AVOIDING SEISMIC PITFALLS IN THE U.S. NORTHEAST

Sept. 17, 1990
James R. Morris Quaker State Corp. Titusville, Pa. Seismic reflection profiling has become a significant part of most exploration programs, with a growing number of geologists, and sometimes petroleum engineers, responsible for the acquisition, processing, and interpretation of seismic data. The purpose of this article is to identify, discuss, and provide solutions to the most common mistakes that I have made and have seen others make during my last 10 years working in the U.S. Northeast. My
James R. Morris
Quaker State Corp.
Titusville, Pa.

Seismic reflection profiling has become a significant part of most exploration programs, with a growing number of geologists, and sometimes petroleum engineers, responsible for the acquisition, processing, and interpretation of seismic data. The purpose of this article is to identify, discuss, and provide solutions to the most common mistakes that I have made and have seen others make during my last 10 years working in the U.S. Northeast. My goal is to help you avoid making these same mistakes.

ACQUISITION

The most common pitfall in seismic acquisition involves shooting a line of insufficient length. As discussed by L.R. Denham in a 1984 paper, this can lead to a distorted structural picture of the subsurface, additional multiples, and low signal to noise ratios.1 These errors occur when the line is not sufficiently extended beyond the areal zone of interest. Proper imaging of dipping events, faults, pinchouts, reef edges, etc., inside our area of interest requires source-receiver positioning outside our area of interest (Fig. 1).

Inevitable line end effects of low fold and loss of longer offsets give rise to automatic static solution errors, insufficient multiple attenuation, and low signal to noise ratios. To avoid these effects in our zone of interest, Denham developed these rules of thumb: "First, establish the edge of your area of interest. Second, extend your seismic line past this point a distance equal to your target depth. Third, add a tail of source or receiver points as long as half the far trace offset."

Line orientation has a very important role in controlling the degree of sideswipe contamination or displacement of depth points to the side of the line of profile. This displacement becomes more severe with steeper dips and deeper depths. 2D migration of lines having sideswipe will produce an equally erroneous profile and a false sense of accuracy. To avoid sideswipe we must determine the regional dip and structural and fault trends of our target zone then do our best to orient our lines parallel to dip and perpendicular to structure. In areas of extreme structural relief and tight folds, strike lines should be avoided altogether.

Proper choice of line location and energy source used in field acquisition can greatly influence the quality of our seismic section. Glacial till and unconsolidated valley fill, very common in the Northeast, will scatter and absorb our shot energy, leaving us with total washouts or weak reflections that are pulled down by the low velocity material. For example, crossing the Allegheny River Valley in Cattaraugus County, N.Y., can result in a change of surface velocity from 13,000 fps to 7,000 fps. This would produce a 13 ms reflection sag per 100 ft of valley fill.

Many of our Vibroseis lines follow river or abandoned glacial valleys, resulting in continuous problems along the entire length of the lines. To make matters worse, our river valleys usually meander, giving us a crooked seismic line with associated common depth point smear, which deteriorates our spacial resolution and stack fold. To avoid these problems, we must keep our lines as straight as possible and away from glacial till or valley fill areas. This often requires cross-country dynamite shooting, which can cost 50-100% more than Vibroseis. Although the cost seems high, it is well worth the price to have increased spacial and temporal resolution, as well as improved signal to noise ratio and structural integrity.

PROCESSING

One of the most common errors made in processing Northeast seismic data is underestimating the velocity used to make surface to datum corrections. Processors usually choose velocities in the range of 10,000-12,000 fps for this correction. Problems can arise when there are elevation changes along the seismic line. For example, if our seismic line runs down a linear slope and we wish to correct our seismic travel times to a flat datum at the base of the slope, we simply multiply our elevation by two and divide by the average rock velocity. This will give us a time correction to subtract at each seismic station on the slope. If we choose a velocity of 12,000 fps when the actual velocity is 14,000 fps, our subsurface reflectors will have an erroneous dip of 2.4 ms/100 ft of elevation change along the seismic line, which will subparallel the surface elevation profile.

It's not uncommon to have elevation changes up to 1,000 ft along seismic profiles, such as those shot over state forest tracts in North Central Pennsylvania. From the above mode, this would give us 24 ms of erroneous dip, easily misinterpreted as structure.

Similar static anomalies occur when we shoot our lines across or along valley fill or glacial till zones, which give us a relative slowdown of the near surface velocity described earlier in the acquisition pitfalls. This results in subsurface reflectors that are easily misinterpreted as synclines. Any abrupt change in near surface conditions can result in static "busts" or "leg jumps" (misaligned reflections), which can be misinterpreted as faults.

The best way to avoid these near surface velocity pitfalls is to always run a full suite of static programs. This would start with a refraction statics routine that is based on absolute first break arrival times. This will determine the near surface velocity and remove long wavelength anomalies (longer than a spread-length).

Follow this with a reflection-based surface consistent statics correction to remove any intermediate and short wavelength static shifts. Since static anomalies are by definition time invariant or independent of time, we should always be suspicious of vertically stacked structures or vertical faults, especially when they occur beneath surface features such as hills, valleys, rivers, swamps, etc.

Beyond the primary processing steps of deconvolution, stacking, and migration, many secondary processing techniques can be applied to improve the quality of the data. Like most things in life, a process that enhances one aspect of the data will often degrade another. For example, if our seismic objective is to locate collapsed Trenton-Black River dolomite zones, we would probably look for a structural sag at the Trenton reflector, boundary faults, and amplitude dimming. Secondary processes like F-K filtering and coherency filtering, while attenuating coherent noise and improving reflector continuity, would enhance horizontal smearing of the data, reducing structural sag, smoothing through boundary faults, and mixing amplitudes. Because the choice of processing flow and parameter selection is very subjective, it is always important that we give our processing contractor as much information as possible. This would include major reflector identification, best undisturbed zones for residual statics and deconvolution windows, and most important, target zones of interest with their expected seismic signatures.

INTERPRETATION

The most notorious pitfall made in seismic interpretation would be misidentifying reflecting horizons. For example, if our target is fractured Oriskany on the high side of a fault block and we mistake the Tully reflector for the Onondaga, we could easily drill the Onondaga on the low block (Fig. 3).

Even more embarrassing is misidentifying a thickened salt section as an Onondaga reef, which has happened on several occasions in New York and Pennsylvania. Since seismic reflectors are usually a composite of several acoustic impedance contrasts which can change over relatively short distances, it is always in our best interests to compare surface seismic data to synthetic seismograms.

These can be made from sonic logs; synthetic sonic logs constructed from gamma, neutron, and density logs; or corridor stacks from upgoing vertical seismic profile wavefields.

Many seismic prospects have ended in failure due to unrestrained interpretation scenarios which were not internally consistent or geologically reasonable. For instance, by integrating known geologic structural styles, gravity maps, magnetic maps, and well log data, certain stratigraphic and structural restrictions will be evident. By interpreting the entire seismic section, instead of a single horizon, the cause and effect relationships existing between the deep and shallow section will put additional constraints on the interpretation (Fig. 4).

With complex structural settings, seismic sections should be checked to see if bedlengths are consistent from one horizon to another horizon.

For example, within the large salt cored anticlines of the Allegheny Plateau, the Onondaga limestone is sometimes more complexly folded and faulted than the overlying Tully Limestone. The interpretation is especially difficult if the Onondaga is stacked. Ensuring total bedlengths for the Tully and Onondaga are approximately equal from syncline to syncline will reduce the possible interpretations.

With few of us being experts in all phases of oil and gas exploration, a synergetic approach that integrates geology, geophysics, and engineering will always enhance our exploration success rate.

REFERENCE

  1. Denham, L.R., "Line Length, The Neglected Parameter," Geophysics: The Leading Edge of Exploration, August 1984.

Copyright 1990 Oil & Gas Journal. All Rights Reserved.