DMO: Dip Moveout Correction, Conflicting-Dip Imaging, and Prestack Partial Migration in WCSB Seismic
DMO, short for dip moveout, is the additional traveltime difference that appears in a reflected seismic wave, measured between receivers at two different source-to-receiver offsets, when the underlying reflector is not flat but dips. It is also the name of the seismic-processing operator that removes that effect. To understand why DMO matters, start with the assumption built into the common-midpoint (CMP) method: that all the traces in a CMP gather share a single subsurface reflection point directly below the midpoint, so they can be corrected for offset by normal moveout (NMO) and then summed, or stacked, to boost signal. That assumption holds only when the reflector is horizontal. When a reflector dips, the actual reflection points for different offsets smear updip and no longer coincide; they fall at a common reflection point only after the dip-dependent component of moveout is accounted for. DMO processing performs exactly this correction, repositioning energy so that dipping reflections from a CMP gather once again map to a common reflecting point before stack. The practical payoff is large in structurally complex parts of the Western Canadian Sedimentary Basin, where steeply dipping fault planes, the flanks of Leduc and Nisku reef buildups, and thrust-sheet geometry in the Alberta Foothills produce reflectors that dip tens of degrees and frequently cross flat-lying events. The classic problem DMO solves is the conflicting-dip-with-different-stacking-velocity issue. At any point where a steep dip and a gentle dip overlap, a single NMO velocity cannot flatten both, so conventional stacking smears one or the other. DMO is velocity-independent in its core formulation, which means it correctly handles all dips simultaneously and frees the stacking velocity to describe the medium rather than to compromise between conflicting events. After DMO, the stacking velocity becomes the true zero-offset, zero-dip velocity, which sharpens the subsequent velocity analysis and feeds a cleaner poststack time migration. DMO is mathematically a form of prestack partial migration: it moves energy partway toward its migrated position, deferring the rest of the imaging to a poststack migration step. It can be implemented in several equivalent ways, including the integral or Kirchhoff summation method, finite-difference offset continuation, a frequency-wavenumber (f-k) Fourier-domain operator, and log-stretch formulations, all developed through the early to mid 1980s following the foundational work of Yilmaz, Claerbout, Hale, and others. Although full prestack time and depth migration have in many modern WCSB workflows superseded the NMO-DMO-stack-migrate sequence, DMO remains conceptually central, computationally cheap, and still widely used for fast, robust imaging of legacy 2D lines and for building accurate velocity models that the heavier migrations depend on.
Key Takeaways
- Dip breaks the CMP assumption: Common-midpoint stacking assumes every offset in a gather shares one reflection point below the midpoint. For a dipping reflector the reflection points smear updip, so NMO alone cannot align them. DMO supplies the dip-dependent moveout correction that restores a true common reflecting point before stack, which is essential on the steep Foothills thrust geometry of the WCSB.
- Velocity-independent operator: The core DMO formulation does not depend on the velocity model, so it corrects all dips at once. This is what lets it resolve conflicting dips that share a midpoint but require different stacking velocities, a situation common where a steep fault plane crosses gently dipping Cardium or Viking strata.
- Cleans the velocity field: After DMO, the stacking velocity reduces to the zero-dip, zero-offset velocity rather than a compromise between competing events. That makes the velocity analysis physically meaningful and improves the accuracy of the poststack migration and any depth conversion built on it.
- Prestack partial migration: DMO moves reflection energy partway to its migrated position, so the workflow becomes NMO, DMO, stack, then poststack migration. It can be implemented by Kirchhoff summation, finite-difference offset continuation, or an f-k Fourier-domain operator, all giving equivalent results within their approximations.
- Still relevant despite prestack migration: Full prestack time and depth migration now handle dip directly in many WCSB projects, but DMO remains cheap, robust, and valuable for reprocessing legacy 2D Western Canada lines and for generating the well-behaved velocity fields that the more expensive migrations require as input.
Conflicting Dips on an Alberta Foothills Line
In the Alberta Foothills, a 2D line shot across a thrust belt routinely images a gently dipping autochthonous Mississippian carbonate beneath a steeply dipping thrust-sheet repeat of the same section. At a shared midpoint the flat event needs a stacking velocity near 4,200 m/s while the steep limb needs an apparent velocity far higher because of the dip. A single NMO pass smears one event into noise. Running DMO between NMO and stack repositions both events to their true reflection points so a single stack preserves both. Interpreters then see the thrust geometry that controls a Turner Valley style structural trap rather than a blurred, misvelocitied section that hides the closure.
DMO Implementation Choices and Trade-Offs
The integral Kirchhoff DMO smears each input sample along an elliptical impulse response and is flexible for irregular acquisition geometry common on rough WCSB terrain, but it can introduce operator aliasing on coarse spatial sampling. The f-k log-stretch DMO is fast and exact for constant velocity but assumes regular trace spacing, which crooked-line vibroseis surveys in the Deep Basin violate. Finite-difference offset continuation steps the data through offset and handles lateral velocity variation gracefully at higher cost. Processors pick the operator by survey geometry and budget; on a tight gas Montney 3D, a turning-wave-aware Kirchhoff DMO often precedes a Kirchhoff prestack time migration in the same pass.
Fast Facts
DMO was effectively invented twice. The dip-correction concept traces to a 1979 prestack partial-migration paper by Judson and colleagues, but the elegant velocity-independent integral operator that made DMO a routine production step came from Dave Hale's 1983 Stanford doctoral work, which expressed the correction in the frequency-wavenumber domain. For roughly fifteen years the NMO-DMO-stack sequence was the default imaging chain for dipping data worldwide, and only the arrival of cheap prestack time migration in the late 1990s pushed it from front-line tool to supporting role.
Related Terms
DMO is applied immediately after normal moveout, which removes the offset-dependent moveout for flat layers, leaving DMO to handle only the dip component. Both are preparatory steps to migration, the imaging process that moves dipping reflections to their true subsurface positions, since DMO is itself a partial migration. The corrected gathers feed a sharper velocity analysis, and the entire sequence operates on data organized by the common midpoint method that DMO exists to repair for dipping reflectors.
Real-World WCSB Scenario: Reprocessing a Legacy Deep Basin 2D Grid
An operator evaluating a Deep Basin Montney and Nikanassin gas play inherits a 1990s 2D seismic grid shot on crooked roads with irregular offsets. The original NMO-stack sections smear the steeply dipping western disturbed-belt structures into incoherent noise, making a CAD 12 million land posting hard to high-grade. A reprocessing contractor applies a Kirchhoff integral DMO tuned to the irregular geometry between refined NMO and a poststack Kirchhoff time migration.
The DMO pass collapses the conflicting-dip smear, and the migrated section resolves a faulted four-way closure that the legacy stack had hidden. The improved image lets the operator rank the parcel confidently and commit to a CAD 9 million horizontal Montney well on the reprocessed structure, a decision the original sections could not have supported.