Full text loading...
To overcome the limitations of conventional Full Waveform Inversion methods for large-scale velocity updates from reflection data, the Reflection Full Waveform Inversion (RFWI) approach has been proposed. RFWI suffers from a slow convergence and potential artifacts related to inconsistencies when alternating background velocity model and reflectivity model updates. Furthermore, the estimation of the reflectivity based on a Reverse Time Migration (RTM) represents a substantial numerical cost in the RFWI workflow. To tackle these difficulties, the Time consistent Waveform Inversion (TWIN) method has been proposed. This method better accounts for the reflectivity-background velocity coupling effect and uses a numerically cheap shot-independent zero-offset migration operator. After summarizing the concept of TWIN, we present the first published real data application of the TWIN method (marine streamer dataset). Our inversion methodology is mainly data-driven and considers a very erroneous initial background velocity model. After 17 iterations, TWIN performs large velocity model updates up to 7 km depth. The inverted velocity model allows for a significant flattening of the RTM-offset gathers, and strong improvements of the RTM image in complex areas. Furthermore, even with a very large smoothing (8 km), we verify that the TWIN and RFWI gradients significantly differ.