Reservoir characterization

Innovative Processing of 3D Land-Seismic Data

The Risha concession, in the desert of eastern Jordan (bordering Iraq, Syria, and Saudi Arabia) contains the partially developed Risha gas field.

The Risha concession, in the desert of eastern Jordan (bordering Iraq, Syria, and Saudi Arabia) contains the partially developed Risha gas field. This has historically been a poor-seismic-data-quality area, mostly because of a complex and variably karstified near-surface region. A high-density wide-azimuth (WAZ) acquisition was carried out in a timely and cost-efficient way, allowing insights into the area’s lithology and potential fracture distribution for the first time.

Seismic Acquisition

A limited exploration and appraisal license period of up to 5 years required that BP acquire 5000 km2 of 3D-seismic data—a very large survey by onshore standards—faster and more cost-effectively than ever before. These goals would mean that BP Jordan would have to achieve production rates and costs per area typical of marine-seismic surveys.

The survey objective was to acquire a high-fold WAZ 3D survey suitable for structural interpretation, attribute analysis, and anisotropy studies. Data quality in Risha is affected by a complex near-surface region with significant karstification, causing a high degree of scattered noise that masks the primary reflection data. Previous 3D vintages were challenging to interpret; they failed to image the Ordovician reservoir adequately, were characterized by poor signal/noise ratio, and lacked resolution and continuity of key reflectors.

To overcome the challenge, 3D-survey parameters were chosen to maximize both quality and productivity. To illuminate the subsurface with a high-fold WAZ ray-path distribution, the design used a 22-live-line recording patch, with 600-m receiver-line spacing. The nominal receiver line was 27.5 km long, resulting in 12,100 traces recorded with each vibrator point (VP), using an active spread spanning from 350 to more than 490 km2. The two-way time of the targets allowed for a relatively wide receiver-line spacing so that the recording patch could sample the entire reflected wave front. Use of single line roll ensured that 11 lines were active on either side of the source swath. Implementing a single roll maximizes both the geophysical quality and the line-crew efficiency, providing full azimuth sampling, maximum surface consistency for statics, and good line-crew efficiency. Receiver groups being spaced 50 m apart with a simple linear array of six geophones was intended to further assist line-crew productivity. Source lines were positioned 50 m apart, parallel to receiver lines with a 50-m VP interval, creating both long source lines to maximize vibrator efficiency and a fully sampled source grid (50×50 m) to attenuate scattered noise and create a high-fold data set.

The long receiver lines (up to 39.5 km) allowed for acquisition of the concession in only two panels, which minimizes repeat VPs (Fig. 1). The use of long source lines extending 7 km outside of the receiver area not only increased vibrator efficiency but also provided increased source separation of the simultaneous distance-separated simultaneous sweeping (DS3) vibrators and interference-free shot records at the zone of interest.

jpt-2014-03-innovativeprocessfig1.jpg
Fig. 1—Acquisition panels and areas of interest (AOIs).

 

Acquisition began in December 2010 with an 18,000-channel recording system and 24 80,000‑lbm vibrator units. Peak acquisition rates reached 1,060 VP/hr. On its best day, the crew acquired 23,851 VPs in a single 24‑hour period on a flat gravel plain, which equates to 59 km2/d. After ramp up, the crew sustained a rate of 18,170 VP/d for the final 50% of the approximately 5000-km2 survey area, which resulted in an average acquisition rate of 45 km2/d. The entire survey was acquired in 186 days, including standby, camp moves, and proactive safety days.

Seismic Processing

The schedule was broadly divided into three phases:

  1. Initial “quick-look” post-stack migrations, by three phases determined by priority of area of interest (AOI) 
  2. A suite of fast-track prestack time-migration (PreSTM) products, also to include initial amplitude vs. offset (AVO) and WAZ products 
  3. A full suite of final prestack time migrations (enhanced noise reduction/refined velocity field) 

This allowed early findings from the survey to be incorporated into well-planning activities. The full-area fast-track PreSTM products were delivered 4 months after receipt of tapes from the field.

Some of the key steps in the processing sequence are discussed in the following subsections. For a discussion of other key steps, please see the complete paper.

Statics. The statics application included the testing of refraction and tomography statics against elevation statics; the results were comparable, and a decision was made to use elevation statics as the simpler solution. The results were also compared with pre-existing uphole measurements, which were incorporated to refine the long period component of the statics model.

Noise Attenuation. The most challenging aspect of the processing project was noise attenuation. Fig. 2 illustrates the high noise levels of the seismic field data. The highest component of the noise is related to seismic scattering. The scattering noise in this area has a much higher amplitude level than the ground-roll noise. Several noise-attenuation methods were used in the data-preprocessing phase of the project. High-amplitude noise bursts and spikes were attenuated using statistical methods that discriminate data on the basis of amplitude and frequency content. For the coherent-noise attenuation, 3D FK and surface-wave-inversion methods were used. These methods provided a good level of noise attenuation, but additional methods were required to handle the random noise. Particular attention was paid to the amplitude fidelity of the noise-attenuation algorithms. Fig. 2b shows the data after application of the noise-reduction process. The flat-lying events are synthetic primary reflectors with known properties, added to the record before noise attenuation and used during the analysis to accurately track the impact of the different filters and their order of application.

jpt-2014-03-innovativeprocessfig2.jpg
Fig. 2—(a) Prestack data before preprocessing. Real noise with addition of synthetic primary (masked) to aid in establishing optimal noise-reduction process. (b) Prestack data after preprocessing. Synthetic primary is now visible.

Seismic Analysis and Interpretation

The phased approach to acquisition and processing meant that an initial post-stack time migration (for the first of three AOIs) was available well before acquisition was complete. The full data area was available 4 weeks from receipt of field tapes; a processing route had been established and applied to preceding AOIs such that the last phase was merely incremental. Using a simplified, extrapolated stacking velocity field also aided in the rapid turnaround of the initial full-area volume.

Coherency. Because of the very low bedding dips, typically less than 1°, the compromise of an initial post-stack time migration (rather than PreSTM) was acceptable and provided a robust initial image; even this preliminary product gave a view-changing first look at the geology. Coherency volumes computed from this product highlighted faults and stratigraphic features not previously identified on heritage 2D or 3D seismic. Initial amplitudes at horizons of interest also appeared highly geological, very clearly delineating the structural grain but also suggesting the possibility of showing a reservoir depositional geometry.

This allowed early survey findings to be incorporated into well-planning activities, and even the earliest products were valuable in this respect. Data quality continued to improve through the processing phases described previously and, in particular, through the range of noise-reduction techniques established for the data.

Knowledge of faulting is critical to well placement in terms of both drilling risk and potential reservoir performance. Faults, even of very small displacement, can now be located and also characterized in terms of structural style and orientation with respect to the local stress direction, and thus inferences regarding the likelihood of open-fracture presence can be drawn.

Spectral Decomposition. The new seismic has revealed the presence of a complex channel system, initially ill-defined on the earliest seismic versions but increasingly apparent on coherency and a range of amplitude attributes as the data improved through noise-attenuation processing. For a given two-way-time window of reflectivity seismic, spectral decomposition analyzes the contribution of single frequencies from the full range of frequencies in the data. Amplitudes at given frequencies respond to impedance changes in the lithologies sampled, but also (because of tuning and interference effects) to variations in thicknesses of those lithologies. In the case of channels, because of rapid thickness change from center to margin, particular frequencies “tune” to different parts of the channel profile, often providing an illuminating view of the depositional system, particularly when the frequencies are scanned or selected to enhance the specific features dependent on their geometry. The impact of these channels on prospectivity is now being assessed.

Rock Properties and AVO. A rock-properties study was carried out to explore the possibility of calibrating AVO seismic to enhance particular interfaces or reservoir characteristics. Because there was an indication from well data and the initial seismic that the reservoir may not be distributed uniformly, the possibility of being able to “tune” the seismic to image reservoir required investigation.

The extensive noise-reduction processing already discussed cleans the data sufficiently to allow derivation of seismic AVO gradient and zero-offset reflectivity, which can be simply inverted to acoustic impedance (AI) and gradient impedance (GI) by a “colored-inversion” technique that shapes seismic reflectivity traces to match the impedance spectrum observed in well control within the seismic bandwidth. Note that the bandwidth limitation means that impedances are relative, not absolute.

Thus, AI and GI seismic traces can be mixed in the appropriate proportions to provide a seismic volume of any extended elastic impedance (EEI) equivalent to different angle projections in AI/GI space. Different projections separate sands and shales to varying degrees and also potentially collapse large ranges of impedances to a much narrower range of EEIs. For example, projecting to approximately 50° in this space provides a separation between sands and shales and also collapses each linear cloud of data to a more restricted range. The result is a seismic volume that represents the sand as low relative impedance (red trough with our color convention) and also reduces the ambiguity caused by different shale properties at the interface and internally within the gross sand section. Elsewhere, in the absence of the reservoir sand, the equivalent interface marks the onset of high relative impedance (black peak).

WAZ and Anisotropy. The DS3 geometry results in the full range of azimuths being recorded, thus allowing the investigation of azimuthal velocity anisotropy in the area. At each common depth point, the azimuthal velocity field is separated into V-fast and V-slow and separate velocity volumes are created for each. The percentage anisotropy is calculated and converted to interval anisotropy. Windowed amplitude extractions from this volume therefore inform on azimuthal velocity anisotropy in any defined time window.

Considerable variation in anisotropy was observed over the license area, and, when compared with both the known stress orientation and structural features from coherency, a broad correlation was observed between higher anisotropy and fault orientations where associated fractures are aligned with the maximum-horizontal-stress direction. The possibility therefore exists that the anisotropy parameter will delineate areas of higher reservoir productivity.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 164016, “Innovative Acquisition and Processing of Low Signal-to-Noise-Ratio 3D Land-Seismic Data Lead to a Leap in Subsurface Understanding: A Case History From Eastern Jordan,” by Chris Pearse, Nicola Adams, Philip Bateman, Jack Bouska, Carlos Duque, Martyn Gravestock, and Johnathan Stone, BP, prepared for the 2013 SPE Middle East Unconventional Gas Conference and Exhibition, Muscat, Oman, 28–30 January. The paper has not been peer reviewed.