Twin Topics on Near-Surface Modeling and Subsurface Imaging

Описание: In this lecture I will present two topics from the new SEG book Land Seismic Case Studies for
near-surface modeling and subsurface imaging. The first topic is: A reality check on full-wave inversion applied to land seismic data for near-surface modeling. In land seismic exploration, data often are acquired over an irregular topographic surface and complex near-surface. In exploration seismology, the near-surface is defined as the depth interval below the free surface with usual thickness up to a few hundreds of meters composed of low-velocity unconsolidated soil and heterogeneous weathered rocks. As such, raypaths are close to vertical incidence within the near-surface --- a requirement for statics corrections to be acceptable. The subsurface is composed of relatively high-velocity consolidated layers of sedimentary rocks, which may be intruded by igneous and metamorphic rocks. As such, the interface between the near-surface and the subsurface often gives rise to a strong impedance contrast, and therefore a strong shallow reflection. Given the low-velocity unconsolidated characteristic of the near-surface, if it is at or near the surface, however, it is not always the near-surface. Specifically, in the absence of a low-velocity medium just below the free surface, there is no near-surface within the context of statics corrections or, for that matter, geotechnical investigations.

We evaluate FWI applied to land seismic data within the context of near-surface modeling. For most cases of near-surface complexity which gives rise to amplitude and traveltime distortions associated with subsurface reflections, traveltime tomography is very robust and far more computationally efficient compared to full-wave inversion, and is more than adequate to resolve the medium- to long-wavelength statics problem as shown at the bottom. Nevertheless, traveltime tomography relies on the accuracy of first-break picking and falls short of resolving velocity reversal --- the case of a high-velocity layer hidden within a low-velocity near-surface.

Practical experience has taught us that, for FWI to likely converge to the global minimum and thus yield a physically plausible near-surface model, it needs an initial model that is closest possible to the near-surface model that can be accepted as a close representation of the ground truth. The initial model that qualifies for this criterion is that estimated by traveltime inversion applied to first-arrival times picked from shot records. The question then is: For near-surface modeling, can we apply full-wave acoustic inversion or is it imperative that we apply full-wave elastic inversion? The answer lies in radiation patterns of seismic source energy and Zoeppritz equations for wave partitioning at the interface between the low-velocity near-surface and the high-velocity subsurface. We conclude that full-wave inversion for near-surface modeling must in principle be performed not as acoustic but elastic-wave inversion. Nevertheless, in a field record, we may safely assume that the early-arrival waveforms associated with the first-break times are essentially associated with P-waves. Thus, we may perform full-wave acoustic inversion of the early arrival waveforms. We may take the bold step and retain the early portions of the shot gathers which we may assume to contain largely P-waves by excluding the surfacewaves and apply full-wave acoustic inversion. However, the retained portions of the shot gather contain a mixture of PP, PS, SP, and SS wave modes trapped in the near-surface. This fact alone disqualifies acoustic-wave inversion applied to land seismic data for near-surface modeling and the resulting near-surface model will most likely be physically implausible and dissimilar to the initial model estimated by traveltime tomography.

We investigate accuracy of traveltime tomography, and full-wave acoustic and elastic inversions applied to near-surface velocity-depth model estimation by a field experiment that involves a shallow seismic survey and borehole drilling. We report the results of the field experiment along one of the line traverses. The comparison of the seismic model of the soil column and the borehole logs gives us the confidence in traveltime tomography for modeling the near-surface. Specifically, we may declare the near-surface model shown here as a close representation of the ground truth.

Next, we perform traveltime tomography to estimate a P-wave velocity-depth model for the near- surface and Rayleigh-wave inversion to estimate an S-wave velocity-depth model for the near- surface, then use the resulting pairs of models as the initial models for the subsequent full-wave elastic inversion. We make the following conclusions from the field experiment: