Introduction

Thermomechanical Analogue Modelling

In analogue models, as in numerical models, the boundary conditions and the inhomogeneities prescribed to the scenario play an important role for the outcome of the experiment. In physical models, the most signicant boundary conditions under control of the experimenter are the velocities or the stresses applied to various parts of the models by controlling motors, the overall "gravity" by performing experiments in a centrifuge, and the temperature at the model boundaries by using circulators and thermostats. The inhomogeneities used in laboratory experiments include the choice of horizontally or laterally different materials, pre-formed or even lubricated thrusts and inclusions of comparatively hard or soft materials.

To simulate the rheological stratification of the earth's crust in physical models, it is necessary to take into account the variations in mechanical properties induced by variations in temperature. Until now this has been done experimentally by using materials such as sand and silicone putty, to model brittle and ductile behaviour, respectively (e. g. Davy & Cobbold, 1991). Major progress has been made in this way in understanding crustal and lithospheric processes (e. g. Chemenda et al., 1995). However, the major drawback with such models is that any material point within the model crust retains its physical properties throughout the experiment, regardless of its position within the model. Thermal readjustment is thus not taken into account, and proper scaling with respect to gravity is therefore not achieved (see Figure 1).

Figure 1. Concept of thermomechanical analogue modelling (bottom) vs. classical analogue modelling (top). In the classical models, material points maintain their physical properties independently of the position within the model. In thermomechanical models, temperature-dependent properties combined with a temperature gradient result in variable rheologies within one material (see tip of footwall block of upper layer). (Click for enlargement)

To investigate tectonic processes in continent-continent collision zones, in particular the role of rheology in the distribution and propagation of deformation, a deformation rig was especially designed for thermomechanical modelling. In this work, analogue materials with a temperature-sensitive viscosity were used in combination with a thermal gradient in the model to simulate the change of mechanical properties with depth.

Analogue experiments to model crustal processes using temperature dependent material properties have been made before. However, none has fully exploited all the possibilities. Oldenburg & Brune (1972, 1975) focused on structures in solidifying wax, but did not use the temperature-dependency of viscosity. Shemenda & Grocholsky (1994) examined extensional regimes using solidifying hydrocarbons and also worked on physical models of continental collision (Chemenda et al., 2000). Brune & Ellis (1997) used a temperature-sensitive viscosity of wax in an extensional set-up, Rossetti et al. (2000) modelled orogenic wedges using similar principles. None of these authors mapped the temperature distribution in the material. Nataf & Richter (1982) examined convection in fluids with temperature-dependent viscosity and mapped isolines of the thermal gradient, but no isotherms, applying their results to the evolution of planets. A historical overview of analogue modelling has been compiled by Ranalli (2001).

In the experiments described by Wosnitza et al. (2001), the temperature distri-bution was mapped using an infra-red camera. The isotherms in the model could be followed through time, and their deformation was visualized. From the mechanical deformation field obtained from optical photographs, the stress field could be recovered using the known temperature-dependence of the viscosity. The combination of applying thermal boundary conditions to the model resulting in a thermal gradient and mapping the resulting temperature field throughout the deformation exploits for the first time all the possibilities of thermomechanical analogue modelling. This is a major progress in technique, since thermal readjustment in the model is not only a process used, but also documented in the experiments.

Data Analysis

Strain analysis developed for eld observations can easily be applied to analogue experiments. In the laboratory, contrary to nature, the exact geometry of the scenario is known not only at the final stage, but also before and during the deformation. Therefore, it is possible to extract more information from an analogue experiment than from analysing deformed rocks from an outcrop. However, analogue models can only offer a unique source of strain and stress data if they are properly scaled.

Mancktelow (1991) describes the analysis of a rectangular grid inscribed to the side or the top of wax models. His method requires rectangular grids, which cannot always be produced. In the experiments presented here, the blocks adjacent to the thrust are initially trapezoid. Furthermore, passive marker particles had to be added in a random distribution. These additional markers did not form a regular grid.

The algorithms developed by Bons et al. (1993) also use rectangular domains and average deformations obtained from several markers around a grid node. Therefore, this software is not applicable to relatively sparse data. The approach presented here gives results even for sparse data in arbitrarily-shaped domains.

In addition to the strain data, the temperature distribution of the experiments was used for data analysis. Such an approach has been made by Nataf & Richter (1982). Their imaging method involved the temperature-dependent refractive index of the analogue materials and mapped isolines of the temperature gradient, but not the temperature itself. As mentioned above, the thermal field can be used in combination with the deformation field to recover the stress distribution within the models.

As the load over a material point could be extracted from the above data, the lithostatic pressure at a given point can be calculated. Together with the temperature information, it was possible in this work for the first time to present pressure-temperature paths for analogue models.

Continental Collision

It is obvious, that continents can not be interpreted as rigid plates (England & Molnar, 1997). If the internal deformation of continental plates concentrates at former suture zones, the examination of processes of continental collision needs to take into account the whole lithosphere.

Building physical models with a temperature-sensitive viscosity has the advantage that the mechanical consequences of thermal readjustment during the experiment can be reproduced for lithospheric processes. This represents a major improvement in analogue models of tectonic processes. It is particularly significant for experiments investigating lithospheric stretching and the stability of mountain belts. For example, in subduction zones, the material balance is strongly controlled by the rheology of the subducted sediments (e.g. England & Holland, 1979; Mancktelow, 1995). This concept has also been applied to exhumation in collision zones (e.g. Grujic et al., 1996).

Although this work focuses on the set-up of the apparatus and the experimental procedures as well as on the data analysis, first experiments were performed to model simple orogenic scenarios. These models included the crust and the lithospheric mantle to a depth of 50 km, on a length scale of ca. 160 km. For a first set of models, the boundary of the Eastern European plate and the Kazakhstan plate at the onset of the Uralian orogenesis about 310 Ma ago (Zonenshain et al., 1994) was simplied to a single thrust. This simple megathrust geometry cutting the whole crust can be assumed for the Urals (Steer et al., 1998).

In a set of more complex experiments, the simple thrust was replaced by a wedge of weak material, as proposed for the Higher Himalayan Crystalline by Grujic et al. (1996). In these experiments, an aspect of the recent collision between India and Asia was modelled.