(ANU Banner)

Extract from RSES annual Report 1997


Research within the geodynamics group covers three areas and this subdivision is followed in this report. These areas are: (i) Modelling of tectonic processes and landscape evolution, (ii) Sea-level change, ice sheets and mantle viscosity, and (iii) Geodetic monitoring of movements and deformations of the Earth's crust. These three areas are joined by the common thread of understanding the processes that shape the Earth's crust and lithosphere through the linking of geological, geophysical and geodetic observations to numerical models.

One of the highlights of this year's research is the development by Dr J. Braun of numerical models for the evolution of the northern margin of the Amadeus Basin and the Arunta Block, models that illustrate well the substantial progress made in modelling of this region since the analytical models of more than a decade ago. Another contribution that addresses problems examined previously within the group is the modelling by Dr P. van der Beek of the landscape evolution of the highlands of southeastern Australia, work that has been supplemented by new fieldwork by honours student Ms A. Pulford in the Blue Mountains, using Miocene-aged basalts to estimate age constraints on the incision of rivers and to infer erosion rates.

The second area of research is concerned with the deformations of the Earth caused by changes in surface loading as ice sheets wax and wane. One of the primary motivations of this work is to estimate the mantle rheology but this work is also central to the understanding of shoreline evolution, present-day sea-level rise and the immediate past history of the ice sheets. One area where progress is being made is in the measurement of the rebound of Antarctica through the work of Dr D. Zwartz and graduate student Ms J. Quinn. The adjustment of the Earth to the changing ice loads manifests itself in other ways, including changes in the Earth's rotation and in the motion of artificial earth satellites. An analysis of these rotational and orbital effects by Dr P. Johnston has produced a coherent set of earth- and ice-model parameters that are consistent with geological inferences and which point to a secular rise in eustatic sea level for the past few decades of about 1-1.5 mm/year. This estimate does not reflect any changes caused by thermal expansion and these measurements therefore open-up the possibility of separating out thermal effects from glacier melt-water contributions.

The principal contribution to the geodetic monitoring of crustal deformation has been the work by Dr P. Tregoning on the deformation in the Papua New Guinea region with a comprehensive kinematic description of the major and minor plate motions. This work has been extended into the Solomon Islands through a joint project with the University of South Australia.

New graduate students who joined the group this year are Mr J. Tomkin and Mr D. Burbidge. Ms A. Pulford completed her honours thesis under joint supervision with the Geology Department. Dr C. Stirling successfully defended her PhD thesis early this year and took up an appointment at The University of Michigan. Mr G. Batt submitted his thesis this year and took up an appointment at Yale University.


Variations in tectonic style along the northern margin of the Amadeus Basin during the Alice Springs Orogeny

J. Braun and R. Shaw[1]

Using a numerical model of the thermo-mechanical lithosphere recently developed at RSES, we have demonstrated the important role of the thermal regime of the continental crust in determining the partitioning of lithospheric deformation between steeply-dipping crustal-scale shear zones and originally flat-lying mid-crustal decollement surfaces. The initial thermal regime also has a profound effect on the patterns of exhumation predicted at the surface of the model which, in turn influences the distribution of isotopic ages.

This work has helped to understand the variations in tectonic style observed between the central and eastern parts of the northern margin of the Amadeus Basin in terms of relatively small variations in geothermal gradient between the two areas. This region of the Australian continent was subjected to relatively large north-south compressional stresses at the time of the Alice Springs Orogeny, a major orogenic event that led to deep exhumation of the southern margin of the mid-Proterozoic Arunta Block, syn-orogenic deposition of a thick conglomerate sequence along the northern margin of the Amadeus Basin and, in it latest stages, pervasive deformation of the Proterozoic and Paleozoic sections in the basin.

Figure 1: Results of the numerical model as contour plots of accumulated strain and predicted basin geometry on which is overlaid a first-order comparison with observed structures along (a) the central and (b) eastern parts of the northern margin of the Amadeus Basin. OTZ and RTZ are Ormiston and Redbank Thrust Zones. In the first model (panel a), the lithosphere is assumed to be very cold whereas in the in the second model (panel b), it is assumed to be characterized by a more "normal" geothermal gradient which allows for the activation of intracrustal low-viscosity decollement levels.

In the central part of the orogen, the Redbank Thrust Zone (RTZ) is the locus of a very steep metamorphic gradient, which implies that maximum exhumation took place just to the north of the RTZ (Figure 1a), whereas in the eastern part of the orogen, the Entia Domal Structure has formed by exhumation of mid-crustal rocks at some distance (>20km) from the crustal-scale Illogwa shear zone (ISZ) (Figure 1b). The first-order distribution of denudation, the geometry of large-scale structures in the Arunta Block and the margin of the Amadeus Basin, as well as the ages of exhumed rocks for a range of isotopic systems (corresponding to a range of closure temperatures) are well predicted by the numerical model as are the well-documented variations between the central and eastern part of the basin margin.

A new model for asymmetric continental extension and its application to the late Paleozoic extension in the Fitzroy Trough, Canning Basin, northwestern Australia

J. Braun and R. Shaw[1]

Classical conceptual models of continental extension (and rifting) assume that, when subjected to large extensional stresses, the lithosphere may respond in one of two ways: either symmetrically (according to the so-called "McKenzie model") or asymmetrically (according to the so-called "Wernicke model"). In the asymmetrical mode, a low-angle lithospheric-scale shear zone accommodates extension leading to a large lateral offset between the loci of crustal and mantle extension (Figure 2a). The Wernicke model may be regarded as the geometric equivalent of the simple compressional plate subduction model (Figure 2b). We propose that extension of the continental lithosphere can also be accommodated by re-activation of a mantle structure that developed during continental collision along a paleo-plate margin (Figure 2c and 2d). This new model is the equivalent in extension of the mantle subduction model for continental collision.

Reactivation of the deeply seated structure leads first to the development of a symmetrical graben bounded by two steeply-dipping crustal-scale shear zones rooting at the top of the inherited mantle structure. Finite extension leads to a profoundly asymmetrical system in which one of the rift basin-bounding, crustal-scale shear zones, (on the so-called retro-side of the rift) remains active at all times of the rift development while the other margin (the pro-side) is characterized by an array of sub-parallel structures active during successive intervals in the basin formation. This simple model also results in a series of unconformities along the pro-margin of the rift basin and may also result in a peculiar Moho geometry.

We have developed this new model for continental extension to explain the apparently complex structural architecture of the Fitzroy Trough of northwestern Australia The trough underwent a phase of extension in the late-Devonian to mid-Carboniferous which led to the deposition of up to 14 km of predominantly marine sediments. Movement along the southwestern margin of the trough appears to have been continuous while the northeastern margin is characterized by a series of sub-parallel faults the southernmost of which roots into a lithospheric-scale south-dipping structure that was clearly imaged to depths of 80 km on a deep reflection seismic section acquired by the Australian Geological Survey Organisation in 1988.

Figure 2: Conceptual models of continental lithosphere deformation: (a) B. Wernicke's model of extension by simple shear along a shallow-dipping through-going shear zone; (b) A. Bally and S. Snelson's model of continental crust "subduction", the equivalent of the simple shear model in compression; (c) S. Willett and C. Beaumont's model of continental convergence accommodated by subduction in the mantle and thrusting and mountain building in the crust; (d) the new model of extension by reactivation of a pre-existing mantle shear zone. The singularity in strain rate (or discontinuity in velocity) is felt at the base of the upper layer as a result of subduction of the lower layer. Wernicke, B. 1985. Can. J. Earth. Sc., 22, 108-125. Bally, A. and Snelson, S., 1980. Can. Soc. Petrol. Geol. Memoir, 6, 9-94, Willett, S., Beaumont C. and Fullsack P., 1993. Geology, 21, 371-374.

The dynamics of a young compressional orogen: the Southern Alps of New Zealand

G. Batt, J. Braun, I. McDougall and B. Kohn[2]

A unique collection of geochronological data has been used to constrain the dynamic development of the Southern Alps in New Zealand. This mountain belt is the surface expression of the collision between the Australian and Pacific plates accommodated by oblique thrusting along the Alpine Fault which runs for several hundred kilometres along the western coast of South Island, New Zealand. The geochronological dataset gives us important information on the time at which rocks presently exposed at the surface of the orogen cooled through a given set of so-called closure temperatures ranging from 100 to 500deg.C.

This dataset has been analysed by comparison with the predictions of a thermo-mechanical model recently developed at RSES. The model clearly demonstrates that the present-day Alpine Orogeny started some 5 Ma ago, when the plate boundary between the Australian and Pacific plates changed from a predominantly strike-slip boundary to one accommodating substantial convergence. The model also shows that the rate of convergence is not uniform along the length of the Alpine Fault due to the position of the relative pole of rotation with respect to the plate boundary. This has led to more important denudation along the northern section of the Alpine Fault, south of the Marlborough Fault System in comparison to the southern section, north of Westland.

An important conclusion of this work is that a proper interpretation of geochronological data can only be achieved if one considers the thermal perturbation caused by tectonic movements inside an active orogen.

On the tectonic interpretation of geochronologically-derived cooling events

J. Braun, G. Batt and P. van der Beek

We have demonstrated from a simple one-dimensional numerical model of heat transport by advection and conduction in an actively eroding orogen that the commonly assumed correspondence between cooling rate and exhumation rate is inappropriate. We have shown that, in most circumstances, periods of accelerated exhumation (ie. at the onset of a tectonic event) have little (if any) effect on cooling patterns derived from geochronological data whereas periods of rapid decrease in exhumation rate are usually associated with a marked increase in cooling.

This is demonstrated in the following diagram in which the results of the numerical model are shown in a situation where the assumed denudation rate is increased abruptly by a factor of 10. The results show that the critical time at which the exhuming rock undergoes rapid cooling corresponds to the time at which it reaches the surface thermal boundary layer, not the time at which the rapid change in tectonic, hence denudation, rate takes place.

The reason for this behaviour is the inefficiency of conductive cooling at dissipating the heat carried by rapidly exhuming rocks.

These simple theoretical considerations have profound consequences on the interpretation of geochronologically-derived cooling curves in regions of active or paleo-tectonics.

Figure 3: Results of a one-dimensional numerical model; the three thin curves represent the thermal history of a rock being exhumed towards the surface following an abrupt change in exhumation rate at three different times; the thick lines are the corresponding cooling rate curves. The black dots indicate when the change in exhumation rate is imposed in the three different calculations.

Post-Cretaceous landscape evolution in the Southeastern Highlands of Australia: Inferences from numerical modelling

P.A. van der Beek and J. Braun

The tectonic and geomorphic evolution of the southeastern Australian highlands have been a controversial issue between geophysicists, geologists and geomorphologists for several decades. The evolution of the highlands may be linked to three distinct tectonic causes: Palaeozoic orogeny, Late Cretaceous opening of the Tasman Sea, or Tertiary volcanism, possibly associated with lithospheric heating. The relative importance of these causes has been vehemently disputed, with parallel controversies concerning the rates of landscape evolution and the long-term stability of the present landforms.

Data pertaining to the landscape evolution and denudational history of the highlands abound: Southeastern Australia is one of the regions with the highest density of fission track thermochronological data in the world, and has also been the focus of intense geomorphological research for nearly a century. The sedimentary record from on- and off-shore basins has also been used to reconstruct the erosional history of the region. However, the interpretation of the different datasets is ambiguous and the inferences are not easily reconciled with each other.

The southeastern highlands are characterised by a prominent (600 to 1000 m high) escarpment, which is located between 15 and 100 km from the present-day coast line. The continental drainage divide does not coincide with the escarpment but is located 50-100 km inland from it. Geomorphological data suggest that denudation of the highlands has taken place mainly by headward erosion of deep river gorges, with the interfluves suffering negligible downwearing. River profiles in the highlands show conspicuous knickpoints which retreat at estimated rates of ~2 km/My. Drainage patterns seaward of the continental divide are conspicuously barbed, but the available geomorphological data favour long-term drainage stability rather than widespread capture.

We employ an updated version of the CASCADE surface processes model (see page 43 of 1995 annual report) in order to quantitatively assess the controls on the landscape evolution and denudation history of SE Australia. The model adopts three main types of processes: `short-range' hill-slope processes, landsliding, and `long-range' fluvial processes. Slow, continuous hill-slope processes are modelled by a linear diffusion law; landsliding is included by imposing a slope threshold; fluvial transport is controlled by the carrying capacity of rivers and an erosion length scale, which is a measure of the `erodibility' of the substratum and is included to model supply-limited behaviour. We aim to assess (1) what processes are required to reconcile the apparently contradictory data, and (2) what are the tectonic and climatic controls that have led to the rather unique morpho-tectonic setting that is observed in the highlands today. As fluvial processes appear to be by far the most important factor in denuding the highlands, we first constrain the fluvial erosion parameters for our model by fitting long river profiles. The morphology of river gorges indicates that rock fall and landsliding are the most important hill-slope processes. We therefore model hill-slope processes by a very low diffusion coefficient and a finite threshold slope. We first model situations with constant tectonic forcing and unique bedrock lithology in order to obtain the first-order controls on landscape evolution. Finally, we proceed to more general situations in order to study lateral variations in tectonic and lithological controls throughout the highlands, which may explain some of the observed lateral morphological variation.

Our modelling results indicate that (1) the observed highlands morphology requires that the drainage divide was established at its present location prior to opening of the Tasman Sea; whether the divide is inherited from Palaeozoic orogeny or resulted from Mesozoic syn-rift uplift cannot be established, however; (2) the apparent stability of the system indicates a subtle balance between the uplift rates of the divide on one hand and the escarpment on the other; (3) the observed barbed river drainage may have been imprinted on the system from the start; it does not necessarily evolve through capture.

Fluvial incision rates in the Finisterre Range, Papua New Guinea

P.A. van der Beek, J. Stone, R.K. Fifield [7]and R.G. Creswell[7]

The Finisterre Range in northeastern Papua New Guinea has been rapidly uplifted since 3-3.5 Ma as a result of continuing collision between the Australian continental margin and the Bismarck Arc terrane. The range forms a prime example of a young collisional orogen that has not yet reached steady-state. Post-Pliocene uplift of the range has been quantified previously by dating uplifted coral terraces to the north of the range and the limestone plateau which caps most of its crest. Uplift of the range has resulted in active incision of rivers draining the southern slope of the ranges, creating deep river gorges with spectacular strath- and fill-terrace remnants along their walls. Dating of the terraces has previously proved extremely difficult because they do not contain any material suitable for 14C analysis. We have attempted to quantify river incision rates by dating the terraces with cosmogenic isotopes (see page 108-109 of 1996 Annual Report). We sampled boulders from the different terraces, carefully selecting sites where boulders appeared in-situ (fluvial bedding visible) and post-depositional erosion of the terrace tops appeared minimal.

In an initial attempt, six basalt and limestone samples were prepared for 36Cl-dating using AMS. 36Cl concentrations in these rocks were remarkably low, suggesting either an extremely young age (< 2 ka) for the terraces, exhumation of the boulders or loss of 36Cl from the samples by weathering. Interpretation of the 36Cl concentrations in terms of terrace age implies fluvial incision rates that are higher than in any other part of the world, and an order of magnitude higher than long-term uplift rates of the range. We are currently preparing additional samples for 10Be-dating, in order to obtain an independent check on these results.

Neogene landscape evolution in the Blue Mountains, NSW

A. Pulford, P. van der Beek and R.A. Eggleton[4]

The Blue Mountains have not previously been studied in detail from a landscape evolution perspective. The Blue Mountains are unique in that they form a plateau, part of the Great Dividing Range, which has been deeply dissected into deep river gorges and relatively uneroded highlands. The landscape evolution of this area was investigated by mapping the Miocene palaeotopography preserved beneath the basalt-capped peaks in the western Blue Mountains and calculating erosion rates for the region since the eruption of these basalts.

Detailed mapping of the basal contact between Miocene aged basalts and the underlying Triassic Hawkesbury sandstone provides an image of the landscape present in Miocene times. Comparison between the present and Miocene topography indicates large differences between landscape forming processes at the two times, with greater incision of valleys seen at present.

K/Ar dating of the basalt has provided a maximum age for incision of present day river gorges in the region. Geochemistry and petrography of the basalts indicate at least seven different flows were present in the region with ages which lie in four clusters of ca. 20 Ma, 17 Ma, 15 Ma and 14 Ma. River incision, which dissects the basalt outcrops, has occurred since the youngest basalts were erupted as there is no evidence of major erosion during eruption of the basalts.

The resulting erosion rates of ca. 1000 m3/a from the major valleys indicates large amounts of erosion focused in the valleys with little erosion seen elsewhere. This information combined with images of the pre-Miocene landscape provide further information on the possible tectonic history and landscape evolution of the region and highlights differences between the Blue Mountains and other areas, in a similar position relative to the eastern coast of Australia.

Modelling the re-organization of drainage networks due to tectonics

J.H. Tomkin and J. Braun

The central Otago region of New Zealand is an area of active continental shortening. Recent folding of this Pliocene peneplain has led to the formation of a series of linear ridges. We have performed numerical and analytical modelling of the complex interactions between vertical tectonic movements and drainage pattern re-organisation in the area, particularly in the vicinity of the Rough Ridge-South Rough Ridge system. The unusual drainage network pattern (particularly the asymmetric water catchments and the creation of "windgaps" or dry valleys) is the result of South Rough Ridge actively propagating past Rough Ridge.

The landscape model successfully recreates the general drainage network features, and supports the analytical results, which suggest that the asymmetric catchments and the presence of regularly spaced windgaps at the top of South Rough Ridge are a response to the rapidly emerging South Rough Ridge along the flank of the pre-existing Rough Ridge. A dimensionless number was found that relates the rate of tectonic uplift with that of erosion, so that if an absolute rate could be found for one of the two processes (such as the age of a windgap), the other (such as the rate of ridge emergence) could be determined. It is also clear that absolute rates cannot be derived from the present-day geometry of the system, only ratios of rates (or relative rates). Another interesting result from the model is that the periodic nature of the river system is achieved using monotonic linear uplift; cyclic driving forces, such as a variable climate, are not required as suggested by other authors.

Figure 4: Model results after 2.5 million years of emergence of South Rough Ridge. Rivers are shown as thick black lines; catchment boundaries are shown as thick dashed lines; circles indicate the position of windgaps.


Modelling of glacial rebound phenomena

P. Johnston and K. Lambeck

The Holocene and present sea-level record shows that the Earth continues to change shape in response to the melting of the Late Pleistocene ice sheets. Ongoing deformation of the Earth changes the mass distribution within the Earth and consequently the Earth's gravity field and moment of inertia. The Earth's hydrological system (oceans, lakes, ice sheets and ground-water) is another source of large scale mass redistribution at the Earth's surface, and therefore also affects the gravitational field of the Earth.

Satellites orbiting the Earth are subject to its gravity field and by observing their motion, we are able to measure the height of the equipotential surface (geoid) relative to a reference ellipsoid and the rate of change of the geoid. Satellite techniques can also be used to measure changes in the position of the Earth's rotation axis, although longer time series are available through astrometric observations. The contribution to the rate of change of the geoid and polar position due to melting of the Late Pleistocene ice sheets is well constrained by the knowledge of the position and size of these ice caps but Earth rheology is the main unknown. Observations of these time-dependent changes in the Earth's gravity field and rotation therefore provide some constraint on the rheology of the mantle. The effect of present melting of the Antarctic and Greenland ice sheets and mountain glaciers is also calculated. Comparison of the predictions with the observations constrains the present rate of sea-level rise to about 1.0 mm/yr and an average lower mantle viscosity of approximately 1022 Pa s, consistent with analyses from sea-level observations. This mechanism fully explains the polar wander signal without the need to invoke other processes such as subduction to explain the present observed polar wander velocity and direction.

Contribution of tectonics, isostasy and natural compaction to vertical land movement in the Netherlands

H. Kooi[5] P. Johnston, K. Lambeck and C. Smither

In recent years, the Survey Department of the (Netherlands) Ministry of Transport, Public Works and Water Management (Rijkswaterstaat) has documented significant rates of vertical ground movements in the Netherlands from an analysis of their long-term record of geodetic levelling data. Rates of vertical movement vary between +10 cm/100y (uplift) in the southeast and -10 cm/100y (subsidence) in the northwest. The spatial pattern reveals an overall tilting of the country towards the NW with superimposed shorter wavelength differential movements which coincide with well known tectonic structures.

Analysis of geological data and modelling of geological process rates in conjunction with the analysis of geodetic levelling data by the Survey Department of Rijkswaterstaat demonstrate that:

(1) The differential movements inferred from levelling can be explained by the combined effects of isostasy, tectonics and compaction. The first two processes account for approximately 35% and 65% of the observed long-wavelength tilting, respectively. In the coastal provinces, compaction and/or tectonics are responsible for < 100 km scale differential movements. In the remaining part of the country such differential movements are largely caused by tectonic deformation.

(2) Average movement of the geodetic benchmarks with respect to the geoid is in between -8 and +5 cm/100y and a significant eustatic sea-level rise >10 cm/100y is required to account for the relative sea-level rise indicated by tide-gauge records.

(3) The isostatic component of vertical land movement will be very stable for the next centuries and gradually decay over the next millennia. Compaction contributions are uncertain, but if present, will decay by probably not more than 30% per century. Tectonic contributions to vertical land movement are highly unpredictable. In all, the coastal zone of the Netherlands will probably continue to subside at approximately the current rate for at least several centuries. However appreciable modification of current subsidence by tectonic change cannot be ruled out.

Benchmark comparison of glacial isostatic adjustment

G. Kaufmann and P. Johnston

The Earth's response to the growth and decay of the large Late Pleistocene ice sheets over northern North America, northern Europe, Greenland, and Antarctica is a useful tool for inferring the rheological structure of the mantle. Here, modelling the glacial isostatic adjustment process as the time-dependent response of a viscoelastic continuum to an applied surface load, which describes the redistribution of meltwater from the ice sheets to the oceans, can be used to infer both the ice sheet history and the structure of the Earth's mantle.

This computational task is based on a theoretical framework and, depending on the assumptions made to solve the problem, requires a stable numerical scheme to solve for the unknowns. Several numerical codes are used to perform these calculations, which encompass a variety of different implementations. Though published results usually agree quite well, a benchmark to qualitatively verify a code and to provide some simple case examples to compare with is still missing. Therefore, Drs G. Kaufmann and P. Johnston have initiated a benchmark comparison by defining some problems typical for modelling glacial isostatic adjustment. The results provided by the participants in this benchmark experiment can then be used as a future guide to verify new numerical codes.

Several contributions have already been received and processed, all of which show good agreement in the results. A preliminary analysis has been performed by the authors and is published on the World Wide Web (http://rses.anu.edu.au/geodynamics/GIA_benchmark) to encourage more researchers to contribute to the benchmark.

Upper mantle lateral viscosity variations and postglacial rebound

G. Kaufmann and P. Wu[6]

This project aims to provide deeper understanding of the physical processes of glacial rebound as affected by lateral variations of physical properties in the upper mantle. Focusing our interest on lateral viscosity variations in the asthenosphere, we have shown that a variety of geodetic signatures related to the glacial isostatic adjustment process are capable of resolving a proposed lateral structure within the observational uncertainties. In particular, relative sea-level changes, present-day velocities and gravity anomalies from the marginal areas of the formerly glaciated regions have the potential to distinguish between different rheologies.

These general results obtained by two-dimensional finite-element modeling have been extended to a three-dimensional case study of the glacial isostatic adjustment of the Barents Sea. The Barents Sea is a shallow ocean basin in northwestern Europe including the arctic islands of the Svalbard Archipelago, Franz Joseph Land and Novaya Zemlya. The area was covered by an essentially marine ice sheet during the Late Pleistocene, which reached glacial maximum some 20 ka ago. The northern and western ice margins were close to the continental margin during that time. The ocean-continent transition from the Greenland deep-ocean basin to the shallow Barents Sea continental shelf occurs in this region, therefore we can expect lateral variations in rheological properties across the ice margins at the Last Glacial Maximum.

A comparison between a laterally homogeneous reference earth model and a laterally heterogeneous earth model has been performed. The results indicate that a change in asthenospheric viscosity of about four orders of magnitude significantly influences predictions of land uplift, present-day velocities, and present-day gravity anomalies in the northwestern part of the Barents Sea region.

The crustal rebound and allied phenomena of Scandinavia

K. Lambeck and C. Smither

Scandinavia remains one of the classic sites for investigating the Earth's response to surface loading because of the large amount of available field evidence for both the ice history across the region and for the changing relationship between land and sea levels. A comprehensive analysis of the geological evidence has been completed (see the 1996 report for a preliminary account) and the more recent emphasis has been on a number of related problems, including the analysis of the geodetic data, the history of the Baltic Sea from the time of the Baltic Ice Lake in Lateglacial time to the Litorina stage when the marine incursion occurred for the last time, and high-resolution modelling of critical regions within Fennoscandia including southern Finland, central-southern Sweden and the Danish Straits.

The tide-gauge data for the Baltic and evidence for tilting of some of the large lakes has provided a particularly good test of the conclusions drawn from the geological data: of a depth dependent viscosity and of an asymmetric ice sheet in which the ice in the east and south was considerably less thick than in western Scandinavia. The tide-gauge data, when corrected for the isostatic contributions, confirm that eustatic sea-level change for the past century has been about 1.0-1.2 mm/year.

Defining the eustatic sea-level curve during Late Pleistocene and Holocene time

K. Fleming and K. Lambeck

Sea-level varies over time for a number of reasons; changes in ocean volume, the shape of the ocean basin, the distribution of water within the ocean basins and the movement of coastal areas relative to sea-level, all of which result from a number of causes, ranging from tectonic activity to the growth and decay of ice sheets. However, the different processes tend to have a noticeable effect over different time scales, for example tectonic activity will alter ocean basin shape over millions of years, while glacial cycles show their influence over periods of 104 to 105 years.

The objective of this work is to estimate the volumes of the continental ice sheets from the time of the Last Glacial Maximum to the present. This is achieved by estimating sea-level changes at sites far from former ice sheets and where the first-order change is the eustatic sea level. The isostatic contributions resulting from the adjustment of the Earth to the changing ice-water load distribution are relatively small at these sites and the numerical rebound models are used to correct the observations for these effects. The so-corrected sea-level curve, the eustatic sea-level function - gives a representation of past changes in the volumes of the ice sheets.

Preliminary results are: i) Late Quaternary ice sheets continued melting after 6,000 years BP, contributing an equivalent of 3 to 4 m of water since that time. This is the result of the analysis of sea-level observations from North Queensland, the Senegal River of West Africa, the Malacca Straits of Southeast Asia and the Caribbean. ii) The rise in eustatic level during Lateglacial time is well constrained by the observational data and consistent with the time history of the global ice models. Data from different localities do not reveal a coherent pattern of variable rates of melting and melt-water pulses cannot be identified. iii) At the Last Glacial Maximum the continent-based ice sheets were sufficiently large to cause, upon melting, a rise in eustatic sea level of about 125-130 m. The observational evidence for this remains however, limited and new field data is required.

Reconstructing Antarctic ice sheet history

D. Zwartz and K. Lambeck

Work has continued on obtaining a reconstruction of Antarctic ice sheet history during the latest Pleistocene and Holocene. The Antarctic is thought to have accounted for up to 35 m of global eustatic sea-level rise since the Last Glacial Maximum, yet there is considerable debate concerning whether this much ice could have been accommodated in the Antarctic ice sheets, and, if it could, where it would have been located.

Reconstructions of the northern hemisphere ice sheets have been based on numerous and spatially dense observations of glacial deposits and sea-level change. Similar observations in Antarctica are sparse and restricted to a few coastal regions where rock is exposed, so ice sheet history cannot be deduced in the same detail as is possible for the Laurentide and Fennoscandian ice sheets. Therefore, regional ice sheet reconstructions have been made, and these have been compiled and extrapolated to create a reconstruction for the whole East Antarctic ice sheet.

Sea-level records have been collected and compiled from 4 sites on the East Antarctic coast: the Vestfold Hills (78deg.E), the Bunger Hills (101deg.E), Windmill Islands (110deg.E), and Skarvsnes (40deg.E). At all sites sea-level fell steadily from 6 ka BP until the present, although the rate of sea-level fall varied considerably between sites, from 1.5 mm/yr in the Vestfold Hills to 4.5 mm/yr at the Windmill Islands. At the Vestfold Hills, the high-precision record derived from lake isolation data also shows a sea-level rise before 6 ka.

Holocene sea-level change on the Antarctic coast is a combination of eustatic sea-level change and isostatic rebound of the crust in response to recent deglaciation. The eustatic contribution to sea-level change at the four sites is the same, so the differences between them must be due to the differing amount of rebound. This can be used to reconstruct the amount of ice which has been removed from each of the region since the Last Glacial Maximum.

Numerical models of the Earth's isostatic response to changing ice sheets were used to calculate the predicted sea-level change for a range of simple ice models. Simplified regional models were used to reconstruct the ice sheet history at each site. The models had the following attributes:

1) The ice sheet is circular, with a parabolic cross-section. The ice sheet height (3700 m) and radius (1100 km) were chosen to approximate the margin of the Antarctic ice sheet.

2) Deglaciation occurred by margin retreat, with height of the ice sheet held constant.

3) The rate of deglaciation was mainly proportional to the global eustatic sea-level curve.

4) A range of Earth rheological models was investigated, and the results presented here were obtained using the following parameters: elastic crust thickness = 100 km, visco-elastic upper mantle viscosity = 21020 Pa s, lower mantle viscosity = 11022 Pa s.

By comparing the predicted sea-level curves with the observations, the amount of ice sheet thinning and margin retreat were estimated for each site (Table 1).

Site                  Thinning (m)  Retreat (km)   
Vestfold Hills            718       45             
Bunger Hills              809       56             
Windmill Islands          1162      122            
Skarvsnes                 1014      91             

Table 1: Estimates of ice sheet thinning and retreat at four coastal sites in East Antarctica.

These regional estimates and the corresponding history of deglaciation were then extrapolated to create an ice sheet reconstruction for the whole of East Antarctica, from 0deg. to 150deg. East (Figure 5). This reconstruction is similar to published reconstructions derived from glaciological numerical models, although significant differences are evident in some regions. Comparison of the sea-level changes predicted by each reconstruction with the sea-level observations at each site is a useful indicator of regions where improvement is necessary.

Figure 5: Change in ice sheet thickness since the Last Glacial Maximum, according to the new reconstruction. The largest change occurs in the coastal regions, and no change has taken place in the large blank region in the centre of the ice sheet.

Late Pleistocene glaciation of the Tibetan Plateau

G. Kaufmann and K. Lambeck

The Tibetan Plateau, covering an area of about 2106 km2 and having an average elevation of 5 km, is the result of the continent-continent collision between India and Asia, which has started roughly 20 myr ago. The present-day uplift of the plateau of a few mm/yr has been attributed to a variety of forces, ranging from tectonic mechanisms as a result of the ongoing continent-continent collision to isostatic rebound in response to the enhanced erosion rates during the Pleistocene ice ages along with their climatic changes, e.g. the weakening of the monsoon.

It has been argued that the melting of a large Late Pleistocene ice sheet covering the Tibetan Plateau with its ongoing glacial isostatic adjustment can contribute to the present-day uplift rate. However, the Late Pleistocene glaciation of the Tibetan Plateau is a very controversial issue in the recent literature and the large uncertainties in our knowledge of the extent and the timing of the last glaciation of the Tibetan Plateau limit a proper determination of this glacial contribution.

We have nevertheless investigated the effects of a glaciation cycle on present-day uplift rates and gravity anomalies and the possible contribution to global sea-level change to obtain order of magnitude estimates of the rebound. To account for the large uncertainties in the proposed Late Pleistocene ice models, we have considered both minimal and maximal ice models. In Figure 6, the extent of the maximal ice model (max) and the minimal ice model (min) is shown for the Last Glacial Maximum. The former model embraces the entire Tibetan Plateau, having an ice thickness of 1000 m, the latter model is restricted to the mountain ranges and reaches a thickness of only about 250 m. Another maximal ice model is max-2, whose deglaciation history is derived from the rise of the equilibrium line altitude, is also illustrated in Figure 6. Here, the ice thickness is spatially variable, with a maximum thickness of 3000 m over the northeastern part of the plateau.

Figure 6: Left: Areal extent of ice models max (a) and min (b) at the Last Glacial Maximum (LGM). Right: Space-time history for two selected epochs of the ice model max-2.

In general, the maximal models indicate that the disintegration of an ice sheet can contribute up to 7 mm/yr of present-day vertical uplift and 2 mm/yr of horizontal extension. The peak free-air gravity anomaly arising from the deglaciation would be -5.4 mGal, and is only a secondary effect when compared to the large positive values observed. In contrast, the smaller ice sheet models do not contribute significantly to the present-day observables.

A comparison of present-day radial velocities along two profiles (Figure 7) reveals that both maximalistic models experience maximum uplift rates over the eastern part of the W-E profile. Of more interest is the greater spatial variability in uplift rates along the W-E profile for ice model max-2, for which the uplift rates are concentrated over eastern Tibet as a result of the large ice cover at this location. The uplift rate resulting from ice model max is much smoother across the profile, resulting in uplift rates of over 2 mm/yr across the entire Tibetan Plateau. Finally, we observe uplift rates of less than 0.6 mm/yr for ice model min, which are negligible relative to the tectonic contribution.

The predictions for the S-N profile essentially confirm the results obtained along the W-E profile. We observe a large updoming of the Tibetan Plateau for ice model max, and a shift of the area around the maximum uplift rates towards the northern end of the profile for ice model max-2. Again, ice model min has an insignificantly small contribution to the present uplift rate.

The meltwater from the Tibetan Plateau can contribute a significant amount to the total eustatic sea-level rise (ESL) resulting from the melting of the large Pleistocene ice sheets. For a complete ice cover at the Last Glacial Maximum, the contribution to eustatic sea-level rise can be more than 5 m for ice models max and max-2, while the contribution arising from the melting of the smaller ice model min is negligible (< 1.5 m) compared to the total ESL rise of about 130 m. While the simple linear deglaciation of ice models max and min contributes to the ESL until Late Pleistocene times, the meltwater peak released by ice model max-2 is sharper and no significant meltwater runs into Bengal Bay after about 12,000 yr BP.

Figure 7: Profiles of present radial velocities across the Tibetan Plateau for ice models max (solid), min (dotted), and max-2 (dashed).

Radiocarbon time scale calibration

Y. Yokoyama, T. Esat and K. Lambeck, K. Fifield [7], R. Cresswell[7], K. Liu[7], and M. di Tada[7]

It is well known that 14C age differs from calendar age. Continuous sequences of radiocarbon ages up to about 8000 years have been calibrated by dating tree rings. At 10,000 calendar years the measured radiocarbon ages are about 9000 years; the difference is mainly caused by an increase in atmospheric 14C production over time. The calibration range has further been extended by comparing ages of 14C dated Barbados corals with thermal-ion mass spectrometer U-series ages up to about 30,000 years. The divergence in ages appears to continue to 30,000 years and is about 3800 years at 20,000 years. However, there are no reliable data beyond 15,000 years and particularly around 30,000 to 40,000 years ago. Tree ring calibrations, where the outer actively growing ring is presumed to be in equilibrium with atmospheric carbon, is not directly equivalent to radiocarbon dating of corals in the marine environment. Apart from the radiocarbon offset between the two reservoirs, perturbations in palaeo-oceanic circulation due to climate change are likely to alter the ocean carbon budget. Atmospheric 14C production is determined by changes in the solar wind and the intensity of the geomagnetic field. Extending the 14C calibration beyond 30,000 years will determine whether the increase in 14C production and divergence in ages continues further in time and may also reveal climate-related changes in ocean circulation.

Figure 8: Background 14C ages, obtained using the new CO2 extraction, graphite conversion line using samples of `old' marble and graphite, both known to be older than
50 ky.

Figure 9: Step-wise dissolution of a Last Interglacial age (125 ka) coral sample. Beyond the first 20% leach, the next 5 aliquots indicate a uniform age of 50 ky; in agreement with the old marble and graphite tests.

Corals which nominally span the time interval 30-70 ky were collected during an international expedition to Huon Peninsula, Papua New Guinea, in 1992. Huon Peninsula is essentially the only place where corals of this age are exposed above the present sea level. For old samples (>20 ky), contamination by modern carbon can be significant. For a 35 ky coral, 5% modern carbon contamination is equivalent to a 6000 year bias in age. One of the significant issues in 14C measurement in corals relates to successful removal of contaminated surface layers of coral samples prior to recovery of CO2. We have built an all metal CO2 extraction and carbon combustion vacuum line that enables stepwise dissolution of coral samples. Base pressures as low as 10-5 torr vacuum can be obtained for clean, high yield, CO2 -graphite conversion. The veracity of each date is confirmed by stepwise leaching of the sample and 14C measurement to confirm a plateau in age. Up to 6 leaching steps can be accommodated. Tests indicate that the total procedural blanks from a 125 ky interglacial coral, old marble and graphite are better than an equivalent of 50 ky in age (Figures 8 and 9). The 14UD accelerator- based mass spectrometer facility at Nuclear Physics, RSPhysSE, is being used to determine the 14C ages which are then compared to U-series ages to provide a calibration for the 14C time scale.


Papua New Guinea

P. Tregoning, K. Lambeck and H. McQueen

The thrust of activity in 1997 has been the analysis of all Global Positioning System (GPS) data observed in Papua New Guinea (PNG) between 1990 and 1996. Site coordinates and velocities were estimated using the GAMIT/GLOBK software and the results have led to new understandings of the present-day plate kinematics in this complex region (Figure 10). The results are the culmination of many years of cooperative research between RSES, the PNG National Mapping Bureau, The University of Technology, Lae (Unitech), The University of New South Wales, the Rabaul Volcanological Observatory, The University of Canberra and the Australian Surveying and Land Information Group.

Figure 10: GPS site velocities and major plate boundaries in Papua New Guinea. Sites observed in the Solomon Islands in 1997 are also shown.

The convergence of the Australian and Pacific plates is accommodated by the South Bismarck, Woodlark and Solomon Sea microplates with active spreading occurring on the Woodlark Basin spreading centre, subduction on the New Britain Trench, spreading across the Bismark Sea Seismic Lineation, and evidence of compression across the Ramu-Markham Fault. The northern tip of New Ireland and Manus Island are both possibly moving slowly northwards relative to the Pacific Plate, suggesting the possible existence of a North Bismarck Plate, but this motion is at the level of uncertainty of our results.

Following on from these results, in conjunction with Unitech we have commenced a monitoring project of a 30 km baseline which crosses the Ramu-Markham Fault near Lae. The expected convergence is about 30 mm/yr which should be detectable in one year. We have also established a new GPS site located close to the pole of rotation of the South Bismarck Plate in conjunction with Unitech and the University of California, Santa Cruz, and are planning future fieldwork for areas now identified as plate boundaries undergoing deformation.

Solomon Islands

P. Tregoning, H. McQueen, D. Zwartz and K. Lambeck

In cooperation with the School of Geoinformatics and Building, University of South Australia (UNISA) a four day GPS survey was conducted in the Solomon Islands this year. UNISA reoccupied three sites previously occupied in 1995 and installed one new site whilst ANU installed two new sites (Figure 10, marked Sol-97). Analysis of the 1995 and 1997 data is currently underway and will lead to velocity estimates for two sites on the Pacific Plate and one on the Australian Plate. This will further enhance the emerging picture of plate kinematics in the western Pacific region. A reobservation of all sites, planned in two to three years, will provide velocity estimates for the six sites and begin to give us a picture of the internal deformation within this complex double-chain island region.

Antarctica - Isostatic Rebound

P. Tregoning, D. Zwartz, H. McQueen and K. Lambeck

One interesting application of GPS for measuring the tectonics of the earth is the detection of post-glacial rebound. In the summer of 1997/98 a GPS receiver will be installed at Beaver Lake, Antarctica, as a pilot project of a long term monitoring program to detect the isostatic rebound of the continent as a result of the melting of the Lambert Glacier. This project is in conjunction with the Australian Antarctic Division and the Australian Surveying and Land Information Group.

The equipment consists of a GPS receiver, laptop computer and two gelcell batteries housed in an insulated pelican suitcase. The equipment will be powered by four 40 W solar panels, which should provide power until early April 1998, but the system will not be retrieved until the beginning of the 1998/99 Antarctic season. This will provide an initial height estimate of the site based on three months of GPS observations. We will monitor the internal temperature of the case and the power created and consumed so that future systems can be improved.

Estimation of atmospheric precipitable water

P. Tregoning, R. Boers [8] D. O'Brien [8] and M. Hendy[9]

Absolute estimates of the precipitable water (PW) in the troposphere can be achieved using GPS observations and local surface pressure and temperature measurements. The propagation of the GPS signals transmitted from the orbiting satellites are delayed by about 0.01 us as they pass through the troposphere, a delay which leads to height errors of up to 2 m if not accounted for. This delay can be estimated and converted into accurate estimates of the amount of PW present in the troposphere.

We have compared GPS estimates of PW with estimates from a microwave radiometer and from radiosonde launches at Cape Grim, Tasmania during the Aerosol Characterisation Experiment in November/December 1995. Results show that GPS estimates of PW agree with the other two systems to within about 2 mm, with no significant bias present. Figure 11 shows the PW estimates of the three systems for a 34 hour time period. There seems to be a systematic offset between the GPS and microwave radiometer estimates for PW values greater than 20 mm. This is not currently understood and work is continuing in order to explain this unexpected result.

Figure 11: Comparison of GPS (with error bars), MWR (solid line) and radiosonde data (triangles) estimates of precipitable water at Cape Grim, Tasmania for days of year 331 and 332, 1995.

The advantages of using GPS for measuring PW is that GPS systems are low cost, automatic systems which do not require calibration. If the communications with the equipment can provide the GPS and surface meteorological data in realtime then PW estimates can also be made in realtime and can be implemented in weather forecasting models.


H. McQueen and K. Lambeck

A Superconducting Gravimeter capable of detecting variations in the strength of earth's surface gravity to one part in 1012 was installed in a basement laboratory at Mt Stromlo Observatory in January. Operating at liquid helium temperatures, the instrument is the most sensitive gravimeter ever operated in Australia. It is built around a superconducting niobium sphere levitated in a magnetic field in an evacuated chamber. Gravity fluctuations down to a few nanogals can be determined from changes in the coil current required to precisely centre the sphere in its chamber.

The installation is a collaboration between the Geodynamics group in RSES and the Japanese National Astronomical Observatory, Mizusawa, which is operating several superconducting gravimeters as part of the Japanese Ocean Hemisphere Project. The site is one of 15 currently reporting to the data centre of the Global Geodynamics Project, a world-wide array making precise observations of faint signals from the interior of the earth in an attempt to detect motions in the deep interior, infer details of earth's internal structure, and provide information on a range of problems in global geodynamics.

Data collected in the first six months operation indicate that the Mt Stromlo site has low noise characteristics relative to others around the world and a relatively low incidence of instrumental offsets. This data is currently being used in studies of the excitation of free oscillations in the earth and analysis of the response of the earth to tidal forcing in the Australian region. Repeated absolute gravity determinations on nearby benchmarks will be necessary for full calibration of the superconducting instrument and to ensure stability for long term observations.

Figure 12: Compact Superconducting Gravimeter CT031 at the Mt Stromlo gravity station.