Workshop 4

Description

For last several decades, the evolution of both hardware and algorithmic advancement has made full-waveform inversion (FWI) computationally affordable and even used in practice. Given the role of FWI in velocity model building and depth imaging, FWI has become a matured and practical technology in industry. Nowadays, there have been many trials to apply the FWI technology for the time-lapse (4D) problem due to its capability of capturing subtle changes of seismic signature caused by the time-dependent subsurface variation. The inherent problems of FWI, such as cycle skipping or data fitting, are still the subject of research activity. However, in 4D FWI, non-repeatability of seismic data acquisition adds additional problematic factors, which bothers practical applications of FWI to capture subtle 4D effects.

A classical method for the 4D analysis investigates differences between seismic data with two different vintages. The post-stack time shift analysis from migrated seismic volumes of baseline and monitor data is a simplistic representation of 4D effects. Also, not taking velocity changes into account can cause additional artifacts into amplitude maps and their 4D differences, which leads to mis-positioning of lateral boundaries of a 4D anomaly. The interpretation of baseline and monitor data often suggests several over- and under-burden changes, which are likely caused by imperfect accounting of time-lapse velocity changes during 4D imaging. Therefore, 4D FWI using pre-stack data can be an alternative, which provides better perspectives for investigating 4D fluid effects and velocity changes. Restoring accurate velocities for both baseline and monitor data changes would help us make further interpretation of data by distinguishing between real and artificial changes around the target reservoirs.

4D FWI is not a just repetitive implementation of conventional FWI. 4D FWI requires special preparation procedures to resolve the non-repeatability issues such as:

• Precise replication of acquisition geometry is not a trivial issue. Appropriate treatment for regularizing locations of sources and receivers is required when there are admissible differences between source and receiver positions of baseline and monitor data. If the differences exceed the threshold, the time-lapse data is unlikely to be used for 4D FWI.
• Source signatures may not be identical between baseline and monitor data. This problem is associated with environmental changes such as water velocity variation, which is uncontrollable and difficult to incorporate in velocity (or impedance) models. Therefore, 4D FWI requires more rigorous static correction to treat the statics of multiple reflections compared to 3D FWI.

In order to highlight 4D subsurface changes via data processing, a number of 4D FWI methodologies have been developed (i.e., the sequential difference, double difference, and central difference methods), which are crucial research topics in 4D FWI. Hence, the maturity of 4D FWI algorithms needs to be discussed as well in the workshop.

The aim of this workshop is to 1) address the challenges of suppressing the effect of non-repeatability in 4D FWI and to 2) brainstorm ideas for capturing subtle changes of seismic signatures hidden behind baseline and monitor data. Also, we will share 3) various case studies of successful 4D FWI applications from the industrial communities:

• What are the current industrial applications?
• Which data is the best for an optimal 4D FWI?
• Can 4D FWI preserve the true amplitude variation?
• How do we overcome the issues of different source signatures and non-repeatability of source and receiver locations in time-lapse data?
• What types of additional pre-processing techniques are required for 4D FWI compared to the conventional FWI methods to amplify the 4D effect?
• What is an optimal 4D FWI method for the industrial-scale practical applications?
• How do we decouple parameters in multi-parameter 4D FWI?

We invite submissions on all aspects of 4D FWI applications and case studies across the fields of hydrocarbon reservoir and CO2 subsurface storage monitoring. Contributions are welcome from industry corporates as well as academia (including non-seismological society) studying fundamental issues of time-lapse inverse problem.

Sub-Topics that will be covered in the workshop:

• FWI, time-lapse, pre-processing