|
<< Click to Display Table of Contents >> Navigation: Geometry > QC geometry by FB |
The QC Geometry by FB module is an interactive geometry quality control tool that detects and corrects errors in source and receiver positions using first-break (FB) arrival times. By comparing theoretical first-break travel times (computed from a linear moveout model) with the actual first arrivals measured on each trace, the module identifies sources and receivers whose coordinates appear inconsistent with the seismic data. This misfit-based approach is highly sensitive to swapped source/receiver positions, mis-wired channels, and incorrect line/station assignments that would be difficult to detect from header information alone.
For each source gather, the module picks or imports first-break times and computes the deviation (misfit) between the observed first breaks and the expected linear moveout curve. Several misfit metrics are calculated per source: peak misfit, average misfit, regression-based misfits, and receiver-line statistics such as the number of bad lines, total trace shift, and maximum trace shift within a line. These metrics are displayed as colour-coded maps (in XY or Line-SP coordinates) and tabular summaries, enabling the geophysicist to quickly identify anomalous sources.
Once a problematic source is identified, it can be selected interactively on the map view. The module displays the corresponding shot gather with overlaid first-break picks, LMO curves, and receiver-line separators, allowing the user to visually inspect and diagnose the geometry error. Source and receiver positions can then be corrected directly: sources may be repositioned by dragging on the map (with optional snapping to a bin grid net), and receiver commutation errors (channel mis-assignments) can be fixed automatically based on a configurable threshold. All changes are tracked and can be saved or loaded as separate correction files for sources, receivers, and killed traces. Corrected geometry is propagated back through the trace headers in the output data stream.
The primary seismic data connection. Connect the upstream module that provides pre-stack shot gathers with geometry already applied to the trace headers.
A handle to the SEG-Y file from which traces are read. This allows the module to access the seismic waveform data for interactive gather display and first-break picking.
The full trace header array for the dataset. Source and receiver coordinates, line numbers, and station point numbers are read from these headers to build the initial geometry model and compute expected offset distances.
The seismic gather item passed through from the input. Traces are read gather-by-gather (source by source) during the QC analysis pass.
Optional 2D stack line definition. Required for 2D crooked-line surveys to correctly associate traces with their line geometry.
Optional crooked line geometry definition for 2D surveys with non-straight acquisition lines.
The 3D survey bin grid definition. Used to convert between inline/crossline bin coordinates and map (XY) coordinates when displaying misfit maps.
A pre-sorted index of trace headers enabling efficient random access to individual shot gathers during interactive inspection without re-sorting the entire dataset.
An optional external SPS receiver point list. When provided and the receiver fixing mode is set to "From sps item", corrected receiver positions are snapped to the nearest SPS-defined point rather than left free in map space. This ensures that any receiver relocation remains physically consistent with the field deployment plan.
An optional set of predefined grid node positions. When provided and the source fixing mode is set to "Bin grid", source relocations during interactive QC are constrained to snap to the nearest node of this reference grid.
An optional first-break or mute picking item. When the LMO mode or first-break source is set to "Picking", this item provides the pre-existing pick times used both for LMO application and for first-break misfit calculation.
This group controls the offset and time range used to compute first-break misfits. Restricting the analysis to a well-defined offset window eliminates near-surface noise or far-offset signal degradation that could contaminate the misfit statistics.
Selects how the offset analysis window is defined. Delta offset computes the misfit for each receiver line separately by comparing the observed first-break slope to the best-fit regression line, using a maximum allowable offset deviation (Max delta offset) to determine which traces belong to each line regression. Min/Max offset restricts misfit calculations to traces within an absolute offset range defined by Min offset and Max offset. Use "Min/Max offset" for surveys where you want to analyse a specific aperture; use "Delta offset" when you want per-receiver-line regression analysis across all available offsets.
The minimum source-receiver offset (in metres) included in the misfit calculation when "Min/Max offset" mode is active. Set this value above zero to exclude very near-offset traces where surface waves or direct-wave interference make first-break picking unreliable. Default: 0 m.
The maximum source-receiver offset (in metres) included in the misfit calculation when "Min/Max offset" mode is active. Traces with offsets beyond this limit are excluded. Set to the maximum reliable first-break offset in your survey. Default: 1000 m.
Active when "Delta offset" mode is selected. Defines the maximum offset difference (in metres) between adjacent traces within a single receiver line that are considered part of the same regression group. This parameter controls how tightly each receiver-line regression is fitted: smaller values create narrower, more localised regression windows; larger values allow the regression to span a wider offset spread. Default: 1000 m.
The maximum record time (in seconds) used during gather loading and analysis. Traces are truncated at this time to speed up I/O and display. Set to cover the expected maximum first-break arrival time at your far offsets. Default: 4 s.
The number of consecutive source gathers to read and process during the analysis pass. Reducing this number limits analysis to a subset of shots, which is useful for fast testing on large 3D surveys. Set to a large value (or the total number of shots) to analyse the full dataset. Default: 100.
Controls how traces are ordered within each displayed shot gather. AS_IS preserves the original trace order from the data file. OFFSET sorts by signed source-receiver offset (negative offsets first). ABS_OFFSET sorts by absolute offset, placing the nearest receiver at the left of the display. Sorting by offset is recommended for first-break inspection because it makes the linear moveout trend immediately visible. Default: AS_IS.
When enabled, traces that have been marked as killed (zeroed or dead) are hidden in the gather display. This reduces visual clutter and makes it easier to see first-break alignments on the remaining live traces. Default: disabled.
The maximum allowable trace position shift (in trace count units) for a receiver-line commutation error to be counted as a "mistake". Receiver traces shifted by more than this threshold within their line are flagged as bad. Increase this value to tolerate larger commutation errors; decrease it for stricter flagging. Default: 1 trace.
Sources whose average first-break misfit (across all receiver lines) exceeds this threshold (in seconds) are highlighted as anomalous on the average misfit map. A value near the expected first-break picking uncertainty (for example 0.01 s for 10 ms sample interval data) is a good starting point. Default: 0.01 s.
Sources whose peak (maximum single-trace) first-break misfit exceeds this threshold (in seconds) are highlighted as anomalous on the peak misfit map. This threshold is typically set equal to or slightly larger than the average abnormal threshold. Default: 0.01 s.
The time origin offset (in seconds) applied when computing the expected linear moveout first-break curve. This accounts for a non-zero recording delay or a bulk time shift in the picks. For most surveys this value is near zero or slightly negative (e.g. -0.048 s to account for a pre-trigger recording delay). Default: -0.048 s.
The constant apparent velocity (in m/s) used to compute the reference linear moveout (LMO) curve: expected first-break time = offset / velocity + T start. This should approximate the near-surface refractor velocity or the average apparent first-break velocity for your area. Typical values range from 1500 m/s to 3500 m/s. Default: 2500 m/s.
Controls whether a linear moveout correction is applied to the displayed gather to flatten the first-break alignment. None shows the raw (uncorrected) gather. V const applies a time shift equal to -offset / (Constant LMO velocity). Picking applies individual per-trace time shifts derived from the imported or auto-picked first breaks. Applying an LMO correction is recommended for visual QC because it aligns first breaks to a nearly flat event, making commutation errors immediately apparent as misaligned traces. Default: None.
Selects the source of first-break times used for misfit calculation. V const computes first breaks from the constant LMO velocity model (offset / velocity + T start). Picking uses pre-existing picks supplied via the Picking item input. Use "Picking" when accurate first-break picks are already available from a prior picking pass, as this gives more reliable misfit statistics. Default: V const.
This group contains parameters that govern how geometry errors are corrected interactively. Both receiver commutation errors and source position errors are controlled here.
Defines the behaviour when the user interactively marks traces as killed (dead) on the gather display. Free kills only the individual traces selected by the user. Magnet to nearest line extends the kill selection to snap to the nearest complete receiver line boundary. Whole line kills all traces belonging to the entire receiver line containing the selected trace. Use "Whole line" when an entire receiver cable is known to be misfired or disconnected. Default: Free.
Specifies where receiver point positions are taken from when applying commutation fixes. From headers uses the receiver coordinates embedded in the trace headers. From sps item uses the external SPS receiver point list connected to the "SPS coordinate points" input. Use the SPS option when the header coordinates are suspected to be erroneous and the original SPS file provides the ground truth. Default: From headers.
The minimum number of traces that must be shifted in a receiver commutation correction for the fix to be applied automatically. Setting this above zero prevents spurious single-trace corrections; set higher values to apply only significant bulk-shift corrections. Default: 0 (all shifts are applied).
The station-point increment (in SP units) between adjacent receiver positions in the commutation grid. This defines the quantisation step when the auto-fix algorithm searches for the best receiver commutation shift. This should match the station interval used in your SPS geometry. Default: 1 SP unit.
Specifies the reference coordinate set used when displaying and relocating source positions. From headers uses the source XY embedded in trace headers. From sps item reads source coordinates from the connected SPS point list. Bin grid snaps source positions to the nearest node on the bin grid net (requires the "Bin grid net points geometry" input to be connected). Default: From headers.
Controls whether source position corrections are unrestricted or constrained to the bin grid net. Free allows a source to be dragged to any XY location on the map. Attach to net snaps the source to the nearest node of the bin grid net after each move, preventing corrections from placing sources at geologically unreasonable positions. Default: Free.
The search radius in the inline direction (in lines) within which candidate replacement source positions are evaluated during automatic source fixing. A larger value allows the algorithm to consider source relocations across a wider inline range, which is appropriate for surveys with coarse inline spacing or large expected displacements. Default: 20 lines.
The search radius in the crossline direction (in lines) within which candidate replacement source positions are evaluated during automatic source fixing. Set this to match the expected magnitude of crossline mis-positioning in your survey. Default: 4 lines.
The misfit metric used to rank and select the best source position when the automatic source fix algorithm searches within the aperture. The algorithm moves the source to the candidate position that minimises the chosen metric. Options: 2 Regressions total — minimises the total regression misfit across two receiver-line groups; 2 Regressions avg — minimises the average regression misfit; Peak — minimises the largest single-trace misfit; Avg misfit — minimises the average misfit; Bad lines count — minimises the number of receiver lines with misfits above threshold; Total traces shift — minimises the total number of shifted traces across all lines; Max trace shift — minimises the largest single-line trace shift. Default: 2 Regressions total.
This group configures the "Table. Bad points" output table that lists anomalous sources. Select the misfit metric to display and set per-metric thresholds to control which sources appear in the table.
Selects which misfit metric is displayed as the primary sort column in the "Table. Bad points" output and which threshold is applied to filter rows. Each option corresponds to a different aspect of geometry quality: peak and average capture absolute timing misfits; regression metrics capture systematic linear-moveout deviations; bad lines count and trace shift metrics capture receiver-line commutation errors; trace count shows per-source coverage. Default: Regressions.
Sources with a peak first-break misfit exceeding this value (in seconds) are listed in the bad-points table when "Peak" metric is selected. Default: 0.01 s.
Sources with an average first-break misfit exceeding this value (in seconds) are listed in the bad-points table when "Average" metric is selected. Default: 0.01 s.
Sources with a total two-regression misfit exceeding this value are listed in the bad-points table when "Regressions" metric is selected. Default: 10.
Sources with an average two-regression misfit exceeding this value (in seconds) are listed in the bad-points table when "Regressions avg" metric is selected. Default: 0.25 s.
Sources with more than this many receiver lines showing commutation errors are listed in the bad-points table when "Bad lines count" metric is selected. Default: 4 lines.
Sources with a total trace commutation shift (summed across all receiver lines) exceeding this count are listed in the bad-points table when "Trace shift total" metric is selected. Default: 20 traces.
Sources with a maximum single-line trace commutation shift exceeding this count are listed in the bad-points table when "Max trace shift in line" metric is selected. Default: 8 traces.
Sources with fewer traces than this threshold are listed in the bad-points table when "Trace count" metric is selected. This helps identify sources with low fold due to missing receivers. Default: 10000 traces.
This group controls the automatic first-break picking algorithm used when the module computes first-break times from the seismic data (as opposed to using imported picks). The parameters define how each trace's first-break arrival is detected and refined.
Defines how the initial pick time is refined (or "magnetised") within the search window. Max energy snaps the pick to the time of maximum amplitude in the window. Nearest keeps the pick at the nearest sample to the initial estimate. First break finds the actual onset of the first arrival within the window, using the QC FB window and threshold parameters. "First break" mode is recommended for geometry QC because it targets the true onset rather than the peak. Default: First break.
Selects the polarity of the first-break onset to pick. Negative phase picks the first negative-going arrival; Positive phase picks the first positive-going arrival. The correct setting depends on the recording polarity convention of your survey. Default: Negative phase.
The half-width (in seconds) of the search window centred on the expected first-break time within which the auto-picker searches for the actual onset. The full search window spans 2 x Half window size. Increase this for noisy data or uncertain velocity models; keep it small to avoid picking multiples or noise. Default: 0.35 s.
An amplitude sensitivity threshold for the first-break auto-picker. Only energy exceeding this fraction of the local RMS amplitude triggers a pick. Higher values reduce false picks on noisy traces but may miss weak first arrivals. Default: 3.
When enabled, a polynomial function is fitted to the automatically picked first-break times as a function of offset, and this smooth curve is used instead of the raw per-trace picks. This suppresses outlier picks due to noise and provides a cleaner misfit calculation. Default: disabled.
When enabled and "Use approximation" is active, a separate polynomial approximation is fitted independently for each receiver line rather than across all traces. This accounts for azimuth-dependent velocity variations and is recommended for wide-azimuth 3D surveys. Default: disabled.
The degree (order) of the polynomial used to approximate the first-break time versus offset curve. A value of 1 fits a straight line (pure linear moveout); higher values allow for curved moveout. Use 1 or 2 for most near-surface velocity structures; use higher values only if the first-break curve is clearly non-linear. Valid range: 1–100. Default: 3.
Active when "Magnet type" is "First break". The time interval (in seconds) before the initial pick estimate that defines the upper boundary of the first-break onset search window. Default: 0.050 s.
Active when "Magnet type" is "First break". The time interval (in seconds) after the initial pick estimate that defines the lower boundary of the first-break onset search window. Default: 0.050 s.
Active when "Magnet type" is "First break". The signal-to-noise amplitude ratio required for the first-break onset detector to confirm a pick within the QC FB window. Increase this on noisy surveys to reduce false detections. Default: 2.
Active when "Magnet type" is "First break". When enabled, an energy-based quality check is applied to each auto-picked first break: picks that fall in a window with insufficient energy (below the QC FB Threshold) are rejected. This removes spurious picks on traces dominated by noise. Default: enabled.
When enabled, any first-break picks that have been imported via the "Import external FB picking" custom action are used in the misfit calculation and displayed on the gather view. Disable this to ignore imported picks and rely only on automatic picking. Default: enabled.
Controls whether input and output data items are connected automatically when the module is added to the processing sequence.
Configuration for the internal SEG-Y read cache. Increasing the cache size improves random-access performance when the user interactively selects different shot gathers for inspection, at the cost of additional memory usage.
Selects the compute device for the analysis pass. CPU execution is standard.
Options for distributing the processing workload across multiple compute nodes in a cluster environment.
The minimum number of gathers processed in each parallel work chunk. Larger values reduce scheduling overhead but may reduce load balancing efficiency.
When enabled, restricts the number of CPU threads used on each distributed compute node to the value specified in "Number of threads".
An optional suffix appended to the distributed job name for identification in cluster job schedulers.
When enabled, allows manual specification of CPU core affinity for the processing threads via the "Affinity" parameter.
Specifies the CPU core mask for thread affinity when "Set custom affinity" is enabled.
The maximum number of CPU threads used for parallel gather processing. Set to the number of logical CPU cores available for best throughput.
When enabled, this module is bypassed and data passes through unchanged. Use this to temporarily disable the geometry QC step without removing it from the processing sequence.
The primary output data connection. Passes through the same gathers as the input but with corrected trace headers reflecting any source or receiver position changes and trace kills applied during the QC session.
The SEG-Y file handle passed downstream for further processing steps that require direct file access.
The updated trace header array with corrected source and receiver coordinates, updated offsets, and kill flags for traces that were manually or automatically killed during the QC process.
The corrected gather item, ready for use by downstream modules such as geometry application, binning, or statics.
The stack line definition passed through unchanged to downstream modules.
The crooked line definition passed through unchanged to downstream modules.
The bin grid definition passed through unchanged to downstream modules.
The updated sorted header index, reflecting any header changes applied during the QC session.
A table listing all source positions that have been relocated during the interactive QC session, showing the original and corrected XY coordinates for each moved source. Use this table for documentation and to verify the scope of geometry corrections before writing the output headers.
A table listing all sources that exceed the configured misfit threshold for the selected Table misfit type. Each row represents one anomalous source, with columns showing its line/SP identifier, map coordinates, and the misfit value. This table is the primary output for identifying geometry problems across the entire survey.
A table recording all traces that have been manually killed during the QC session, including their source and receiver identifiers. This can be saved to disk and reloaded in future sessions or shared with the data processing team.
A table listing the source point positions as read from the input trace headers. This provides a reference snapshot of the original source geometry before any corrections are applied.
A table listing the receiver point positions as read from the input trace headers, providing a reference for the original receiver geometry.
A table showing the original SPS geometry entries for the currently selected source gather. This is updated interactively as the user selects different sources on the map.
A table showing the corrected (updated) SPS geometry entries for the currently selected source gather after any fixes have been applied. Comparing this with the "Current SRC original SPS" table allows the user to review the net effect of corrections on the current source.
A table listing all sources for which receiver commutation corrections have been applied, showing the nature and magnitude of the commutation shift for each affected source-receiver-line combination.
Applies all accumulated geometry corrections (source relocations, receiver commutation fixes, and trace kills) to the trace headers and writes the updated header array to the output. Run this action after completing the interactive QC session to commit the corrections into the data stream.
Removes all manually edited LMO pick times, reverting to the automatic or imported picks.
Loads a previously saved LMO picking file from disk, restoring manual pick edits from a prior session.
Saves the current LMO pick times to a file on disk so they can be reloaded in a future session or shared with other users.
Saves the current set of killed traces to a text file. This file can be loaded in a subsequent processing run to re-apply the same kill list without repeating the interactive QC session.
Loads a previously saved killed-traces file from disk, restoring the kill list from a prior QC session.
Removes all trace kills accumulated in the current session, restoring all traces to live status.
Saves all receiver commutation corrections to a file on disk so they can be reloaded or applied in a future processing run.
Loads a previously saved receiver corrections file from disk, restoring commutation fixes from a prior session.
Removes all receiver commutation corrections accumulated in the current session, reverting receivers to their original header positions.
Saves all source position corrections to a file on disk so they can be reloaded or applied in a future processing run.
Loads a previously saved source corrections file from disk, restoring source position fixes from a prior session.
Removes all source position corrections accumulated in the current session, reverting sources to their original header coordinates.
Imports first-break pick times from an external file (for example, picks produced by a dedicated first-break picking module or a third-party software). Once imported, these picks can be used for misfit calculation by enabling "Use imported picks" and selecting "Picking" as the first-break source.
Automatically detects and kills all zero-amplitude (empty) traces across the entire dataset. This is a batch operation that does not require manual trace selection.
Automatically detects and kills all zero-amplitude (empty) traces within the currently displayed shot gather only, allowing selective cleaning of a single problematic source without affecting the rest of the dataset.