FORWARD NUMERICAL MODELLING

As we know, geological systems are dynamic structure in which several elements and processes take place and interact at the same time. Understanding the dynamics of these systems is important in geological research as well as in industry to better predict their distribution and geometry.

The traditional approach to studying geology includes field work, data obtained from boreholes, geophysics, etc. However, the geological record is incomplete, partial or, at least, limited and alternative methods are necessary to complement and to quantify the processes that are not observable using the abovementioned techniques. One alternative tool that can be useful in completing the approximation is geological modelling.

Now we could fall into a discussion spiral about what exactly is or is not a model and their types and classification. But from my point of view, all models represent an advance in the geosystem knowledge and have helped and continue to help to understand the behaviour of the earth system.

One kind of geological modelling is the forward numerical modelling, used for example for the study of sedimentary basins. It allows us to experiment directly by playing with different parameters and interactions to reproduce the temporal and spatial evolution of a basin. There are a lot of them developed by several universities, research groups and companies.

I was in a team developing a forward numerical model during my thesis. This code is a 2.5D forward process-based numerical model to simulate a marine basin at a geological time scale. It models the fluid flow, the siliciclastic transport and sedimentation, and the carbonate production using an ecological model.

I am still working on it in my spare time developing a fork of this model. The first step has been translating and updating a relatively old fortran code to python. You can find a working version of this code in Github. But, you can run models and obtain results like these (the pictures and videos obtained from the VTK files have been made with ParaView):

Why python? because it is a multiplatform scripping language. It is easy to learn. And there are a lot of powerful libraries, including VTK for a 3D visualization, or Scipy to solve ODE's. One of the main problems is the computation time: a simple model running on pure python needs minutes while the same model needs seconds in fortran.

The next step is being convert the procedural code to Object Oriented and add new functionalities.