E-K. Potter, K. Lambeck and T. Esat
The timing and magnitude of sea level oscillations during the last glacial cycle can be calculated from uplifted coral reefs such as those found on the island of Barbados. Using Radtke and Schellmann’s (2001) revised morphostratigraphic analyses of the Barbados coral terraces, as well as a new set of over 80 U-Th coral age measurements, we have re-calculated paleo-sealevels for the oxygen isotope stage (OIS) 5c and 5a sea level maxima. The sea levels associated with the “classic” stage 5c (102 ka B.P.) and 5a (83 ka B.P.) maxima are both calculated to be around 22 m below present. We have also identified a third, previously poorly described, sea level feature with an age of around 76 ka B.P. at a paleo-sealevel of around 24 m below present.

A comparison of the OIS 5a paleo-sealevels calculated for Barbados with estimates of stage 5a sea level at other sites in the Caribbean region show an apparent disagreement. In the Bahamas and Florida stage 5a sea-level estimates range between 5 and 10 m below present. At Bermuda and on the US East Coast, stage 5a age marine deposits can be found above present day sea level. These apparently conflicting observations can be reconciled by taking into account the deformation of the earth in response to changing ice and water surface loads, a process termed glacio-hydro-isostasy. The sites mentioned lie in the “intermediate zone” of the Laurentide Ice Sheet, and so a gradient of stage 5a relative sea level observations is expected. The gradient in stage 5a sea level across this region is sensitive to the rheology of the earth and the melting history of the Laurentide during the time leading up to the peak of stage 5a and during the last deglaciation (stage 2-1).

K. Lambeck
Observed relative sea-level change reflects changes in ocean volume, glaciohydro-isostasy, vertical tectonics and redistribution of water within ocean basins by climatological and oceanographic factors. Together these factors produce a complex spatial and temporal sea-level signal. For the tectonically stable Australian margin, geological evidence indicates that sea-levels at 7000-6000 years ago were between 0 and 3 m above present level, due primarily to glacio-hydro-isostatic effects of the last deglaciation. The spatial variability of this signal determines the mantle response to the surface loading and leads to an effective lithospheric thickness of 75-90 km and an effective upper mantle viscosity of (1.5-2.5) x 10²⁰Pa s. Compared with results for other regions this is indicative of regional variation in upper-mantle response. Also, ocean volumes continued to increase after 7000 years ago by enough to raise global mean sea level by about 3 m. Much of this increase occurred between 7000 and 3000 years ago. Because of the spatial variability in mantle response, isostatic corrections to tide-gauge records of recent change should be based on regional model-parameters rather than on global parameters. The two longest records from the Australian margin give an isostatically corrected rate of regional sea-level rise of 1.40±0.25 mm/year. Comparisons of this rate with rates from other regions indicates that the spatial variability in secular sea-level is likely to be significant, with estimates of regional rates ranging from about 1mm/year to 2 mm/year. These rates of secular change cannot have persisted further back in time than a few hundred years without becoming detectable in high-resolution geological and archaeological indicators of sea-level change.

P. Tregoning, H. McQueen, R. Stanaway, S. Saunders^[1], R. Curley^[2], and K. Lambeck
GPS fieldwork has continued during 2001 to monitor the post-seismic deformation of the Gazelle Peninsula and southern New Ireland in Papua New Guinea after the sequence of three major earthquakes which occurred in November 2000. To date, up to 300 mm post-seismic motion has occurred at the continuously-observing GPS sites operated by the Rabaul Volcano Observatory. Quasi-continuous observations have been made on a bi-weekly basis at several sites near Rabaul while two campaign-style occupations have been carried out on the New Ireland sites.
The information obtained from the GPS monitoring of the post-seismic deformation shows clearly that the region has not yet returned to its pre-earthquake stress state. (Figure 1). Numerical modelling techniques are being applied to extract material parameters and fault geometry from the observations on deformation proceeds. At this stage it is not known how long it will take for this to occur; however, continued monitoring of this lithospheric relaxation/stress cycle will provide considerable insight into the tectonic stress regime that affects the Rabaul and New Ireland regions.

Figure 1: Time series of in local north and east coordinates of the position estimates at the Rabaul Volcano Observatory continuous GPS site with respect to the rigid South Bismarck Plate. Co-seismic offsets and the post-seismic relaxation are clearly evident.

P. Tregoning, H. McQueen and K. Lambeck
The Antarctic GPS program to monitor glacial isostatic adjustment near the Lambert Glacier continued in 2001. In November 2000 our field party returned to the Prince Charles Mountains to check the operation of the Beaver Lake system, install a new suite of equipment at Landing Bluff (at the Amery Ice Shelf coast) and upgrade the installation at Dalton Corner. The Landing Bluff installation was completed successfully and data from this site were transmitted to Canberra on a daily basis from December 2000 to May 2001. Figure 2 shows the calculated daily position estimates of the Landing Bluff GPS site.
At this stage this project still requires additional data before being able to produce accurate estimates of vertical uplift rates at the remote sites. Horizontal velocity estimates at Mawson, Davis and Beaver Lake and show no significant horizontal motion with respect to a rigid Antarctic Plate.

Figure 2: Daily position estimates of the Landing Bluff benchmark during the summer of 2000-2001.

P. Tregoning
An investigation into the rigidity of the Australian Plate was performed using position estimates derived from GPS observations spanning several years. The site positions were calculated in the International Terrestrial Reference Frame 2000 (ITRF2000) and velocity estimates have a precision of typically < 1 mm/yr. These velocities were used to estimate an Euler vector from which the motion of any point on the Australian Plate can be predicted. To within the level of uncertainty of the velocity estimates, no motion of any sites on the Australian Continent can be detected relative to this model of the Australian Plate (Figure 3). That is, there is no large scale deformation within the Australian continent that can be detected at current levels of instrument precision. However, significant relative motion can be seen at Port Moresby, suggesting either that the site is undergoing local deformation or that there is relative motion occurring between mainland Papua New Guinea and the Australian Plate.

Figure 3: GPS site velocities relative to the Australian Plate. 66% confidence error ellipses are plotted.

A simulation of the Fennoscandian Ice Sheet through a glacial cycle
D. Fabel, J,Harbour^[3], A. Stroeven,^[4] J. Kleman,⁴ . C. Hättestrand ,⁴ and D. Elmore^[5]

Critical tests of global climate models include their ability to reconstruct environmental conditions during former periods of distinctly different or rapidly changing climate, such as glacial maxima and periods of large-scale ice-sheet growth and decay. Global climate signals and ice volume estimates have been extracted from deep-sea, coral reef, and ice sheet proxies. Over the past 20 years, since the CLIMAP reconstruction of the Last Glacial Maximum (LGM), including LGM ice volumes, there has been growing concern that ice sheet thicknesses and volumes suggested by CLIMAP may have been overestimated, and that new reconstructions are needed. However, CLIMAP estimate of LGM ice volumes are broadly consistent with some recent new evidence: (i) far-field evidence of sea level low-stands imply more ice on earth than formerly thought, (ii) relative sea level curves from glaciated regions do not normally cover the late glacial period, and, hence, potentially underestimate the volume of ice at LGM, (iii) the Arctic ocean may have been covered by an ice shelf, and (iv) direct glacial-geological observations of ice sheet extent have modified prevailing perceptions of ice sheet volume (e.g. the Innuitian Ice Sheet is now commonly accepted). However, key aspects such as the height and evolution of individual paleo-ice sheets cannot be traced in proxy sea level data, yet have a significant impact on climate models. Potential records of the paleotopography of Northern Hemisphere ice sheets are found in formerly glaciated areas in northern Sweden, western Norway, Scotland, and eastern Canada. In these locations, features such as the upper limit of glacially eroded bedrock surfaces (trimlines) and differences in rock weathering (weathering zones) have been interpreted as indicators of former ice sheet height. This interpretation has been questioned by others who have argued that such features may reflect internal thermal boundaries between wet (warm-) based erosive ice and dry (cold-) based ice that is effectively non-erosive. Resolving this issue is critical because of differences in predicted ice sheet configurations; a large and thick late Weichselian Fennoscandian ice sheet (FIS), as opposed to a much thinner, and possibly multidomed, late Weichselian ice sheet.

Over the past three years we have worked collaboratively to examine deglaciation chronology and patterns of erosion and landscape preservation in the northern Swedish mountains, the core area of the Fennoscandian ice sheet using in situ produced cosmogenic ¹⁰Be and ²⁶Al. The accumulation of in situ produced cosmogenic ¹⁰Be (half-life = 1.51 ± 0.05 x 10⁶ yr) and ²⁶Al (half-life = 7.1 ± 0.2 x 10⁵ yr) in quartz exposed to cosmic radiation provides a means of determining the amount of time the rock has been at or near the ground surface. Because nuclide production decreases with depth, removal of two or more meters of irradiated rock during one glacial event will create a zero age surface. In this context, areas known to have been ice covered should have exposure ages equivalent to deglaciation if they were significantly eroded by ice and older exposure ages if they suffered limited erosion or were completely protected. Ice cover of 10m or more shields the underlying rock surface from most cosmic radiation, so areas that undergo multiple cycles of ice sheet overriding but no erosion should accumulate ¹⁰Be and ²⁶Al cosmogenic nuclide concentrations equivalent to the sum of the ice free periods, minus decay during periods of ice shielding.

Patches of relict landform assemblages were identified in the northern Swedish mountains through extensive field and air photo mapping. Quartz-rich samples for cosmogenic radionuclide analysis were collected from bedrock outcrops and erratics in mapped relict patches. Erratics confirm that the sites were overridden by ice and were dated to determine whether they were deposited during the last glaciation. Bedrock in relict patches was sampled to determine whether these sites were in fact eroded during the last glaciation (to give a deglacial exposure age) or whether they are relict (exposure ages reflect both post-glacial time and inheritance from one or more previous ice-free period).

Our results from erratics provide deglaciation ages and indicate that relict patches have been covered by ice during the last glaciation. Bedrock samples from the same relict patches provide much older cosmogenic ages suggesting insufficient glacial erosion and hence preservation of these patches during multiple ice overriding events. Relict patches therefore appear to represent areas of frozen bed conditions at the base of the FIS supporting the current swing in opinion toward the idea that, under given boundary conditions, ice sheets are landscape 'preservers' rather than 'destroyers’.

The results so far provide the groundwork for the next phase of this work, in which we will reconstruct Fennoscandian ice sheet (FIS) thickness, extent, and dynamics (including total ice volume-induced sea level change) over critical periods of the last glacial cycle (Marine Isotope Stage [MIS] 5d or 5b, inception phase; MIS 2, LGM phase, and; MIS 1, deglaciation phase). The paleotopography (height) of the FIS through a glacial cycle will be simulated using a state-of-the-art thermomechanical numerical ice sheet model, with key boundary conditions constrained both by cosmogenic nuclide-based reconstructions of subglacial conditions, and by an isostatic model. Our team for this work has been expanded to include Hubbard (University of Edinburgh), Lambeck (ANU), and Näslund (Stockholm University) in the areas of glaciological and isostatic modeling. Our proposed modelling is significantly different from prior work, in part because the timing of ice sheet inception and deglaciation, and patterns of subglacial erosion, deposition, and preservation in the mountains and lowlands will be constrained using in situ produced cosmogenic nuclides, with particular focus on the tempo of ice sheet retreat along major deglacial lineation corridors. Our new reconstructions will likely have wide significance, both to European ice sheet and paleoclimate reconstructions, evaluations of far-field evidence of sea level low-stands (proxy for total land-based ice on earth), and also in encouraging re-evaluation of the dynamics of inception, growth, and decay of other major ice sheets worldwide.

Jean Braun
Extracting independent information on mean exhumation rate and the rate of surface relief change from thermochronometric datasets is essential to improve our understanding of the complex coupling between tectonics and surface erosion, i.e. the time scale over which landforms react to changes in uplift rate and/or climate.

We have developed a new method, based on spectral analysis of age-elevation data collected along one-dimensional profiles, that provides independent estimates of the mean exhumation rate and the relative change in surface relief. The method relies on the strong dependency of the thermal perturbation caused by surface topography on the wavelength of the relief. Short wavelength topography affects the geometry of low temperature isotherms, whereas long wavelength topography affects the temperature field to much greater depths.

The topography of the Earth’s surface is self-similar, i.e. relief is usually distributed among a broad range of wavelengths. Collecting age data along a one-dimensional profile thus ensures that the relationship between age and elevation is sampled at a range of wavelengths. At short wavelengths, the relationship between age and elevation provides direct constraints on the mean exhumation rate (i.e. the tectonic signal) whereas at long wavelengths, it provides information on the relative change in surface relief amplitude (i.e. the geomorphic signal). It can be shown that the information that can be derived from age-elevation relationships is independent of the assumed past or present geothermal gradient.

We have applied the method to existing age datasets from the Sierra Nevada, California (Figure 1), and have derived robust constraints on the antiquity of the present-day relief in the region, demonstrating that it is likely to be the result of the slow decay of a topography created during the Laramide Orogeny. Most importantly, the spectral analysis demonstrates how current sampling strategies should be modified to optimise the tectonic and geomorphic information that can be retrieved from a thermochronometric dataset.

Figure 4: (a) Elevation and apatite (U-Th)/He age data are from a transect through the Sierra Nevada, from House et al (1998). A mean value of 2 km and 67 Myr has been subtracted from the elevation and age datasets, respectively. (b) Gain function representing the relationship between age and elevation as a function of topographic wavelength. At short wavelengths, gain contains information on exhumation rate; at long wavelengths, it contains information on surface relief change.

S. Frederiksen, J. Braun
The two end member models of strain geometry are pure shear and simple shear and have been introduced to explain the large-scale morphology of rift zones and passive margins. As the main difference between the two rift morphologies is directly related to the presence (or absence) of a lithospheric-scale shear-zone, it is reasonable to state that simple shear deformation of the lithosphere is most likely to take place when conditions are favourable for strain localisation (i.e. where deformation leads to the formation of narrow shear zones). Conversely, if the lithosphere is extended under conditions, which do not favour strain localisation, it tends to stretch uniformly (i.e. by pure shear). Thus, to understand the reasons for the evolution of a rift system by either pure shear or simple shear requires a better understanding of the mechanical behaviour of mantle rocks under conditions of temperature, pressure and strain rate representative of the uppermost mantle.

There is also direct evidence that strain localisation plays a role during continental extension. This includes well-documented fabrics, such as tectonite and mylonite shear zones that developed in mantle peridotite under sub-Moho P-T conditions and are now exposed at the surface following exhumation by movement along low-angle normal faults. Furthermore strong dipping reflectors observed on reflection seismic lines recorded beneath continental rift zones have in some cases been interpreted to be mantle shear zones.

Strain localisation in the ductile regime is often accompanied by grain size reduction. This has led many to postulate that strain localisation is likely to take place during the transition between dislocation creep and grain-size-sensitive creep. This transition is triggered by grain-size reduction following dynamic recrystallization during deformation in the dislocation creep regime. Under the P-T conditions characteristic of the uppermost mantle, the change in deformation mechanism is accompanied by a large drop in material strength that can lead to strain localisation.

We have developed a first-order rheological model in which the strength of rocks is not only strain-rate-dependent and thermally activated but also affected by the total accumulated strain.

Figure 5: The distribution of total strain in the lithosphere for the same amount of horizontal extension for a model with and without strain localisation.
Results have shown that the lithosphere may evolve from a pure shear into a simple shear system during extension. The onset of simple shear deformation is dependent upon whether or not the conditions in the lithosphere are favourable for strain softening. An example of the difference between a model with and without localisation is shown in Figure 1. A deeper and narrower sedimentary basin has developed in the model with strain localisation. The total amount of horizontal strain is clearly more evenly distributed in the lithosphere when no localisation takes place.

It has also been shown that the isostatic state of the lithosphere during rifting (which will control the vertical deflection of the surface and, hence, the amount of sediments being deposited during and after rifting) is controlled by the presence and nature of a so-called "strong fibre" that usually lies beneath the Moho. The thermal state of the lithosphere as well as the crustal thickness and composition determine the position and relative strength of this fibre. It is interesting to note that strain localisation in the ductile regime predominantly takes place in the coldest regions of the mantle, i.e. where the largest stress levels can be sustained. This implies that strain localisation is likely to affect the strength of the strong fibre and thus the isostatic response of the lithosphere and sedimentation patterns during and after rifting.

J. Braun and R. Shaw^[6]
A thin plate model of the continental lithosphere in which deformation is driven by velocities specified along the plate boundaries has been developed and applied to the Australian continent. The geometry of the model, the strength of each lithospheric block, and the boundary conditions have been chosen to reproduce the major tectonic episodes experienced by the Australian continent during a 200 Ma time period starting in the Ordovician (i.e. 470 Ma). The model’s focus is on the reactivation and/or reworking of zones of weakness within the continent that have either been set a priori or developed in response to previous tectonic regimes. Using the tectonic history of the Australian continent as a natural laboratory in which hypotheses on the nature and style of intracratonic deformation can be tested, the following conclusions can be made: (I) intracratonic deformation results from the concentration of strain into regions of decreased lithospheric strength; these weak zones are often caused by previous intracratonic deformation and/or develop at the interface between regions of contrasting strength; (ii) repeated deformation episodes lead to strain localisation; (iii) localised deformation may also take place as the result of the constructive interaction between two tectonic regimes originating on separate margins; and (iv) there are mechanisms that operate within the lithosphere by which deformation leads to local strengthening.

^[1] Rabaul Volcano Observatory
^[2] Department of Surveying and Land Studies, The Papua New Guinea University of Technology.
^[3] Department of Earth and Atmospheric Sciences, Purdue University, U.S.A.
^[4] Department of Physical Geography and Quaternary Geology, Stockholm University, Sweden
^[5] PRIME Lab, Purdue University, U.S.A.
^[6] formerly at Geoscience Australia