Varve-Teleconnection Across the Baltic

Abstract

In Sweden there are over 3900 varves which have been measured and numbered (-V), in the New Swedish Time Scale. In Finland 2200 varves have been measured and numbered in the Finnish scale (F). The 2200 years of the Finnish scale are known to lie within the 3900 years of the Swedish, and one hypothesis suggests that when F = 0, V = -8213.

Varve-curves are published for both Sweden and Finland and patterns have sometimes been “identified” visually across the Baltic, but the matching of such curves by eye has not led to any accepted correlation. Here we have assumed that non-random features, found in summer weather to-day, existed also in Late Glacial times, and then we have searched for the best fit statistically. For any given non-random element we constructed “skeleton plots”. Anomalies, positive and negative, were weighted on a scale of 1 to 4 (one representing σ or the standard deviation). These weights were multiplied together for all plausible points of correlation. Scores, plotted against the lag (the difference F - V), oscillated about zero except near the correct position. Successive approximations enabled this to be specified, first within a quarter-century, then within a decade or less.

A major discrepancy occured between positive and negative Finnish dates, that is, dates respectively after and before the end of Salpausselkä II. Positive dates (zone IV) required a formula F-V < 8160, negative dates required a lag > 8212. A breakdown by smaller periods was therefore attempted. For F > 0, independent tests for each century revealed c. 8160 > F - V > c. 8130 and precisely 8149 in the 6th Finnish century. For F < 0, (F - V) was c. 8217 in the period -600 > F > > -900 and precisely 8226 in the period -1000 > F > > -1400. The errors are within the limits expected by the Swedish geologist Fromm (1970).

The major discrepancy (c. 8226-c.8140 = 86 years) could be explained by errors

(a) in the Finnish series

(b) in the Swedish series.

In the Finnish series minor errors appear to be associated with those periods in which the varves were narrow.

In the Swedish series there are major errors in the difficult period of the Central Swedish moraines. Some of Nilsson's varves dated -8080 (and some C.-8071) appear to belong to the same year as others dated -8166. Earlier dates are thus nearly a century too long.

The author's methods of teleconnections cannot easily be used in the period C.-8000 to 8250 and the advice of Scandinavian field-workers is sought.

Index numbers are being prepared by independent collaborators for computer study of meteorological oscillations. Suggestions are made for testing hypotheses for discrepancies by analysis of aeolian content, micro-varves, X-ray structure, etc. The same methods have been used to cross-correlate the Baltic varves with the several series numbered by Antevs in the United States and Canada. The principal correlations are as follows: The New England varves (B) agree with the Nilsson (N) and (negative) Finnish chronology:

B - N = 17027 ± 5 (Pollen Zone I)

B - N = 16960 ± 10 (Pollen Zone II)

The Canadian varves (C) likewise agree with the (positive) Finnish (F) and the New Swedish Varve Chronology (V):

F - C = 52 to 63, varying in different periods C - V = 8083 to 8090

The Döviken zero-varve in northern Sweden (V = -6923) corresponds with a formula C-V = 8089 ± 1 or C-D = 1166 ± 1 and (D) dates relative to this zero are given by the formula

C-D = 1166 ± 10

Transatlantic correlations are also discussed, and, on certain assumptions, about 4200 conventional C¹⁴ years must be equivalent to 3900 tree-ring years in pollen-zones I to V.