Late Devensian and Holocene records of relative sea-level changes in northwest Scotland and their implications for glacio-hydro-isostatic modelling

Raised tidal marshes and isolation basins (lakes that were once connected t o the sea) in northwest Scotland record changes in relative sea level follo wing deglaciation during the Late Devensian to the present. The Kentra to A risaig area, which was covered by relatively thick ice (c. 900 m) at the La st Glacial Maximum (LGM), shows a regression from a marine limit between 36 .5 m OD and 40 m OD at c. 15.9 kyr cal BP (range 15.6 - 16.3 kyr cal BP) th rough to an early-Holocene minimum. A range of sites in the same area recor d a mid-Holocene maximum, indicative of mean sea level c. 6.5 m above prese nt. The maximum is not a well-developed and short duration highstand as pre dicted by a number of models, but is an extended period, similar to 8.0 - 5 .0 kyr cal BP, with sea level within c. 1 m of the maximum. Sites to the no rth, in Kintail, show no Late Devensian record because much of the area lie s within the Younger Dryas ice limit. The altitude of the mid-Holocene maxi mum in Kintail is not well constrained, but occurred 7.9 - 8.1 kyr cal BP. Further north, sites on the Applecross peninsula record a Late Devensian fa ll in sea level and a Holocene maximum for mean sea level no higher than c. 3.0 In above present. In Coigach, the furthest north of the new sites and well outside the Younger Dryas Ice limit, there is no evidence recorded of Late Devensian sea levels above present. The Holocene maximum here was arou nd c. 2.5 m above present. These observations of sea-level change, all stan dardised to change in mean sea level relative to present, constrain the gla cio-hydro-isostatic rebound model parameters. Earth models comprising three mantle layers, with lateral viscosity and elastic parameters, give a satis factory description of rebound. The parameters H-1 (lithosphere thickness) = 65 km, eta(um) (upper mantle viscosity) = 4 x 10(20) Pa C-14 seconds and eta(lm) (lower mantle viscosity)= 10(22) Pa C-14 seconds give the best over all agreement but discrepancies between observations and predictions remain . hn increase of 10% in ice thickness north of the Great Glen, compared to the previous optimum ice model, provides good agreement for many sites but important discrepancies remain for the northern sites and indicate inadequa cies in the model of the British ice sheet. Several alternative ice models are examined but the various combinations of earth and ice-model illustrate the non-uniqueness of the solution. A combination of more extensive ice li mits, especially onto the Hebridean Shelf and West Shetland Shelf, and some changes to ice thicknesses over the mainland should produce a better agree ment, but the spatial coverage of observations remains a limitation to prod ucing a unique solution. The characteristics of the Holocene highstand, age , duration and amplitude, at the different sites refutes the assumption tha t globally deglaciation ceased abruptly 7000 yr ago. The observations are c onsistent with an ice model that includes c. 3 m of melting over the last 7 000 yr. (C) 2000 Elsevier Science Ltd. All rights reserved.