|
<< Click to Display Table of Contents >> Navigation: »No topics above this level« Diffraction |
The Diffraction group contains modules for edge-diffraction imaging — a specialized technique used to detect and image subsurface structural edges, faults, fracture zones, and other discontinuities that generate diffracted wave energy. Unlike conventional reflection imaging, diffraction imaging focuses on scattered energy emanating from point-like or linear scattering bodies, providing higher-resolution images of geologic boundaries that may be invisible or poorly resolved in standard stack sections.
The diffraction imaging workflow in g-Platform follows a three-step pipeline: geometry preparation, diffraction engine execution, and interactive visualization. An optional amplitude correction step is available to normalize the output image amplitudes for improved interpretability. The engine supports both 2D and 3D survey geometries and can use GPU acceleration for large datasets.
A typical diffraction imaging project consists of the following steps, corresponding to the modules in this chapter:
Step 1 — Edge Diffraction Geometry Input Data By Azimuth. This preparatory module reads the input SEG-Y dataset and organizes traces into directional azimuth sectors defined by a start azimuth, end azimuth, and angular half-step. For each sector, it builds the CMP geometry (using the inline and crossline CMP spacing) and computes which input traces fall within the spatial aperture of each output bin. The result is a set of pre-organized gather structures — one for duplicate traces and one for unique traces — ready to be fed into the diffraction engine. Running this step first reduces computation time in the main engine by pre-sorting and filtering the input data.
Step 2 — Engine - Diffraction imaging 2D/3D. This is the core processing module. It performs Kirchhoff-based diffraction imaging by applying moveout corrections and coherency (semblance) analysis across the defined azimuth range. For each output bin, the engine computes diffraction semblance and stacks the traces within the spatial aperture, applying an optional rho-filter for amplitude balancing and anti-alias protection. The results are written to a binary database file (.kdb) that stores the imaged diffraction energy organized by azimuth, offset, and spatial position. The engine supports 2D, 2.5D, and full 3D modes, and can distribute computation across GPUs and remote nodes for large 3D surveys.
Step 3 — Imaging - Diffraction imaging 2D/3D. This interactive visualization module reads the .kdb database produced by the engine and displays diffraction images along inline and crossline sections. It provides views of semblance, field amplitude, and fold panels for a user-selected azimuth index. Polar azimuth maps show the directional distribution of semblance energy at the selected bin, enabling geophysicists to identify the orientation of diffracting edges. The module supports point-and-click navigation on a map view and allows full-cube export to SEG-Y format.
Step 4 (optional) — Diffraction amplitude correction. This post-processing module normalizes the amplitude of the diffraction image gather to compensate for uneven illumination caused by variable fold, geometric spreading, and lateral amplitude trends. It applies three independent normalization mechanisms: fold-weighted normalization (which scales each sample by the local trace count), vertical normalization (which equates amplitude levels along the time axis within a sliding window), and lateral normalization (which removes large-scale amplitude trends across the inline and crossline directions). Use this module after the imaging step when the raw diffraction image shows strong amplitude variations unrelated to geology.
Quality Control — Azimuth Selection Visualization. This companion module provides a fold map and azimuth coverage display for quality control of the survey geometry before or after diffraction imaging. It is also listed separately in the Azimuth chapter.
Azimuth sectors. The diffraction imaging workflow analyzes data in multiple directional sectors simultaneously. Each sector is centered at an azimuth angle and has a half-width defined by the angular half-step parameter. Imaging in multiple azimuths enables detection of the orientation of structural edges — for example, a fault striking at 45° will produce maximum diffraction energy in that azimuth sector. The default azimuth range is 10° to 170° with a half-step of 10°, which produces 16 azimuth sectors covering a half-space (0°–180°, exploiting the symmetry of P-wave diffraction).
Aperture. The imaging aperture controls the lateral extent (in metres) of the input traces gathered around each output bin location. A larger aperture includes more traces and improves the imaging of steeply dipping or deep diffractors, but increases computation time. For the geometry preparation module, the aperture is used to pre-select which traces are relevant to each output bin. For the engine, the aperture also controls the maximum migration distance from each diffraction point.
Storage database (.kdb). The engine writes all computed diffraction results — semblance gathers, field stacks, and fold maps organized by azimuth — into a proprietary binary database file with the .kdb extension. This file is the input to the imaging visualization module. It is also shared with the DB Connect module, which exposes the database as a data connector for downstream modules in the processing flow.
Velocity model. The engine requires a VRMS (root-mean-square) velocity model to apply the correct time moveout corrections for each source-receiver pair and each diffraction point in time. The accuracy of the velocity model directly affects the sharpness and resolution of the resulting diffraction image. Use the same velocity field employed for pre-stack time migration of the same dataset.
The following modules are available in the Diffraction chapter:
Edge Diffraction Geometry Input Data By Azimuth — Sorts and filters input seismic traces into directional azimuth sectors and spatial aperture bins to prepare data for the diffraction engine. Outputs pre-organized gather structures (duplicate and unique gathers). Run this module before the engine to reduce engine computation time.
Engine - Diffraction imaging 2D/3D — The main diffraction imaging engine. Applies Kirchhoff moveout correction and semblance-weighted stacking over a range of azimuths and offsets. Supports 2D, 2.5D, and 3D modes. GPU and distributed execution available. Writes results to a .kdb database. Key parameters include the imaging aperture (default 3000 m), azimuth range (default 10°–170° in 20° steps), CMP spacing, maximum frequency for the rho-filter (default 120 Hz), and offset range for common-image gathers.
Imaging - Diffraction imaging 2D/3D — Interactive visualization module that reads the .kdb database and displays inline and crossline diffraction semblance sections for a selected azimuth. Provides polar azimuth maps of semblance energy at any selected bin. Supports full-cube export to SEG-Y. Use this module to interactively browse the diffraction image and identify fracture orientations and structural edges.
Azimuth Selection Visualization — Displays a fold map and azimuth coverage distribution for the survey geometry. Use this module for quality control to verify that the azimuth and offset coverage is adequate before running the diffraction engine.
Diffraction amplitude correction — Post-processing normalization of diffraction image amplitudes. Applies fold-weighted, vertical (time-domain), and lateral (spatial) normalization to remove illumination imbalances from the imaged result. The vertical normalization window defaults to 0.3 s; the lateral normalization windows default to 150 bins in inline and crossline directions. Apply after the imaging module when amplitude variations mask geological features of interest.