Friday, 21 September 2007

Analogue modeling: keep it simple, stupid

Geology is difficult to study. In my chosen field of basin analysis, it's very rare that you have complete information on which to base an interpretation. You can only directly observe the final configuration of the basin, and have to use inference and educated guesswork to understand how the basin evolved through time. You usually have incomplete exposure, and even where exposure is 100% it rarely gives good 3D coverage. If you're using subsurface data, you can only observe objects that are large enough to be resolved in the data (for seismic data, faults that displace strata less than 20 m are typically not resolved).

So, to look at how a whole basin evolves through time, geologists build analogue models. At the simple level, you put some sand in a box, extend the box, and look what happens to the sand. This sounds simple, but actually scaling these models correctly is quite complicated. Still, geologists have produced sandbox models that look remarkably like rift basins using these kind of methods.

A model is essentially a simplification of the system you're trying to study, in order to make it tractable. So you start off with simple models, and gradually add more and more complexity as you try to understand the complexity of the real system. In basin research, analogue modeling hit the buffers a few years ago as geologists tried to incorporate reactivated faults into their models. Generally, once a fault forms it remains a relatively weak point in the crust for a very long time. This tends to mean that when you apply tectonic forces to the crust, old faults will become active again. In analogue models, this was usually addressed through cuts in a plate at the base of the model. But, as was pointed out by C.K. Morley [1], it is the sand itself that should contain the pre-existing faults, if the model is to accurately mimic the natural system. This is hard to do with sand, because it has no tensile strength. So what next?

The answer is provided by Bellahsen and Daniel [2], and it's remarkably simple. They created 'faults' in the sand by introducing a piece of card into the sand layer, then removing it. This created a zone of disturbed sand that had a lower coefficient of internal friction than the undisturbed sand. The authors went on to create models that look remarkably like real basins that contain reactivated faults. Sometimes apparently complex problems have simple solutions.

[1] Morley, C.K. 1999. How successful are analogue models in addressing the influence of pre-existing fabrics on rift structure? Journal of Structural Geology, 21, 1267-1274.

[2] Bellahsen, N. and Daniel, J.M. 2005. Fault reactivation control on normal fault growth: an experimental study. Journal of Structural Geology, 27, 769-780

No comments: