Peter Haffinger

Broadband Full Waveform Inversion by Shot/Receiver Re-focusing

The resolution of FWI schemes is largely limited by the huge computational resources which are involved in the process and current implementations use only low frequencies to obtain a background velocity model for migration. In this paper we present an innovative scheme to split up the large scale inversion problem into many small scale local inversions. The method is based on back-propagation of seismic sources and receivers into the subsurface which allows us to use the full seismic bandwidth to obtain inversion results with a resolution far beyond current practice. By combining the local results and performing a global wave field update nonlinear effects, e.g. internal multiple scattering, can be estimated and a high resolution nonlinear inversion result is obtained. We successfully demonstrate our scheme on an acoustic 2-D synthetic dataset while inverting for acoustic velocity only and assuming density to be known. Extension to a multi parameter inversion will be the subject of future research.

Localised Linear 2-D Inversion by Data Pre-conditioning in the Time-domain

In seismic, full-waveform inversion is a rather new approach since the scale of the problem to date, has not allowed for really satisfactory implementations yet. This is mainly due to the size of seismic data sets and the limitation in computational resources, as well as to the high ill-posedness of the problem. In this paper we propose a data pre-conditioning scheme that allows splitting up of the large scale inversion problem into many local inversions. In this way, it is expected that the computational limitation and the ill-posedness can be handled by present-day computers. By combining the local inversion results, a global property model of the subsurface is obtained.