coherence vector map

A coherence vector map in 3D seismic interpretation is a directional attribute display derived from a coherence volume that shows not only the magnitude of lateral waveform dissimilarity at each sample location in the seismic cube but also the azimuth and dip of the discontinuity surface causing the coherence anomaly, rendered as arrows or vectors on a time slice or horizon extraction where arrow direction indicates fault or fracture strike perpendicular to the vector and arrow length indicates the strength of the discontinuity; the vectors are computed by taking the spatial gradient of the coherence volume at each sample, which points in the direction of steepest coherence decrease and aligns perpendicular to the local fault plane if a fault is present within the analysis aperture, or by computing the principal eigenvector of the covariance matrix of trace-to-trace waveform differences in a local neighborhood (a more statistically robust approach using the dominant eigenvalue direction). In the Western Canada Sedimentary Basin, coherence vector maps are used primarily in two geological contexts: fracture network characterization in Montney siltstone, Duvernay shale, and Muskwa-Otter Park Formation unconventional plays in west-central Alberta and northeast British Columbia, where the NE-SW maximum horizontal stress direction drives predominantly NE-SW-trending natural fracture systems and coherence vector maps provide quantitative rose diagram statistics on fracture azimuth used to orient horizontal wells perpendicular to the dominant fracture population for maximum fracture intersection; and structural mapping in WCSB Foothills thrust-fold belt programs covering the Rocky Mountain Trend and Grande Prairie areas, where faults dipping at 20 to 60 degrees with strikes that vary spatially across thrust sheets require the directional information of the vector map to distinguish conjugate fault sets, separate thrust and back-thrust families, and determine fault kinematics from dip direction. WCSB coherence vector map computation is performed in Petrel, Kingdom, and OpendTect using C3 eigenstructure coherence as the preferred input volume because C3 provides the most noise-resistant coherence anomalies from which to compute spatial gradients, and WCSB 3D surveys of 100 to 5,000 km2 over Montney and Duvernay plays produce vector maps summarized as rose diagrams to quantify fracture orientation for horizontal well trajectory selection and completion design.

• Coherence vector computation methods and the eigenstructure approach in WCSB 3D seismic programs: Three methods compute the coherence vector at each sample point in a seismic volume: finite-difference gradient (simple, fast, sensitive to noise), principal eigenvector of the local trace-difference covariance matrix (more robust, used in WCSB Montney programs where signal-to-noise ratios of 3 to 8 dB would alias a finite-difference approach), and gradient structure tensor methods that average the spatial gradient over a small neighborhood before eigendecomposition (intermediate accuracy and cost). The eigenvector approach constructs a 3x3 covariance matrix from the inline gradient, crossline gradient, and time gradient of the seismic amplitude field in a local window of 3x3 to 5x5 traces and 8 to 16 ms; the eigenvector corresponding to the largest eigenvalue points in the direction of maximum waveform dissimilarity, which aligns perpendicular to the fault plane if a fault dominates the local neighborhood. In WCSB Foothills programs where dips reach 30 to 60 degrees and coherence anomalies must be extracted from steeply inclined reflections, the eigenvector approach tolerates larger dip-estimation errors than the finite-difference gradient because the covariance matrix eigenstructure is invariant to moderate rotations of the coordinate frame, allowing accurate vector computation even when the dip-steering used for coherence filtering has residual errors of 5 to 10 degrees.
• Coherence vector map display conventions, rose diagram analysis, and fracture azimuth statistics in WCSB programs: Coherence vector maps are displayed as arrow overlays on coherence amplitude slices, with arrow length scaled to coherence anomaly magnitude after applying a threshold to suppress vectors below the noise floor (eigenvalue magnitude less than 0.05 to 0.10 of the maximum in WCSB data sets), and arrow direction plotted perpendicular to inferred fault or fracture strike so that a NE-SW striking fracture zone produces NW-SE pointing arrows. Rose diagrams of vector azimuths computed over geological domains (individual WCSB Montney drainage units, Duvernay wells pads, Foothills thrust blocks) quantify the statistical distribution of fracture orientations and are the primary deliverable from coherence vector map analysis for horizontal well planning; WCSB Duvernay shale programs in the Kaybob, Edson, and Fox Creek areas show a dominant NE-SW fracture population (azimuth 030 to 060 degrees) confirmed by coherence vector rose diagrams that is consistent with the regional maximum horizontal stress direction (SHmax 045 to 060 degrees across most of west-central Alberta), validating the vector map as a reliable fracture orientation indicator. Conjugate fault systems produce two distinct azimuthal populations in rose diagrams separated by 60 to 90 degrees, enabling statistical separation of fault families in WCSB Foothills programs that cannot be distinguished on a standard scalar coherence map where both fault sets appear as coherence lows regardless of orientation.
• Coherence vector maps for Montney and Duvernay fracture corridor characterization in WCSB horizontal well programs: In WCSB Montney siltstone programs at Groundbirch, Dawson, Progress, and Tower in northeast British Columbia, coherence vector maps are used to identify NE-SW trending fracture corridors at 50 to 500 m spacing that produce both a coherence low (from the impedance contrast between the fracture zone and intact matrix) and a consistent vector azimuth population (NE-pointing arrows perpendicular to NE-SW fracture strike), providing higher-confidence fracture identification than coherence amplitude alone; wells drilled with horizontal trajectories oriented perpendicular to the dominant coherence vector azimuth (east-west laterals crossing NE-SW fractures) show 30 to 50 percent higher 12-month cumulative production than wells drilled parallel to the fracture corridors (north-south laterals), demonstrating the commercial value of vector map-based trajectory selection at the Montney scale. In WCSB Duvernay programs, coherence vector maps extracted from 3D seismic at 3,200 to 3,800 m depth with dominant frequencies of 25 to 40 Hz provide fracture orientation data at lateral scales of 100 to 400 m that complement the meter-scale fracture information from borehole image logs (FMI, OBMI) at specific well locations, enabling fracture orientation mapping across the inter-well space that is unavailable from well data alone and allowing completion design teams to adjust perforation cluster spacing and hydraulic fracture stage count based on the seismically-imaged fracture density.
• Vertical resolution limits and sub-seismic fracture detection thresholds for WCSB coherence vector applications: The vertical resolution of coherence vector maps is controlled by the dominant seismic wavelength at the target depth, typically one-quarter wavelength, which corresponds to 10 to 30 m at WCSB Montney and Duvernay target depths of 1,500 to 4,000 m with dominant frequencies of 25 to 50 Hz; sub-seismic fractures with individual apertures below 1 to 5 mm are not resolved individually but can produce a detectable coherence vector anomaly if the fracture density within a seismic resolution cell (12.5 to 25 m inline by 12.5 to 25 m crossline by 10 to 30 m vertically) is high enough to alter the bulk seismic impedance by more than approximately 0.5 to 1.0 percent from the unfractured matrix. Fracture swarms (corridors of 10 to 100 closely spaced fractures within a zone 1 to 10 m wide) are detectable on WCSB Montney and Duvernay coherence vector maps when the swarm width exceeds approximately 15 to 25 m, while isolated fractures below approximately 10 m width are generally below the detection threshold at typical WCSB seismic signal-to-noise ratios of 3 to 8 dB and require borehole image log confirmation; this resolution limit is communicated to WCSB horizontal well planning teams as a minimum fracture corridor width threshold below which coherence vector map predictions are unreliable and microseismic or production logging data from offset wells must be used instead.
• Coherence vector maps in WCSB structural interpretation and Foothills fault kinematics: In the WCSB Foothills thrust belt from the Rocky Mountain Trend in southern Alberta through the Deep Basin and Grande Prairie area in northwest Alberta, coherence vector maps computed from 3D seismic surveys help distinguish thrust faults (shallow east-dipping planes, vector arrows pointing east perpendicular to north-south strike) from back-thrusts (steeper west-dipping planes, vector arrows pointing west), conjugate strike-slip faults (diagonal vectors crossing the main thrust trend), and tear faults (vectors oriented along the thrust transport direction, perpendicular to the main fault strike). WCSB Foothills programs at Deep Basin Cretaceous and Jurassic plays (depths of 2,000 to 6,000 m) use coherence vector maps to identify fault relay ramps between overlapping thrust segments, which represent zones of elevated fracture intensity where permeability is enhanced relative to the footwall and hanging wall blocks; production data from WCSB Foothills wells at known relay ramp positions (identified by convergence and crossing of vector arrows from two adjacent thrust fault segments) confirm 20 to 40 percent higher initial gas rates compared to equivalent wells in unfaulted portions of the same formation, supporting the use of coherence vector maps to rank and sequence drilling targets in structurally complex WCSB Foothills programs.

Coherence Vector Map Guiding Duvernay Fracture Corridor Targeting at Kaybob

A WCSB Duvernay operator in the Kaybob area of west-central Alberta extracted a coherence vector map from a 280 km2 3D seismic survey at the Upper Duvernay interval at 3,500 m depth. C3 eigenstructure coherence was computed using a 5x5 trace aperture and 16 ms window. Vector arrows were computed by eigendecomposition of the local gradient covariance matrix and displayed with a length threshold corresponding to eigenvalue magnitude above 0.08. Rose diagram analysis over four drainage units identified a dominant NE-SW fracture population (azimuth 042 degrees, 65 percent of vectors) and a minor NW-SE conjugate (azimuth 315 degrees, 22 percent). Four horizontal wells were drilled east-west (perpendicular to the 042-degree fracture corridors imaged by the vector map) while two offset wells drilled north-south served as controls. After 12 months, east-west wells averaged 2,800 m3 condensate equivalent versus 1,550 m3 for the north-south wells, a 81 percent production difference consistent with the fracture intersection model derived from the coherence vector map azimuth statistics.

Related Terms

Coherence map is the scalar parent display from which coherence vector maps are derived; the coherence map shows where discontinuities exist while the vector map shows their azimuth and dip orientation. Coherence attribute computation (C1, C2, C3) provides the input volume; C3 eigenstructure is preferred for WCSB Montney and Duvernay vector maps due to superior noise rejection producing cleaner spatial gradients. Fracture corridor characterization in WCSB Montney and Duvernay uses coherence vector rose diagram statistics to orient horizontal wells; east-west laterals perpendicular to the NE-SW fracture population produce 30-50% more than north-south wells at Groundbirch and Kaybob. Seismic attribute workflows combine coherence vector maps with curvature and spectral decomposition to confirm fracture corridors with multiple independent indicators. Discrete fracture network (DFN) models for WCSB Montney and Duvernay reservoir simulation use coherence vector azimuth statistics as the primary fracture orientation input, supplemented by borehole image logs.