Climate and environmental evolution scenarios for the Meuse/Haute-Marne region, France

Transcription

1 Mém. Soc. géol. France, n.s., 2007, n 178, pp Climate and environmental evolution scenarios for the Meuse/Haute-Marne region, France VALÉRIE ANDRIEU-PONEL 1,DELPHINE TEXIER 2,JACQUES BRULHET 2,JACQUES-LOUIS de BEAULIEU 1, CAROLE BÉGEOT 3,RACHID CHEDDADI 4,EMMANUEL GANDOUIN 1,CLAIRE GOODESS 5,FRÉDÉRIC GUITER 1, BRUNO HAMELIN 6,MARIE-FRANCE LOUTRE 7,MICHEL MAGNY 3,VALÉRIE MASSON-DELMOTTE 8, DIDIER PAILLARD 8,NICOLE PETIT-MAIRE 9,PHILIPPE PONEL 1,EDWIGE PONS-BRANCHU 6*, MAURICE REILLE 1,PASCALE RUFFALDI 3,MIKE C. THORNE 10 and BRIGITTE VAN VLIET-LANOË 11 Keywords. Palaeoclimate, Palaeoenvironment, Periglacial, Climate modelling, Feasability of a radioactive waste repository in deep geological formations, France. Abstract. This document assesses the climate and environmental evolution of the Meuse/Haute-Marne (M/HM) site and of neighbouring regions from the end of the Saalian glaciation (ca. 150,000 years Before Present (B.P.), OIS 6) to 1 million years After Present (A.P.). It relies not only on the multidisciplinary study of a number of climatic and palaeoenvironmental archives, but also on various numerical climate simulations. The biological, sedimentological and geochemical indicators used showed that, during the last 150,000 years, significant climatic and environmental variations have occurred over pluri- to infra-millennial timescales. The mean annual temperature of the M/HM region (currently 10 o C) has varied from interglacial maxima in the order of 12 o C to glacial minima ranging from 6 o C to 12 o C. The vegetation has oscillated between stages of forest cover during interglacial periods to stages of herb cover with scarce trees (tundra to bush tundra) that were adapted to the cold and dryness of glacial maxima. The soil and the subsoil, ice-free during interglacial periods, were subject to deep multi-year freezing as soon as average annual temperatures fell below 0 o C for several months. During the last glacial maximum, some 18,000 years B.P., the permafrost may have reached a depth of 120 m in the M/HM region [Van Vliet Lanoë, 2004]. With regard to the future climate evolution of the region, several scenarios are possible. If the disturbance caused to the climate system by the combustion of fossil fuels is not taken into account, it is likely that the current interglacial period will continue for several thousands or even tens of thousands of years. Nevertheless, the initiation of our current climate cooling, between 5,000 and 2,000 years B.P., as recorded by the dynamics of the vegetation, shows that we are moving toward the end of an interglacial period. Numerical simulations carried out under the EC BIOCLIM Project [Texier et al., 2003; Texier et al., 2007] have indicated that our interglacial period will last for another 50,000 years A.P., whereas chronological models bring that figure down to 3,000-20,000 years A.P. However, the prospects change if the impact of anthropic disturbances on the climate, which are thought to have become significant since the beginning of the 20 th century, are taken into account and if those disturbances are extrapolated to the future. According to the BIOCLIM simulations, which represent the impact of a substantial injection of carbon dioxide into the atmosphere as a result of fossil fuel consumption, the end of our interglacial period may be delayed by 100,000 or even 180,000 years. Other contributions, e.g. methane and sulphate aerosols, are likely to modify this conclusion to only a limited degree. During the next few centuries ahead, the climate of the M/HM region would therefore move towards warmer temperatures than the current temperate ones. The consequence of that climate-warming process would be felt over several hundreds of thousands of years. In particular, it would take at least 600,000 years, if not 850,000 years for the regional climate to undergo a cooling sufficient to induce a glaciation similar in magnitude to those peak glacial episodes that have occurred over the last 850,000 years. Scenarii d évolution climatique et environnementale pour la région Meuse/Haute-Marne, France Mots-clés. Paléoclimat, Paléoenvironnement, Périglaciaire, Modélisation du climat, Faisabilité d un stockage de déchets radioactifs en formations géologiques profondes, France. Résumé. Ce document présente l évolution du climat et des conditions environnementales du site Meuse/Haute-Marne (M/HM) et des régions voisines, depuis la fin de la glaciation saalienne (ca ans Before Present (BP), OIS 6) 1 IMEP (Institut méditerranéen d écologie et de paléoécologie), CNRS-UMR 6116, Europôle méditerranéen de l Arbois, BP 80, Aix-en-Provence cedex 04, France. 2 Andra, Agence nationale pour la gestion des Déchets Radioactifs, Direction Scientifique, Service Milieu géologique, 1-7 rue Jean Monnet, Châtenay-Malabry cedex, France. 3 Laboratoire de Chrono-écologie, CNRS-UMR 6565, Université de Franche-Comté, La Bouloie, 16, route de Gray, Besançon cedex, France. 4 ISEM (Institut des sciences de l évolution), CNRS UMR 5554, Université Montpellier II, Place Eugène-Bataillon, Case postale 61, Montpellier cedex 05, France. 5 University of East Anglia, Abbotsleigh Kebroyd Mount, Ripponden, Halifax, West Yorkshire HX6 3JA, U.K. 6 Centre Européen de Recherche et d Enseignement des Géosciences de l Environnement CEREGE, CNRS-UMR 6635, Europôle méditerranéen de l Arbois, BP 80, Aix-en-Provence cedex 04, France. 7 Université catholique de Louvain, Chemin du Cyclotron, 2, 1348 Louvain-la-Neuve, Belgique 8 Laboratoire des sciences du climat et de l environnement LSCE, IPSL-CEA/CNRS, Orme des Merisiers Bâtiment 701, CEA Saclay, Gif-sur-Yvette cedex, France. 9 Maison Méditerranéenne des Sciences de l Homme MMSH, 5, rue du Château-de-l Horloge, BP 647, Aix-en-Provence cedex 2, France. 10 Mike Thorne and Associates Limited, Abbotsleigh, Kebroyd Mount, Ripponden, Halifax, West Yorkshire HX6 3JA, UK. 11 Processus et bilans des domaines sédimentaires SN5, CNRS-UMR 8110, Université des sciences et techniques de Lille, Villeneuve-d Ascq cedex, France.

2 116 ANDRIEU-PONEL V. et al. jusqu à la fin du prochain million d années. Il repose non seulement sur l étude multidisciplinaire d un ensemble d archives climatiques et paléoenvironnementales, mais aussi sur des simulations numériques du climat. Les marqueurs biologiques, sédimentologiques et géochimiques étudiés montrent qu au cours des derniers ans, d importantes variations climatiques et environnementales se sont produites, en des laps de temps pluri- à infra-millénaires. La température annuelle moyenne en M/HM (actuellement de 10 o C) est passée par des optima thermiques interglaciaires de l ordre de 12 o C et des minima glaciaires de l ordre de 6 o C à 12 o C. Le couvert végétal a oscillé entre des états de végétation forestière pendant les périodes interglaciaires et des états de végétation herbacée quasi dénuée d arbres (toundra à toundra buissonnantes) qui étaient adaptés au froid et à la sécheresse des maxima glaciaires. Le sol et le sous-sol, libres de glace pendant les périodes interglaciaires, ont été sujets à un gel pluriannuel profond dès que les températures annuelles passaient sous le seuil de 0 o C et restaient négatives pendant plusieurs mois. Lors du dernier maximum glaciaire, il y a ans, le pergélisol serait descendu jusqu à 120 m environ en région Meuse/Haute-Marne [Van Vliet Lanoë, 2004]. En ce qui concerne l évolution future du climat de M/HM, plusieurs scénarii sont possibles. Si l on ne prend pas en compte la perturbation que représente pour le système climatique la combustion d énergies fossiles, il est probable que notre interglaciaire persiste encore quelques milliers voir quelques dizaines de milliers d années. Toutefois, l amorce d un refroidissement climatique, entre et ans B.P., enregistrée par la dynamique de la végétation, indique une fin prochaine pour la période interglaciaire actuelle. Les simulations numériques issues du projet européen BIOCLIM [Texier et al., 2003; Texier et al., in prep.] donnent une durée de ans pour la persistance de notre interglaciaire, tandis que les modèles chronologiques donnent une durée de ans à ans. Cependant, si l on prend en compte la perturbation anthropique, dont l effet sur le climat est estimé être significatif depuis le début du 20 ème siècle, et si l on extrapole cette perturbation dans le futur, les perspectives sont différentes. D après les simulations BIOCLIM, qui ne concernent que l impact d une injection importante de dioxyde de carbone dans l atmosphère, la fin de notre interglaciaire pourrait être retardée à ans voire ans dans le futur. Prendre en compte l impact d autres gaz à effet de serre, tels que le méthane ou les aérosols sulfates, ne pourrait modifier que de peu cette perspective. Au cours des prochains siècles, le climat de M/HM basculerait vers des états plus chauds que l état tempéré actuel. L écho de ce réchauffement climatique serait ressenti pendant plusieurs centaines de milliers d années. En particulier, ans voire ans pourraient passer avant que le climat régional ne connaisse un maximum glaciaire d une amplitude équivalente à ceux rencontrés pendant les derniers ans. INTRODUCTION As soon as consideration was given to the feasibility of developing a deep geological repository for high-level long-lived waste (HLLLW) in the Meuse/Haute-Marne (M/HM) region, the issue was raised as to the impact of climate variations on the geological environment of the repository and on the overlying biosphere. Geoprospective studies co-ordinated by Andra were launched with an aim to identify not only the maximum amplitude of past and future climate episodes likely to modify significantly the M/HM landscapes, but also the time frames of those episodes. A major question was the possible difference between the future and the past, considering that the Quaternary period would cease to be a suitable analogue for the future. With regard to a deep geological repository, glacial periods are the main focus, because they are characterised by particularly efficient erosion processes and development of permafrost. Interglacial periods, with their mild climate conditions and suitable conditions for the settlement and development of human communities, also needed to be documented in detail. Through various studies, it was possible to summarise all available information on the history of past environments and climates in mid-european France, to understand the mechanisms underlying climate change and to simulate the future evolution of both climate and landscape. Those studies also helped to identify not only the local consequences of the major glacial-interglacial cycles prevailing over the last 3 million years, but also the influence of rapid climate events occurring in the North Atlantic during glaciation and deglaciation periods [Imbrie and Imbrie, 1979]. Cross-analyses of the different types of climate archive and numerical simulations relating to potential future evolution were instrumental in developing a scenario for the climate history of the M/HM region between 150,000 years Before Present (B.P.) and 1 million years After Present (A.P.). Although French palaeoclimate sequences for the last 450,000 years were available, those dates were selected because the last 150,000 years is considerably better documented than the earlier periods. As far as the future is concerned, international recommendations [IAEA, 1995] for safety studies relating to HLLLW repositories suggest that a period of about 1 million years be typically taken as a basis for performance assessment. Andra s geoprospective investigations were conducted in the context of theses [Pons-Branchu, 2001; Danis, 2003; Gandouin, 2003], multi-year and multidisciplinary working groups (ARMINES/ Andra ; Groupement de Laboratoires en iogéoprospective : Brulhet [2001]) and through the EC BIOCLIM Project [ co-ordinated by Andra between 2000 and 2003 [Texier et al., 2003; Texier et al., in prep.]. Those investigations were carried out by the partner research laboratories listed in table I. The information collected or simulated over a large-scale area (i.e., western Europe, the northern Hemisphere, or the world, depending on tools used) helped to quantify the local climate evolution for the specific case of the M/HM region and to situate that regional evolution in a larger-scale climate framework. The reconstitution of the past climate evolution of the M/HM region relies on the multidisciplinary study of a series of different climate archives pertaining to other sites near that region (Lorraine Plateau) or further away (Massif central, Alps, Vosges) (see fig. 1). The nature and geographical context of the palaeoclimate archives used are presented in Past climate evolution of the M/HM region-climate archives. The past climate and

3 CLIMATE AND ENVIRONMENTAL EVOLUTION SCENARIOS (MEUSE/HAUTE-MARNE, FRANCE) 117 TABLE I. List of Andra s scientific partners for climate and environmental studies. TABL. I. Liste des partenaires scientifiques de l Andra, en ce qui concerne les études portant sur le climat et l environnement. Laboratory CEREGE (Centre européen de recherche et d rnseignement des géosciences de l environnement), UMR 6635 du CNRS, BP 80, Aix-en-Provence cedex 04, France National French Committee of the International Union for Quaternary Research; Commission de la carte géologique du monde, 77, rue Claude-Bernard, Paris, France École des mines de Paris, CGEJ 35, rue Saint-Honoré Fontainebleau, France IMEP (Institut méditerranéen d écologie et de paléoécologie), UMR 6116 du CNRS, Europôle méditerranéen de l Arbois, BP 80, Aix-en-Provence cedex 04, France ISEM (Institut des sciences de l évolution de Montpellier), UMR 5554 du CNRS, Université Montpellier II, Place Eugène-Bataillon, Case postale 61, Montpellier cedex 05, France Laboratoire de Chrono-Ecologie, UMR 6565 du CNRS, Université de Besançon, La Bouloie, 16 route de Gray, Besançon cedex, France UMR 8110, Processus et bilans des domaines sédimentaires SN5, Université des sciences et techniques de Lille, Villeneuve-d Ascq cedex, France LSCE: Laboratoire des sciences du climat et de l environnement, UMR CEA/CNRS, Centre d étude de Saclay, Orme des Merisiers, Gif-sur-Yvette cedex, France UEA: University of East Anglia, University PlainNorwich, Norfolk NR4 7TJ, United Kingdom UCL/ASTR: Université catholique de Louvain, Chemin du Cyclotron, 2, B-1348 Louvain-la-Neuve, Belgium Contributions and tools Dating U/Th, TIMJ of speleothems of the M/HM Region and of the Gard Department. Palaeoclimatic applications. Summary map. Reconstructions of palaeoenvironments in France during the last climate extremes. CLIMEX Project. Summary map of the work during the last climate extremes. Work of the Andra-École des mines Geoprospective Group: phenomenological analyses of past climate evolutions, chronology, beginnings and ends of glaciation (Ordovician, Plio-Quaternary, etc.) Palaeoclimate, palaeoenvironment, biological markers (pollen, diatoms, coleoptera), lake sites (palaeo) in the Vosges, Dombes and Massif central, the last 410,000 years Climate reconstruction, current pollen analogues Palaeoclimate, palaeoenvironment, pollen, river and peat-bog sites in Lorraine, the last 20,000 years Palaeoclimate, palaeoenvironment, geomorphology, periglacial Climate modelling and climate reconstructions Climate modelling Climate modelling environmental evolution scenarios of the M/HM site and of surrounding regions are presented in Past climate evolution of the M/HM region-plurimillenial variations and Inframillenial variations. FIG. 1. Location of data sites used for palaeoenvironmental and palaeoclimate reconstructions: 1. M/HM Site; 2. Marly; 3. Harmignies; 4. La Grande Pile; 5. Les Echets; 6. Velay and Ribains. FIG. 1. Localisation des sites de données utilisées pour les reconstructions paléoclimatiques et paléoenvironnementales. The projection of the future climate evolution in the M/HM region relies on a series of numerical simulations performed under the BIOCLIM Project. In this document, only a selection of the results is presented for three potential climate evolution scenarios in the M/HM region over the next million years. Those scenarios are taken from the regional outputs of the global LLN-2D-NH model of the Université catholique de Louvain [Gallée et al., 1991; BIOCLIM, 2001; Loutre and Berger, 2005]. The model and the related method are described in Future climate evolution of the M/HM region-lln-2d-nh model and regionalisation method. Plurimillennial climate oscillations described by those scenarios are driven by future variations in insolation and carbon-dioxide concentrations in the atmosphere (pco 2 ). Extrapolations made for pco 2 are based on the hypotheses proposed by the International Panel on Climate Change [IPCC, 2001; BIOCLIM, 2001; Paillard et al., in prep.]. Three possible future pco 2 evolutions were tested ( Future climate evolution of the M/HM region-input data: insolation and pco 2 ). The first, called the natural-evolution scenario, does not take into account the disturbance caused by the combustion of fossil fuels on the climate system. The other two, called disturbed-evolution scenarios, correspond to two different intensities of anthropic disturbance. The three BIOCLIM climate-evolution scenarios are discussed in Future climate evolution of the M/HM region Plurimillennial variations. The

4 118 ANDRIEU-PONEL V. et al. potential occurrence of rapid (infra-millennial to infra-secular) climate variations in the climate-warming context is dealt with in Future climate evolution of the M/HM region-inframillennial variations. PAST CLIMATE EVOLUTION OF THE M/HM REGION Over recent years, Andra has co-ordinated several summaries of collected information on past climate changes over the entire world and in France [ARMINES/Andra partnership work, , unpublished; Brulhet and Petit-Maire, 1999; Petit-Maire and Bouysse et al., 1999; Keller et al., 2001; Guiter et al., 2003; Andrieu-Ponel et al., 2003]. Those summaries provided a snapshot view of the state of the environment at certain key periods of the past and a better comprehension of climate dynamics over the last glacial/interglacial cycles in mid-european France. The focus here is on the last 150,000 years, since that period is best documented. Figure 2 shows a simplified evolution of the climate in the M/HM region, and references are made to it throughout this section. This schematisation illustrates mainly the plurimillennial variations discussed in Past climate evolution of the M/HM region. Climate archives The climate archives supporting the palaeoclimate reconstructions extrapolated to the M/HM region are derived from ancient lakes, alluvial deposits, peat bogs or speleothems. They are located at distances varying from a few kilometres to 400 km from the M/HM site (fig. 1). Information on the last climate cycle is taken from the sequence of La Grande Pile [Woillard, 1975, 1978; Beaulieu and Reille, 1992; Ponel, 1995], Les Échets [Beaulieu and Reille, 1984 and ongoing studies], and Velay [Reille and Beaulieu, 1990; Andrieu et al., 1995; Reille et al., 2000; Rioual et al., 2002]. For the last 20,000 years, data were collected in the wet environments of Lorraine [Bégeot et al., 2001]. Rapid climate fluctuations with a strong amplitude were assessed in studies dealing with the Massif central [Rioual et al., 2001], the Alps [Danis, 2003; Jouzel, 1999; Von Grafenstein et al., 1998] and the Nord Department [Gandouin, 2003]. The information provided by periglacial structures found in the Paris Basin, as well as in Lorraine, Sologne and Brie, were taken into account (ARMINES/Andra scientific watch conducted between 1991 and 2002). Lastly, the speleothems of the Lorraine and Gard Plateaux proved to be good climate markers [Pons-Branchu, 2001; Pons-Branchu et al., 2003; Losson et al., submitted]. Most of those sequences were subjected to multiproxy analyses, including various markers: biological (pollen, spores, plant and animal macro-remains), sedimentological (sedimentary facies, grain-size analysis) and geochemical (isotope, carbonate and organic-matter studies). The influences on climate archives of their geographical context (altitude, continentality, topography, rock substratum, etc.) and of the analytical methods selected were taken into account by using various transfer functions [Guiot, 1990; Defaut and Courbouleix, 1991 and 1992; Lachenbruch and Marshall, 1986; Lachenbruch et al., 1988; Koster and Judge, 1994; Rousseau, 1991; Rousseau et al., 1990; Van Vliet-Lanoë, 1988, 1989 and 1996]. Plurimillennial variations The last 150,000 years include the end of a climate cycle (end of the Saalian glaciation), a full climate cycle (Eemian and Weichselian) and the current climate cycle that started ca 10,000 years B.P. All climate cycles comprise both an interglacial and a glacial; by definition they always start with an interglacial period. For the last 850,000 years, each climate cycle spread over ca 100,000 years. Every interglacial consists of a progressive phase during which the various vegetation types move from a pioneer stage (with Pinus, Betula and steppic herbs) to a forest stage characterised by the presence of a diversified Quercus woodland with or without Taxus. That forest phase corresponds to a climate optimum, and is followed by a temperature decrease marking the beginning of a regressive phase characterised by the settlement of a Carpinus woodlands first and of a Fagus forest afterwards (except during the Eemian) and then by coniferous trees dominated mostly by Abies, but also by Picea and Pinus. The end of any interglacial is always marked by the expansion of an open pine woodland comprising Pinus, Betula and Juniperus. The duration of interglacial periods is still being debated: proposed values vary between 13,000 [Turner, 2002] and 17,000, or even 23,000 [Kukla et al., 1997, 2002b] and 30,000 years [MacManus et al., 2002]. Glacial periods last much longer than interglacial periods and, contrary to a long-standing belief, are far from being uniform from either a climatic or an environmental standpoint. The beginning of a glacial period (Preglacial) alternates between two cold and wet stadial (during which ice accumulates in high and very high altitudes) and two temperate interstadials. It is the case for the last years at least as proved in the French Massif Central long sequences [Reille et al., 2000]. During the maximum ice-sheet period (Pleniglacial), the climate is not only very cold, but fluctuates very significantly. It oscillates between glacial maxima and less cold phases during which a pioneer and sometimes forest (taiga-type) vegetation may temporarily arise. During the Lateglacial period, climate conditions become milder and a boreal vegetation develops, either mixed or not with thermophilous elements. The end of a glaciation period is marked by a short phase (a few decades) of rapid warming, announcing the beginning of a Postglacial (or interglacial) period (at least for the Eemian and the Holocene). End of the Saalian glaciation During the Saalian glaciation (Saalian II, Riss II, OIS 6), the glaciers located in the valleys of the major French mountains spread over their respective piedmonts, especially in the case of the Rhône Glacier which pushed its glacial front to Lyon, but without covering the Dombes Plateau. At that stage, glaciers had a larger extension greater than that achieved during the next glaciation (Weichselian glaciation), but that was smaller than that achieved during past glaciations [Brulhet et al., 1993]. In the plains of northern France, the soils were in the discontinuous permafrost domain [Loyer et al., 1995; Van

5 CLIMATE AND ENVIRONMENTAL EVOLUTION SCENARIOS (MEUSE/HAUTE-MARNE, FRANCE) 119 Vliet-Lanoë, 1996] as is shown by the presence of frost-shrinkage cracks, with annual temperatures varying between 3 o C and 6 o C. Pollen assemblages dating from that glaciation reflect the existence of vast areas of Poaceae and Artemisia. The end of the Saalian glaciation was marked by a strong cooling associated with the release of icebergs in the North Atlantic [Kukla et al., 2002a], known as Heinrich 11. That event was also recorded on the continent, particularly in the new Les Échets profile (EC1) investigated by the Les Échets Working Group (valerie.andrieu@cezanne.fr) financed by Andra, the French CNRS, as well as North American (NSF) and Swedish funds. The glacial/interglacial transition is marked by a very substantial increase in winter temperatures (January) varying between 15 o C and 20 o C [Cheddadi et al. 1998; Klotz et al., 2003a; Seidenkrantz et al., 1996]. Annual precipitation increased from ca 400 mm to ca 800 mm. Eemian The Eemian (OIS 5e, 126, ,000 year BP, Event 1 in figure 2) was marked by the early settlement of a pioneer arboreal vegetation, followed first by a pure Quercus forest and mixed later with Corylus, Taxus and Carpinus. The climate was typically caracterised by mild winters and hot summers with heavy rainfall (fig. 3c) and by a low annual temperature range [Rioual et al., 2000; Cheddadi et al., 1998; Reille et al., 2000]. The summer insolation was particularly high in the northern Hemisphere due to the orbital configuration (the maximum occurred at 128,000 years B.P. [Berger, 1978a, b] and the annual temperature was more than 2 o C higher than at the present day [Cheddadi et al., 1998]. That climate optimum occurred at the beginning of the Eemian (fig. 3). As soon as Carpinus forest was fully developed, climate conditions had begun to deteriorate. Temperature and annual precipitation decreased and the seasonal variation in temperature increased [Cheddadi et al., 1998; Rioual et al., 2001]. The disappearance of Corylus from the vegetation was accompanied by a new phase of climate cooling allowing a fir forest to grow. The climate became much FIG. 2. Evolution scenario of mean annual temperatures in the M/HM region during the last 120,000 years. FIG. 2. Scenario d évolution de la température moyenne annuelle de la région M/HM au cours des derniers ans. cooler (especially in the summer) and wetter [Rioual et al., 2001; Klotz et al., 2003b] when the first traces of Picea appeared. Whereas the vegetation has started to open up, the Abies forest underwent a new expansion phase on all French Eemian sites; those facts suggest a last increase episode of winter temperatures. At the end of the interglacial, an open taiga spread out accompanied by severe climate-cooling conditions and a decrease in rainfall over the Massif Central (fig. 3c). During the Eemian, the mean annual temperature was estimated at ca 11 o C in the M/HM region, that is, slightly in excess of the current annual temperature evaluated at ca 10 o C (fig. 2). In the Massif Central, at an altitude of ca 1,100 m, the mean annual temperature decreased from ca 12 o C at the beginning of the Eemian to ca 6 o Cattheendof the interglacial (fig. 3c). Sea temperature was much higher, since Lusitanian molluscs were found as far north as Jutland (Denmark). The average sea level was ca 5 m above the current level. Pre-Weichselian stadials and interstadials During Melisey 1 (OIS 5d, 110, ,000 years B.P., Event 2 in figure 2), the climate was cold and wet. The preservation of a snow cover on the Barrois Plateau and of a bush tundra seems to limit the development of permafrost. The M/HM region was only affected by seasonal frost as opposed to the nearby site of Harmignies (Belgium, south of Mons, a windy cuesta) where traces of ice segregations have been detected [Van Vliet-Lanoë, 1986; Courbouleix et al., 1998]. Throughout that stadial, lasting for ca 6,000 years, the estimated average temperature varied between ca 0 o C and 2 o C in the M/HM region (fig. 2). The climate became milder during the Saint-Germain 1 interstadial (OIS 5c, 104,000-92,000 years B.P., Event 3 in figure 2), which was marked by the development of a coniferous forest with Pinus, Betula and Picea, succeeded by a regional and non-local diversified deciduous Quercus woodland in the Velay and Vosges areas [Beaulieu and Reille, 1992]. The absence of thermophilous elements (Hedera, Ilex, Viscum) implies low winter temperatures, thus allowing the existence of Carpinus, Abies and Picea forests especially in low-altitude areas. The Saint-Germain 1 was interrupted by a short climate cooling, called the Montaigu [Reille et al., 1992; Beaulieu and Reille, 1992, Reille et al., 2000], inducing a spectacular destruction of all thermophilous and mesophilous forests (deciduous Quercus Carpinus and Abies forests) and the development of steppic grassland. According to Beaulieu and Reille [1984], the climate was colder and dryer at the beginning of the Montaigu than at the end. Those cooling events were also recorded in the soils from the coast throughout Europe [Van Viet-Lanoë, 1986, 1988], especially at Harmignies (Belgium) and Achenheim (eastern France). During the upper part of the Saint-Germain 1, a mixed Quercus forest (without Taxus) developed once again at low altitude [Beaulieu and Reille, 1984], followed first by a Carpinus woodland and later by a mixed forest with coniferous and broad-leaved trees. The Fagus forest present at Les Échets [Beaulieu and Reille, 1984] failed to develop in regions where the climate was more rigorous (Velay and Vosges). At the end of this interstadial, a forest started to grow, first with Picea, and later mostly with Pinus and

6 120 ANDRIEU-PONEL V. et al. FIG. 3. Climate and vegetation during the Eemian according to Cheddadi et al. [1998] and Rioual et al. [2001]. FIG. 3. Climat et végétation durant l Eémien, d après Cheddadi et al. [1998] et Rioual et al. [2001]. Betula. On average throughout the St Germain 1 that lasted ca 12,000 years, the estimated annual temperature ranged between 3 o Cand6 o C in the M/HM region (fig. 2). During the Melisey 2 (OIS 5b, 93,000-84,000 years B.P., Event 4 in fig. 2), the climate was cold but less wet than during the previous stadial, thus allowing the first loess of the Weichselian glaciation to settle in Belgium. Ice wedges developed at Harmignies but, in the Somme Department, only frost-shrinkage cracks and soil wedges have been detected. On the Barrois Plateau, the estimated annual temperature ranged between 6 o C and 8 o C (fig. 2). The vegetation was prostrate and unable to hold back the snow. All suitable conditions for the development of continuous permafrost existed over the Barrois Plateau. Modelling work performed by Courbouleix et al. [1998] with the software GELSOL 12 [Blanchard and Frémond, 1982, 1985a-b; Blanchard et al., 1990] shows that the 0 o C isotherm may have reached a depth of 72 m, a figure that matches those currently observed in Lapland and in eastern Canada. In the bottom part of valleys, however, the soils remained unfrozen. In the rest of France, a steppe with Poaceae and Artemisia developed, whereas in the least exposed areas to severe climate conditions (piedmonts), a shrub steppe with Juniperus and Betula developed. According to Klotz et al. [2003b], the annual temperature during Melisey 2 was ca 1 to 4 o C below that of Melisey 1, but summer temperatures of both stadials were comparable. On average, during the 12 GELSOL was initially used by the LCPC (Laboratoire Central des Ponts et Chaussées, France) to simulate the impact of freezing and thawing on roads structure. Melisey 2 that lasted ca 9,000 years, the annual temperature varied from 5 o Cto 3 o C in the M/HM region (fig. 2). During the Saint-Germain 2 (OIS 5a, 84,000-72,000 years B.P., Event 5 in fig. 2), a mixed forest with Quercus, Carpinus and Picea started to grow, followed first by a spruce stand mixed with Carpinus and later a pine forest. At the end of that interstadial, a steppe developed. The mean annual temperature throughout St Germain 2 ranged between ca 5 o Candca 2 o C in the M/HM region. Weichselian Pleniglacial The long Weicheselian Pleniglacial started with a maximum cooling phase (OIS 4; 72,000-60,000 years B.P., Event 6 of fig. 2) during which French glaciers and North-European ice sheets reached a maximum expansion. The Netherlands, Belgium and northern France were subject to a periglacial climate causing the soils to freeze. The work compiled by Courbouleix et al. [1998] shows that, in northeastern France, the climate was continental, cold and dry. At Harmignies, Belgium, the soils would have been frozen for a large part of the year and would have thawed during the summer throughout a maximal thickness of 2 m. The annual temperature was ca 5 o C at Harmignies and 8 o Cinthe M/HM region. In the Barrois, a permafrost seemed to extend, continuous on the plateau and discontinuous in the valleys. According to numerical-simulation applications [Courbouleix et al. 1998] the 0 o C isotherm may have reached a depth of 112 m, a figure consistent with present-day analogues. On average throughout that cold period spreading over ca 12,000 years, the annual temperature ranged between 4 o C and 8 o C in the M/HM region (fig. 2).

7 CLIMATE AND ENVIRONMENTAL EVOLUTION SCENARIOS (MEUSE/HAUTE-MARNE, FRANCE) 121 At the transition between OIS 4 and OIS 3 (ca 60,000 years B.P., Event 7 in fig. 2), the climate became dryer and favoured the development of real allochthonous loess. The periglacial structures observed in the field attest an annual temperature in the order of 7 o C at Harmignies and of 9 o C on the Barrois Plateau. The values obtained from simulation studies based on physical data from the M/HM site [Courbouleix et al., 1998] are lower and varied between 2 o Cand0 o C (fig. 2). The permafrost was continuous on the Plateau and the 0 o C isotherm was able to reach a depth ranging between 55 m and 100 m, according to the different estimations [Courbouleix et al., 1998]. In the valleys, the permafrost was discontinuous. The OIS 3 phase (60,000-25,000 years B.P., Events 8 and 9 in figure 2) was characterised by climatic instability during which cold and less cold phases alternated in a glaciation context. At Harmignies and in northern France, loess settled and ice-wedge networks developed. Event 8 corresponds to a period of cold and dry climate. The estimated mean annual temperature ranged from 5 o Cto 9 o C. Those conditions favoured the preservation of a continuous permafrost over the entire Barrois and the 0 o C isotherm was able to reach a depth of ca 200 m [Courbouleix et al., 1998]. During Event 9, a relative warming (of 1 o Cto2 o C in annual average) allowed the development of a scattered shrub vegetation. On the Barrois Plateau, the permafrost was discontinuous and the 0 o C isotherm penetration was estimated at ca 100 m in depth [Courbouleix et al., 1998]. The last 10,000 years of glaciation (OIS 2, 25,000-15,000 years B.P., Event 10 in figure 2) were the most dramatic for the living world in medium and high altitudes. A maximum cold and dryness was reached ca 25,000 years B.P. after the Heinrich 3 event [Frechen et al., 2001]. Precipitation started to increase again ca 23,000 years B.P. At ca 18,000 years B.P., a new cooling period was recorded (fig. 4) in response to a decrease in insolation and to a new iceberg-release phase in the North Atlantic (Heinrich 2). That cooling was not sufficient to cause the Scandinavian ice sheets [Mangerud et al., 1998] and the French glaciers [Monjuvent and Nicoud, 1988; Andrieu et al., 1988] to spread over the limits reached 5,000 years earlier during the late Weichselian glacial recurrence [between 25,000 and 23,000 years BP: Andrieu, 1991, Triganon, 2002; Guiter, 2003; Guiter et al., 2005]. In France, the climate was of dry-periglacial type and it was at that time that the main Weichselian loess formations started to deposit [Frechen et al., 2001]. Permafrost was regionally continuous and reached its maximum thickness. Numerical simulations show that the 0 o C isotherm may have reached a depth of ca 125 m to ca 315 m, depending the scenarios, therefore an average depth of 220 m [Courbouleix et al., 1998], for an effective frost limit at ca 120 m [Van Vliet Lanoë, 2004]. In such a severe and dry environment, only a steppe with Poaceae and no trees could resist. Estimated annual temperatures were ca 10 o C at Harmignies and ranged between 12 o C and 6 o C on the M/HM site. At the same time, sea level was at its lowest (ca 120 m), with the maximal continentalisation of the climate. End of the Weichselian glaciation The Lateglacial (OIS 2, 15,000-13,000 years B.P., Event 11 in figure 2) was marked by significant climate fluctuations. Pollen and isotope data show that between 15,000 and FIG. 4. France during the Last Glacial Maximum: 18,000 ± 2,000 years B.P. according to Brulhet and Petit-Maire [1999]. FIG. 4. La France au dernier maximum glaciaire: ± ans B.P., d après Brulhet et Petit-Maire [1999].

8 122 ANDRIEU-PONEL V. et al. 13,000 years B.P., the temperature rose [Andrieu, 1991], thus allowing the expansion of grasslands with Artemisia in western Europe and with Poaceae and Cyperaceae on the Lorraine Plateau and on low-altitude sites in the Haute-Saône Department [Bégeot et al., 2001; Ruffaldi, 2000 and Woillard, 1975]. That phase of plant colonisation on melting soils was temporarily interrupted by a brief cooling period between 14,000 and 13,000 years B.P. (Pre-Bölling, Henrich 1). About 13,000 years B.P. ago, the Lateglacial interstadial began. This lasted from 13,000 to 10,700 years. The initial phase of the Lateglacial was characterised by a rapid temperature rise generating values close to current ones. Those conditions favoured the development of shrub vegetation (high frequencies of Salix in Lorraine, associated with Betula and Juniperus). The presence of grasslands with Poaceae within which Artemisia was spreading, shows that the vegetation was very open. During the Older Dryas (12,300-12,000 years B.P.), a brief climate cooling (no more than 300 years) was detected in the biological sequences studied at high time resolution [Massif Central: Ponel and Coope, 1990; Jura: Bégeot et al., 2000] and in global records [Johnsen et al., 1992; Lehman and Keigwin, 1992]. At the low-altitude site of Marly (192 m, Moselle Department), the Betula forest regressed [Bégeot et al., 2000; Ruffaldi, 2000]. The forested optimum at the end of the interstadial (Alleröd, 12,000-10,700 years B.P., Event 12 in figure 2) was marked by the development of a taiga with Betula and Pinus and the decline of open spaces. That was the case for France as a whole, but not for the M/HM region [Bégeot et al., 2001] where grasslands persisted. Similarly, the thermic decline at the end of the Alleröd is not recorded in Lorraine sequences, probably due to sampling resolution. The Recent Dryas (10,700-10,300 years B.P., Event 12 in figure 2, Heinrich 0) represents the last climatic crisis of the Weichselian glaciation. On the pollen records of Lorraine Plateau [Bégeot et al., 2001 and Ruffaldi, 2000], Pinus frequencies decline and steppic formations are better represented. The preservation of eutrophic soils [Van Vliet et al., 1992] indicate that the region was not totally deforested during this cold episod. The preservation of low-altitude vegetation explains the absence of soil-erosion phenomena. In altitude, the presence of firns probably limited the development of forests [Van Vliet et al., 1992]. The estimated temperature in July, deduced from pollen and insect assemblages, ranged from 10 o Cto11 o C in the Netherlands. In the M/HM region, the estimated mean annual temperature varied between ca 3 o C to ca 1 o C during the Younger Dryas [Courbouleix et al., 1998] (fig. 2). It is possible that a discontinuous permafrost formed during that cold event in the M/HM region. Holocene At the beginning of the Holocene interglacial (OIS 1, 10,300 years B.P., Event 13 in figure 2) vegetation changes were mainly regulated by the climate and the migration dynamics of forest trees benefiting from the increase in temperatures and precipitation to leave their European refuges (eastern Europe and Mediterranean countries). At that time, late-palaeolithic and Mesolithic human populations had a very limited impact on the natural environment. During the Preboreal (10,300-9,000 years B.P.), Betula stands mixed with Pinus startedtogrow.ulmus and deciduous Quercus were also present, but survived only in areas where climate and edaphic conditions would allow it. As soon as those soils reached a sufficient level of maturity, large Quercus forest associated first with Corylus and dominated later by Corylus (during the Boreal: 9,000-8,000 years B.P.) started to develop. During the Atlantic (8,000-4,700 years B.P.), a mixed diverse Quercus forest with Acer, Tilia and Ulmus spread in low to mid-altitude regions of northern France. Riparian forests with Alnus also spreaded because, due to the rise of sea level, precipitation rose, streams reached their equilibrium profile, while constantly humid environments suitable for the devlopment of a riparian forest established along the streams. The beginning of the Atlantic (8,000-6,000 years B.P.) corresponds to the climate optimum of the Holocene (fig. 5). During that period, the estimated annual temperature was 2 o C above the current temperature. The impact of human populations on forest vegetation is globally sensitive from 5,500 years B.P. But on the Lorraine Plateau, traces of early agriculture have only been detected around ca 7,200 years B.P. and ca 6,000 years B.P. [Ruffaldi, 1999]. The most visible evidence of the Neolithic Revolution on ecosystems was the significant reduction of forest areas, with the immediate consequence of an instability of bare slopes, soil erosion and more violent floods than in the past, when vegetation stabilised slopes. Neolithic farmers were not the single responsible for those changes, since the climate also started to evolve and to cool as early as 5,000 years B.P. in boreal and medio-european Europe [Van Vliet-Lanoë, 1998; Seppa and Poska, 2004]. This climate inflection, contemporary with the beginning of the Subboreal (4,700-2,700 years B.P.) was marked in its initial phase by the reappearance of Fagus woodlands in low to mid-altitude regions. The arrival of Fagus in northern France dates to ca 4,000 years B.P. During the Bronze Age, ca 3,600 years B.P., a large clearing made by farmers through the Quercus forest of the Aisne Valley favoured the spreading of Corylus. The increasing human impact on the natural environment as early as 5,000 years B.P. resulted from the evolution of agricultural techniques, particularly the use of metals that accelerated deforestation by their enhanced efficiency. In Lorraine, during the Bronze Age, forests still covered large areas [Gauthier et al., 2000; Koenig and Ruffaldi, in prep.]. Human pressures increased during Gallo-Roman times and in the Middle Ages: Quercus woodlands drastically regressed in favour of cereal crops. The previously wooded and diversified landscape within which Mesolithic populations used to live had disappeared and nowadays, current populations live in a man-made rural landscape shaped by centuries of agricultural activities. Since the end of the 19 th century, however, a new afforestation policy has been launched in response to the awareness of the destabilisation hazard that vegetation-free slopes represent for rural communities. Inframillennial variations In order to improve information on the consequences of past rapid climate variations due to North-Atlantic conditions in France, Andra has commissioned several studies on

9 CLIMATE AND ENVIRONMENTAL EVOLUTION SCENARIOS (MEUSE/HAUTE-MARNE, FRANCE) 123 FIG. 5. France during the Holocene Optimum: 8,000 ± 1,000 years B.P. according to Brulhet and Petit-Maire [1999]. FIG. 5. La France au moment de l Optimum Holocène: ± ans B.P., d après Brulhet et Petit-Maire [1999]. the continent [Bégeot et al., 2001; Jouzel, 1999; Beaulieu, 2003]. Some of them are still under way, for example the 10-year timescale resolution study of the Annecy Lake sequence [Jouzel, 1999 and Von Grafenstein, 2005] or the 10-to-50-year timescale resolution study of Les Échets sequences [Andrieu-Ponel et al., 2003 and 2005]. Brief climate variations (between 10 and 500 years) affected France, including the M/HM region, especially during the last glaciation (OIS 4 to 2). The most severe cooling periods on continents correspond to well-known Heinrich events in marine sediments [Heinrich, 1988; Bond and Loti, 1995]. Those events were associated with iceberg-release phases from Laurentide, British and Fenno-Scandinavian ice sheets towards low latitudes (to 40 o N at least). Six events of that type have been detected during the last 75,000 years: the Younger Dryas or H0 (10,700-10,300 years B.P.), H1 (14,000-13,000 years B.P.), H2 (ca 21,000 years B.P.), H3 (ca 28,000 years B.P.), H4 (ca 40,000 years B.P.), H5 (ca 51,000 years B.P.), H6 (ca 63,000 years B.P.). Those events occurred at the end of slow plurimillennial cooling phases and were followed by rapid warming phases. Those cycles (slow plurimillennial cooling and rapid warming) are called Bond cycles. Superimposed on Heinrich events, more rapid cycles with slow cooling and faster warming, called Dangaard- Oeschger event and spreading over 10-to-100-year timescales, were identified. In NGRIP core (Greenland), 25 D/O events have been evidenced for the last 90,000 years B.P. [North GRIP Members, 2004; Landais et al., 2004]. Palaeo-oceanographical data have revealed that massive iceberg releases modified the salinity of sea water, its density, as well as thermohaline circulation in the North Atlantic, with large-scale consequences [Voelker and workshop participants, 2002]. The switching mechanism between the northern and southern hemispheres, detected by comparing climate records taken at both poles [e.g., Blunier et al., 1998] resulted from those changes in oceanic circulation. During cold events, when heat transport by the North Atlantic was considerably reduced, the extension of the ice field was phenomenal and extended down to temperate latitudes. In reaction to those conditions, atmospheric circulation changes have modified the European climate as indicated by marine pollen off the coast of Portugal [Sanchez Goni, 2000] and showed that even the most rapid changes, such as the last cold spell that occurred 8,200 years B.P. [Von Grafenstein et al., 1998], were synchronous. Nevertheless, we have to emphasize that very few is known of the relation global climatic changes and their consequences on the continent in spite of the attempt recently published by Genty et al. [2005] and not based on continental pollen sequences but on marine pollen record loess soils or speleothems. An important work of comparison and modelisation of climate response of the continent to the global climatic variability ought to be undertaken from long continental vegetational record of France, a country widely open the North Atlantic that remains with polar ices, one of the main control of the Quaternary climatic dynamics. In the climate archives for northern France, it seems that temperate phases of OIS 3 (Dansgaard-Oeschger events: DO) were too brief to allow the forest to start growing again and, at most, a tundra park (Juniperus, Betula, Salix and rare Pinus) settled. Those phases of relative warming associated with an increase in precipitation were suitable for the accumulation of sub-arctic organic matter in the sedimentation zones. In the endokarstic

10 124 ANDRIEU-PONEL V. et al. formations of northeastern France, DO-15 and DO-14 events (55,000-53,000 years B.P.) as well as DO-12 events (45,000 years B.P.) have been recorded and correspond to stalagmite-growth phases that occurred between 53,300 ± 700 years B.P. and 55,400 ± 900 years B.P. in the case of DO-15 and DO-14 events, and ca 45,900 ± 400 years B.P. for DO-12 events [Pons-Branchu et al., 2001; Losson et al., submitted]. The presence of organic matter in the speleothem contemporary with DO-15 and DO-14 events indicates the presence of a soil and of a vegetation cover on the Lorraine Plateau at that time. According to Pons-Branchu [2001], the growth of speleothems throughout both DO events during OIS 3 also shows that there was no superficial permafrost in Lorraine, with the thawing probably extending to a depth of several metres in the top of the former permafrost due to the increase in precipitation. During interglacial periods, the phases of rapid and substantial temperature decline may have had a very negative impact on thermophilous or mesophilous species, since they are particularly sensitive to the thermal variations in winter extremes. During the Montaigu cooling (Saint-Germain 1, OIS 5c), the broad-leaved Quercus forest disappeared completely from all French pollen records. It was replaced by a tundra Arctic park with Betula established on oligotrophic soils. It is possible to suppose that the return to glacial climate conditions at that time favoured the development of soils affected by seasonal frost at shallow depth on the Lorraine Plateau. At the end of that episode, the return to interglacial climate conditions allowed the Quercus forest to spread once again over reworked mesotrophic materials generated by solifluction or runoff. Similar conclusions may be derived for the cold spell of the Recent Dryas that interrupted the interstadial forest dynamics of the Weichselian Lateglacial and delayed the migratory dynamics of thermophilous trees throughout France, this time on an eutrophic parent material. That climate decline did not disturb plant dynamics permanently and, as early as 10,000 years B.P., interglacial broad-leaved vegetation started to spread in combination with boreal heliophilous species during the first millennium of the Holocene. From Atkinson et al. [1987], we know that in Britain winter temperature changes were very strong and rapid at the onset of the last Lateglacial interstadial, between 13,300 and 12,500 years B.P., with a warming of 7 to 8 o C in summer and ca 25 o C in winter. Recent climatic reconstruction at Ammersee (German Alps) shows a rise in temperature of 5.5 ± 1.5 o C at the transition Younger Dryas/Holocene while higher values (10 ± 3 o C) are yielded in Greenland [Masson et al., 2005]. In Britain [Atkinson et al., 1987], at the same time, the temperature warming is evaluated at 1.7 o C per century. These divergences underline again that new multiproxy (biological and geochemichal markers) and high resolution studies must be accomplished on the continent, especially in France, on long continental record and not marine pollen sequences only (deep water and continental margin included), to better understand the impact, on the continent, of the climatic variability recorded in global sequence. The palaeoenvironmental data collected for the upper half of the Pleistocene helps to show that rapid climate changes may generate a radical modification in vegetation structure and flora composition, that those changes were rarely redhibitory because plants were able to survive in refuges and that a return to initial conditions was possible over time. No forest taxa have disappeared from the Palaeoarctic world as a consequence of climate changes during the last five climate cycles and even less over the last 150,000 years. FUTURE CLIMATE EVOLUTION OF THE M/HM REGION This section presents some of the transient climate simulations performed within the BIOCLIM project [Texier et al., 2003; Texier et al., in prep.]. A more complete description of the work produced by the project can be found in a special issue of the Journal of Radiological Protection [Thorne et al., in prep., Degnan et al., in prep., Paillard et al., in prep., Loutre et al., in prep., Lunt et al., inprep.and Goodess et al., in prep.], or in the deliverables of the project [BIOCLIM, 2001 and 2003a, b, c, d]. The transient climate simulations presented here are those executed by the LLN-2D-NH model [Gallée et al., 1991]. LLN-2D-NH model and regionalisation method The LLN-2D-NH model [Gallée et al., 1991] is an Earth model of intermediate complexity [see Claussen et al., 2002 for review] representing the northern Hemisphere atmosphere, ocean mixed layer, sea ice, ice sheets and continents. Climate simulations performed with the model have already shown its ability to reproduce glacial-interglacial cycles with characteristics similar to those reconstructed from proxy data. Although far from being comprehensive, the model has been able to represent several gross features of the climate during the last 3 million years: the entrance into glaciation around 2,750,000 years B.P. [Li et al., 1998a], the late Pliocene-early Pleistocene obliquity cycle [Berger et al., 1999], the emergence of the 100,000 year cycle around 900,000 years B.P. [Berger et al., 1999], the glacial-interglacial cycles of the last 600,000 years [Li et al., 1998b] and the climatic variations over the last 200,000 years [Berger and Loutre, 1996]. As early as the late 1980s, the LLN-2D-NH model was used to discuss climate issues over the next millennia, including possible impacts of human activities during the next few centuries on climate evolution over millennial timescales [Gallée, 1989; Loutre, 1995]. The model structure is altitude-latitude dependent. In each latitudinal belt, the surface is divided into at most seven oceanic or continental surface types, each of which interacts separately with the subsurface and the atmosphere. The oceanic surfaces may be ice-free ocean or sea-ice covered, whereas the continental surfaces may be snow-covered or snow-free and include up to three northern Hemisphere ice sheets. Atmospheric dynamics are represented by a zonally averaged quasi-geostrophic model, which includes a parameterisation of the meridional transport of potential vorticity and of the Hadley sensible heat transport. The atmosphere interacts with the other components of the climate system through vertical fluxes of momentum, heat and water vapour. The model explicitly incorporates detailed radiative transfer, surface energy balances, as well as snow and heat budgets. The vertical profile of the upper ocean temperature is

11 CLIMATE AND ENVIRONMENTAL EVOLUTION SCENARIOS (MEUSE/HAUTE-MARNE, FRANCE) 125 computed using a mixed-layer model, which takes into account meridional heat transport. It is important to note that an ocean mixed layer approach is unable to represent the thermohaline circulation. A thermodynamic model including leads and a parameterisation of lateral accretion represents sea ice. Simulation of the present climate shows that the model is able to reproduce the main characteristics of the general circulation [Gallée et al., 1991]. The present-day features of the seasonal cycles of the oceanic mixed layer, sea ice, and snow cover are also fairly well produced. The atmosphere-ocean model is asynchronously coupled to a model of the three main northern Hemisphere ice sheets and their underlying bedrock. Since the model does not contain any representation of the carbon cycle, the atmospheric CO 2 concentration is considered as an external forcing in addition to the astronomically derived insolation. A rule-based downscaling approach from the LLN-2D-NH model of intermediate complexity has been developed at the University of East Anglia in order to provide consistent estimates of monthly temperature and precipitation, in particular for the M/HM region [BIOCLIM, 2003d]. Such an approach had been developed and used in a previous study funded by Nirex [Burgess, 1998]; it provided later the starting point for the BIOCLIM work, but the details of the methodology differed substantially. The rule-based methodology assigns climate states or classes to each point in time for a region according to a combination of simple threshold values, which are selected from the coarse-scale climate model. Once climate states or classes have been defined, monthly temperature and precipitation climatologies are constructed using analogue stations identified from a database of present-day climate observations. The most appropriate climate classification for BIOCLIM purposes was the Køppen/Trewartha scheme [Rudloff, 1981]. The scheme has the advantage of being empirical, and only requires monthly temperature and precipitation averages as input variables. The BIOCLIM rule-based downscaling procedure is fully described within BIOCLIM [2003d]. The locations of the analogue stations selected for the M/HM site are shown in figure 6. Input data: insolation and pco 2 For the climate system, insolation and greenhouse-gas concentrations represent two essential driving forces. Although the variations of plurimillennial insolation, induced by the changes in the Earth s orbital characteristics are understood, the driving mechanisms for the variations in concentrations of greenhouse gas variations are only known in part. For the time being, there is no full theory to explain all observations and to develop a solid basis for extrapolations of the future evolution of greenhouse gas concentration [Paillard et al., in prep.]. The BIOCLIM Project was limited to the study of the impact of future anthropic pco 2 variations on climate. The other greenhouse gases were not taken into consideration. The key consideration was whether an anthropic pco 2 disturbance occurring over the next few centuries world have any consequences on the climate for thousands to hundreds of thousands of years? Archer et al. [1997] showed that, although a large part of the pco 2 resulting from the combustion of fossil fuel would be absorbed by the oceans during the next few centuries, a non-negligible fraction (5-10%) may remain in the atmosphere for hundreds of thousands of years. Depending on the total increase of pco 2 due to anthopic emissions, that remaining fraction could have a significant effect on future glacial-interglacial cycles. The summer insolation in high northern latitudes is a key variable for the astronomical climate theory [Milankovitch, 1941]. According to that theory, minimal values of that variable coincide with glacial advances in the northern Hemisphere. In general, variations in the insolation distribution (in latitude and during the year) control or regulate the succession of glacial-interglacial cycles. Figure 7 shows that, during the next few hundreds of thousands of years, insolation will vary with a lower amplitude than in the last 200,000 years. At least 600,000 years are needed for insolation to reach as low a value as during the last entrance into glaciation, 115,000 years BP. The moderate nature of insolation variations during the next few hundreds of thousands of years combines with other factors and, particularly, with the projected atmospheric concentrations of greenhouse gases, to suggest that the occurrence of a severe glacial episode over that period is unlikely, though some more limited glacial episodes may occur. Three future pco 2 evolution scenarios were selected and tested within the BIOCLIM Project. The natural evolution scenario postulates that pco2 may evolve in the future with the same cyclicity as during the Quaternary and oscillate between interglacial values in the order of 280 ppmv and glacial values in the order of ppmv [EPICA Community Members, 2004] (fig. 8, fine dash). In both other scenarios (called disturbed evolution ), the pco 2 disturbance that the combustion of fossil fuel represents for the climate system is taken into account and added linearly to the natural pco 2 variations (fig. 8, bold and dotted-bold dashes). The result is a curve reaching its peak during the 22 nd century, before decreasing towards the natural cyclicity, but only after several hundreds of thousands of years (ca 600,000 years). Between 1990 and 2100, the BIOCLIM scenarios lie within the spectrum of the IPCC scenarios [IPCC, 2001] according to which the estimated quantity of carbon injected into the atmosphere ranges from 700 to 2,600 gigatonnes (GtC). Beyond 2100, BIOCLIM extrapolates the IPCC low and high scenarios and reaches a total quantity of emissions ranging from 2,160 and 3,660 GtC between 2100 and However, if the interval is examined, BIOCLIM provides a total estimate lying between 3,160 and 5,160 GtC. Those values are consistent with the current estimate of the potential of global fossil-fuel reserves (close to 5,000 GtC according to Archer et al. [1997]). Plurimillennial variations Figure 9 describes the future evolution of the climate in the M/HM region from the regional output of the LLN-2D-NH model as forced by future insolation and pco 2 variations. Indices shown as ordinates correspond to climate states resulting from the Køppen/Trewartha climate classification [Rudloff, 1981] used in the regional characterisation technique. Figure 9 should not be considered as a forecast, but rather as a simplified potential scenario for climate evolution in the M/HM region. The staircase-aspect of the curves is the result of temperature and precipitation thresholds

12 126 ANDRIEU-PONEL V. et al. FIG. 6. Location of meteorological stations used in the rule-based downscaling regional method to characterise climate states in the M/HM region. Three geographical windows: (top) European Sector, (bottom) Eurasian Sector, (following page) Northern Atlantic Sector. FIG. 6. Localisation des stations météorologiques utilisées par la technique de régionalisation dite de rule-based appliquée au cas de la région M/HM. Trois fenêtres géographiques: (haut) Secteur européen, (bas) Secteur eurasien, (page suivante) Secteur nord atlantique. used in the classification method. Inframillennial events are not shown, the time interval on which the results of the LLN-2D-NH model were saved being 1,000 years. In response to the sole natural of pco 2 During the next million years, the M/HM region is projected to go successively through various states: temperate oceanic and temperate continental, boreal, periglacial and tundra (Køppen/Trewartha classification, fig. 9a). The sequence is dominated for more than 81% of the time by cold-climate states (mostly boreal and periglacial ), whereas the temperate oceanic state characterising the M/HM region only represents 18%. A particularly long periglacial climate period is simulated at ca 400,000 years A.P. The current interglacial, i.e. the current temperate oceanic state, would persist for ca 50,000 years A.P., a much longer period than the last Quaternary interglacials. It should be noted that Loutre and Berger [2000] have also simulated an extended present interglacial of ca 50,000 years by forcing the LLN-2D-NH model with past

13 CLIMATE AND ENVIRONMENTAL EVOLUTION SCENARIOS (MEUSE/HAUTE-MARNE, FRANCE) 127 FIG. 6. Follows/Suite. FIG. 7. Insolation evolution (at 65 o N and during the summer solstice) from 200,000 years B.P. to 1 million years A.P. according to Berger [1978a and b], for the last 20,000 years, and BIOCLIM [2001], for the next million years. FIG. 7. Evolution de l insolation (à 65 o N et au moment du solstice d été), depuis ans B.P. jusqu à 1 million d années A.P., d après Berger [1978a-b] pour les derniers ans, et d après BIOCLIM [2001], pour le prochain million d années. pco 2 variations determined by Jouzel et al. [1993] and used as a surrogate for potential variations over the next 130,000 years. The persistence of the current interglacial is primarily due to the low variations in insolation characterising the next few tens of thousands of years. Annual average temperature and precipitation of the temperate oceanic state are 11 o Candca 800 mm, respectively (averages of the 14 DO stations, fig. 6). Those values are consistent with the local climate data of Météo-France (ca 10 o C and mm). After 50,000 years A.P., the climate of the M/HM region should start to deteriorate and reach a maximally cold episode, corresponding to a tundra state ca FIG. 8. Evolution of the carbon-dioxide concentration in the atmosphere (in parts per million volume: ppmv) taken into account in BIOLCIM simulations for the next 5,000 years (left) and the next 200,000 years (right). Three scenarios: natural when no antropogenic perturbation is accounted for, B3 and B4 when an antropogenic perturbation (moderate or stronger) is accounted for. FIG. 8. Evolution de la concentration atmosphérique du dioxyde de carbone (en parties par million de volume: ppmv) prise en compte pour les simulations climatiques du projet BIOCLIM pour les prochains ans (gauche) et les prochains ans (droite). Trois scenarii: «naturel» quand aucune perturbation anthropique n est prise en compte, «B3» et «B4» quand une perturbation anthropique (plus ou moins forte) est prise en compte. 100,000 years A.P. That state would recur several times with an increasing frequency after 500,000 years A.P., but would only represent 7% of the million years. At the scale of the northern Hemisphere, that state corresponds to a total ice volume in the order of 22 to 35 million cubic kilometres, a figure that is less than the 50 million cubic kilometres simulated by the LLN-2D-NH model for the glacial maximum that occurred 18,000 years B.P. As in the case of the last glacial maximum, the M/HM region would remain free of ice sheets. The mean annual temperature (based on

14 128 ANDRIEU-PONEL V. et al. FIG. 9. Climate evolution scenarios for the M/HM region over the next million years according to BIOCLIM [2003a]. In ordinates, the climate classes are taken from the Køppen/Trewartha classification: 0 = tundra (FT); 10 = periglacial (EC); 15 = boreal (EO); 20 = temperate continental (DC); 25 = temperate oceanic (DO); 30 = humid subtropical (Cr); 35 = subtropical with dry summer (Cs). FIG. 9. Scenarii d évolution climatique pour la région M/HM pour le prochain million d années, d après BIOCLIM [2003a]. En ordonnées figurent les classes climatiques de la classification de Køppen/Trewartha: 0 = toundra (FT) ; 10 = périglaciaire (EC) ; 15 = boréal (EO); 20 = tempéré continentale (DC) ; 25 = tempéré océanique (DO) ; 30 = subtropicale humide (Cr) ; 35 = subtropicale à été sec (Cs). all relevant FT stations, fig. 6) would be in the order of 2 o C, with 6 o C in summer, 10 o C in winter and under 0 o C between October and May. The most frequent cold-climate states during the next million years are of the boreal and periglacial types. Those states correspond to average annual temperatures of 3 o C and 1.8 o C, respectively, with annual precipitation values of ca 650 and ca 607 mm, respectively (based on all EO and EC stations, fig. 6). Under a boreal climate, the temperature is 10 o C in summer, 1.5 o C in winter and remains below 0 o C between November and March. Under a periglacial climate, the temperature is 12.5 o C in summer, 16.8 o C in winter and remains under 0 o C between November and April. The values of cumulated annual precipitation mentioned above seem overestimates in comparison with the range of precipitation values observed by different authors in various boreal climate ( mm) and periglacial climate (> 400 mm) regions.