Full waveform inversion (FWI) has recently been considered for extracting detailed near surface information through the surface waves inversion. But the need of long computing time and the risk to get trapped into local minima make FWI cumbersome. Both these shortcomings are strongly dependent on forward modeling. In fact, forward modeling must yield very accurate seismograms and be computationally feasible. In this study we use an elastic finite difference modeling and discuss means to address, at least partially, these two conflicting requirements. In particular, we illustrate the possibility of attenuating the computing time problem by implementing the convolutional perfectly matched layers (CPML), and by performing the 3D to 2D correction on the observed data, thus allowing for using simple 2D forward modeling. It turns out that CPML has the ability to accelerate the computations compared with the standard tapering method, while maintaining the same efficiency in attenuating the unwanted artifacts from the model boundaries, and that 3D to 2D correction gives satisfactory results. Both of them result very useful in the perspective of surface waves FWI applications, also in view of the fact that other features, such as irregular topography, which cannot be neglected, require additional computing efforts.


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