In this abstract, we highlight the performance of PreStack Depth Migration (PSDM) on an ION GeoVentures multi-client survey from northwest Pennsylvania, the GroundhogSCAN 3D. The survey area of interest and a representative crossline are shown in Figure 1. Compressional tectonics during the Appalachian Orogeny shortened the geology above the Silurian-age Salina Salt; here the salt acted as a detachment surface. The overlying Devonian-age Marcellus Shale is heavily faulted and folded, but the underlying Ordovician-age Utica Shale appears relatively undeformed.

First, we show how conventional seismic time processing (PSTM) can increase drilling risk by causing false seismic structures and positioning errors, and by blurring faults and steep dips. We show how PSDM overcomes these hurdles. PSDM can also improve the accuracy of azimuthal velocity anisotropy (HTI) analysis, an indicator of differential horizontal stress (and/or natural fractures), by removing false HTI effects caused by lateral velocity variation. We show evidence that we have detected – only by using PSDM – a vertical change in regional horizontal stress across the Salina Salt.

Fig. 01
Figure 1. GroundhogSCAN survey (data courtesy ION Geophysical). The 378 square-mile survey is located east of Pittsburgh, Pennsylvania, in Westmoreland and Armstrong Counties. The results shown in the presentation focus on XL 1452; a PSDM section through XL 1452 shows how the Salina Salt varies in thickness considerably (and in internal composition – see blue circle). These variations in salt thickness drive lateral velocity variations that distort the Marcellus and Utica horizons in time.

PSDM Improves Structural Imaging

PSDM reduces the risk of dropping out of zone during horizontal drilling. Lateral variations in seismic velocity cause distortions on PSTM images. As Figure 2 shows, if velocity anomalies fit between existing well control, a driller might believe that the time structure represents actual geology. While corrections could be made with LWD data, it’s likely that the confusion caused by the false seismic structure would lead to “porpoising” and reduce the well’s productivity. On the other hand, PSDM can, with an accurate velocity model, remove false seismic structures and help keep the drill bit in zone.

Fig. 02
Figure 2. Low velocity anomaly causes time sag (top), which is corrected by PSDM (bottom)
Fig. 02a
Figure 2a. Simple raytracing experiment. Top: near-, mid-, and far-offset rays (sharing a common midpoint) traced from surface to Utica. Bottom: Zoom on the reflection points. Though the Utica Shale dip is only 0.5°, the lateral velocity variations cause over 700 ft of reflection point dispersion.

Lateral velocity variation also blurs the imaging of fault planes and steep dips and causes lateral mispositioning of these key features. Figure 2a illustrates how the reflection points for the traces in a common midpoint gather are actually spread over 700 ft. In this case, when we stack the Utica Shale event on a PSTM offset gather, we’re also stacking across 700 ft of midpoint. We stand the risk of smearing steep dips and faults, as Figures 3-5 show. PSDM overcomes this deficiency.

Fig. 03
Figure 3. XL 1452 zoom above the Marcellus. Left: PSTM. Right: PSDM. PSDM shows improved fault focussing. Data courtesy ION Geophysical.
Fig. 04
Figure 4. XL 1452 zoom around the Marcellus. Left: PSTM. Right: PSDM. Approximate Marcellus pick is shown in green. Note how the Marcellus horizon appears more geologically “simple” on the PSDM than on the PSTM. Also, note how the deep folded structures – imaged well by PSDM – are distorted and smeared on the PSTM image. Data courtesy ION Geophysical.
Fig. 05
Figure 5. XL 1452 zoom inside the Salina Salt. Left: PSTM. Right: PSDM. PSDM images apparent internal salt thrust structures, probably interbedded clastics. Data courtesy ION Geophysical.

PSDM Improves Azimuthal Anisotropy (HTI) Analysis

Idealized vertical fractures and/or differential horizontal stress are known to cause seismic velocities to vary as a function of azimuth. This is known as azimuthal anisotropy (HTI). By measuring HTI with seismic data (azimuthal traveltime variations), we can infer the presence and direction of the fractures or differential horizontal stress. This has obvious implications for the optimization of hydraulic fracturing.

In theory, vertically-oriented fractures will produce a sinusoidal traveltime pattern versus azimuth. We typically measure two attributes from OVT (Offset-Vector Tile) gathers – “Vfast-Vslow”, an estimate of the magnitude of the HTI effect, and “Vfast azimuth”, the direction of the fast velocity.

However, lateral velocity variation can cause significant false HTI effects to manifest on OVT PSTM gathers, as Figure 6 below shows. An operator might mis-interpret the false HTI effect as a localized zone of fractures and/or mis-interpret the direction of maximum horizontal stress. Luckily, PSDM correctly handles lateral velocity variations, and can remove false HTI effects from OVT PSDM gathers.

Fig. 06
Figure 6. 2.5D isotropic raytracing experiment from West Texas. Left: velocity model with “wedge” of high velocity (orange) corresponding to Central Basin Platform (carbonate), juxtaposed against lower-velocity sediments (green). Black line = experiment location. Center: PSDM image of shelf edge. Raytrace from the surface to the flat red horizon. Right: Raytraced traveltime versus azimuth. Although the target is flat and the model isotropic, we see a significant false 10 ms HTI response.

Figure 7 below displays a map, centered on XL 1452, comparing the RMS Vfast-Vslow HTI attributes at the Marcellus Shale, generated by PSTM and PSDM. Although the calculations are made in exactly the same way, the magnitude, on average, is smaller on the PSDM-derived attribute. This indicates that PSDM has likely removed some false HTI effects, leaving behind the true HTI response.

Fig. 07
Figure 7. RMS Vfast-Vslow (magnitude of HTI effect) analysis at the Marcellus Shale for a 3 square mile area centered on XL 1452. Top: derived from OVT PSTM. Bottom: derived from OVT PSDM.

Originally, most HTI analysis didn’t correctly handle vertical variations in HTI orientation – the calculations were “RMS” in nature. The “Generalized Dix Equation” was introduced in 1999 by Grechka et al., and enables us to derive interval HTI attributes, as Figure 8 illustrates.

Fig. 08
Figure 8. Justification for interval HTI analysis. We performed azimuthal HTI raytracing in a two-layer model (left), with a 45° change in HTI orientation (ϕ). Right: Computed azimuthal traveltime, from the surface to base of bottom layer. The curve forms a sinusoid with a 7.6 ms magnitude. The apparent Vfast azimuth, 65°, is different from the Vfast azimuth of either layer! An interpreter would infer an incorrect fracture orientation. Interval HTI analysis is robust to this issue.

Even when we use interval HTI analysis, the HTI response measured from PSTM will be contaminated by lateral velocity variation. This problem will become acute when we perform HTI analysis on the deep Utica Shale, because of the significant variations in Salina Salt thickness above the Utica. Figure 9 below illustrates interval HTI analysis at the Utica Shale, using both PSTM and PSDM inputs. PSTM provides a muddled view of Vfast azimuth, whereas PSDM provides a far more consistent Vfast azimuth, differs from the shallow Vfast azimuth.

Fig. 09
Figure 9. Interval HTI analysis at Utica Shale on the same 3 square-mile area shown in Figure 6. Each of the three marked subdomains contains about 3,000 analysis points. Therefore, in each subdomain, we have 3,000 measurements of Vfast azimuth, which we can use to construct a rose diagram. Top-left: rose diagrams from PSTM. There doesn’t seem to be any dominant Vfast azimuth direction in the PSTM-derived attribute. Bottom-left: rose diagrams from PSDM. The PSDM-derived Vfast azimuth attribute appears to consistently indicate a NW-to-SE direction. Right: rose diagram displaying the shallow dominant stress orientation for the Marcellus Shale, the orthogonal “J2” fracture direction, and the “Deep SHmax” direction measured at the Utica.


Figure 10 compares the Vfast azimuth direction computed at the Marcellus (green) and at the Utica (blue). In general, Vfast azimuth is believed to align with present-day SHmax. Vfast azimuth from the surface to the Marcellus is, interestingly, perpendicular to the direction of compression. The green arrow aligns with the SHmax from the World Stress Map. Although the Appalachian Orogeny no doubt caused compression of the shallow section, if the shallow section was currently under compressive stress, we’d expect shallow SHmax to be parallel to the compression; instead, it is roughly perpendicular. Intriguingly, the deep Vfast azimuth direction that we measured at the Utica is aligned with the direction of paleo compression.

Fig. 10
Figure 10. World Stress Map around Pennsylvania. Yellow line illustrates trace of Appalachian Mountains. The state of Pennsylvania and the GroundhogSCAN survey are outlined in red. Green arrow illustrates regional SHmax at the Marcellus. Blue arrow illustrates the deep SHmax measured from interval HTI analysis at the Utica.


We demonstrated that PSDM has significant utility in the Western Pennsylvania Shale Basin, although the technology has not (to date) been frequently utilized. PSTM appears to yield a “good image” in most areas, but there are subtle-but-crucial differences in the structural image quality (faults and steep dips), caused by lateral velocity variations. Additionally, PSDM appears to “simplify” local structures around the Marcellus Shale that are horizontal drilling targets.

Secondly, we demonstrated how PSDM provided a superior input for HTI analysis. By removing false HTI effects from OVT gathers, PSDM produces more reliable azimuthal attributes. Intriguingly, by utilizing interval PSDM HTI analysis at the Utica Shale, we detected a consistent Vfast azimuth direction (not interpretable on the PSTM attributes) which differed from the shallow Vfast azimuth. From available well data, we observe a change in the differential stress across the salt, so it is reasonable to see a stress rotation as well. For what it is worth, the measured deep Vfast azimuth is perpendicular to the trace of the Appalachian Mountains, and presumably parallel to the paleo SHmax direction.



We thank the management of NEOS for permission to publish, and to ION Geophysical for the right to show the data examples. We also thank for their valuable feedback many colleagues at NEOS, as well as countless attendees of geophysical luncheons where this work has been previously presented.

About the Author(s)

Dr. Morgan Brown holds degrees in applied mathematics (BA, 1997) from Rice University and in geophysics (PhD, 2004) from Stanford University. He worked in geophysical R&D at Hess Oil and 3DGeo, before joining a depth imaging startup, Wave Imaging Technology Inc.. He served as CEO from 2008 to the company’s sale in 2013 to GeoCenter. After consulting for two years, he recently joined the NEOS Seismic Imaging Group in Denver as a Geophysical Advisor in Depth Imaging.



Join the Conversation

Interested in starting, or contributing to a conversation about an article or issue of the RECORDER? Join our CSEG LinkedIn Group.

Share This Article