• Author:

    Daniel Köhn, Lisa Groos, Martin Schäfer

  • IFOS is a 2D elastic full waveform inversion code.


Project Description

IFOS (Inversion of Full Observed Seismograms) is a 2D elastic full waveform inversion code. It uses the full information content of seismic data by fitting the whole seismograms in the time domain. The solution of this highly nonlinear inverse problem is done by a preconditioned conjugate gradient method where the gradients are calculated in the time domain with the numerically very efficient adjoint method. IFOS applies a multi parameter inversion of a set of three elastic parameters e.g. the P-wave velocity vp, the S-wave velocity vs and the density or the two Lame parameters lambda and mu and the density. Forward modeling is done in the time domain by a finite difference algorithm with a 2D Cartesian standard staggered grid. Viscoelastic forward modeling is implemented by using a generalized standard linear solid as rheological model, though inversion for viscoelastic parameters like Q values is not implemented. This code is efficiently parallelized by a domain decomposition using MPI. In the inversion one can use different preconditioning tapers that are applied to the gradients. Furthermore, the inverted models can be smoothed after each iteration step. Frequency filtering and time windowing are implemented to do a multi-stage inversion. Starting at low frequencies and increasing the frequency content of the data step by step helps to reduce the nonlinearity of the inverse problem. The program is able to determine an optimal source time function which eliminates the characteristic of the field-equipment from the inverse problem.