The LGM interval used here encompasses the 16-20 kyr interval of the conventional but uncalibrated 14C chronology. It includes the time slice selected by PMIP for paleoclimate model experiments.
Care was taken to avoid data from Heinrich layers 1 and 2, which actually correspond to colder conditions than the 16-20 kyr interval by itself [Bond et al., 1993. Nature, 365: 153]. Yet, the 18±2 kyr interval includes data from three more discrete fast-deposited units [Stoner et al., 1996. Paleoceanography, 11: 309]. In most cores, high resolution 14C and stable isotope stratigraphies as well as exhaustive sedimentological data sets, allowed to delimit the interval.
The reference dinoflagellate cyst data base includes 439 sites from middle to high latitudes of the North Atlantic and its adjacent polar and subpolar basins. It also includes reference points from Arctic polynia, from the Beaufort, Siberian and Barents seas. The environmental parameters considered include sea-surface temperatures and salinities in February and August (i.e., coldest and warmest months of the year), in addition to the seasonal duration of sea-ice cover.
Modern sea-surface (0 m) temperatures and salinities are compiled after the World Ocean Atlas 1994 data set (NODC 1994). The seasonal duration of the sea-ice cover corresponds to the number of months per year, with 50 % or more of sea surface covered by sea-ice. It is compiled after the 1953-1990 data set provided by the National Climate Data Center (NCDC at Boulder).
A total of 27 dinoflagellate cyst taxa were used for statistical treatments. The procedures, which rely on methods and programs by Guiot [1990; Palaeo3, 80: 49], were basically the same than the ones described by de Vernal et al. [1994; Can. J. Earth Sci., 31: 48]. The search for modern analogues was done after logarithmic transformation of percentage data, using an even weight for all taxa. A set of ten analogues was used to estimate past sea-surface conditions.
The last validation runs (April 1996) yielded accurate results. More than 95% of reconstructions are included within one standard deviation of instrumental data. The range of environmental values, the coefficients of correlation (R) between estimations and instrumental values, and the limit of accuracy (s), are as follows:
Reference data bases and validation of transfer functions
| *Temperature-August (-0.9-21.2°C) | R = 0.965 | s = 1.4°C |
| *Temperature-February (-1.8-15.2) | R = 0.968 | s = 1.1°C |
| Salinity-August (25-36.1 o/oo) | R = 0.951 | s = 0.7 o/oo |
| Salinity-February (27.9-36.1 o/oo) | R = 0.958 | s = 0.6 o/oo |
| Sea-ice cover (0-11.7 months/yr.) | R = 0.936 | s = 1.2 months/year |
Sea-ice cover is a parameter which is characterized by a particularly high interannual variability. The standard deviation around the mean extent of sea-ice is 3.2 months/yr. according to observations. Therefore, in the case of sea-ice, the accuracy of reconstructions can be considered excellent.
At the moment, the proposed reconstructions rely on 31 sites from the northwest North Atlantic. At each site, series of samples (N = 3 to 24) from the LGM-interval were analysed.
At some sites, variable sea-surface estimates are indicated by the studied assemblages (see Table 1). The possibility of a bias linked to low cyst counts (productivity was low during the LGM) or to distant analogues in the reference data base, cannot be totally discarded. However, the distribution of these sites is consistent geographically and probably reflects a "real variability" of the sea-surface conditions at the limit of sea-ice. Modern analogues for such highly variable environments are found in the Northwest Greenland Sea, in the Frame Strait region notably.
At most sites, LGM estimates of paleo-SS conditions show noticeable fluctuations (median, mean, and standard deviation are reported in Table). To draw the maps, we used the mean values.
The results presented here are preliminary. From day to day, complementary analyses improve the statistical value of the reconstructions.
LGM sediments of the North Atlantic yield sparse microfossil assemblages.(dinoflagellate cysts, coccoliths, diatoms, foraminifers [e.g., Hillaire-Marcel et al., 1994; Can. J. Earth Sci., 31: 139]). The possibility of an over-representation of long-distance transported specimens cannot be excluded although it is not probable in view of the distinct characteristics of the assemblages on regional scales.
These LGM reconstructions will be regularly updated until completion of the work. Please quote the date of the version if using this information.
also from GEOTOP: Guy Bilodeau, Linda Levesque, André Rochon and Sylvain Vallières
and from abroad: Jean-Louis Turon (Université de Bordeaux I), Jens Matthiessen (Alfred Wegener Institute, Bremerhaven),
Joël Guiot (Faculté de Saint-Jérôme) and many others who help with data acquisition and/or sampling.
