Waveform inversion of seismic data involves optimising a function, usually a misfit, which compares observed data with synthetic seismograms generated from a chosen model withm a specified model space. Unless we begin with a starting model which is close to the real velocity structure, any linearised search will fail to find the global minimum of the misfit, as this misfit will be non-linear. As we tend to have little prior knowledge of the velocity/depth structure of the area we are interested in and typically have a large parameter space, global optimisation procedures have to be implemented to explore the entire model space. Genetic algorithms and simulated annealing are two such techniques. They are both very robust, bearing resemblance to the natural systems of biological evolution and thermodynamics. We apply these techniques to real wide-angle seismic data and compare their performance and efficiency.


Article metrics loading...

Loading full text...

Full text loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error