I think that what this guy claims.
http://www.vukcevic.talktalk.n...
Direct relationship between GMFz, Arctic temperature anomaly and the Atlantic multidecadal oscillation index is established, with the GMFz in the lead on the time scale. It should be borne in mind that the AMO was identified in 2001 by Goldenberg et al and since it is correlated to air temperatures and rainfall of the North America and Europe it was reconstructed back to late 1850s, based on available instrument data .
Note: There is no delay between the Arctic temperature and GMFz. However there is variable delay between the Arctic temperature and the AMO, this is possible due to variable velocity of the Arctic ice drift into the Northern Atlantic.
On the average, it takes ice about 6-10 years (min 6 years) to drift from the Beaufort Sea to the Fram Strait.
Sea ice from the Kara Sea region reaches Fram Strait from 2 to 4 years (min 2 years) on average, while sea ice from the Laptev Sea takes roughly 4 to 6 years (min 3 years) to reach Fram Strait (see the Arctic map above).
Journal of Marine Systems, Volume 48, Issues 1-4, July 2004, Pages 133-157
This is reflected in the optimal correlations calculated for the 3 indices:
Excel calculated correlations:
Arctic temper anomaly - GMFz R = 0.9434
AMO ( advanced 7 yr ) - GMFz R = 0.8246