Researchers Explore Climate Variations Over the

Last 80 ka on Decadal-to-Millennial Time Scales

 

Introduction

           

Drilling of two ice cores on the summit of Greenland in the early 1990s provided dramatic confirmation of numerous large and abrupt reversals of climate, especially in the equivalent of marine oxygen isotope stages (MIS) 3-4 , about 25-80 ka. These findings provide support for the conjecture of Broecker et al. (1985, and earlier comments by Stommel) of possible multiple stable states in the ocean-atmosphere system that could be due to abrupt changes in the strength of the North Atlantic thermohaline circulation (THC), "The Great Ocean Conveyor".

            Since these findings a wealth of new information has provided consid­erable addi­tional insight into the patterns of climate change during these inter­vals. Many records of environmental change covering the last 80,000 years from polar, marine and terrestrial systems show that there were rapid changes with "Dansgaard-Oeschger" (DO) -style and higher-frequency oscillations in, for example, atmospheric temperature as inferred from d18O in ice records, sea sur­face temperature as inferred from planktic foraminif­era faunas in ma­rine sediments, and in terrestrial vegetation as deduced from lacustrine pollen. DO cycles occur in subpolar North Atlantic marine records nearby to Greenland and European pollen profiles as well as in a number of more remote regions (South Atlantic, circum-North Pacific, and south and east Asia) and in atmospheric trace gas (CH4 and CO2) records of Antarctica.

            The pattern of climate change observed in a number of sites supports the idea of a north-south anti-phasing of responses in the Atlantic Basin. In the north ~700-1500 year long DO stadials, including the large Heinrich Events, are marked by massive meltwater and ice­berg discharges. In the last phase of the stadials brine discharges may have trig­gered the subsequent prominent, but only 3-15 years long DO warming events. There is also an unusual and not-well-understood ~7200 year oscillation, the saw-toothed Bond Cycles of Heinrich events in many of these records. These cyclicities show the largest ampli­tudes during cold marine isotope stages, when the sea level dropped below -45 m. In most cases the stadials appear to greatly reduce North Atlantic Deep Water (NADW) production rates and the global conveyor, associated with major changes in atmospheric 14C. It has also been suggested that the cycles may be due to low-frequency variations in the equatorial Pacific.

Prerequisites for the study of the mechanisms which control rapid decadal to mil­lennial environmental changes and leads and lags between different global environ­mental systems are high-resolution stratigraphic records with accurate calendar year chronolo­gies. Precisely this issue of creating a joint global time scale of decadal-to-millennial-scale climate oscillations was discussed by more than 50 top scientists at the SCOR-IMAGES Workshop in Trins, Austria, February 16-19, 2000. Their broad spectrum of scientific expertise comprised spanned from climate modelling, paleoceano­graphy, ice core and terrestrial paleoclimatology, trace gas analysis to radiometric and geomagnetic dating).

 

Task Group 1:  Calendar year chronologies of the last 80,000 years

           

Task Group 2:  Physical dating techniques

 

            The techniques for establishing calendar year chronologies for stratigraphic records of the last 80 ka fall into three categories:

 

1.Incremental dating techniques which provide continuous and independent calendar year chronologies throughout a record.

 

- Annual rings laid down by long-lived trees in temperate regions of the world provide one of the best annual records - dendrochronology. Ring-width variations can be cross-matched between trees and a composite dendro­chronological (calendar year) timescale can be constructed. A continuous dendrochronology now extends to 11,800 calendar years BP. Older "floating" sections of dendrochrono­logical records have yet to be tied to the continuous record; once this is achieved, a detailed calibration curve will extend to ca 40 ka. However, it is rarely possible to apply dendrochronological dates to other records without recourse to cross matching between tree-ring measurements and other proxies.

- The ice cores from the polar ice sheets have yielded, and will continue to yield incred­ibly valuable palaeoclimatic archives. One great strength of the records is their extremely high stratigraphic resolution (for the upper parts, sub-annual) and that many different types of record may be derived from the same core. Thus there is not the phase uncertain­ty between the various climate records from a single ice core that so often affects different geological records (although uncertainties exist in transferring a solid ice time scale to the atmospheric gas records). The stratigraphically best-resolved records for the last climate cycle come from two cores drilled at the summit of Green­land. The GISP2 and GRIP data sets were published together in a single issue of J.G.R. (Vol 102, 1997). How­ever, sev­eral different time scales exist for each ice core and to date no attempt has been made to develop an optimal timescale based on the data in both the ice cores. 

            The GISP2 time scale of Meese et al. (1997) was derived by continuous counting of several properties exhibiting a seasonal cycle; conservatively estimated errors range from 2% over the last 40 kyr to about 20% at 110 ka. Older than about 50 ka the "offi­cial" GISP2 timescale uses a comparison of d18O of oxygen in trapped air with that in the Vostok record and is consis­tent with the marine SPECMAP timescale. This provides a time scale with probably somewhat smaller uncertainties at 100 ka. An independent time scale for the GISP2 core older than 40 ka and back to the last interglacial was based on the record of laser light scattering by dust particles, which preserves annual layering.

            The GRIP time scale back to ca 7000 years BP is based on a correlation with the counted Dye-3 record using volcanic acidity layers as tie lines. Continuous layer counting in GRIP was used be­tween 7000 and the Oldest Dryas/Bolling transition close to 14,500 BP. Ice flow modeling, incor­porating smoothly varying accumulation estimates based on the d18O record, and using age control points at the Younger Dryas/Pre­boreal transition (11 500 BP) and MIS 5d at 110 ka, was used to con­struct the remainder of the GRIP "ss08 timescale" (Dans­gaard et al., 1993). Hammer et al. (1997) have given an independent time scale for GRIP back to ~60,000 years, based on annual dust layers.

-  Annually laminated or varved deposits occur in both the marine and lacustrine realms. Marine varved chronologies are usually fragmentary covering only parts of the last 80 ka, e.g. 9-15 calendar ka in the Cariaco Basin record. Lacustrine envi­ronments can provide ideal conditions for varved sedimentation throughout their exis­tence. Few such sites are Lake Gosciaz (Poland), Lake Holzmaar, Meerfelder Maar (Germany), Lake Monticchio (Italy), Elk Lake (USA), and Lake Suigetsu (Japan), but none of them covers the complete last glacial cycle.

-  Corals and speleothems can have annual growth bands which provide a chronological framework for subsequent analyses. The counting of growth bands, however, produces calendar years only for modern corals and speleo­thems, usually restricted to the last few centuries. Fossil corals can provide records from any older time window providing "floating" chronologies that need to be dated by other techniques such as 14C and U-series dating. Speleothems might capture much longer time intervals but in long speleo­thems the growth bands tend to fade and loose their annual resolution.

            The consensus of the Trins workshop is that the records from the Greenland ice cores have a value that extends far outside the "ice core community" for many reasons:

1. These two records certainly provide the longest archives within which features such as the 11-year sunspot cycle can be studied.

2.  Some hypotheses for explaining climatic variability rely on time-series analyses which are critically dependent on the integrity of the timescale.

3. The important paleoceanographic records available can only obtain a precise timescale sufficient for time-series analysis through correlation to the Green­land ice core record.

4. Correlation to Greenland obviously provides an essential link for ensuring the accuracy of the Antarctic ice core chronologies.

5. The accuracy of time scales based on annual layer counting in lake sediments and tree rings can ultimately only be checked by compar­ison with other annually resolved time scales. Again the Greenland records will probably remain an ulti­mate means of checking.

6. More and more important paleoclimatic data series are based on "floating" annually resolved timescales: stalagmite sections, pieces of wood, lake sediments not extending to the present, and so on. Their value will be enhanced if they can be fixed in time through correlation to the Greenland ice record with its broad diversity of proxy data sets.            Thus it is imperative that a significant coordinated effort should be made to develop a new calendar time scale for the Greenland GISP2 and GRIP records, that is documented in sufficient detail that it can in the future be readily upgraded incrementally as new informations (e.g. from NGGRIP) become available.

 

2. Radiometric techniques which provide discrete dates for sub­samples from a record, with the continuous chronology being constructed by interpolation between these dates.

 

-  Uranium-Thorium (230U/234Th) dating using thermal-ionisation mass spectrometry (TIMS) can be used to date materials from speleothems, corals, carbonate deposits, peat and bones from 5-350 ka. The crucial  prerequisite for this technique is that there has been no loss or gain of nucleides since depostion, i.e. the sys­tem is closed. Unlike 14C the formation of these nuclides is not dependent on varia­tions in solar activity or reorga­nizations in the global carbon cycle. When correcting for various sources of error (devia­tion from the closed system, inherited Th) this technique provides calendar chronologies.

-  Radiocarbon (14C) dating is the much widely used technique for dating materials up to ca. 50 ka. 14C dating by decay counting or Accelerator mass spectrometry (AMS) has an age uncertainty ranging from about ñ20-30 years for samples back to 10.5 ka, up to ~100 years back to 20 ka, and to >~1000 years near to the limits of the counting statistics and background due to counting statistics depending on the age and size of a sample. Since the atmospheric 14C/12C ratio was not constant in the past as a consequence of short-term changes in the 14C pro­duction rate and/or in the properties of the oceanic carbon system (e.g. deep-water formation and upwelling) the radiocarbon age has to be calibrated.

            This calibration which is necessary to compare 14C dated records to other calen­dar year records leads to a further increase of the error range. The problem is exacer­bated by plateaux in the calibration curves resulting from a great temporal variations in the atmo­spheric 14C content. Back to 11,800 years the calibration uses 14C datings of dendro­chronologically dated tree rings and will be extended into the late Weichselian when floating Late Glacial dendrochronologies are tied to the Holocene master record. Beyond these limits back to about 55,000 years the attempts to extend the calibration curve are based on (1.) 230U/234Th ages of corals, (2.) annually laminated sediments from marine (Cariaco Basin, off Venezuela) and lacustrine realms (Suigetsu, Japan; Lisan, Israel). (3.) An indirect calibration approach is to measure high-resolution 14C ages on records for which a climate proxy exist, that allows detailed cross-matching with well-dated climate records, e.g. from Greenland ice cores.

            In addition, the dates in marine sediment records may be biased by the local 14C reservoir effect, i.e. the 14C/12C ratio of CO2 in ambient sea water, that is different from the atmospheric ratio. This local effect may vary by the same magnitude as potential leads and lags between climate signals of interest and thus add large uncertainties to the inter­pretation of marine proxy records.

            To overcome these calibration problems we need to measure 14C ages at high resolution in further annually laminated records (e.g. Lake Monticchio, Italy; Lake Suigetsu, Japan) which have independent calendar-year chronologies and finally, to assemble a single 14C calibration curve from these records for the past 50 ka.

 

3. Correlation techniques which facilitate the synchronous cross-matching of proxies that can be used to transfer the chron­ology of a dated to an undated climate record.

 

-  Geomagnetic palaeointensity is a global and synchronous signal in many geological records and is used to correlate them. A prime candidate for this purpose is the newly available high-resolution pale­ointensity record from the North Atlantic (NAPIS-75) cov­ering the interval 10-75 ka. The stacked record has been dated by detailed correlation of one of the NAPIS-75 cores to the GISP2 ice core via d18O- meltwater signals. Geomag­netic excursions and events are used as master tie points. Refined techniques are neces­sary for precise and objective correlations using the entire shape of the paleointensity records which will soon enable global correlations with an uncertainty of less than 5 ky.

-  Volcanic ash layers, once well defined in terms of geochemistry and mineralogy, can be also utilized as event markers to correlate ice and sediment cores. Furthermore, such events are suitable for 39Ar/40Ar dating with an error of 2 ka or less for the last 80 ka.

-  Changes of sea level directly express changes in the volume of conti­nental ice sheets. Such changes can be determined from coral reef deposits, dated by 14C and/or 230U /234Th methods. Coral terraces at sites with high tectonic uplift  provide detailed records of past sea level changes. However, sea level stands as global chronological markers may be prob­lematic because of local tectonics, effects of storm events, and tsunamis.

            Based on the outlined dating techniques the Trins Workshop is recommending a number of joint research efforts across various disciplines, necessary in the near future to

-  establish a common 14C-time scale based on the existing annually lamina­ted sediment and ice records and to locate additional laminated lacustrine and marine records,

-  calibrate the common 14C-time scale back to 55 ka,

-  define variations in the 14C reservoir effect in various ocean basins,

-  validate and promote the geomagnetic palaeointensity record as dating tool,

-  establish a set of global time markers.

 

Task Group 3: Sites and possiblities to test mechanisms proposed to explain centennial-millennial-scale Dansgaard-Oeschger-Bond oscillations

 

            At present there are a number of sites of varying resolution (Figure 1) that record, or seem to record, the centennial-millennial-scale DO and Bond oscillations. These records describe vari­ability in atmo­spheric/ocean circulation through proxies reflecting changes in polar tempera­ture, SST, sea ice and salinity (meltwater), productivity, inter­mediate/deep-ocean convection, monsoon intensity and precipi­tation. Figure 2 shows examples of some particularly high-resolution time series.

            Several different hypotheses have been proposed to explain, at least in part, such

phenomena, the most prominent being variations of the Atlantic conveyor and low-fre­quency variability in the eastern equatorial Pacific upwelling system. Different mecha­nisms may be triggering such variations Ð in the case of the conveyor e.g. meltwater pul­ses linked to internal (binge-purge) ice sheet instabilities, variations in the tropical Atlantic sector, or solar irradiance changes. The equatorial Pacific mode might be some resonant time scale for the Pacific Basin. The two mechanisms may be interdepen­dent Ð variations in the equatorial Pacific should affect the North American and Greenland ice sheets which, in turn, may affect the conveyor. Conversely, changes in the conveyor may trigger trade wind changes in the eastern equatorial Pacific or changes in the strength of the Siberian high may influence westerly wind bursts in the western equatorial Pacific.

            Each of the above mechanisms has a different "fingerprint" in the time- and space-domain. For example, the conveyor is associated with an anti-phase SST pattern in the North and South Atlantic, possibly also in the northwestern Pacific, with variations in the southern Indian Ocean and the Pacific south of 30-40 degrees N being muted. There are some of the same regions that should record the strong imprint of equatorial Pacific warming.

 

            The consensus of the Trins Workshop supports two major initiatives:

1.  Acquisition of new records to test proposed mechanisms

            One of the goals of future high-resolution climate studies should be to test among these different mechanisms by more specific targeting of "high payoff" sites. Although additional projects of opportunity may justify addition of samples from other regions, the tremendous time and costs of acquiring and analyzing such records requires some priority be given to sites that yield the highest payof for the effort involved. However, because our understanding of such processes is still at an early stage, there should be a delicate bal­ance between these tar­geted sites and 'serendipity'. One criterion to select other sites would be the sheer quality of proposed sites. As geoscience investigations have repeated­ly shown, high quality records con­sistently yield surprises as to how the climate system works. "Incomplete" records of climate change such as varves and speleothems should also be collected for the last 80,000 years. Such information can often be very valuable.

            Some specific tests of the conveyor hypothesis would involve:

-  North Atlantic studies examining phase relations between deep- and surface water hydrography and origin, ice-rafting events, meltwater and brine pulses, and indications of tropical humidity changes (e.g. at tropical Atlantic/Indian Ocean and Mediterranean sites);

-  South Atlantic sites where the Atlantic see-saw would predict increased SST and southern African precipitation;

-  Regions affected by meltwater-induced changes in the strength of the Siberian high and its effect on the south and east Asian monsoons, e.g. to test whether subpolar North Pa­ci­fic SSTs are controlled by coeval atmospheric forcing or rather by thermo­cline ventilation at the terminus of the 'global conveyor' in anti-phase with North Atlantic changes;

-  High-latitude sites in the Southern Ocean and Antarctica, that record anti-phase infor­mation; since the 'conveyor' also predicts a west-east see-saw in Antarctic Circumpolar flow, cooling in the Ross Sea may be in-phase with North Atlantic cooling; additional testing from these sites would be most useful.

            Equatorial Pacific fluctuations can be tested with key records of

-  Eastern equatorial Pacific warming;

-  Drying in Australia that should be detectable with pollen records;

-  SST increases along the west coast of the USA (e.g., Santa Barbara Basin);

-  Increased precipitation in the U.S. southwest;

-  Peru margin upwelling;

-  Potential links to central Canada.

            Holocene studies imply some support for the role of low-frequency solar irradi­ance changes as contributing to decadal-to-millennial-scale variability. If so, such vari­ability may be also detected in MIS 2-4. Better testing of the solar hypothesis requires:

-  More 10Be and 36Cl data that would enable separation of the cosmo­genic effect from the 14C record, which is also influenced by THC changes;

-  More detailed records and improved age control of continental climate change, particularly in the southern hemisphere, for models suggest that the solar signal is stronger over land, with the same sign of response in both hemispheres.

 

2.  Analysis of new and existing records as a guide to testing mechanisms

             A number of complementary / overlapping approaches can provide a better understanding of the processes involved in centennial/millennial-scale climate change:

-  "Zereo'th order" best-fit correlations between different time series to determine more objectively (independent of phase) the linear relationship of different records with the Greenland ice core "type sections". 

-  Time slice studies of selected DOB oscillations to map the spatial and temporal evolu­tion of patterns associated with the atmosphere and ocean (both surface and subsurface). ù  Effect of such changes on methane and carbon dioxide variations and potential feed­backs of atmospheric trace gas changes on centennial-to-millennial-scale oscillations.

-  Intensive statistical examination of the 1500-year cycle to determine its robustness, band with, and coherence and phase relationships between different regions, moreover, an examination of periodicities of the DOB band other than the 1500-year cycle.

-  Analysis of centennial-millennial-scale fluctuations in non-glacial sections (Holocene, MIS 5.5 and 11). Even though such oscillations are of lower amplitude than their glacial cousins, the removal of the greatly complicating effect of most of the northern hemisphere ice sheets may enable isolation of the relative importance of other sources of forcing espe­cially during the last 1000 years, where time and space control of samples is very good.

-  Intercomparisons of climate models to assess robustness of model predicted responses to e.g., meltwater pulses or equatorial Pacific oscillations.

 

Authors:

M. Sarnthein, J.P. Kennett,

MAIN CO-AUTHORS: J. Chappell, T. Crowley, W. Curry, P. Grootes, I. Hendy, C. Laj, J. Negendank, M. Schulz, N.J. Shackleton, A. Voelker, B. Zolitschka

For more information contact M. Sarnthein, Institut fuer Geowissenschaften, University of Kiel , Olshausenstr. 40, D-24098 Kiel, Germany; E-mail: ms@gpi.uni-kiel.de