delta-t stretch, Oil & Gas Glossary

What Is Delta-T Stretch in Seismic Processing?

Delta-t stretch (also written delta-T stretch or dT stretch) is a seismic processing artefact in which the application of normal moveout (NMO) correction to remove the offset-dependent travel time difference between traces progressively elongates (stretches) the seismic wavelet at large offsets and shallow depths, shifting the apparent frequency content of stretched traces toward lower frequencies and changing the shape of the waveform relative to the unstretched near-offset traces, requiring mute of heavily stretched traces before stacking to prevent frequency distortion and amplitude artefacts in the final seismic image.

Key Takeaways

NMO correction flattens seismic reflection events by shifting each time sample on a far-offset trace upward by the calculated moveout time; the shift is larger for shallow events at large offsets, causing the wavelet to be spread over a longer time interval (stretched).
The stretch factor f_stretch = t_NMO_corrected / t_original = (offset² / (v² × t₀²) + 1)^0.5; at large stretch factors (greater than 1.3-1.5), traces are typically muted before stacking.
Delta-t stretch reduces the dominant frequency of stretched traces, causing them to appear as a lower-frequency noise contribution in the stack that can obscure thin-bed resolution.
Stretch mute is applied to remove the most-stretched traces from the stack; the mute pattern follows the stretch factor threshold across offset and time, removing more of the far-offset traces at shallow times.
Optimal NMO mute decisions involve a trade-off between signal-to-noise (more traces = better S/N) and frequency content preservation (fewer stretched traces = better bandwidth).

The Physics of NMO Stretch

Normal moveout (NMO) correction is applied to seismic shot records to align reflection events from a common reflecting point that arrive at different times on different offset traces due to the longer travel path at larger source-receiver offsets. The NMO correction shifts each time sample on each trace by subtracting the calculated moveout time: t_NMO = sqrt(t₀² + x²/v²) - t₀, where t₀ is the zero-offset two-way travel time, x is the source-receiver offset, and v is the NMO velocity. For deep reflectors (large t₀), the moveout time is a small fraction of t₀, and the correction shifts the entire event by a small, approximately constant amount — the wavelet shape is preserved. For shallow reflectors (small t₀), the moveout time can be a large fraction of t₀ — approaching or exceeding t₀ at large offsets — and different parts of the wavelet (beginning, middle, end) are shifted by increasingly different amounts because they occur at slightly different times with slightly different NMO corrections.

This differential shift stretches the wavelet: the start of the reflection arrives earlier (the beginning of the wavelet is shifted less) than the end (which is shifted more because it is at a slightly later time where the NMO correction is slightly different). The result is a wavelet that is wider in time — and correspondingly lower in dominant frequency — after NMO correction at far offsets for shallow events. A wavelet that was 40 milliseconds wide before NMO correction might be stretched to 60 or 80 milliseconds after correction, shifting its 50 Hz dominant frequency to 33 or 25 Hz. When this stretched, low-frequency trace is stacked with unstretched near-offset traces, the frequency content of the stack at that shallow time is degraded, and amplitude artefacts can appear because the stretched wavelet no longer correlates well with the unstretched traces.

Delta-T Stretch Effects Across International Jurisdictions

In Canada, NMO stretch mute is a standard processing consideration for WCSB reflection seismic surveys targeting shallow gas sands (100-600 metres depth) in the Mannville Group and Horseshoe Canyon Formation, where the combination of shallow targets and long offsets (necessary for AVO analysis) creates significant NMO stretch in the shallow section. AER seismic data quality requirements for exploration licence submissions specify minimum data quality standards; seismic processors document the stretch mute applied to confirm that shallow target frequencies meet resolution requirements after muting. For the Grand Banks offshore Newfoundland, shallow hazard identification above Jurassic reservoirs requires high-resolution shallow imaging — the stretch mute must preserve near-offset data quality while removing far-offset stretch artefacts to maintain resolution.

In the United States, delta-t stretch is a significant practical issue in Gulf of Mexico shallow (Pleistocene and Pliocene) AVO analysis, where the shallow gas hazard identification for geohazard assessment and the characterisation of biogenic gas intervals require accurate far-offset amplitude information. Excessive stretch mute removes the far-offset data needed for AVO analysis; insufficient muting leaves stretch artefacts that distort amplitude versus offset patterns and can create false AVO anomalies in the shallow section. BSEE shallow hazard surveys for deepwater drilling programmes require high-quality shallow reflection data where these stretch considerations are practically important. In Norway, Sodir seismic data standards for NCS exploration surveys require documentation of stretch mute parameters applied during processing; the shallow section resolution above Quaternary gas pockets must be adequate for safe drilling hazard identification. In the Middle East, shallow high-resolution seismic for near-surface mapping above Arab Formation targets uses very short offsets to minimise stretch and preserve resolution in the shallow (<300 metre) section.

Fast Facts

The maximum stretch factor typically accepted before muting in most seismic processing flows is approximately 30-50% (stretch factor 1.3-1.5), meaning the wavelet has been elongated by 30-50% compared to its unstretched form. For a 60 Hz dominant frequency wavelet, a 50% stretch reduces the dominant frequency to 40 Hz — a significant reduction that degrades vertical resolution from approximately 12 metres (at 60 Hz in 1,500 m/s formation) to approximately 18 metres (at 40 Hz). In AVO-sensitive surveys targeting thin gas sands near or below the resolution limit, even a 20% stretch (1.2 stretch factor) may be unacceptable, requiring the mute to be applied more aggressively and accepting the signal-to-noise reduction from fewer contributing traces in the stack.

Stretch Mute Design and AVO Trade-offs

The stretch mute applied to remove heavily stretched far-offset traces before stacking defines a mute function in the offset-time domain: for each time sample, all traces beyond the offset where the stretch factor exceeds the threshold are zeroed before stacking. The mute is time-variant — at early times (shallow reflectors), even moderate offsets cause large stretch so the mute kills far-offset traces aggressively; at late times (deep reflectors), the mute is lenient because stretch is small even at large offsets. Designing the mute requires balancing two competing objectives: preserving frequency content (favouring aggressive muting of stretched traces) and preserving signal-to-noise ratio (favouring lenient muting to include more traces in the stack). For a target where AVO analysis is the primary goal, the mute must preserve sufficient far-offset data to measure the amplitude variation with offset reliably — this may require accepting some stretch artefacts that can be partially corrected by stretched-trace spectral equalisation. For a target where resolution and thin-bed detection are the primary goals, aggressive muting to preserve the highest possible frequency bandwidth is preferred even at the expense of some signal-to-noise reduction from fewer contributing traces.

Tip: When the seismic data shows a frequency loss in shallow section stacks compared to the near-offset shot records, check whether the stretch mute has been applied appropriately before investigating other causes such as near-surface absorption or source signature changes. Compute the theoretical stretch factor at the shallow target depth for the maximum offset included in the mute, and compare this to the frequency degradation observed. If the computed stretch is 1.4 (40% wavelet elongation) and the observed dominant frequency on the stack is 30% lower than on the near-offset panel, the stretch mute is likely the dominant cause rather than intrinsic formation absorption. Applying a more aggressive stretch mute (reducing the maximum stretch threshold from 1.5 to 1.3) may restore most of the frequency loss at the cost of a modest decrease in the number of contributing traces and a small signal-to-noise ratio reduction.

Delta-t stretch is also referenced as:

NMO stretch — the most common alternative name, emphasising that the stretch arises specifically from the NMO correction step; "NMO stretch" is used interchangeably with "delta-t stretch" in seismic processing documentation and is arguably more widely used in the current literature
Stretch mute — refers specifically to the processing step that removes stretched traces; "applying a stretch mute" is the corrective action taken to mitigate delta-t stretch, as distinct from delta-t stretch itself which is the problem being corrected
Far-offset mute — a less precise term sometimes used for the stretch mute; "far-offset mute" correctly describes the location of the muted traces (far offset) but does not specify the physical reason (stretch) for the muting, and may be confused with other reasons for far-offset muting such as noise contamination or ground roll

Related terms: NMO correction, stacking, AVO, seismic resolution, mute

Frequently Asked Questions

Why does NMO stretch cause frequency reduction in the wavelet?

The frequency reduction from NMO stretch arises because the NMO correction applies a non-uniform time shift to different parts of the wavelet. At a fixed offset and shallow two-way time, the beginning of a seismic wavelet (which arrives at time t₁) is shifted upward by a slightly smaller amount than the end of the wavelet (which arrives at time t₂ > t₁) because the NMO correction decreases with increasing two-way time: t_NMO = sqrt(t₀² + x²/v²) - t₀, where dt_NMO/dt₀ is negative. This means the end of the wavelet is advanced by less than the beginning, stretching the wavelet over a longer time interval. If the original wavelet was a 20-millisecond duration pulse with dominant frequency 50 Hz, and NMO stretch elongates it to 30 milliseconds, the dominant frequency decreases to approximately 50 × 20/30 = 33 Hz. The frequency reduction is proportional to the stretch factor: a stretch of 1.5 (50% elongation) reduces the dominant frequency by a factor of 1.5. This is why the threshold for acceptable stretch (typically 1.3-1.5) directly translates to an acceptable frequency degradation (23-33% reduction in dominant frequency) — the processing team must decide whether this frequency loss is acceptable for the interpretation objectives before setting the mute threshold.

How does delta-t stretch affect pre-stack amplitude analysis?

Delta-t stretch creates systematic offset-dependent frequency and amplitude distortions that affect pre-stack amplitude analysis (AVO) if not adequately controlled. The frequency reduction at far offsets means that the amplitude of far-offset traces in a given time window depends on the wavelet overlap between adjacent events as well as on the AVO signal itself. If two closely-spaced reflectors (below tuning thickness) are interfering, their combined amplitude changes with offset partly because the stretched wavelet changes the constructive/destructive interference pattern with offset — this is a processing artefact, not a formation property. Additionally, the amplitude normalisation applied during processing (trace equalization, noise attenuation) may partially compensate for or amplify the stretch-induced amplitude changes in an offset-dependent way. Best practice for pre-stack AVO studies includes: applying an aggressive stretch mute to remove traces with stretch factor above 1.2-1.3; applying a spectral equalisation filter to stretch-corrected traces to restore frequency balance across offsets; and verifying AVO gradients against well synthetic gathers at the well locations to confirm that the observed amplitude gradient reflects formation properties rather than processing artefacts.

Delta-T Stretch: Definition, NMO Stretch, and Seismic Processing Artefacts