and eld reconstruction ap proaches concerns the observation op erator, xx, which often involves biological quantities of proxy data that have uncer tain relationships to climate. Statistical re constructions directly relate proxy data to the set of spatial patterns, which is essen tially an empirical estimate of the inverse of xx, and therefore subject to similar un certainty. Current and future directions Research on paleoclimate data assimila tion is rapidly developing in many areas. For climate state estimates, a wide range of methods are currently under explora tion (see Brnnimann et al., this issue), in cluding nudging climate models to largescale patterns derived from proxy data (Widmann et al. 2010), and variational (Gebhardt et al. 2008) and ensemble ap proaches (Bhend et al. 2012). Ensemble approaches involve many realizations of climate model simulations, each of which is weighted according to their match to the proxy data, either in the selection of members (Goosse et al. 2006) or through a linear combination. Among the important obstacles to progress in paleoclimate data assimila tion, some challenges are generic, such as improving the chronological dating quality of proxy records and reducing the uncertainties of the paleoclimate data. Other problems are more specic to data assimilation, such as the development of proxy forward models. Moreover, proxy data typically represent a time average, in contrast to instantaneous weather obser vations, although solutions that involve assimilating time averages have been pro posed to tackle this problem (Dirren and Hakim 2005; Huntley and Hakim 2010). Model bias is also problematic for paleo climate data assimilation, especially for regions with spatially sparse proxy data. While the eld of paleoclimate data assimilation is still in its infancy, these challenges are all under active research. Merging climate models and proxy data has a bright future in paleoclimate re search (e.g. the P2C2 program of the U.S. National Science Foundation), and it is likely that paleoclimate data assimilation will play a central role in this endeavor. Selected references Full reference list online under: Annan J, Hargreaves J, Edwards N, Marsh R (2005) Ocean Modelling 8(1): 135-154 Bhend J et al. (2012) Climate of the Past 8: 963-976 Goosse H et al. (2006) Climate Dynamics 27: 165-184 Widmann M et al. (2010) Climate of the Past 6: 627-644 Wikle CK, Berliner ML (2007) Physica-D 230: 1-16 Figure 2: Data assimilation for scalar variable x assuming Gaussian error statistics. Prior estimate, given by the dashed blue line, has mean -0.25 and variance 0.5. Observation y given by the dashed green line, has mean 1.0 and variance 0.25. The analysis, given by the thick red line, has mean 0.58 and variance 0.17. The parabolic gray curve denotes a cost function, J which measures the mist to both the observation and prior; it takes a minimum at the mean value of x a From Holton and Hakim 2012.


74 74 PAGES news Vol 21 No 2 August 2013 Transient state estimation in paleoclimatology using data assimilation STEFAN BRNNIMANN 1 J. FRANKE 1 P. BREITENMOSER 1 G. HAKIM 2 H. GOOSSE 3 M. WIDMANN 4 M. CR UCIFIX 3 G. GEBBIE 5 J. ANNAN S AND G. V AN DER SCHRIER 7 1 Institute of Geography and Oeschger Centre for Climate Change Research, University of Bern, Switzerland; 2 Department of Atmospheric Sciences, University of Washington, Seattle, USA; 3 Earth and Life Institute and Georges Lematre Centre for Earth and Climate Research, Universit catholique de Louvain, Belgium; 4 School of Geography, Earth and Environmental Sciences, University of Birmingham, UK; 5 Department of Physical Oceanography, Woods Hole Oceanographic Institution, USA; 6 Research Institute for Global Change, JAMSTEC, Yokohama Institute for Earth Sciences, Japan; 7 KNMI, De Bilt, The Netherlands Data assimilation methods used for transient atmospheric state estimations in paleoclimatology such as covariance-based approaches, analogue techniques and nudging are briey introduced. With applications diering widely, a plurality of approaches appears to be the logical way forward. R eliable estimations of past climate states are the foundations of paleoclimatology. Traditionally, statistical reconstruction tech niques have been used, but recent develop ments bring data assimilation techniques to the doorstep of paleoclimatology. Here we give a short overview of transient atmo spheric state estimation in paleoclimatology using data assimilation. An introduction to data assimilation as well as applications for equilibrium state estimation and parameter estimation are given in the companion pa pers to this special section (see also Wunsch and Heimbach 2013). Data assimilation combines information from observations with numerical models to obtain a physically consistent estimate (termed analysis) of the climate state. It has been hugely successful in generating three-dimensional atmospheric data sets of the past few decades. The Twentieth Century Reanalysis Project (Compo et al. 2011) extended the approach as far back as 1871, but there is a limit to further extension, because conventional data assimilation re lies on the availability of state observations. Paleoclimate proxies do not capture atmo spheric states, but time-integrated functions of states, such as averages, in the simplest case. Therefore, for assimilating proxies, other methods are required than those applied in atmospheric sciences. We briey present be low, three groups of assimilation methods for transient atmospheric state estimation in paleoclimatology: Classical covariancebased approaches such as the Kalman Filter or variational techniques; approaches based on analogues such as Particle Filters; and nudging techniques. A schematic view of these methods is given in Figure 1. Note that other methods may be used for the ocean (see Gebbie 2012). Covariance-based approaches The assimilation problem can be formulated as a cost function J, assuming Gaussian prob ability distributions: J( x ) = ( x x b ) T B -1 ( x x b ) + ( y H [ x ]) T R -1 ( y H [ x ]) (1) where x is the analysis, x b is a model forecast, y are the observations (or proxies), H is the observation operator that mimics the obser vation (or proxy) in the model space, B is the background error covariance matrix and R is the observation error covariance matrix (of ten assumed to be diagonal). The solution to (1) in the classical Kalman form is: x = x b + BH T ( R + HBH T ) -1 ( y Hx b ) (2) where H is the Jacobian of H Variational ap proaches can be used to approximate the so lution. In the Ensemble Kalman Filter (EnKF), B can be estimated from the ensemble, and each member is updated individually. Normally x is a state vector. However, Dirren and Hakim (2005) have successfully extended the concept to time averages. Data assimilation entails that x serves as an initial condition for the next forecast step. Focusing on the seasonal scale, Bhend et al. (2012) use the EnKF without updating the initial conditions (termed EKF here), which are no longer important on this scale (rather, predictability comes from the boundary conditions, including sea-surface tempera tures). This conveniently allows one to use pre-computed simulations. Because x does not serve as new initial condition, it can be small and can be a vector of averaged model states (e.g. all monthly averages of a season for three variables). H can be a simple proxy forward model, i.e. a time-integrated function of elements of x Covariance-based approaches are pow erful but computationally intensive and can be sensitive to assumptions (e.g. of Gaussian distributions), to the treatment of covariance matrices, or to the behavior of the observa tion operator. Figure 1: Schematic overview of assimilation approaches. Arrows denote steps in the procedure.


75 75 PAGES news Vol 21 No 2 August 2013 approaches require a huge pool of possible analogues (Annan and Hargreaves 2012). Nudging approaches Nudging approaches (Widmann et al. 2010) do not explicitly minimize a cost function. The distance between model state and observa tions is reduced by adding tendencies to (a subspace of) the model state at each time step, similar to an additional source term in the tendency equations. Following our nota tion: x = x b + G ( F [ y ] x ) (4) where F [ y ] represents the target eld (derived using up-scaling method F from observations or proxies y ) in the dimension of the model (sub-)space. G is a relaxation parameter. The Forcing Singular Vectors method (van der Schrier and Barkmeijer 2005) ma nipulates the tendency equations as well, but adds a perturbation, which modies the model atmosphere in the direction of the tar get pattern only. Examples Figure 2 shows April-to-September averages of surface air temperature obtained from two assimilations approaches (EKF and BEM) for the year 1810 relative to the 1801-1830 mean. Both approaches are based on the same ensemble of simulations described in Bhend et al. (2012). The ensemble consists of 30 simulations performed with ECHAM5.4 at a resolution of T63/L31 (ca. 2 x 2), with seasurface temperatures and external forcings as boundary conditions. The unconstrained ensemble mean (Fig. 2 top) shows the eect of boundary condi tions, here resulting in cooler than average summer temperatures following the large, but not yet localized volcanic eruption in 1809. Anomalies are small and smooth which is typical for an ensemble mean. The EKF analysis was constrained by historical instru mental observations using Eq. (2). The EKF en semble mean suggests a more pronounced cooling over northern Europe, but over most regions (due to lack of observations) it is close to the unconstrained ensemble mean. BEM was constrained with tree rings from 35 locations. The VS-lite tree growth model (Tolwinski-Ward et al. 2011) was used as H and Eq. (3) was minimized. BEM identi es member 01 as the best tting one. This member exhibits large anomalies in Alaska and Eurasia, but due to the small ensemble size little regional skill is expected (Annan and Hargreaves 2012). For instance, it does not t well with instrumental observations over Europe. The same member in the EKF analy sis (Fig. 2, bottom) shows a better correspon dence, but we loose the advantage of having the full 6-hourly model output available. Limitations and future directions Paleoclimatological applications are much more disparate than atmospheric sciences in terms of time, time scales, systems analyzed, and proxies used. Therefore, a plurality of data assimilation approaches is a logical way forward. However, all approaches still suer from problems and uncertainties. Ensemble approaches (PF, EnKF, EKF) provide some information on the methodological spread, which however represents only one (dicult to characterize) part of the whole uncertainty. Further uncertainties are related to model biases, limited ensemble size, errors in the forcings and proxy data. Validation of the ap proaches using pseudo proxies in toy models and climate models and validation of the re sults using independent proxies is therefore particularly important. Any approach, how ever, fundamentally relies on a good under standing of the proxies. Selected references Full reference list online under: Bhend J et al. (2012) Climate of the Past 8: 963-976 Dirren S, Hakim GJ (2005) Geophysical Research Letters 32, doi:10.1029/2004GL021444 Goosse H (2010) Journal of Geophysical Research 115, doi:10.1029/2009JD012737 van der Schrier G, Barkmeijer J (2005) Climate Dynymics 24: 355-371 Widmann M et al. (2010) Climate of the Past 6: 627-644 Figure 2: Northern hemisphere temperature anomalies for April to September 1810 (relative to 1801-1830) from the unconstrained ensemble mean, the EKF ensemble mean, BEM (member 01), and the EKF analysis for the BEM member. Circles indicate locations and anomalies of the assimilated instrumental measurements; red squares the locations of tree-ring proxies. Analogue approaches Reverting to cost function (1), we can also look for an existing x e.g. by choosing among dierent ensemble members. The cost func tion (1) reduces to: J( x ) = ( y H [ x ]) T R -1 ( y H [ x ]) (3) for x { x 1 x 2 ,..., x n } New ensemble members are then generated for the next time step by adding small pertur bations to x and the nal analysis is a continu ous simulation. The Particle Filter (PF, Goosse et al. 2010) approach uses a distribution of x to calculate a weighted sum of cost function contributions to (3). In the Proxy Surrogate Reconstruction approach (PSR, Franke et al. 2011) and the Best Ensemble Member approach (BEM, Breitenmoser et al., in preparation) pre-com puted simulations are used with { x 1 x 2 ,..., x n } denoting dierent slices of a long simulation for PSR or in the case of BEM, the same slice in an ensemble of simulations. The analysis in both cases is a sequence of short, discon tinuous simulations. In contrast to EnKF, H may be non-dierentiable (e.g. H can be a complex forward model driven by the full simulation output). R may be non-diagonal, and x may be very large (e.g. six-hourly model output over a 6 month period). However, to reconstruct the state of systems including a large number of degrees of freedom, these


76 76 PAGES news Vol 21 No 2 August 2013 Best-of-both-worlds estimates for time slices in the past TAMSIN L. EDWARDS 1 J. ANNAN 2 M. CR UCIFIX 3 G. GEBBIE 4 AND A. PA UL 5 1 Department of Geographical Sciences, University of Bristol, UK; 2 Research Institute for Global Change, JAMSTEC, Yokohama Institute for Earth Sciences, Japan; 3 Earth and Life Institute and Georges Lematre Centre for Earth and Climate Research, Universit catholique de Louvain, Belgium; 4 Department of Physical Oceanography, Woods Hole Oceano graphic Institution, USA; 5 MARUM Center for Marine Environmental Sciences and Department of Geosciences, University of Bremen, Germany We introduce data assimilation methods for estimating past equilibrium states of the climate and environment. The approach combines paleodata with physically-based models to exploit their strengths, giving physically consistent reconstructions with robust, and in many cases, reduced uncertainty estimates. T hose seeking to understand the Earths past usually take one of two approaches: reconstructing paleocli mate and paleo-environmental states from proxy data derived from natural archives such as ice cores and trees; or simulating them with earth system models that contain theoretical knowl edge of physical processes. Proxy-based reconstructions are based on observations of the real world, but most consider data points indepen dently rather than accounting for cor relations in space, time and between climate variables. Therefore they risk being physically inconsistent. Models incorporate aspects of physical consis tency, but are imperfect and are tested during development only with present day observations. Data assimilation produces bestof-both-worlds estimates that combine observational and theoretical informa tion while not ignoring their limita tions. We discuss data assimilation for estimating past equilibrium states of the earth system such as climate and vegetation. We use the term paleodata for measurement-based data: either the observations of proxies or the statistical reconstructions derived from them. Aims and methods Equilibrium state, or time slice, data assimilation is estimation of a snapshot in time during which it is assumed the state variables are not changing. The state estimates may be the primary sci entific aim or simply a bonus of cali brating model parameters (Annan et al., this issue). Time slice estimation is a natural starting point in data assimilation be cause it is more straightforward than es timating a transient state (Brnnimann et al., this issue) and is particularly ap propriate if spatial patterns are more important than temporal changes or if the model is computationally expen sive. For a given computational resource time slice estimation permits more complete exploration of model uncer tainties in parameters, structure, and inputs. Another advantage of a focus on time slices is that for eras studied by the Paleoclimate Model Intercomparison Project (PMIP) relatively large quantities of paleodata and simulations are avail able. Most data assimilation estimates of equilibrium paleo-states are there fore of the Last Glacial Maximum (LGM: 21 ka cal BP), the most recent era for which annual mean climate is substan tially different to the present that also has a long history of study by PMIP. We use model simulations in pa leo-state estimation because models provide links across different locations, times (relevant to transient or multistate estimation) and state variables. This has two advantages: it helps ensure the resulting state is physically consis tent, and it also means we are not limit ed to assimilating the same variables we wish to estimate. We could assimilate data in one place to estimate another, or assimilate temperature data to estimate precipitation, or assimilate variables corresponding to the outputs of a mod el to estimate variables corresponding to the inputs. The last are termed in version methods, such as estimating atmospheric variables or terrestrial car bon from paleo-vegetation records (e.g. Guiot et al. 2000; Wu et al. 2007; Wu et al. 2009; Pound et al. 2011) or estimat ing oceanic variables from paleo-tracer records (e.g. LeGrand and Wunsch 1995; Roche et al. 2004). Data assimilation requires the fol lowing ingredients: paleodata with un certainty estimates, simulations with uncertainty estimates, and a metric to quantify the dissimilarity, or distance, between the two. Climate state esti mates are obtained by searching for the simulation(s) closest to the paleodata (optimization) or calculating a weight ed combination of the two (updating). Distance is usually measured with the standard metric for normally dis tributed model-data differences, i.e. the sum of squared differences weighted by the uncertainties, though some use ad-hoc or fuzzy metrics (e.g. Guiot et al. 2000; Wu et al. 2007; Gregoire et al. 2010). For non-continuous variables, for example with a threshold, variables must be transformed or a non-Gauss ian metric chosen (e.g. Stone et al. 2013). Optimisation methods search for the simulation with the minimum dis tance from paleodata. One approach uses numerical differentiation of the model with respect to the parameters, essentially least-squares fitting of a line or curve to one-dimensional data (e.g. LeGrand and Wunsch 1995; Gebbie and Huybers 2006; Marchal and Curry 2008; Burke et al. 2011; Huybers et al. 2007; Paul and Losch 2012). Another approach generates an ensemble of simulations using many different parameter values and then selects the members with the smallest model-data distance (per turbed parameter ensemble methods; e.g. Gregoire et al. 2010). Updating methods combine model and paleodata estimates. Typically the model estimates are generated with a perturbed parameter ensemble, which permits well-defined sampling of pa rameter uncertainties; the model esti mates are reweighted with the modeldata distance using Bayesian updating (e.g. Guiot et al. 2000; Wu et al. 2007; Wu et al. 2009; Holden et al. 2009; Schmittner et al. 2011). Interpretation Figure 1 illustrates some strengths of data assimilation. The model propa gates information from LGM surface air temperature (SAT) reconstructions over land to other regions, and to sea surface temperatures (SST). In this ex ample assimilating SAT reconstructions produces an SST estimate with a warm ing at the LGM in the northern North Atlantic, which is consistent with the SST reconstructions. Uncertainties are reduced relative to the model estimate


77 77 PAGES news Vol 21 No 2 August 2013 in most locations (grayed out areas are reduced). How should we interpret assimi lated paleo-states? Optimization meth ods select a single best simulation so the state estimate is physically selfconsistent according to the model. But the state estimate from updating meth ods is a combination of multiple model simulations and paleodata, therefore interpretation requires more care. An ensemble mean anomaly of zero might correspond to a wide spread of posi tive and negative results; this would be reflected in large model uncertain ties. A spatially coherent signal with small uncertainty might emerge from an ensemble after assimilating a single pinning point from paleodata; this sig nal should be physically consistent be cause it arises from the model physics. Such considerations are common to all multi-model ensemble summaries and reanalyses. For statistically meaningful results it is essential to use a distance met ric grounded in probability theory, i.e. corresponding to a particular distribu tion of model-data differences (likeli hood function in Bayesian terms). This might preclude the use of non-standard variables such as biomes. Data assimilation is a statistical modeling technique and should be evaluated. Testing the method with pseudo-paleodata can help avoid the (literal) pitfalls of finding local rather than global minima in high-dimension al spaces. Future directions Data assimilation is a formal method that not only highlights model-data discrepancies but also corrects them. It can be challenging, because it requires a process-based model and reliable esti mation of uncertainties for both paleo data and simulations. For paleodata, difficulties may arise from dating and time averaging. But im provements in estimating reconstruc tion uncertainties can be made by us ing forward modeling approaches (e.g. Tingley et al. 2012). These approaches allow greater freedom in specifying the behavior of climate-proxy relationships (such as nonlinearity and multi-modal uncertainties) and enables uncertain ties to cascade through the causal chain to allow full probabilistic quantification of the unknown state. Using physically -based forward models for reconstruc tion, i.e. data assimilation, incorporates information about the relationships between locations, times and variables and therefore minimizes the risk of physical implausibility. The long-term goal may be forward physical modeling of the whole causal chain from radiative forcings to proxy archives (e.g. Roche et al. 2004; Stone et al. 2013). For paleo-simulations, we do not need models to be complex or stateof-the-art, but we do need to estimate their uncertainties. If they are complex it is difficult to generate their deriva tives with respect to the parameters. If they are expensive it is difficult to sample, and therefore to assess, their uncertainties. Thoughtful experimental design with statisticians, and perhaps also statistical modeling of the physi cal model (known as emulation; e.g. Schmittner et al. 2011), can help in this regard. A research priority is to estimate the discrepancy between a model and reality at its best parameter values, and how this varies across different eras. New updating methods are emerging that use the PMIP multi-model ensem ble to explore structural uncertainties. For example, Annan and Hargreaves (2013) use the linear combination of en semble members that best matches the paleodata. These challenges are worth tackling for the substantial benefits. Information from paleodata can be extrapolated to other locations, times and state vari ables, and uncertainties are smaller (or at worst, the same) than those of the individual model or proxy-based esti mates. Selected references Full reference list online under: Bartlein P et al. (2010) Climate Dynamics 37(3-4): 775-802 Guiot J et al. (2000) Ecological Modelling 127(2-3): 119-140 Schmittner A et al. (2011) Science 334: 1385-1388 MARGO et al. (2009) Nature Geoscience 2: 127-132 Stone EJ, Lunt DJ, Annan JD, Hargreaves JC (2013) Climate of the Past 9: 621-639 Figure 1: LGM annual mean temperature anomalies from: A) surface air temperature (SAT) reconstructions based on pollen and plant macrofossils (Bartlein et al. 2010); B) sea surface temperature (SST) reconstructions based on multiple ocean proxies (MARGO et al. 2009); C, D) simulations from the HadCM3 general circulation model (mean of 17 member perturbed parameter ensemble; Edwards, unpublished data). Data assimilation estimates generated by updating with SAT reconstructions (E, F) and both SAT and SST reconstructions (G, H) Gray areas indicate regions with low signal-to-noise: magnitude of temperature anomaly is less than 3 of uncertainty estimates.


78 78 PAGES news Vol 21 No 2 August 2013 Parameter estimation using paleodata assimilation JAMES D ANNAN 1 M. CR UCIFIX 2 T .L. EDWARDS 3 AND A. PA UL 4 1 RIGC/JAMSTEC, Yokohama, Japan; 2 Georges Lemaitre Centre for Earth and Climate Science, Universit Catholique de Louvain, Belgium; 3 School of Geographical Sciences, Bristol University, UK; 4 MARUM, University of Bremen, Germany In addition to improving the simulations of climate states, data assimilation concepts can also be used to estimate the internal parameters of climate models. Here we introduce some of the ideas behind this approach, and discuss some applications in the paleoclimate domain. E stimation of model parameter values is of particular interest in paleoclimate and climate change research, since it is the formulation of model parameterizations, rather than the initial conditions, which is the main source of uncertainty regarding the climates long-term response to natu ral and anthropogenic forcings. We should recognize at the outset that the question of a correct param eter value might in many cases be quite contentious and disputable. There is, for example, no single value to describe the speed at which ice crystals fall through the atmosphere, or the background rate of mixing in the ocean, to mention two parameters which are commonly varied in General Circulation Models (GCMs). Generally the best we can hope for is to nd a set of parameter values, which per form well in a range of circumstances, and to make allowances for the models inad equacies, i.e. structural errors due to inad equate equations and parameterizations. However, inadequacies will always be present no matter how carefully param eter values are chosen: this should serve as a caution against over-tuning. It may not be immediately clear how one can use proxy-derived observational estimates of climatic state variables such as temperature or precipitation to es timate the values of a models internal parameters. However, from a suciently abstract perspective, the problem of pa rameter estimation can be considered as equivalent to state estimation, via a stan dard approach in which the state space of a dynamical model is augmented by the inclusion of model parameters (Jazwinski 1970; Evensen et al. 1998). To see how this works, consider a system described by a dynamical model f, which uses a set of internal parameters and propagates a state vector x through time through a set of dierential equations: x = f (x) (1) We can create an equivalent model g(x, ) which takes as its state vector (x, ) (in which the parameter values have simply been concatenated onto the end of the state vector), and propagates this vector through the augmented set of equations x = f (x) (2) = 0 (3) Thus, the existing methods and technol ogy for estimating the state x can, in prin ciple, be directly applied to the estimation of (x, ), or in other words, the joint estima tion of state and parameters. While this approach is conceptually straightforward, there are many practi cal diculties in its application. The most widespread methods for data assimila tion, including both Kalman ltering and 4D-VAR, rely on (quasi-)linear and Gaussian approaches. However, the aug mented model g is likely to be substan tially more nonlinear in its inputs than the underlying model f due to the presence of product terms such as i x j (Evensen et al. 1998). Further challenges exist in applying this approach due to the wide disparity in relevant time scales. Often the initial state has a rapid eect on the model trajectory within the predictability time scale of the model, which is typically days to weeks for atmospheric GCMs. On the other hand, the full eect of the parameters only becomes apparent on the climatological time scale, which may be decades or centuries. Applications Methods for joint parameter and state es timation in the full spatiotemporal domain continue to be investigated for numerical weather prediction, where data are rela tively plentiful. But identiability, that is the ability to uniquely determine the state and parameters given the observations, is a much larger problem for modeling past climates, where proxy data are relatively sparse in both space and time. Therefore, data assimilation in paleo climate research generally nds a way to reduce the dimension of the problem. One such approach is to reduce the spatial di mension, even to the limit of a global av erage. For example, a three-variable glob ally averaged conceptual model for glacial cycles has been tuned using exible and powerful methods such as Markov Chain Monte Carlo (Hargreaves and Annan 2002) and Particle Filtering (Crucix and Rougier 2009). Figure 1 presents the results of one parameter estimation experiment by Hargreaves and Annan (2002). In the case of more complex and higher resolution models, the problems of identiability and computational cost are most commonly addressed by the use of equilibrium states. Here, the full initial condition of the model is irrelevant, at least within reasonable bounds, and the dimension of the problem collapses down to the number of free parameters; typically ten at most, assuming many boundary conditions are not also to be es timated. With this approach, much of the detailed methodology of data assimilation as developed and practiced in numerical weather prediction, where the huge state dimension is a dominant factor, ceases to be so relevant. While some attempts at using stan dard data assimilation methods have been performed (e.g. Annan et al. 2005), a much broader range of estimation methods can also be used. With reasonably cheap models and a suciently small set of pa rameters, direct sampling of the param eter space with a large ensemble may be feasible. A statistical emulator, which pro vides a very fast approximation to running the full model, may help in more compu tationally demanding cases (e.g. Holden et al. 2010). One major target of parameter esti mation in this eld has been the estima tion of the equilibrium climate sensitivity. This may either be an explicitly tunable model parameter in the case of simpler models, or else an emergent property of the underlying physical processes, which are parameterized in a more complex global climate model. The Last Glacial Maximum is a particularly popular inter val for study, due to its combination of a large signal to noise ratio and good data coverage over a quasi-equilibrium in terval (Annan et al. 2005; Schneider von Deimling et al. 2006; Holden et al. 2010; Schmittner et al. 2011; Paul and Losch


79 79 PAGES news Vol 21 No 2 August 2013 2012). The methods used for studying the LGM in order to estimate the equilibrium climate sensitivity have covered a wide range of techniques including direct sam pling of parameter spaces (with and with out the use of an emulator), Markov Chain Monte Carlo methods, the variational ap proach using an adjoint model, and the Ensemble Kalman Filter. In general, more costly models require stronger assump tions and approximations due to compu tational limitations. Approaches which aim at averaging out the highest frequencies of internal variability while still retaining a transient and time-varying forced response may make use of temporal data such as tree rings over the last few centuries (Hegerl et al. 2006). In that case, the spatial dimen sion can still be reduced, e.g. by averaging to a hemispheric mean. A similar approach was used by Frank et al. (2010) to estimate the carbon cycle feedback. Paleoclimate simulations provide the only opportunity to test and critically eval uate climate models under a wide range of boundary conditions. This suggests that we need to continue to develop a broad spectrum of methods to be applied on a case-specic basis. Selected references Full reference list online under: Annan JD et al. (2005) SOLA 1: 181-184 Crucix M, Rougier J (2009) The European Physical Journal-Special Topics 174(1): 11-31 Hargreaves JC, Annan JD (2002) Climate Dynamics 19: 371-381 Holden PB et al. (2010) Climate Dynamics 35: 785-806 Paul A, Losch M (2012) Perspectives of parameter and state estimation in paleoclimatology : In, Berger A et al. (Eds) Climate Change: Inferences from Paleoclimate and Regional Aspects, Springer, 93-105 Figure 1: Experiment with 350 ka of data assimilated. The red line at the top is the normalized summer solar insolation forcing at 65 N. The black dot-dashed lines are normal ized proxy data from Vostok (ice volume and atmospheric CO 2 concentration) and SPECMAP (deep ocean temperature) cores. Data to the left of the vertical magenta line were used to tune parameters, with the right hand side used as validation of the model forecast, which (over a range of experiments) shows substantial skill for a duration of around 50-100 ka. The dark blue lines show the mean of the ensemble and the light blue lines show one standard deviation of the ensemble. Modied from Hargreaves and Annan (2002).


80 Impact of climate and sea level change on coastal evolution KWASI APPEANINGADDO, S.W LARYEA, B.A. FOLI AND C.L. ALLOTEY Department of Marine and Fisheries Sciences, University of Ghana; T he third West African Quaternary Research Association (WAQUA) work shop, hosted by the Department of Marine and Fisheries Sciences, University of Ghana, was held at the Institute for Local Government Studies in Accra. The objec tive of the workshop was to identify how humans adapted to past climatic and sea level changes, and to discuss future adapta tion strategies through a multidisciplinary approach. Twenty-seven scientists from ve countries attended the workshop. The in augural lecture was given by Dr. Thomas Kwasi Adu of the Geological Survey Department of Ghana. He spoke about the causes of sea-level rise and coastal change, and their implications for coastal regions. His lecture stressed the fact that important planning decisions for sea-level rise should be based on the best available scientic knowledge and careful consideration of long-term benets for a sustainable future. He recommended that decisions on adap tation or mitigation measures should also take into consideration economic, social, and environmental costs. Sixteen scientic papers were pre sented and discussed on various topics covering sea level rise, coastal erosion and climate change issues. In particular, the presentations addressed the impact of sealevel changes on coastal tourism develop ment; the linkages between sea-level rise and ground water quality, hydrodynam ics, upwelling and biogeochemistry in the Gulf of Guinea; paleoclimatic evidences from the quaternary coastal deposits from Nigeria; dynamics of ocean surges and their impacts on the Nigerian coastline; and the eect of climatic extreme events on res ervoir water storage in the Volta Basin in Ghana. The workshop provided a platform for scientists to share knowledge and informa tion on their respective areas of research. At the end of the presentation sessions, the plenum agreed that research on sea-level rise should be particularly encouraged and that further activities to bring together scientists working in this area should be organized. To stimulate interest and ex pose students to new methods, regular in ternational workshops or summer schools will be held to bring together students and experts from within and outside the subregion. The meeting participants then went on a guided tour of coastal communities along the eastern coast of Ghana. One of them was Keta, a coastal town in the Volta Region that was partly destroyed by sea erosion at the end of the 20 th century. Keta is situated on a sandspit separating the Gulf of Guinea from the Keta Lagoon. Due to this double waterfront, the city area is particularly vul nerable to erosion. It is ooded from the ocean front during high tides and from the lagoon front during heavy runo, especial ly in the rainy seasons. During devastating erosion events between 1960 and 1980, more than half of the town area has been washed away. The photo (Fig. 1) shows Keta in 1985. Since 1999 more than 80 million US$ have been invested to protect, restore and stabilize the coast of Keta. The next WAQUA workshop will be held in Senegal in 2014. Acknowledgements We express our sincere gratitude to INQUA, PAGES, PAST and the University of Ghana for sponsoring this workshop. We are also grateful to the local hosts, the Department of Marine and Fisheries Sciences, University of Ghana and the Ghanaian Institute of Local Government Studies. Figure 1: This photo from 1985 shows a section of the Keta town destroyed by sea erosion. Photo by Beth Knittle. Accra, Ghana, 8-12 October 2012


81 Holocene land-cover change in Eastern Asia for climate modeling MARIE-JOS GAILLARD 1 AND QINGHAI XU 2 1 Department of Biology and Environmental Science, Linnaeus University, Kalmar, Sweden; 2 Key Laboratory of Environmental Change and Ecological Construction, Hebei Normal University, Shijiazhuang, China T his workshop was held at Hebei Normal University in China, and contributed to the PAGES Focus 4 theme Land Use and Cover (LUC). A ma jor goal of LUC is to achieve Holocene land-cover reconstructions that can be used for climate modeling and test ing hypotheses on past and future ef fects of anthropogenic land cover on climate. Collaborations initiated be tween Linnaeus University (Sweden; M.-J Gaillard), University of Hull (Britain, Jane Bunting), Hebei Normal University (China; Q. Xu and Y. Li), the French Institute (Pondicherry, India; A Krishnamurthy) and Lucknow University (India; P. Singh Ranhotra) are in line with the goals of Focus 4-LUC and follow the strategy of the European LUC-relevant LANDCLIM project (Gaillard et al. 2010). The aim of these collaborations is to develop quantitative reconstructions of past vegetation cover in India and China using pollen-vegetation model ing and archaeological/historical data together with other land-cover model ling approaches. The objective of this workshop in China was to initiate the necessary col laborations and activities to make past land-cover reconstructions possible in Eastern Asia. The three major outcomes of the workshop are (i) the building of a large network of experts from Eastern Asia, Europe and USA now working to gether to understand past changes in land use and land cover in Eastern Asia, (ii) that all existing East Asian pollen databases will be integrated into the NEOTOMA Paleoecology Database by 2014 (coordinated by Eric Grimm, Illinois State University, USA), and (iii) that a re view paper will be prepared with the working title Past land-use and anthro pogenic land-cover change in Eastern Asia evaluation of current achieve ments, potentials and limitations, and future avenues. It is planned to submit this article this fall. It will include 1) a re view of Holocene human-induced veg etation and land-use changes in Eastern Asia based on results presented at the workshop; 2) a discussion of the exist ing anthropogenic land-cover change scenarios (HYDE, Klein Goldewijk et al. 2010; KK 10, Kaplan et al. 2009) in the light of the reviewed proxy-based knowledge; and 3) a discussion on the implications of these results for future climate modeling and the study of past land cover-climate interactions. These significant outcomes are the result of three intense days of presen tations and discussion sessions. During the lectures, results were presented for historical, paleoecological and pollenbased reconstructions; pollen-vegeta tion relationships; past human-impact studies; and model scenarios of anthro pogenic land-cover changes in the past. Some of the conference was also dedi cated to reviewing the state of existing pollen databases and more generally, database building. Plenary and group break-out dis cussions focused on more technical aspects such as improving pollen da tabases, especially for the East Asian region; planning the review paper; and scientific topics such as the use of pollen-based and historical land-cover reconstructions for the evaluation of model scenarios of past anthropogenic land-cover change (Fig. 1), and the inte gration of model scenarios with paleo ecological or historical reconstructions. References Gaillard MJ et al. (2010) Climate of the Past 6: 483-678 Kaplan J, Krumhardt K, Zimmermann N (2009) Quaternary Science Reviews 28: 3016-3034 Klein Goldewijk K, Beusen A, Janssen P (2010) The Holocene 20: 565-573 Olofsson J, Hickler T (2008) Vegetation History and Archaeobotany 17: 605-615 Figure 1: Anthropogenic deforestation in Eastern Asia at AD 1 simulated by three dierent approaches: A) Olofsson and Hickler (2008); (B) HYDE (History Database of the Global Environment) version 3.1 (Klein Goldewijk et al. 2010); and (C) Kaplan et al. (2009). Note the large dierences in land-cover between the dierent models used. Shijazhuang, China, 9-11 October 2012


82 Palaeo50: The priority research questions in paleoecology Oxford, UK, 13-14 December 2012 ANSON W MACK AY 1 A.W .R. SEDDON 2,3 AND A.G. BAKER 2 1 Environmental Change Research Centre, University College London, UK; 2 Oxford Long-Term Ecology Laboratory, University of Oxford, UK; 3 Department of Biology, University of Bergen, Norway P aleoecological studies provide in sights into ecological and evolu tionary processes, and help to improve our understanding of past ecosystems and human interactions with the envi ronment. But paleoecologists are often challenged when it comes to processing, presenting and applying their data to im prove ecological understanding and in form management decisions (e.g. Froyd and Willis 2008) in a broader context. Participatory exercises in, for example, conservation, plant science, ecology, and marine policy, have developed as an ef fective and inclusive way to identify key questions and emerging issues in sci ence and policy (Sutherland et al. 2011). With this in mind, we organized the rst priority questions exercise in paleoecol ogy with the goal of identifying 50 priori ty questions to guide the future research agenda of the paleoecology community. The workshop was held at the Biodiversity Institute of the University of Oxford. Participants included invited experts and selected applicants from an open call. Key funding bodies and stakeholders were also represented at the workshop, including the US NSF, IGBP PAGES, UK NERC, and UK Natural England. Several months prior to the work shop, suggestions for priority questions had been invited from the wider commu nity via list-servers, mailing lists, society newsletters, and social media, particu larly Twitter (@Palaeo50). By the end of October 2012, over 900 questions had been submitted from almost 130 indi viduals and research groups. Questions were then coded and checked for dupli cation and meaning, and similar ques tions were merged. The remaining 800 questions were re-distributed to those who had initially engaged in the process. Participants were asked to vote on their top 50 priority questions. At the end of November the ques tions were grouped into 50+ categories, which in turn were allocated to one of six workshop themes to be chaired by an ex pert: Human-environment interactions in the Anthropocene (Erle Ellis, University of Maryland, USA); Biodiversity, conser vation and novel ecosystems (Lindsey Gillson, University of Capetown, South Africa), Biodiversity over long time scales (Kathy Willis, University of Oxford, UK), Ecosystems and biogeochemical cy cles (Ed Johnson, University of Calgary, Canada), Quantitative and Qualitative reconstructions (Stephen Juggins, University of Newcastle, UK), and Approaches to paleoecology (John Birks, University of Bergen, Norway). Each working group also had a cochair, responsible for recording votes and editing questions on a spreadsheet, and a scribe. Workshop participants were allocated into one of six parallel working groups tasked with reducing the number of questions from 180 to 30 by the end of day one. This was an intensive process involving considerable debate and edit ing. During day two, these 30 questions were winnowed down further with each group arriving at seven priority ques tions. The seven questions from each group were then combined to obtain 42 priority questions. Each working group had a further ve reserve questions, which everyone voted on in the nal plenary. The eight reserve questions that obtained the most votes were selected to complete the list of 50 priority ques tions. Working group discussions were of ten heated and passionate. Compromises won by the chairs and co-chairs were dif cult but necessary. It is important that the nal 50 priority questions are not seen as a denitive list, but as a starting point for future dialogue and research ideas. The nal list of 50 priority questions and full details of the methodology is currently under review, and the publi cation will be announced through the PAGES network. Acknowledgements We thank PAGES for generously providing sup port. We also acknowledge sponsorship from the British Ecological Society, the Quaternary Research Association and the Biodiversity Institute, University of Oxford. References Froyd CA, Willis KJ (2008) Quaternary Science Reviews 27: 1723-1732 Sutherland WJ et al. (2011) Methods in Ecology and Evolution 2: 238-247 Figure 1: Flowchart of the selection process for the 50 priority research questions.


83 The Agulhas System and its role in changing ocean circulation, climate, and marine ecosystems Stellenbosch, Republic of South Africa, 8-12 October 2012 RAINER ZAHN 1 W .P.M. DE RUIJTER 2 L. BEAL 3 A. BIASTOCH 4 AND SCOR/WCRP/IAPSO WORKING GROUP 136 5 1 Institut de Cincia i Tecnologia Ambientals, Universitat Autnoma de Barcelona, Bellaterra, Spain; 2 Institute for Marine and Atmospheric Research, Utrecht University, The Netherlands; 3 Rosenstiel School of Marine and Atmospheric Science, University of Miami, USA; 4 GEOMAR Helmholtz Centre for Ocean Research, Kiel, Germany; 5 T his conference was held in recogni tion of the signicance of the Agulhas Current to ocean physical circulation, marine biology and ecology, and climate at the regional to global scale. The con ference centered on the dynamics of the Agulhas Current in the present and the geological past; the inuence of the cur rent on weather, ecosystems, and sher ies; and the impact of the Agulhas Current on ocean circulation and climate with a notable focus on the Atlantic Meridional Overturning Circulation (AMOC). 108 participants from 20 countries, includ ing seven African countries, attended the conference, of which a quarter were young researchers at the PhD student level. Participants came from the areas of ocean and climate modeling, physical and biological oceanography, marine ecology, paleoceanography, meteorology, and ma rine and terrestrial paleoclimatology. The Agulhas Current attracts interest from these communities because of its signi cance to a wide range of climatic, biologi cal and societal issues. The current sends waters from the Indian Ocean to the South Atlantic. This is thought to modulate convective activ ity in the North Atlantic. It is possible that it even stabilizes the AMOC at times of global warming when freshwater pertur bation in the North might weaken it. But these feedbacks are not easy to trace, and direct observations and climate models have been the only way to indicate the possible existence of such far-eld tele connections to date. This is where marine paleo-proxy proles prove helpful. They reveal the functioning of the Agulhas Current under a far larger array of climatic boundary conditions than those present during the short period of instrumental observations. For instance altered condi tions in the past with shifted ocean circu lation and wind elds stimulated Agulhas water transports from the Indian Ocean to the Atlantic, the so-called Agulhas leak age, at very dierent rates from todays. A number of the paleo-records that were shown at the meeting demonstrated a link between Agulhas leakage and Dansgaard/ Oeschger-type abrupt climate changes in the North Atlantic region, suggesting that salt-water leakage may have played a role in strengthening the AMOC and sudden climate warming in the North. Marine ecosystems were also shown to be measurably impacted by the Agulhas system. Notably, the high vari ability associated with the prominence of mesoscale eddies and dipoles along the Current aect plankton communities, large predators, pelagic sh stocks, and possibly even facilitate the northward sar dine runs swimming against the vigorous southward ow of the Agulhas Current. The meteorological relevance of the Agulhas Current was also demonstrated, for example its role as a prominent source of atmospheric heat and its signicance in maintaining and anchoring storm tracks. Among other things, these aect the atmospheric westerly Polar Front Jet and Mascarene High, with onward conse quences for regional weather patterns, in cluding extreme rainfall events over South Africa. The conference developed a number of recommendations; two key ones being that eorts should be made to trace the impacts of the Agulhas leakage on the changing global climate system at a range of timescales, and that sustained obser vations of the Agulhas system should be developed. Implementing these recom mendations would constitute a major chal lenge logistically and the Western Indian Ocean Sustainable Ecosystem Alliance (WIOSEA) was identied as a possible inte grating platform for the cooperation of in ternational and regional scientists toward these goals. This would involve capacity building and training regional technicians and scientists, which could be coordinat ed through partnerships with the National Research Foundation in South Africa. The conference was held under the auspices of the American Geophysical Union Chapman Conference series and organized by the Scientic Committee on Oceanic Research (SCOR)/ World Climate Research Program (WCRP)/ International Association for the Physical Sciences of the Oceans (IAPSO) Working Group 136. Additional sponsorship came from the International Union of Geodesy and Geophysics; US NSF; NOAA; PAGES; Institute of Research for Development (IRD) France; and the Royal Netherlands Institute for Sea Research. References Biastoch A, Bning CW, Schwarzkopf FU, Lutjeharms JRE (2009) Nature 462: 495-499 Figure 1: A model perspective of Agulhas leakage and the interbasin water transports between the Indian Ocean and Atlantic. A high-resolution Agulhas model (1/10, gray box), is nested in a (1/2) global ocean/sea ice model to simulate temperature and magnitude of currents (see Biastoch et al. 2009).


84 IPICS First Open Science conference JRME CHAPPELLAZ 1 ERIC WOLFF 2 AND ED BROOK 3 1 Laboratoire de Glaciologie et Gophysique de lEnvironnement, Universit Joseph Fourier Grenoble, St Martin dHres, France; 2 British Antarctic Survey, Cambridge, UK; 3 College of Earth, Ocean, and Atmospheric Sciences, Oregon State University, Corvallis, USA I PICS (International Partnerships in Ice Core Sciences) is the key planning group for international ice core scientists. Established in 2005, it now includes sci entists from 22 nations and aims at den ing the scientic priorities of the ice core community for the coming decade. IPICS lies under the common umbrella of IGBP/ PAGES, SCAR (Scientic Committee on Antarctic Research) and IACS (International Association of Cryospheric Sciences). IPICs First Open Science conference, organized by the European branch of IPICS (EuroPICS), took place in a beauti ful setting on the French Cte dAzur. 230 scientists gathered from 23 nations, with a good mix of both junior and senior sci entists present. While most of the partici pants work on ice cores, a signicant num ber were scientists working on marine and continental records as well as on climate modeling. The sponsorship received from several institutions, agencies and projects, enabled us to invite ten keynote speak ers as well as six scientists from emerging countries. The program followed IPICs main scientic objectives as outlined in four white papers ( Notably, it covered questions of climate variability at dierent time scales (from the last 2000 to 1 M years), biogeochemi cal cycles, dating, and ice dynamics. New challenges, such as studying the bacterial content of ice cores, and new methodolo gies were also the focal point of specic sessions. Over the ve days of the confer ence, all attendees gathered for the ple nary sessions combined with long poster sessions. These sessions oered valuable and ecient networking opportunities. The full program can be found at: www. Among the various results presented at this occasion, signicant information was provided on two big recent proj ects of the ice core community: the WAIS Divide deep drilling in West Antarctica, and the NEEM deep drilling in North-West Greenland. Ice core projects from outside polar regions were also well represented, with results obtained from the Andes, the Alps and the Himalayas. The beautiful and peaceful setting of the conference center enabled strong and ecient networking; no doubt, in the future we will see that many new ice core drilling projects had their roots at IPICS 1 st OSC. Proving that ice core scientic out puts remain of prime importance to highimpact journals, the Chief Editor of Nature as well as an editor of Nature Geoscience attended the full ve days of the event. IPICS next OSC will take place in 2016. An open call for bids to organize it will be launched in 2013. Acknowledgements The organizers thank the generous nancial sup port of many sponsors: European Descartes Prize of the EPICA project, ERC project ICE&LASERS (Grant # 291062), LabEX OSUG@2020 project of the Grenoble Observatory for Earth Sciences and Astronomy, European Polar Board, SCAR, PAGES, IACS, Aix-Marseille University, Versailles St Quentin University, Fdration de Recherche ECCOREV, IRD, LGGE, LSCE, Picarro, LabEX proj ect L-IPSL, Collge de France and IPEV. Ecient administrative and nancial handling of the conference was provided by the Floralis com pany. Special thanks to Stphanie Lamarque and Audrey Maljean from Floralis for their help over the two years of conference preparation. References NEEM community members (2013) Nature 493: 489-494 Figure 1: Inside the drilling trench at NEEM, Greenland. The North Greenland Eemian (NEEM) ice drilling is an international project managed by the Centre for Ice and Climate, Denmark, involving 14 nations. In Summer 2010, it recovered ice from the Eemian dating back from 130 ka BP, helping to describe the warming and ice sheet shrinking at a time of unusually high Arctic summer insolation (NEEM community members 2013). Photo: Jrme Chappellaz. Presqule de Giens, France, 1-5 October 2012


85 Analyzing paleolimnological data with R Isle of Cumbrae, Scotland, 16-20 August 2012 DIEGO NA V ARRO 1 AND STE VE JUGGINS 2 1 Instituto de Investigaciones Marinas y Costeras, CONICET, Universidad Nacional de Mar del Plata, Argentina; 2 School of Geography, Politics and Sociology, Newcastle University, Newcastle upon Tyne, UK P aleolimnology has grown rapidly over the last two or three decades in terms of the number of physical, chemical, and biological indicators analyzed and the quantity, diversity and quality of data gen erated. Such growth has presented paleo limnologists with the challenge of dealing with highly quantitative, complex, and multivariate data to document the timing and magnitude of past changes in aquatic systems, and to understand the internal and external forcing of these changes. To cope with such tasks paleolimnologists continuously add new and more sophis ticated numerical and statistical methods to help in the collection, assessment, sum mary, analysis, interpretation, and com munication of data. Birks et al. (2012) sum marize the history of the development of quantitative paleolimnology and provide an update to analytical and statistical techniques currently used in paleolimnol ogy and paleoecology. R is both a programming language and a complete statistical and graphi cal programming environment. Its use has become popular because it is a free and open-source application, but above all because its capability is continuously enhanced by new and diverse packages developed and generously provided by a large community of scientists. This recent workshop, held in the com fortable facilities of the Millport Marine Biological Station, trained researchers on the theory and practice of analyzing pa leolimnological data using R. The course was led by Steve Juggins (Newcastle University) and Gavin Simpson (University College London, recently moved to the University of Regina), two of the research ers that have contributed in the develop ment and application of dierent statisti cal tools and packages for paleoecology within the R community. A total of 31 participants from a range of continents (North and South America, Europe, Asia, and Africa), career stages (PhD students to faculty) and scientic backgrounds (paleolimnology, palynolo gy, diatoms, chironomids, sedimentology) enjoyed four long days of training in statis tical tools and working on their own data. Initially, participants were introduced to R software and language, tools for summa rizing data, exploratory data analysis and graphics. The following lectures and prac tical sessions focused on simple, multiple and modern regression methods; cluster analysis and ordination techniques used to summarize patterns in stratigraphic data; hypothesis testing using permuta tions for temporal data, age-depth mod eling, chronological clustering, smoothing and interpolation of stratigraphic data; and calculation of rates of change. The nal lectures dealt with the application of techniques for quantitative environmental reconstructions. The theory and assump tions underpinning each method were introduced in short lectures, after which the students had the opportunity to ap ply what they had learned, to data sets and real environmental questions, during practical sessions. There was also time in the evenings for sessions on important R tips, advanced R graphics, special topics proposed by the assistants, and for the students to work on their own data. The course was conveniently orga nized just prior to the 12 th International Paleolimnology Symposium (Glasgow, 20-24 August 2012), which enabled all of the workshop participants to attend the symposium and encouraged further discussions throughout the following week. PAGES covered travel and course costs for ve young researchers from de veloping countries (Turkey, South Africa, Macedonia, and Argentina) all of who were very grateful for the opportunity to attend. References Birks HJB, Lotter AF, Juggins S, Smol JP (2012) Tracking Environmental Change Using Lake Sediments: Data Handling and Numerical Technique, Developments in Paleoenvironmental Research 5. Springer, 745 pp Figure 1: Participants during the R workshop. Photo by S. Juggins.


86 The Sun and its role in climate change Workshop of the PAGES Solar Working Group Davos, Switzerland, 5-7 September 2012 JUERG BEER Swiss Federal Institute of Aquatic Science and Technology (Eawag), Dbendorf, Switzerland; B etter understanding the Sun and its role in climate change is an important but dicult goal. It is important because to properly assess the anthropogenic ef fect on climate change an accurate quan tication of the natural forcing factors is required. But it is dicult because: (1) natural forcing records are generally not well quantied; (2) the response of the climate system to forcings is non-linear due to various feed back mechanisms and can only be esti mated using complex climate models; (3) in spite of their complexity models may not comprise all relevant processes and have to be validated, but the instrumental records of climate forcing and climate re sponse are generally too short for this pur pose and the data set needs to be comple mented by proxy data; (4) proxy data are derived from natural archives and are only indirectly related to the physical parameters of interest, and their calibration is based on assumptions that may not be fully valid on longer time scales; (5) instrumental and proxy data reect the combined response to all forcings, and not only the inuence of the Sun. Furthermore, the climate system shows internal unforced variability. All this makes separation and quantication of the indi vidual forcings very dicult. The main aim of the rst workshop of the solar forcing working group was to as sess the present state of the art and iden tify knowledge gaps by bringing together experts from the solar, the observational and paleo-data, and modeling communi ties. The workshop was organized jointly with FUPSOL (Future and past solar inu ence on the terrestrial climate), a multidisciplinary project of the Swiss National Science Foundations that addresses how past solar variations have aected climate, and how this information can be used to constrain solar-climate modeling. FUPSOL also aims to address the key question of how a decrease in solar forcing in the next decades could aect climate at global and regional scales. Here are some examples of open questions and problems that were identied in this workshop, and will be addressed in more detail in subsequent meetings: Physical solar models are not yet capable of explaining many observed features such as solar cycles and changes in total solar irradiance (TSI) and solar spectral ir radiance (SSI). There are still unresolved discrepancies between dierent composites of TSI based on the same satellite data. Semi-empirical models are relatively suc cessful in explaining short-term changes in TSI and SSI on time scales of days to years. However, on multi-decadal time scales input data and instrumental TSI and SSI data for comparison are missing. TSI and SSI reconstructions based on proxies suer from large uncertainties in their amplitudes. The most recent minimum, between so lar cycle 23 and 24, and probably also the upcoming minimum provide a glance of the Sun at its lowest activity level ever ob served during the satellite era. UV forcing and possibly also precipitat ing particles have signicant impacts on atmospheric chemistry and dynamics and need to be included in models. Detection and attribution of solar forc ing is often hampered by volcanic erup tions occurring simultaneously. Strategies to separate solar and volcanic forcings could be to select periods of low volcanic activity (e.g. roman period), to consider regional eects that dier for dierent forcings, and to look for multi-decadal to centennial solar cycles with well-dened periodicities. As an opening spectacle to the workshop, a medium-sized are initiated a long, mag netic lament burst out from the Sun (Fig. 1). Viewed in the extreme ultraviolet light, the lament strand stretched outwards until it nally broke and headed o to the left. Some of the particles from this erup tion hit Earth in September 2012, generat ing a beautiful aurora. Figure 1: Image of a solar magnetic lament burst. Image by NASA.


87 The backbone of PAGES 2k: data management and archiving LUCIEN V ON GUNTEN, D .M. ANDERSON 2 B. CHASE 1 M. CURRAN 1 J. GERGIS 1 E .P. GILLE 2 W GROSS 2 S. HANHIJR VI 1 D .S. KA UFMAN, T KIEFER, N.P. MCKAY 1 I. MUNDO 1 R. NEUK OM 1 M. SANO 1 A. SHAH 2 J. TYLER 1 A. VIA U 1 S. WAGNER 1 E .R. WAHL 2 AND D WILLARD 1 PAGES 2k data managers 2 NOAA Paleoclimatology branch members Full aliations listed here: T he PAGES 2k Network has formed to study climate change over the last two millennia at a regional scale, based on the most com prehensive dataset of paleoclimate proxy-re cords possible. In 2011, at its second network meeting in Bern, Switzerland (von Gunten et al. 2012) the network formally acknowledged that the envisioned data-intensive multiproxy and multi-region study must be built on the foundations of ecient and coordinat ed data management. In addition, the group committed to PAGES general objective to pro mote open access to scientic data and called for all records used for, or emerging from the 2k project to be publicly archived upon publi cation of the related 2k studies. Architects of the home for 2k data The National Climatic Data Center at the National Oceanic and Atmospheric Administration (NOAA) oered to host the primary 2k data archive. They set up a dedi cated NOAA task force to tailor the 2k data archive to the specic needs of the 2k project and to coordinate archiving with NOAA's data architecture and search capabilities. The 2k groups nominated regional data managers to provide input from the users end. Over the last two years, the regional 2k data managers have worked closely with NOAA to tailor the database infrastructure and prepare the upload of the 2k data. In addi tion, they provided expertise to help promote improvements in NOAAs archival of paleosci entic data in general. Since the data manag ers of the regional 2k groups are spread across the globe, the collaboration was organized around bi-monthly teleconference meetings under the lead of NOAA. In spite of the occa sional unearthly meeting hours for some, the interaction between the 2k data and NOAA database groups has worked fruitfully, as the following achievements show. The 2k database The paleoclimatology program at NOAA has set up a dedicated 2k project site with subpages for all regional groups (www.ncdc. html). This page was created early in the project to provide the regional groups with a central place to continuously compile datas ets considered relevant to their studies. Populating the database A two-step approach was applied for entering the 2k data into the database in order to serve demands for both speediness and thorough ness. First, all records used for the rst syn thesis article on regional temperature recon structions (PAGES 2k Consortium 2013) were made available on a data synthesis products page dedicated to the paper (hurricane.ncdc. ID:14188). This ensured that the records were made publicly available exactly at the time of publication and in a format that will remain identical with the data les supplementing the article. In a second step, all these records are currently being (re)submitted to NOAA with more detailed metadata information than before using a new submission protocol. Additionally, many new and already stored records that were not used for the PAGES 2k temperature synthesis are (re)formatted to the new submission protocol. This will allow improved search and export capabilities for a wealth of records that can currently only be accessed individually. Improved data submission protocol The data submission process is a crucial step for the long-term success of a database. On the one hand, it should contain as much relevant information as possible in order to maximize the value of the data. On the other hand, it should remain simple enough to keep the threshold for data providers as low as pos sible. The NOAA task force and 2k data man agers therefore created a substantially revised submission template le. This new protocol allows including more comprehensive in formation relating to the proxy records, and, crucially, is organized in a structured format that allows machine reading and automated searching for dened metadata information. This is critical in order to maximize the useful ness of the data to other scientists, as it addi tionally allows them to reprocess underlying features of the records such as the chronology or proxy calibrations. The new data submission template is also op timized for taking advantage of NOAAs archi val structure, which follows international con ventions for data description and archiving, and the Open Archive Initiative Protocol for Metadata Harvesting. This allows PAGES 2k data to be visible beyond the NOAA web site. A new feature of the NOAA-Paleo archive is the search capabilities that allow for projectspecic searches by a logical operator (e.g. PAGES 2K AND Monsoon). Additional func tionalities are planned that will, for example, allow the user to select a subset of proxy data for a region, and generate a single download able le of the requested data in NetCDF, ASCII, or Excel TM formats. 2k data management next steps In the next phase of the project starting now, the 2k network will work on completing the database of paleoclimate records of the last 2000 years to eventually produce new synop tic climate reconstructions. The proxy records will be collected according to the new NOAA data submission protocol. The prior use of this template during the collection and analysis phases of the project has the following advan tages: 1) all records are collected in the same, uniform format allowing for the inclusion of all relevant information, 2) the les are easily computer readable for data analyses, and 3) no additional formatting is required for the subsequent submission to the NOAA-Paleo archive. For large, data-intensive studies a good data management strategy is crucial. The ex perience from the PAGES 2k project suggests that setting up a data manager team and in volving archivist partners such as NOAA at an early stage of the project is key to handling data eciently. This new collaborative data storing ef fort is only possible thanks to the members of the regional 2k groups who provide their data and metadata inventories with the aim to make the global network of paleoclimate datasets publicly available. References PAGES 2k Consortium (2013) Nature Geoscience 6: 339-346 von Gunten L, Wanner H, Kiefer T (2012) PAGES news 20(1): 46 Program News The PAGES 2k Network and NOAA collaborate closely to optimize data compilations, and to build structures to facilitate ongoing supply and dynamic use of data. It is thus an successful example for a large trans-disciplinary eort leading to added value for the scientic community.


88 Updated Latin American Pollen Database: Version 2013 in preparation for NEOTOMA SUZETTE G.A. FLANTU A 1,2 H. HOOGHIEMSTRA 1 E GRIMM 3 AND V. MARK GRAF 4 1 Institute of Biodiversity Ecosystem Dynamics, University of Amsterdam, Netherlands; 2 Palinologa y Paleoecologa, Universidad Los Andes, Bogot, Colombia; 3 Illinois State Museum, Springeld, USA; 4 Institute of Arctic and Alpine Research, University of Colorado, Boulder, USA Figure 1: NEOTOMA and LAPD pollen sites. Red dots: existing NEOTOMA database; blue dots: African Pollen Database; yellow dots: inventoried sites for inclusion from Latin America and Japan. (Figure from Grimm et al. 2013). T he Latin American Pollen Data Base, better known as the LAPD, is an extensive data base of pollen from peat and lake cores and surface samples. It covers the areas of Central America, the Caribbean, and South America. The database was launched in 1994 by a re search group headed by Vera Markgraf at the University of Colorado, USA, and its manage ment moved to the University of Amsterdam in 1998, where it was hosted by Robert Marchant at the Institute of Biodiversity Ecosystem Dynamics (IBED). The LAPD started as a website where palynologists were encouraged to share their data. Unfortunately, after the project ended in 2003 no further updates were made to the website and the related database, and the list of pollen data of Latin America quickly became outdated. Recognizing the urgency for an updated LAPD, the group at IBED de cided to revive it. Supported by three grants from the Amsterdam-based Hugo-de-VriesFoundation (Van Boxel and Flantua 2009; Flantua and Van Boxel 2011), Suzette Flantua initiated a search for studies published after 2003. While the exploration for pollen records continues, this new LAPD 2013 inventory contains 1478 pollen sites throughout Latin America, multiplying by a factor of three the number of sites compared with the last update in 1997 (463 sites). The number of countries represented has increased from 15 to 29. In the meantime, the NEOTOMA data base ( was developed through an international collaborative eort of individuals from 23 institutions. This cyber infrastructure was designed to manage large multiproxy datasets, which makes it easy to explore, visualize, and compare a wide vari ety of paleoenvironmental data. This is why it was chosen as the primary archive site for the Global Pollen Database. The LAPD data in the 1997-list is now available through NEOTOMA (Fig. 1), and the complete 2013 LAPD inven tory with metadata will soon be available through NEOTOMA Explorer (http://ceiwin5. Although all LAPD data will be incorpo rated in NEOTOMA, the LAPD database will still exist as an independent entity. There will be a LAPD page within the existing NEOTOMA website which can be used to obtain informa tion on updated pollen sites, the pollen data, and events related to LAPD. It was also pro posed to form a Latin American NEOTOMA group of paleoecological researchers to man age the database, and upload and control the quality of the data. As almost no new data has been contrib uted to LAPD since 2003, IBED is in the pro cess of digitalizing and uploading their pollen database to make a signicant contribution to LAPD and thus stimulate global collabora tion, and data input and use by other research groups. The data of the newest pollen sites will be kept on standby in an oine database and made publicly available once the related research papers are accepted for publication. We would like to make researchers aware of the much richer palynological information now available for Latin America and we hope the pollen community will recognize LAPDNEOTOMA as an important archive, where the original authors are cited and acknowledged to have contributed to the database. We em phasize the great opportunity to promote multidisciplinary research on a continental scale and international scientic cooperation. We invite anyone with questions, doubts, or relevant information to contact us through Suzette Flantua and to support this global palynological initiative for an improved inte gration of knowledge and eorts. References Flantua SGA, Van Boxel JH (2011) Reviving the Latin American Pollen Database (LAPD) IBED, University of Amsterdam; document/447756 Grimm EC et al. (2013) Pollen Methods and Studies. Databases and their application In: Elias SA Mock CJ (Eds) Encyclopedia of Quaternary Science 2, Elsevier, 831-838 Van Boxel JH, Flantua SGA (2009) Pilot study of the geographical distri bution of plant functional types in the Amazonian lowlands. IBED, University of Amsterdam; Program News


89 A brief report on the 2 nd PAGES Young Scientists Meeting in Goa, India RA JEE V SARASWAT 1 AND BRITTA JENSEN 2 1 Micropaleontology Laboratory, National Institute of Oceanography, Goa, India; 2 School of Geography, Archaeology, Palaeoecology, Queens University, Belfast B ack in 2009, PAGES experimented with a dierent type of meeting for the rst time the inaugural Young Scientists Meeting (YSM) in Corvallis, USA. Recently several 1 st YSM alumni worked together with PAGES to build on the success of that inaugural meet ing with another YSM. The 2 nd YSM took place from the 11-12 February 2013 at the International Centre Goa in India. It brought together graduate students, post-doctoral fel lows and early career scientists from around the globe to share their research, network, present and attend workshops and panel discussions designed to address the specic challenges and opportunities facing early ca reer paleoscientists. A total of 79 participants from over 27 countries attended the meeting. Participants were welcomed by S. Rajan, Director of the National Center for Antarctic and Ocean Research, the Goan host insti tution. Thorsten Kiefer, PAGES Executive Director, then outlined the rationale behind the meeting and expressed the hope that the YSM would foster multi-disciplinary, interna tional interaction and collaboration amongst the next generation of paleoscientists. The meeting was structured around sev en themes: Climate Forcings; Regional Climate Dynamics; Global Earth-System Dynamics; Human-Climate-Ecosystem Interactions; Chronology; Proxy Development, Calibration and Validation; and Modeling. Twenty participants gave oral presen tations and many others presented post ers around each of these themes. A written peer-feedback activity provided presenters with valuable feedback on their presentation and ways to improve. The best presentations received an award, including one year of free online access to the Nature Geoscience jour nal: Ilham Bouimetarhan (Bremen, Germany) and Vladimir Matskovsky (Moscow, Russia) received prizes for the best oral presentations, and Jesper Bjrklund (Gteberg, Sweden), Gayatri Kathayat (Xian, China), and Timothe Ourbak (Niamey, Niger), for the best poster presentations. In the keynote talk Alan Mix of Oregon State University reected back upon his career as a climate scientist, which began during a time of discovery dened by a paucity of data a stark contrast to the present, with its wealth of data and the commensurate need for new approaches to interpreting it. He emphasized the need for more interaction among paleo scientists and the increased need for more quantitative climate data, which can be better utilized by the modeling community. The three The Art of sessions were a newly framed item in the YSM program and aimed to provide young scientists with practi cal information about data sharing, reviewing and communicating science (see the follow ing articles). In The Art of Sharing Data, David Anderson from the Institute of Arctic and Alpine Research, and National Climatic Data Center, highlighted the importance of sharing data, and in particular, making data publicly available through archiving. He discussed the data-rich world we live in where the sharing and archiving of data can increase the visibil ity of an individuals research tremendously, and how easy accessibility to datasets will en courage the community to develop new and novel quantitative approaches to interpreting them. In The Art of Communicating Science Gavin Schmidt from the NASA Goddard Institute for Space Studies suggested ways to convey science to dierent audiences and how to tackle controversies and criticism. He recommended the use of simple language with common examples and as many pic tures and graphs as possible, instead of tables and technical jargon. The Art of Reviewing panel included Alicia Newton, Editor of Nature Geoscience ; Denis-Didier Rousseau, Co-Editor-in-Chief of Climate of the Past ; Chris Turney, Asian and Australasian Regional Editor for the Journal of Quaternary Science ; and moderator Alberto Reyes, of Queens University, Ireland. They elded many questions from the audience, addressing various topics such as signing re views vs. double-blind reviews and what edi tors expect in a good review. It became clear during this session that many YSM partici pants did not feel their training had prepared them adequately for the peer-review process. During breakout sessions, participants divided into groups and deliberated on four challenges facing young paleoscientists. More detailed summaries of the breakout and The Art of sessions are reported else where in this issue of PAGES news. A key theme emerged across all groups: Many of the big important issues and problems fac ing early-career paleoscientists are similar the world over and may be dealt with through in ternational eorts. However, just as many are specically local, and therefore will require lo cal solutions. There was, of course, plenty of opportu nity for social interaction with one another, as well as with the organizers and guests. Communal meals, the icebreaker, and a din ner on one of the famous Goan casino boats featuring a Bollywood show, gave partici pants an opportunity to bond with each oth er while enjoying Indian culture and food. To conclude, participants thanked PAGES for taking the initiative to hold the YSM, and requested such meetings be convened more often. We would also like to gratefully thank the generous sponsors, listed below, whose contributions directly assisted many young scientists travel to and attend the meeting. Sponsors included: The Ministry of Earth Sciences, Government of India; National Centre for Antarctic and Ocean Research, India; US National Science Foundation; and the Swiss National Science Foundation; Asia-Pacic Network for Global Change Research; System for Analysis, Research and Training; IGBP, Brazil Regional Oce; National Oceanic and Atmosphere Administration, USA; National Institute of Ocean Technology, India; Indian National Centre for Ocean Information Services, India; Indian Institute of Tropical Meteorology, India; Oeschger Centre for Climate Change Research, University of Bern, Switzerland; and the International Association of Sedimentologists.


90 The Art of Communicating Science: traps, tips and tasks for the modern-day scientist HEIDI ROOP 1 G. MARTNEZ-MNDEZ 2 AND K. MILLS 3 1 Antarctic Research Centre, Victoria University of Wellington, New Zealand; 2 MARUM Center for Marine Environmental Sciences, University of Bremen, Germany; 3 Department of Geography, Loughborough University, UK I n a true test of modern-day communi cation, the participants at the 2 nd PAGES Young Scientists Meeting travelled virtually from the 26C heat of a Goan afternoon to a brisk -2C morning in New York City to join Gavin Schmidt (NASA, USA) for a lesson in the art of science communication. During this session, Gavin then delved into the nuts and bolts of why scientists are ethically obli gated to publicly communicate their science and how communicating it well is an increas ingly challenging but important aspect of our profession. This article highlights Gavins tips for eective public communication, some common traps scientists fall into, and tasks or next steps our community needs to take to improve the publics access to accu rate, high-quality scientic information. Despite the general public's interest in science, it is often hard to know where to go for accurate (and understandable) scientic information. In a world of rapid and wide dissemination of knowledge and opinions, it is increasingly important to communicate outside the scientic community. Not only do we have an obligation to communicate broadly, due to the typically high proportion of science funding coming from the tax payer, but broad communication is essential to avoid misuse or misinterpretations of our work and to slow the propagation of scien tic misconceptions. Crucially, many of the important scien tic concepts that need to be conveyed are simply not news. For example, the physics of greenhouses gases will undoubtedly nev er make the headlines yet it is a fundamental building block to being literate in the issue of climate change. Communicating these types of facts requires scientists to step be yond traditional avenues of communication. Gavin emphasizes that we, as a community, need to engage with social media and webbased communications, in addition to tradi tional means of communication (e.g. press releases, interviews and essays). People in creasingly rely on the internet as a primary source of information, which means there is a need to provide more accurate and appro priate information online through scientists blogs, videos, and social media platforms. We need to use this diverse set of tools to not only convey our expertise but impor tantly, to engage with dierent audiences. The challenge of clearly communicat ing the intended scientic message to the public is not insurmountable but requires an understanding of what works and what does not work. Falling into typical science communication traps can quickly turn an interview, article or outreach event into a counterproductive debate or an unintend ed source of misleading information. Here are some common traps and points on what does not work: Avoid talking too much about technical de tails and avoid technical debates Avoid using jargon that you dont take time to explain Avoid scientic stereotypes, e.g. arrogance or elitism Avoid triggering issues of free speech, data access, and secrecy Do not respond poorly to criticism by get ting angry or taking it personally Always distinguish between personal opin ion and scientic consensus Try to understand the context in which your statements will be heard or read Try to defuse pseudo-debates but do not ignore them Avoid sensationalism and over-extrapolat ed conclusions In order to get your audience to engage with your science, you have to engage with your audience. Several easy tips that can improve the strength and resonance of your message include: Listen to what your audience is interested in or concerned about Use imagery and animations. These can of ten create a deeper connection for the audi ence (e.g. see Fig. 1) Make it personal: use personal, relatable stories that remind people that we are nor mal. Personal anecdotes often generate ex citement and engagement Always remember the big picture and the reasons why you are a scientist Be a credible and trustworthy guide and take advantage of the generally held feeling of public admiration towards scientists Promote the investigative nature of science Provide accessible context outside of the technical literature Always underscore what can and cannot be concluded from your work As mentioned above, we live in an increas ingly connected world with growing ac cess to information (and misinformation). Our collective task is to ensure access to appropriate and understandable scientic information. The key to success relies on a collaborative approach in which we all rally together and begin to communicate the importance of science and the scientic pro cess. En masse we can begin to change the public perception of science, and can con tribute to developing a more scientically literate society. Making use of the above tips, learned from the frontlines of public science communication, can help to initiate this pro cess and move our community in the right direction. Now it is time to get blogging, writing, and tweeting! For further reading, see Gavins recommen dation: Debunking_Handbook.pdf Figure 1: The use of imagery is an effective way to convey messages to the general public. Which illus tration would you choose for a public lecture? Image sources: US National Snow and Ice Data Center.


91 The Art of Reviewing: Holding up quality in the scientic quality control system RA JANI PANCHANG 1 A. GO VIN 2 AND C. OMUOMBO 3 1 Geology & Palaeontology Group, Agharkar Research Institute, Pune, India; 2 MARUM Center for Marine Environmental Sciences, University of Bremen, Germany; 3 Department of Geology, University of Nairobi, Kenya E ver since the publication of the Philosophical Transactions of the Royal Society by Henry Oldenburg (1665, Fig. 1), scientists have acknowledged the im portance of relaying research ndings to a wider scientic community. Currently, publishing scientic ndings in peerreviewed journals, brands the research work as credible. These peer-reviewed scholarly articles improve the quality of science and are used as a metric for sci entic performance, which is essential for career advancement. The Program Committee of the 2 nd PAGES YSM recog nized the need to contribute to building capacity in the art of reviewing among young paleo-scientists. To this end, edi tors from prominent paleoscience jour nals, namely Nature Geoscience (Alicia Newton), Journal of Quaternary Science (Chris Turney) and Climate of the Past (Denis-Didier Rousseau) were invited to be part of a panel discussion highlighting the aims and process of peer-reviewing. The discussion commenced with the importance of peer reviewing in science and the responsibility journal editors have to maintain the integrity and quality of sci ence. Accordingly, the nominated review ers bear the responsibility of scrutinizing the quality and clarity of the scientic con tent in the manuscript and providing con structive criticism to help authors improve it. From the editors perspective, feedback on a manuscript should indicate if the conclusions are new, if the study builds upon the existing literature, and above all assess if the evidence presented supports the conclusions. On the question of How to dier entiate a poor review from a good one?, the panelists collective reply was: A good review demands useful and constructive criticism that identies the strengths and weaknesses of a manuscript and gives rec ommendations with supporting justica tions. As a reviewer, never go by dogmas and dont reject new ideas, if they are supported with data and systematic methodology. If the work is promising, but has not attained its best form, insist that authors take on the burden of more work to justify their novel ideas and research. Always be polite, objec tive and respectful to the author(s), regard less of whether you recommend the editor accepts or rejects the manuscript. A major issue for editors is nding suitable reviewers, as often the best-known ex perts in the eld lack time. Nevertheless, it is ne to politely turn down the oppor tunity of reviewing a paper if you lack the time or the expertise the paper demands, or in the event of conicting research in terests. On the other hand, if you agree to do a review, you should keep your word and not decline the responsibility after several months of doing nothing with the manuscript. The experienced referees on the panel reminded young scientists to be aware of the time pressures involved in reviewing. A thorough review of a paper takes a mini mum of two to three days depending on ones eciency and ability. Anticipating this pressure allows one to set aside ample time to read the paper and let the ideas sink in for a few days, before writing the review. This process helps sharpen ones ideas and thinking on the issues covered in the manuscript. While it remains important that the authors cite the most recent papers rele vant to the study, it is equally important to check if the authors cite the original pio neer works. In case you do not have access to the references the authors cite, you can ask for the papers from the editors rather than provide a poorly informed review. Therefore, working on the review of a manuscript ahead of the deadline allows time to build initial impressions, verify cit ed references and reserve sucient time for constructive feedback, steps that en sure a thorough review. Finally, submitting the report on time also ensures a smooth and ecient evaluation, and makes the Figure 1: Introduction by Henry Oldenburg to the rst issue of the Philosophical Transactions of the Royal Society (Oldenburg 1665).


92 The Art of Data Sharing: key in future climate science AURORA ELMORE 1 F LEHNER 2 AND J. FRANKE 2 1 Department of Geography, Durham University, UK; 2 Oeschger Centre for Climate Change Research, University of Bern, Switzerland A t the PAGES Young Scientists Meeting, 11-12 February 2013, in Goa, India, 79 young researchers from around the world gathered to discuss research, to network, and to exchange ideas for the future of cli mate research. Initiated by a talk on The Art of Data Sharing given by David Anderson, head of the World Data Center for Paleoclimatology at the National Oceanic and Atmospheric Administration (NOAA), a lively discus sion arose on the benefits and potential of data sharing for future research. Fortunately, many researchers al ready upload their data and computer code to an Internet database to be available for future projects. Therefore, a wealth of databases and software exist that are open and easily acces sible (see Box 1). These include data from classical proxy archives such as tree rings, ice cores, lake and marine sediments, as well as model output, re analysis, observations and a multitude new results swiftly available, to the benet of science overall. Ethics are a critical aspect in the art of reviewing, and an area in which young reviewers may need to develop. As a re viewer, one has the moral responsibility to put aside ones own research and publica tion interests, be honest and fair with the authors and consider the interests of the respective journal. The same is true for an editor or reviewer while dealing with con icting reviews or accepting risky ideas. A test of ones integrity as a reviewer is whether or not you are able to put your name to the review, i.e. disclose your iden tity. To ensure the editor can build an adequate assessment of the reviews, it is crucial that the reviewer clearly states which topics are outside his or her area of expertise. This allows the editor to identify additional experts, if needed. One should not shy away from providing detailed re views, so that a paper can be improved to near perfection. This can include provid ing grammar and language amendments if possible. However, the main task of a re viewer is to evaluate the science of a study and not necessarily correct the language of the manuscript. Instead of spending time correcting spelling mistakes at the expense of evaluating the scientic con tent, the reviewer can mention the need for copy-editing to the editor. Although demanding and time-con suming, reviewing manuscripts provides a unique opportunity to improve ones criti cal thinking and writing skills, stay updat ed on cutting-edge research techniques and ensure the quality and integrity of published science. This interactive session on The Art of Reviewing brought to light that about half of the ~80 early-career participants had reviewed papers either on behalf of their supervisors or directly for journals; however, most of them had never received formal training in review ing. Consequently, the young scientists unanimously expressed the wish for for mal training in reviewing as part of their doctoral education. To conclude, the guiding line for a scien tist should be publish or perish ! But at the end of this session, we found a new one: Peer review: love it or hate it, its an integral part of every scientists life. (Welsh 2010). So do not panic if an editor picks you to be the chosen one! Additional information on the art of review ing is available in published articles, e.g. Smith 1990; Provenzale and Stanley 2006; Rosenfeld 2010, and on these dedicated websites: users.;;; cfm?pageid=3331 References Oldenburg H (1665) Philosophical Transactions 1: 1-22 Provenzale JM, Stanley RJ (2006) Journal of Nuclear Medicine Technology 34: 92-99 Rosenfeld RM (2010) Otolaryngology Head and Neck Surgery 142: 472486 Smith AJ (1990) Computer 23: 65-71 Welsh J (2013) Discoblog: Year's Best Peer Review Comments: Papers That "Suck the Will to Live" coblog/2010/12/15/years-best-peer-review-comments-papersthat-suck-the-will-to-live/ Box 1: Examples of databases, software, and sample repositories. The number of databases, open-source software and repositories is growing, providing ex tensive resources for scientists to engage in data-intensive research. Databases Pangaea Data publisher for Earth & Environmental sciences, World Data Center for Paleoclimatology, Neotoma A paleoecology database and community, JANUS Data from the Integrated Ocean Drilling Program, Core Curators Database, the Index to Marine and Lacustrine Samples, www.ngdc.noaa. gov/mgg/curator PAGES list of databases, Software Calib the radiocarbon calibration program, Analogue, Analogue and weighted-averaging methods for paleoecology, Singular Spectrum Analysis A toolkit for spectral analysis, Ocean Data View a software package for the exploration and analysis of oceanographic and other data, KNMI Climate Explorer an online tool to visualize and analyze climate data with a large ready-to-use database,


93 of free algorithms, scripts and software packages. Not only can researchers use these archives to compare with their own new data, but also groundbreaking studies seeing the large-scale picture can result from compiling or reanalyzing existing data sets (e.g. Lisiecki and Raymo 2005; Mann et al. 2008; Andrews et al. 2012). These kinds of data compilation proj ects are time-intensive. However, when fed back to the database, the resulting data can be highly beneficial for the paleoscience community as it avoids duplication of effort. Data compilation efforts can also be funding-efficient, as some funding agencies have already requested proposals specifically based on that approach. Some general ideas for future compilation-style research include time-slice reconstructions and comparison between transient model simulations and sensitivity experiments conducted by different institutes. So, why do not all researchers share their data? The idea of being scooped seems to be one of the most important fears preventing researchers from upload ing their published data; this seems to be particularly important for young researchers who are working towards establishing their careers and thus can not afford to not be credited for their work. But in reality, such cases occur rarely and the vast majority of scientists sharing their data experience only ben efits from it, including more citations and often even additional co-author ships. Another barrier to uploading re search data seems to be the author's worry about data being misused or misrepresented. Providing detailed me ta-information about the data and cor responding error bars greatly reduces the risk of inappropriate use of data. However, the largest hindrance seems to result from confusion about which data repository to use and how to for mat the data. To this end, many data repositories have helpful read-me files and staff support to help with the up loading, so that the researcher hardly has to spend much additional effort. Some journals and funding agen cies are now mandating that authors archive data that appear in publications and discussants at the YSM were united in their hope that this trend towards open access continues. One way to encourage and credit data sharing in the future could be in the form of a data citation index. Usually, data compilation studies do not cite every individual data paper that went into the compilation mainly to prevent the bibliography from ex ploding. A data citation index, follow ing the example of the classical citation index for papers, could provide an ef ficient way of crediting the papers un derlying data compilation studies with out generating lengthy bibliographies. With limitless potential for compi lation studies to generate truly inno vative science and the relative ease of uploading data, we hope many readers consider using these great resources and, of course, helping them to grow. References Andrews T et al. (2012) Geophysical Research Letters 39, doi: 10.1029/2012GL051607 Lisiecki LE, ME Raymo (2005) Paleoceanography 20: 1-17 Mann ME et al. (2008) PNAS 205: 13252-13257 Mann ME et al. (2009) Science 326: 1256-1260 Figure 1: Sharing and combining data and model simulations allows for a better understanding of past climate variability. Temperature anomalies relative to 1961-1990 in a reconstruction (Mann et al. 2009) and a Last Millennium simulation with a comprehensive coupled model (Community Earth System Model). White areas indicate no reconstruction data; the same areas have been masked in the model output for comparability.


94 Breakout Group A: What should the research questions and priorities in paleoscience be for the next 10 years? SYL VIA DEE 1 AND FRANCESCO MUSCHITIELLO 2 1 Department of Geological Sciences, University of Southern California, Los Angeles, USA; 2 Department of Geological Sciences, Stockholm University, Sweden H ow can we improve model-based estimates and predictions? How can we improve the production of paleo data? How can we better constrain past rates of change in the Earth system? These ques tions, among others, were identied as key priorities for future paleoscience dur ing our breakout sessions at the YSM. We identied that model-based cli mate sensitivity estimates and the abil ity to correctly capture climate feedbacks, abrupt transitions, and threshold behavior in models are key to predicting climate and associated changes. Integrated earthsystem modeling with improved feed back interactions will be required to study whole-earth system dynamics. Furthermore, assessing climate mod el performance requires better datasets of high-resolution proxy reconstructions: We need more high-quality data from under-represented regions. We also need new proxies for several climate variables. Also high on our wish list are solid con straints upon previously unresolved cli mate system components such as clouds and aerosols. We require better solar and volcanic forcing reconstructions, and we should strive to understand the underly ing causes of discrepancies between the dierent forcing reconstructions available. To improve the quality of our proxy networks we need to employ replication, high-resolution dating, statistical analysis and multi-proxy approaches in our re search. Data uncertainty estimates should always be clearly stated. Process stud ies and controlled experiments must be used to establish regional calibrations and transfer functions to allow proxy-based reconstructions to capture not only highfrequency climate variability, but also a quantiable climatic parameter such as temperature or precipitation. Finally, we need to compile datasets and make them available in a quality-con trolled, well-documented and easy-to-use form. Strict formats for big data should be employed in a globally acknowledged framework. The eld could vastly benet from larger collaboration with computer software engineers and informatics sci ence to improve eciency and manage ability of earth science datasets. Breakout Group B: Advocating the relevance of paleoresearch to a funding agency or policy maker ILHAM BOUIMETARHAN 1 AND HANS CHRISTIAN STEEN-LARSEN 2 1 MARUM-Center for Marine Environmental Sciences, University of Bremen, Germany; 2 Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, USA D ear policy makers and funders of sci ence, We understand your need to base your decisions and investments on stron ger arguments of the important role that paleo-research plays in international ef forts to understand and emphasize the social, economic, and geopolitical implica tions of Earths changing climate. We are also acutely aware of many unanswered questions and substantial uncertainties that currently exist, and always will exist, in paleo-research, as in any other eld of science. However, it is now evident that better projections of future climate and environmental change, which form the basis of decisions on national and interna tional greenhouse gas emission policies, require consultation with information from Earths past. Repeatedly during its history, Earths climate has changed abruptly within just a few decades. Climatic variability including changes in the magnitude and frequency of extreme events such as droughts and oods has often had a devastating impact on local societies, and events in the future are likely to bring greater environmental risks as the environment is already subject to substantial stress. Furthermore, due to the increased complexity of modern soci eties extreme events are likely to become more costly, as recently illustrated by Hurricane Sandys eect on New York City. Paleoscientists are undertaking enor mous eorts to assess the high complexity of Earths climate system and gain a better understanding of the general forces con trolling global climate change. Integrating local and regional climate information from marine and terrestrial environ ments all over the world has allowed, for example, the identication of important feedbacks in the climate system; this is crucial if we are to avoid being surprised by abrupt climatic events. The results of this research have helped us to better un derstand the role human activities play in causing a large part of the changes in the Earths climate system, namely the sig nicant increase in global temperatures. Moreover, many paleoclimate results are now being eectively assimilated with climate models in order to provide better future projections and predictions of po tential impacts likely to aect people in the short-term and in coming decades. While paleoscientic research is able to highlight some of the environmen tal risks threatening our planet, much remains to be learned, and this kind of research will still need more nancial in vestment in order to thoroughly examine other scientic questions. While it might not provide direct applications, or solu tions for engineering a better future, pa leoscientic research makes a substantial contribution to constraining the possible scenarios of future environmental and cli matic changes. We encourage you there fore to let paleoscience evidence guide you towards wise decisions on policy and science funding in the context of the high priority challenges facing humanity. Yours sincerely, Ilham Bouimetarhan and Hans Christian Steen-Larsen On behalf of the PAGES Young Scientists 2013


95 Breakout Group C: Challenges and solutions for enhanced paleoscience communication HEIDI ROOP 1 AND ELISABETH DIETZE 2 1 Antarctic Research Centre, Victoria University of Wellington, New Zealand; 2 Climate Dynamics and Landscape Evolution, German Research Centre for Geosciences, Germany A t the 2 nd YSM there was consensus among the young international paleoscientists that there is a great need to better develop skills for communicating with dierent types of non-academic audiences. Clear and eec tive communication to the public is becoming increasingly important as current and future climatic and environmental changes are fre quently a major focus in the media and poli tics. However, for the paleoscience community there remains the challenge of properly con veying the concept of past change on longer timescales. Facilitating better public under standing of the scientic process is required to break down barriers and have objective discussions, especially regarding the issue of future climate change. A productive discussion at the YSM about how to address the challenges we face in com municating paleoscience resulted in two po tential solutions: First, as scientists, we need to be proac tive in making our research available in our local communities. Creating connections with internet platforms, classrooms, media outlets and other informal science education venues can be highly productive and rewarding, but also dicult and time consuming to develop. To address this problem and to facilitate pa leoscience communication, we propose to link with the PAGES scientist database and outline researchers availability for specic outreach activities (e.g. classroom visits, blog articles, Skype calls, laboratory tours, radio interviews). This type of additional database should be communicated through educational networks such as Polar Education International (PEI). The ultimate hope is that this freely accessible da tabase can begin building lasting relationships between the public and local researchers by making it easy for the public to nd local sci entists. An encouraging example is the Social Media Knowledge Exchange (, which provides a platform for early career sci entists in history and archeology to share their research with non-academic audiences. Second, the current lack of formal train ing opportunities in science communication is a major obstacle preventing the eective com munication of our research. Coursework and other training opportunities, beyond short workshops, are needed to instruct researchers how to eectively and concisely communicate the signicance of their research to any audi ence. In a highly inter-connected world, it is critical that scientists develop an appropriate level of uency and understanding of how to use communication tools ranging from social media to informal writing. Ultimately, communicating our scientic results should become a regular and profes sionally recognized part of the scientic pro cess. Developing the skills to eectively share our science will undoubtedly increase our broader impacts, and PAGES is in a unique position to facilitate this development in the paleoscience community. Already the YSM has stimulated discussion, and we hope that this dialogue can continue in the broader PAGES network to strengthen and broaden our sci ence communication skills into the future. Breakout Group D: Key educational ingredients to ensure the success of future paleoscientists RA JANI PANCHANG 1 AND JULIE RICHEY 2 1 Agharkar Research Institute, Pune, India; 2 U.S. Geological Survey, St. Petersburg, USA T he breakout sessions at the 2 nd YSM proved to be an extremely useful exercise that re sulted in concrete suggestions for future direc tions for PAGES and the broader scientic com munity. The topic of our group was discussed in two subgroups by a total of 21 participants from 12 countries. One of the key recommendations was that future paleoscience students need bet ter computational skills. In the early days of paleoclimate research, students could turn 50 analyses into a dissertation, but with modern methodological advances, students can now produce hundreds or thousands of geochemi cal measurements. In addition to expertise in micropaleontology, palynology, organic geochemistry, etc., students need to have the quantitative skills to statistically analyze that data, and eectively put it in the context of a wealth of other paleoclimate archives. Paleoclimate modelers should have more training in geosciences so that they can better understand the value and limitations of proxy records. Conversely, those generating proxy records need to be capable of understanding and using model results to make proxy-model comparisons. Students should be encouraged to com plete a small research project before opting for a doctoral program so that they can as sess their interest as well as aptitude for re search. This could be oered as a Bachelors or Masters dissertation, as is already common practice in some countries, e.g. the USA. This also led to the idea of oering supervisors more incentives (e.g. research assistance or teaching time exemption) for investing time and energy in short-term (i.e. masters-level) re search students. Inspired by the panel discus sion on peer reviewing, it was also suggested that reviewing should be made a formal part of graduate education. Over time English has become the sin gle global language bridging international language borders, and thus its knowledge facilitates the eective communication of sci ence. Accordingly, some participants from countries in which English is not the rst lan guage did express the need for formal train ing in reading and writing English. They also wished that some of their science education had been in English. Earth science is currently not part of the required curriculum in many countries, and the need and importance for elevating earth science education at the primary and second ary school level was expressed. Regarding the much more advanced career stage paleocli matologists, participants expressed the con cern that paleoclimate is not as lucrative as mining and petroleum! Although not related to education, the breakout group argued that better incentives and job opportunities will be key conditions to ensure the success of the next generation of paleoscientists.


96 Impressum PAGES International Project Oce Zhringerstrasse 25 3012 Bern Switzerland Tel.: +41 31 631 56 11 Fax: +41 31 631 56 06 Editors: Series Editors: Lucien von Gunten and Thorsten Kiefer Guest Editors: Pascale Braconnot, Chris Brierley, Sandy P. Harrison Text Editing: Leonie Goodwin Layout: Lucien von Gunten Cover image: Photo of a Tridacna spp. giant bivalve by Kevin Welsh. Hardcopy circulation: 2550 ISSN 1811-1602 Printed on recycled paper by Lderach AG Bern, Switzerland The PAGES International Project Oce and its publications are supported by the Swiss and US National Science Foundations and NOAA. 2013 Contents Announcements Inside PAGES 46 20 years of PAGES news 47 PAGES calendar 47 Special Section: El Nio-Southern Oscillation Editorial: El Nio-Southern Oscillation observation and modeling 48 Science Highlights: Myanmar monsoon drought variability over the past 300 years: linkages to ENSO 50 R. DArrigo, J. Palmer, C. Ummenhofer, N.N. Kyaw and P. Krusic Coral microatoll reconstructions of El Nio-Southern Oscillation 52 H.V. McGregor, C.D. Woodroe, M. Fischer, M.K. Gagan and D. Fink Giant clam recorders of ENSO variability 54 M. Elliot, K. Welsh and R. Driscoll Reconstructing ENSO in the Eastern Tropical Pacic from short-lived marine mollusks 56 M. Carr, S. Purca and J.P. Sachs Challenges in understanding and modeling ENSO 58 A. Capotondi, E. Guilyardi and B. Kirtman Coral-model comparison highlighting the role of salinity in long-term trends 60 D.M. Thompson, T.R. Ault, M.N. Evans, J.E. Cole, J. Emile-Geay, A.N. LeGrande The response of pseudo-corals to ENSO in an isotope enabled climate model 62 T. Russon, A.W. Tudhope, G.C. Hegerl, M. Collins and J. Tindall ENSO and changes in the mean state in Holocene simulations 64 P. Braconnot and Y. Luan A controversial insight into Southwest Pacic mid-Holocene seasonality 66 C.E. Lazareth, M.G. Bustamante Rosell, N. Duprey, N. Pujol, G. Cabioch, et al. Interannual variability in the Tropical Pacic during the last glacial period 68 U. Merkel, M. Prange and M. Schulz ENSO behavior before the Pleistocene 70 C. Brierley Mini Section: Data assimilation methods Overview of data assimilation methods 72 G.J. Hakim, J. Annan, S. Brnnimann, M. Crucix, T. Edwards, H. Goosse, et al. Transient state estimation in paleoclimatology using data assimilation 74 S. Brnnimann, J. Franke, P. Breitenmoser, G. Hakim, H. Goosse, et al. Best-of-both-worlds estimates for timeslices in the past 76 T.L. Edwards, J. Annan, M. Crucix, G. Gebbie and A. Paul Parameter estimation using paleo-data assimilation 78 J.D. Annan, M. Crucix, T.L. Edwards and A. Paul Workshop Reports: Impact of climate and sea level change on coastal evolution 80 Holocene land-cover change in Eastern Asia for climate modeling 81 Palaeo50: The priority research questions in paleoecology 82 The Agulhas System: its role in ocean circulation, climate, and marine ecosystems 83 IPICS First Open Science conference 84 Analyzing paleolimnological data with R 85 The Sun and its role in climate change 86 Program News: The backbone of PAGES 2k: data management and archiving 87 Updated Latin American Pollen Database 88 Young Scientists Meeting 2013 A brief report on the 2 nd PAGES Young Scientists Meeting 89 R. Saraswat and B. Jensen The Art of Communicating Science 90 H. Roop, G. Martnez-Mndez and K. Mills The Art of Reviewing 91 R. Panchang, A. Govin and C. Omuombo The Art of Data Sharing 92 Aurora Elmore, F. Lehner and J. Franke Reports from the YSM breakout groups 94 S. Dee, F. Muschitiello | I. Bouimetarhan, H-C. Steen-Larsen | H.Roop, E. Dietze | R. Panchang, J. Richey

PAGES (Past Global Changes) supports research aimed at
understanding the Earth's past environment in order to make
predictions for the future. We encourage international and
interdisciplinary collaborations and seek to promote the
involvement of scientists from developing countries in the
global paleo-community discourse.
PAGES scope of interest includes the physical climate
system, biogeochemical cycles, ecosystem processes,
biodiversity, and human dimensions, on different time
scales Pleistocene, Holocene, last millennium and the
recent past.
Founded in 1991, PAGES is a core project of the
International Geosphere-Biosphere Programme (IGBP) and is
funded by the U.S. and Swiss National Science Foundations,
and the National Oceanic and Atmospheric Administration
It is overseen by a Scientific Steering Committee (SSC)
comprised of members chosen to be representative of the
major techniques, disciplines and geographic regions that
contribute to paleoscience.


Download Options

Choose Size
Choose file type
Cite this item close


Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.


Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.


Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.


Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.