Réf. Bertolini & al 2004 - A

Référence bibliographique complète
BERTOLINI, G., CASAGLI, N., ERMINI, L., MALAGUTI, C. Radiocarbon Data on Lateglacial and Holocene Landslides in the Northern Apennines. Natural Hazards, 2004, 31, 645–662.

Abstract: The Emilia Romagna slope of the Northern Apennines is strewn with over 32,000 landslides, 5,000 of which are larger than 1 million cubic metres. They represent the remains of geomorphic agents that shaped the Apennines during the Holocene. Dating them by means of radiocarbon methods adds a contribution to the knowledge about the last period of the geological geomorphological history of the Apennines. They can also be used to examine the influence of Quaternary climatic changes on the instability of slopes and, for practical or planning functions, to assess the periodicity of activity phases of the landslides. The dating has been carried out on wood remnants buried under the landslide bodies. In some cases the entire tree trunk was found. In this paper we present radiocarbon dating of 20 case studies in the Northern Apennines. Results range approximately from 13790–13670 cal y BP to 950–790 cal y BP. The oldest case is that of the Morsiano earth-flow, while the younger dated event is represented by the Marano case that represents an example of how radiometric analyses can further enhance the available historical data. In the Cavola case, wood remnants of different ages were found at different depths (from 9 to 45 m), allowing the dating of the first and following periods of activity of the landslide. The results are discussed and some considerations on the correlation between landslide occurrence and Holocene climate changes are proposed.

Mots-clés
Holocene, landslide, Northern Apennines, radiocarbon dating

Organismes / Contacts

Regione Emilia-Romagna, Servizio Difesa del Suolo di Reggio Emilia, Servizio Difesa del Suolo di Parma, Dipartimento di Scienze della Terra, Università degli Studi di Firenze.
gbertolini@regione.emilia-romagna.it; ncasagli@geo.unifi.it; erminil@geo.unifi.it; spdspr1@regione.emilia-romagna.it


(1) - Paramètre(s) atmosphérique(s) modifié(s)
(2) - Elément(s) du milieu impacté(s)
(3) - Type(s) d'aléa impacté(s)
(3) - Sous-type(s) d'aléa
    Mass movements Landslides (earth flow...)

Pays / Zone
Massif / Secteur
Site(s) d'étude
Exposition
Altitude
Période(s) d'observation
Italy Northern Apennines 20 landslides from the Northern Apennines, Emilia Romagna slope     Holocene

(1) - Modifications des paramètres atmosphériques
Reconstitutions

There is general agreement that a warming phase characterized the beginning of the Holocene through the Pre-Boreal (11,600–10,200 cal y BP), Boreal (10,200– 8,200 cal y BP) and Atlantic periods (8,200–5,700 cal y BP), known as the warmest phase of the entire Holocene, when the Climatic Optimum was also reached about 5,750 cal y BP. The mean annual temperature was probably 3–5 °C degrees higher than the present one and, starting from the available data (Starkel, 1966; Lamb, 1977; Panizza, 1985; Orombelli and Ravazzi, 1996) it was also one of the wettest periods of the entire Holocene.

Successively new cold oscillations followed in the Sub Boreal and Sub Atlantic periods. The first one, known as the “Piora oscillation”, occurred soon after the Climatic Optimum. The last one, known as “Little ice age”, occurred in historic time in the period ranging from 1550 to 1850. There is evidence of the worsening of the climate during this phase, from different parts of Europe.

The available and preliminary data are in agreement with the paleo-climate reconstruction carried out in the same area from pollen stratigraphy studies (Magri, 1999) and seem to suggest that climate variation in the Northern Apennines generally followed a trend similar to that of other parts of Europe.

Observations
 
Modélisations
 
Hypothèses
 

Informations complémentaires (données utilisées, méthode, scénarios, etc.)

A great part of the terrestrial Holocene climate knowledge is based on the reconstruction of the movement of glacier fronts and on pollen stratigraphy studies [see references]. Many efforts have also been developed for getting data about the oscillations of the upper tree-line border that corresponds to the mean annual isotherm of + 2°.

Most of the paleo-climatic data that support the reconstruction of the Holocene climate have been collected in the Alps chain and in Northern Europe. Except for the Late Glacial and Early Holocene periods (Lowe and Watson, 1993; Magri, 1999) no detailed pollen stratigraphy studies exist in the Northern Apennines area, because of the scarcity of good sites for carrying out detailed analyses and the human impact on the environment that started from Neolithic time. However, there are studies directed at determining the sea level variations, that, for short time periods, even in tectonically active areas, are strongly related to temperature variations. These works are based on the analysis of the growing position of speleothemes (Alessio et al., 1992). In an active marine cave, speleothemes grow up at the equilibrium between air and water and “die” when they are submerged. The radiocarbon dating of the calcareous nucleus of the speleothemes dates their death. This permits indirect but very detailed reconstructions of climate oscillations for the last 40,000 years, the present limit for radiocarbon dating.


(2) - Effets du changement climatique sur le milieu naturel
Reconstitutions

The most striking vegetation feature that characterized this phase [Climatic Optimum] was the great spread of the hazel tree throughout Europe. Successively new cold oscillations followed in the Sub Boreal and Sub Atlantic periods. The first one, known as the “Piora oscillation”, occurred soon after the Climatic Optimum was reached and, in alpine environments, provoked the disappearance of the spruce forest that was replaced by open meadows.

Observations

 

Modélisations
 
Hypothèses
 

Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

 

(3) - Effets du changement climatique sur l'aléa
Reconstitutions

Most current landslides in the Apennines are reactivations of pre-existing ones, which occurred in periods of climatic and geomorphic conditions different to those of the present.

The oldest dated landslide event (~13,500 cal y BP), placed at the transition between the Pleistocene and the Holocene, is one (maybe the first) among those large earth-flows that formed the landslide system known as “Morsiano landslide”, while the youngest one belongs to the Marano landslide that occurred ~850 years ago.

From the available climatic and radiocarbon data, it seems reasonable to gather the dated landslide events in three main groups. The first group refers to landslides that occurred near the transition between the Late Pleistocene and the Holocene, probably as a consequence of the rapid climatic changes that preceded (placed approximately in the Alleröd stage 13,350–12,650 cal y BP) a warm climate oscillation occurred between Dryas II and Dryas III (12,650– 11,600 cal y BP) and followed the Dryas III.

Successively, the presented database shows a lack of events for 1,500 years (6,500–4,900 cal y BP). This is the warmest phase of the entire Holocene known as the Atlantic period where also the climatic optimum was reached about 5,750 years before present.

Landslide activity took place again at the passage between Atlantic and Sub Boreal periods. In particular it is possible to separate two main periods of activity: the first one (4,900–3,900 cal y BP) during the “Piora oscillation” and the second one from 3,195 to 2,000 cal y BP.

Other promising results can be obtained by analysing data extracted from the AVI Project database (Guzzetti et al., 1994) that reports more than 3,000 landslide events recorded from all the Italian peninsula. There are several peaks of landslide activity that partly coincide with periods of climatic deterioration that cyclically took place during the last five centuries.

The correlation between landslide activity and colder climatic periods finds confirmation also in recent studies carried out in the Eastern Italian Alps where Corsini et al. (1999, 2000) got very similar results in an area (mostly) moulded by Würmian glaciers.

These data are in partial disagreement with other investigations carried out in Central and Northern Europe (González Díez et al., 1996) that show an increase in landslide activity in wetter and maybe warmer periods. As underlined also by Dikau and Schrott (1999) the unconformity of landslide activity across Europe might possibly be explained by both local climatic and geomorphologic factors.

Relationship between Spatial and Temporal Occurrence of Landslides:
The spatial distribution of the main landslides in the Emilia Romagna Region seems to strengthen the hypothesis that climate changes, that occurred between the Pleistocene and the Holocene, with particular reference to the sharp oscillations that characterised the beginning and the end of Dryas III, could be the triggering factor of those landslides.

In general, the spatial distribution of landslides is strongly related to the lithology of the outcropping rocks and terrains, but it could also give useful indication regarding the geomorphologic and climatic history of the investigated territory.

There are 684 cases of landslides with an area of more than 400,000 m², representing 2% of all the existing landslides in that area (Regione Emilia-Romagna, 1999). It can be seen that the inventoried landslides are concentrated above the 1,300 m contour line.

There are 130 cases of landslides with an area of more than 700,000 m², representing 0.5% of all existing landslides in that area. A comparison of the two suggests that the selection of landslides with an area of more than 700,000 m² from the whole number of landslides, produces a cluster of cases more concentrated around the main mountains than the first class.

55% of the landslides are placed in an elevation range between 400 m and 800 m contour lines. The 29% of the presented landslides that occurred in the area between the 800 m and 1,200 m contour lines, spatially represent only 23% of the entire hilly and mountainous territory of the Emilia Romagna Region.

The highest concentration of large landslides is in the area between the 600 and 800 m contour lines, that represents less than 20% of the entire hilly and mountainous territory of the Emilia Romagna Region, yet contains about 30% of the total number of large landslides. Instead, the low frequency of large landslides in areas placed over the 1,200 m contour line probably depends on geologic factors, such as the presence of Tertiary flysch formation, not very prone to sliding, that outcrop around the Northern Apennines main water divide of Parma and Bologna provinces.

It is also interesting to note that small landslides are more randomly distributed over the whole Emilian Apennines, including the hilly area that appears almost completely free from large landslides.

Gathering together all these observations it might be understood that landslides characterised by larger dimensions are, as a general rule, placed at elevations between 600 and 1,200 m, in those areas that were more affected by the climate variations that took place at the end of the last glacial phase (Val Parma stage, Pellegrini et al., 1998). During this phase the maximum expansion of glaciers was reached about 20,000 y BP, with an upper snow limit placed around 1,250 m a.s.l., while glaciers also reached lower elevation ranges (Pellegrini et al., 1998; AA.VV., 1999). In areas below the 1,250 m contour line limit, physical weathering processes were probably particularly severe. As a consequence, ancient huge earth-flows, like Morsiano (~13,500 cal y BP) and Succiso (~9,500 cal y BP) landslides, were probably triggered by the combined action of permafrost thawing and new morphogenetic factors. In particular, as affirmed by Pellegrini et al. (1998), in that phase the passage from typical physical weathering processes to chemical ones was registered. Unfortunately, from the available paleo-climate data, it is difficult to reconstruct rainfall and temperature trends with respect to altitude.

Observations
 
Modélisations
 
Hypothèses
Starkel (1966) affirms that the melting and disappearance of frozen ground could lead to deeper water tables and to changes in water circulation. In particular, it can be hypothesised that, under permafrost conditions, groundwater recharge was not possible, while, after permafrost disappearance, groundwater recharge occurred and reached high values as a consequence of the rainfall and snowmelt increase that occurred during the improvement of climatic conditions following the last glaciations. For all these reasons Canuti et al. (1998) assumed that the first activation of major landslides in the Apennines, at the end of the Pleistocene and in the early Holocene, has to be correlated to the combined effects of tectonic uplift and wetter periods.

It seems reasonable to affirm that the landslide activity in the Northern Apennines reached a maximum during colder periods, characterised by a general deterioration of the climate, known as resistasia regime. Colder periods may explain more landslide activity due to less evapotranspiration that determines an increase in the groundwater recharge.

Improvements to this hypothesis can be found by zooming on the part of the diagram where historic data are also available. Most of the cases of landslide dams over the Northern Apennines occurred from the end of the 16th century to present (Casagli and Ermini, 1999). The sudden increase of historic chronicles in the second half of the 16th century can probably be related to the beginning of a period characterized by a colder and wetter climate known as the “Little Ice Age”. Another important triggering factor that can explain the increase in landslide activity during the “Little Ice Age”, is probably the extensive deforestation that took place since the 15th century. However the peak of case records at the beginning of the 20th century is probably due to the large amount of information collected in the report of Almagià (1907).

Except for some partial reactivations these landslides are prevalently dormant or inactive. This fact strengthens the hypothesis that they have been triggered in climatic conditions different from those at present. Because of this prevalent dormancy, since ancient times, several villages have been built on the debris displaced by the landslides while these encountered their most prolonged dormant phases. However these landslides are not completely stabilized and the low frequency of their reactivation (very often partial) makes them very difficult to be detected and to be managed with urban planning purposes.

Paramètres de l'aléa
Sensibilité du paramètre de l'aléa à des paramètres climatiques et du milieu / Facteurs de contrôle
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
Landslide occurrence
 

Peat bogs may develop on the landslide surface because the movement leads to the formation of counter slopes and depressions on the hummocky surface of the landslide, provoking problems in land water overflow. This situation does not permit a direct radiometric dating of the landslide event because it is impossible to determine if the datable material was really displaced when the movement took place. If the depression is due to the landslide, the radiocarbon data can give a minimum age of the movement. An example of this type of recovery is that of Valestra.

Landslide occurrence
 

Tree stumps were found buried by the landslide debris. This occurrence can produce accurate radiometric dating of the landslide event, if the stump is still in place. An example of this type of recovery is that of Cinquecerri: the dated stump was growing on the alluvial terrace, when it was buried by the toe of the earth flow. It was recovered with its roots still placed in the bedrock under the sliding surface.

Landslide occurrence
 

Logs were found mixed with the original earth flows that constitute the landslide body. This is the most frequent case in the database of Table I. In a few cases, whole logs were found due to excavations carried out for remedial works or to river erosion. In the case of Romanoro landslide, several buried tree logs were found mixed with the landslide debris. These were placed at the same level within the landslide debris, and thus they belong to the same landslide event that, from radiocarbon dating, took place in a period 3195–2845 cal y BP. In the Cavola landslide event, in a test borehole, five different organic levels were found. From radiocarbon dating, it was possible to reconstruct the activity of this landslide starting from its first event (the deepest sample was recovered on the sliding surface) and lasting over a period of 1000 years. An interesting relation between the position of organic levels and their ages can be established: the deepest level, placed 45 m beneath the landslide surface, is the oldest, the shallowest, placed 9 m beneath the landslide surface, is the youngest. This particular situation could be explained with a mechanism of several earth flow movements that piled up the present landslide body, each time burying the vegetation that was growing on top of the material displaced by the previous movement.

Landslide occurrence
 

The landslide provoked a damming event of a river and sometimes datable elements were found in the peat bogs that formed upstream from the dammed river section. This recovery case permits only a minimum age for the landslide event. However it must be underlined that the results are very useful to date those types of landslides (e.g., rockslides) for which it is difficult to find organic matter within the landslide body. Good examples of this type of occurrence are the Crespino rockslide [ID10] and the Marano earthslide that is briefly described. The landslide took place during February 1996 in Marano (BO) causing the partial blockage of the Reno river. From geomorphologic surveys it was established that this landslide was a reactivation of an ancient movement. The only available historical reference is that of Calindri (1781) who affirms that in an unknown time (maybe Medieval) a landslide intersected the confluence between Marano Creek and Reno River provoking the destruction of an antique thermal spa. In a test borehole drilled through the toe of the landslide an organic level, located under the debris and maybe displaced by the last movement, was recovered. It seemed reasonable to correlate this level to the presence of a peat bog that probably developed in that site as a consequence of a blockage of the Reno river. The radiocarbon dating of a charcoal fragment from the organic level gives the minimum age of the landslide movement at 850 cal y B.P., in agreement with Calindri’s information. However the charcoal fragment could be also older than the landslide and its radiocarbon dating could represent its maximum age. Gathering together all these information, it could be affirmed that the Marano event represents an example of how radiocarbon dating could be crosschecked with the available historic references, leading to the conclusion that, by more chances, landslide occurred in Medieval time.

Recovery of Datable Elements: The accuracy of a radiocarbon dating depends on many factors (mainly related to possible environmental contamination of the sample) and, in the particular case of a landslide, on the position of the datable matter with respect to the landslide body. In order to obtain an accurate dating it is essential to find the least disturbed datable elements.

Lang et al. (1999) presented a scheme that, reviewing the original work proposed by Corominas et al. (1993), classifies the datable elements in seven classes on the basis of their recovery location with respect to the landslide body. Following this classification it is possible to group the selected landslides from the Northern Apennines in four different classes [that are presented in the box above]. The classification, following the scheme of Lang et al. (1999), of each organic sample that was subject to radiocarbon dating shows also other important attributes of the selected landslides. Ages are presented both by conventional age and calibrated age. The calibration has been carried out using Oxcal 3.4, a software, free on the Net and developed by Bronk Ramsey (1998).


(4) - Remarques générales

 


(5) - Syntèses et préconisations

Radiocarbon dating of landslides leads to a better grasp of landslide activity in the Northern Apennine, during late Pleistocene and Holocene.

Large landslide (area > 400,000 m²) were probably triggered as a consequence of the sharp climate variation that characterised the beginning of the Holocene and took place with a maximum intensity in areas located around the upper snow limit. In the Holocene periods these large landslides went through partial reactivations that seem to be related to colder conditions. As a whole, Holocene climatic oscillations played an important role in moulding the Northern Apennine landscape, that was probably not a continuous processes, but occurred mostly by impulses depending on elevation range and climate phases.

Références citées :

AA.VV.: 1999, I Beni Geologici della Provincia di Modena, Univ. di Modena e Reggio Emilia e Amm. Provinciale di Modena, Artioli ed., Modena.

Alessio, M., Allegri, L., Antonioli, F., Belluomini, G., Ferranti, L., Improta, S., Mandra, L., and Proposito, A.: 1992, Risultati preliminari relativi alla datazione di speleotemi sommersi nelle fasce costiere del Tirreno Centrale, Giornale di Geologia, serie 3° 54(2), 165–193.

Almagià, R.: 1907, Studi Geogra?ci sopra le frane in Italia,Mem. Soc.Geogr.It. 13, I Vol., 342 pp.

Ambrosetti, P., Carraro, F., Deiana, G., and Dramis, F.: 1982, Il sollevamento dell’Italia centrale tra il Pleistocene inferiore ed il Pleistocene medio, In: CNR (ed.), Progetto Finalizzato “Geodinamica”, Vol. 1, Contributi conclusivi per la realizzazione della Carta Neotettonica d’Italia, CNR, Roma, pp. 219–223.

Bartolini, C., Bernini, M, Carloni, G. C., Costantini, A., Federici, P. R., Gasperi, G., Lazzarotto, A., Marchetti, G., Mazzanti, R., Papani, G., Pranzini, G., Rau, A., Sandrelli, F., Vercesi, P. L., Castaldini, D., and Francavilla, F.: 1982, Carta Neotettonica dell’Appennino Settentrionale. Note illustrative, Boll. Soc. Geol. It. 101, 523–549.

Bertolini, G., Canuti, P., Casagli, N., De Nardo, M. T., Egidi, D., Garberi, M. L., Genevois, R., Mainetti, M., Palumbo, A., Pignone, R., Pizziolo, M., and Tomassetti, C.: 1999, La Carta della Pericolosità Relativa da Frana ai fini di Protezione Civile della Regione Emilia-Romagna, Atti. Conv. Lincei. 154, 367–373.

Björk, S., Kromer, B., Johnsen, S., Bennike, O., Hammarlund, D., Lemdahl, G., Possnert, G., Rasmussen, T. L., Wohlfarth, B., Hammer, C. U., and Spurk, M.: 1996, Syncronized terrestrial atmospheric deglacial records around the Atlantic, Sciences 274, 1155–1160.

Bronk Ramsey, C.: 1998, Probability and Dating, Radiocarbon 40(1), 461–474.

Calindri, S.: Dizionario corografico, georgico, orittologico, storico della Italia. Bologna, 6 volumes, pp. 1781–1785.

Canuti, P., Casagli, N., and Ermini, L.: 1998, Inventory of landslide dams in the Northern Apennine as a model for induced flood hazard forecasting, In: K. Andah (ed.), CNR-GNDCI-UNESCO (IHP): Managing Hydro-Geological Disasters in a Vulnerable Environment for Sustainable Development, Publ. No. 1900 CNR-GNDCI, pp. 189–202.

Casagli, N. and Ermini, L.: 1999, Geomorphic analysis of landslide dams in the Northern Apennine, Japanese Geomorphological Union Transactions 20(3), 219–249.

Corominas, J., Weiss, E. J., Van Steijn, H., and Moya, J.: 1993, The use of dating techniques to assess landslide frequency, exemplified by case studies from European countries’, In: R. Casale, R. Fantechi, and J. C. Flageollet (eds), Temporal Occurrence and Forecasting of Landslides in the European Community, Contract Epoch No. 90 0025 (Dtee) – Final Report, pp. 71–91. - [Fiche Biblio]

Corsini, A., Pasuto, A., and Soldati, M.: 1999, Geomorphological investigation and management of the Corvara landslide (Dolomites, Italy), Japanese Geomorphological Union Transactions 20(3), 169–186.

Corsini, A., Pasuto, A., and Soldati, M.,: 2000, Landslides and climate change in the Alps since the late Glacial: Evidence of case studies in the Dolomites (Italy), In: E. Bromhead, N. Dixon, and M. Ibsen (eds), Landslides in Research, Theory and Practice, Proc. of the VIII Symp. on Landslides, Thomas Telford, London, pp. 229–234. - [Fiche Biblio]

Dikau, R. and Schrott, L.: 1999, The temporal stability and activity of landslides in Europe with respect to climatic change (TESLEC): Main objectives and results, Geomorphology 30, 1–12.

Gams, H.: 1937, Aus der Geschichte der Alpenwälder, Zeitschrift des deutschen und österreichischen Alpenvereins 68, 157–170.

González-Díez, A., Salas, L., Díaz de Terán, J. R., and Cendrero, A.: 1996, Late Quaternary climate changes and mass movement frequency and magnitude in Cantabrian region, Spain, Geomorphology 15, 291–309.

Guzzetti, F., Cardinali, M., and Reichenbach, P.: 1994, The AVI Project: A bibliographical and archive inventory of landslides and floods in Italy, Environmental Management 18(4): 623–633.

Huntley, B.: 1900, European vegetation history: Paleovegetation maps from pollen data – 13,000 y BP to present, J. Quat. Sci. 5, 103–122.

Kerschner, H.: 1985, Quantitative palaeoclimatic inferences from late-glacial snow line, timberline and rock glacier data, Tyrolean Alps, Austria, Zeit. Gletscherkd., Glazialgeol. 21, 363–369.

Lamb, H. H.: 1977, Climate: Past, Present and Future. T. II, Climate History and the Future, Mathuen, London.

Lang, A., Moya, J., Corominas, J., Schrott, L., and Dikau, R.: 1999, Classic and new dating methods for assessing the temporal occurrence of mass movements, Geomorphol. 30, 33–52.

Lowe, J. J. and Watson, C.: 1993, Lateglacial and early Holocene pollen stratigraphy of the Northern Apennine, Italy, Quat.Sci.Revs. 12, 727–738.

Magri, D.: 1999, Late Quaternary vegetation history at Lagaccione near Lago di Bolsena (central Italy), Review of Paleobotany and Palinology 106, 171–208.

Marckgraff, V.: 1974, Paleoclimatic evidence derived from timberline fluctuations, In: J. Labeyrie (ed.), Les Methods quantitative climate ou corse du Pléistocène, Colloques internat du CNRS No. 219, Paris (CNRS), pp. 67–83.

Orombelli, G. and Ravazzi, C.: 1996, The late glacial and early Holocene: Chronology and Paleoclimate, Il Quaternario 9(2), 439–444.

Panizza, M.: 1985, Schemi cronologici del Quaternario, Geografia fisica e Dinamica Quaternaria 8, 44–48.

Pellegrini, M., Tellini, C., Vernia, L., Larini, G., and Marchi, G.: 1998, Caratteristiche geologiche e morfologiche della Grande Frana di Corniglio (Appennino settentrionale, Provincia di Parma), Mem. Soc. Geol. It. 53, 543–561.

Pinna, M.: 1996, Le variazioni del Clima. Dall’ultima grande glaciazione alle prospettive per il XXI secolo, Franco Angeli, 214 pp.

Regione Emilia-Romagna: 1999, I numeri sulle frane. A cura di Garberi, M. L., Palumbo, A. and Pizziolo M., con la collaborazione di Baldelli, C., Barchiesi, P., Bertolini, G. and De Nardo, M.T.; Servizio Cartografico, Pendragon ed, Bologna.

Roberts, N.: 1998, The Holocene, Blackwell, 260 pp.

Starkel, L.: 1966, Post-glacial climate and moulding of European relief, Proc. Int. Symp. World Climate 8000 to 0 B.C. Royal Meteorological Soc., London, pp. 15–32.

Stuiver, M. and Reimer, P. J.: 1993, Extended 14C data base and revised CALIB 3.0 14Cage calibration program, Radiocarbon 35, 215–230.

Stuiver, M., Grootes, P. M., and Braziunas, T. F.: 1995, The GISP2d18O Climate Record of the past 16,500 years and the Role of the Sun, Ocean, and Volcanoes, Quat. Res. 44, 341–354.