The FISSSA module (developed by the RING Team in collaboration with IECL lab) provides functionalities to load, manipulate and save fault network models in two dimensions. In this work, we sample more or less realistic fault networks to assess the uncertainty when interpreting two dimensional seismic datasets. Fault description is formulated in terms of an explicit mathematical a priori model thanks to the marked point process theory. This model is conditioned to data (the fault likelihood attribute) thanks to a Gibbs probability distribution.

The code is primarily being developed by Fabrice Taty-Moukati. It is a C++/Python library.

The proposed model to interprete faults is an adaptation of the Candy model. Another aspect is to test if the constructed model is able to generate patterns that are consistent with the input seismic data, using distance based methods, such as non parametric measures. These non parametric measures, which include the empty space, the nearest neighbor distance and the bivariate J functions are used to estimate how far the proposed model (Candy model) deviates from a purely random point process in terms of the output realizations.

The code is avalable on Github.


Related publications: