Page 18 - Sea-level change_2004

ARTICLE IN PRESS
1986 K. Lambeck, A. Purcell / Quaternary Science Reviews 24 (2005) 1969–1988

earlier than inferredfrom the far-fielddata or, because Observations from these sites provide the means to
this location is sensitive to the choice of model effect a direct separation of the isostatic and ocean-
parameters (c.f. Figs. 8 and10), the values for the mantle volume signals. At other locations, the predictions are
viscosities shouldbe decreased by a small amount. insensitive to some of the earth-model parameters,
However, until the analyses of a larger Mediterranean particularly to H 1 and Z um , or to both of these
data set is completed, and because the global esl model is parameters (Fig. 8). Other important locations include
consistent with data from outside the Mediterranean, we the north coast of Africa, particularly between and
adopt the global estimate for the esl function. including the Gulfs of Sirte and Gabe ` s, andthe coast of
The resulting sea-level predictions are illustrated in Tunisia. Other potentially useful sites include some of
Fig. 4 for the LGM, the earliest Holocene, andfor the the small (e.g. Malta and Lampedusa) or intermediate
two Late Holocene epochs. An essential characteristic of (e.g. Sardinia) sized islands.
the predictions is that—in the absence of tectonics— The predictions for 2000 years ago (Fig. 14d) are of
well-developed highstands do not occur at any time particular interest because of the substantial amount of
within the Mediterranean basin because of the compet- archaeological sea-level markers available for this
ing effects of the two isostatic contributions Dz I-g and period(e.g. Flemming, 1969). This includes the tidally
Dz I-h andbecause of the small ocean volume increase controlledpiscinae or fish tanks constructedbetween
between 7 and3 ka BP introducedby the esl correc- about 100 BC and100 AD (Schmiedt, 1972). The model
tion. These predictions are model-parameter sensitive predictions exhibit a spatial variability of 1 m at this
andwhile the nominal values adoptedgive a good first- time between sites in North Africa, Italy, andFrance
order description of the observed change, there is where Roman periodconstructions provide sea-level
potential for parameter improvements using fielddata information (e.g. Flemming, 1969; Caputo andPieri,
from the region. The analysis presentedhere provides 1976; Pirazolli, 1976; Morhange et al., 1996) which,
guidelines as to the ideal locations for such data. where the markers bear a precise relation to the mean
Across the region for each epoch, there are locations level, can provide precise constraints on the predictive
where the isostatic corrections vanish (see Figs. 2d–f), model (Lambeck et al., 2004b). Alternatively, compar-
where the relative sea level equals the equivalent value. isons of the model predictions with the archaeological

Fig. 14. Predicted relative sea levels and shorelines across the Mediterranean region at four epochs: (a) 20 ka, (b) 12 ka, (c) 6 ka, (d) 2 ka. The palaeo-
shoreline positions are defined by the green–blue transition. For 20 and12 ka BP, the contour intervals are 5 m. For 6 and2 ka BP, the redcontours
denote negative values, the orange contours denote positive values, and the yellow contour corresponds to zero change. The contour intervals are 1 m
at 6 ka and 0.25 m at 2 ka. The predictions are based on the nominal model parameters (E2, NE-2, NA-2), Antarctic and Alpine deglaciation with on-
going deglaciation of the former defined by the esl function for the Late Holocene period (Fig. 12).