Download

The data used in the code below can be downloaded here. The directory has two sub-folders:

  • DAT folder containing files with .dat extensions
  • MAT folder containing files with .mat extensions

The code below uses data from the MAT folder.

Abstract

MarchenCode contains GPU-optimized Python code to compute the Green’s functions from single-sided surface reflectivity data. Learners will also find code to perform Marchenko-based imaging using decomposed Green’s functions (upgoing). I embarked on this project in an effort to understand the Marchenko methods. For learners proficient in MATLAB, you can download the original version of the code in MATLAB here by Angus Lomas and Andrew Curtis

Figure 1: Input Shot Gather

Figure 2: Focusing Functions (Initial and After 5 Iterations)

Figure 3: Estimated Green’s Function

Figure 4: Marchenko Imaging

A look at the output from the tqdm (taqaddum, which mean progress in arabic) progress bar 64/64 [23:06<00:00, 22.04s/it] indicates:

  1. There are 64 iterations..

  2. Total Runtime: 0 hours, 23 minutes, and 06 seconds.

  3. Per-Iteration Time: ~22.04 seconds per point.

Key Reference

Lomas, Angus, and Andrew Curtis. “An introduction to Marchenko methods for imaging.” Geophysics 84, no. 2 (2019): F35-F45.