Frequency-Domain Common Image Gathers for Quickly Checking the Accuracy of Migration Velocity

Haemin Kim,Yunseok Choi,Changsoo Shin,Yongchae Cho,Seungwon Ko

doi:10.3389/feart.2021.669686

Abstract

The common image gather (CIG) method enables qualitative and quantitative evaluation of the velocity model through the image. The most common such methods are offset-domain common image gather (ODCIG) and angle-domain common image gather (ADCIG). The challenge is that it requires a great deal of additional computation besides migration. We, therefore, introduce a new CIG method that has low computational cost: frequency-domain common image gather (FDCIG). FDCIG simply rearranges data using a gradient (partial image) calculated in the process of obtaining a migration image to represent it in the frequency-depth domain. We apply the FDCIG method to the layered model to show how FDCIGs behave when the velocity model is inaccurate. We also introduced the 3-D SEG/EAGE salt model to show how to apply the FDCIG method in the hybrid domain. Last, we applied 2-D real data. These sample field data also indicate that even in a complex velocity model, deviant behavior by FDCIG appears intuitively if the background velocity is inaccurate.

Highlights

Reverse time migration (RTM) produces a high-fidelity subsurface image from seismic data for identification of complex subsurface structures (Baysal et al, 1983; McMechan, 1983; Whitmore, 1983)
We briefly summarize the theory of frequencydomain RTM, a means of calculating frequencydomain common image gather (FDCIG), and related postprocessing flow
We examined how FDCIGs appear in the true velocity model through the first example and examined how the behavior of the FDCIGs changes when the background velocity is slightly changed

Summary

Introduction

Reverse time migration (RTM) produces a high-fidelity subsurface image from seismic data for identification of complex subsurface structures (Baysal et al, 1983; McMechan, 1983; Whitmore, 1983). The RTM implementation in the time domain is often preferred due to its lower memory consumption than the frequency domain, and this is a critical factor in handling 3-D problems. Calculating both wavefields and imaging conditions in the frequency domain has advantages over time-domain implementation (Pratt, 1999; Wu and Alkhalifah, 2018).

Objectives

Methods

Conclusion