As the adage says, you can't compare apples and oranges. Recording the right baseline and monitor data is vital in production monitoring. Safeguarding repeatability is equally important during seismic processing.
What we track is the 4D difference, obtained by subtracting the baseline data from the monitor data. Even if we have perfectly repeated acquisition parameters, there may still be ‘4D noise’ or ‘non-repeatability’ in the result. Seismic processing workflows can mitigate dynamic causes of the non-repeatable signal. Where variances are unavoidable, they should be unambiguous. Accuracy is not enough, answers must be delivered on time for maximum value, and a robust 4D toolkit must be able to deal with different types of data.
Time-Lapse is Time Critical
As there is often a looming intervention deadline, and the meaningfulness of the 4D information decays as production continues, 4D seismic processing products are often time-critical. Accelerated ‘fast-track’ 4D products, using abbreviated workflows are often delivered within weeks of the last shot, followed later by ‘full integrity’ 4D deliverables.
QC is the Key to 4D Repeatability
Quality control (QC) is essential as a 4D project progresses, highlighting the factors causing non-repeatability of either acquisition or processing, and used to guide future processing steps.
The direct 4D difference is a useful but subjective QC. A number of quantitative QC attributes are used to categorize the energy seen in the 4D difference. The most common approach is based upon root-mean-square (RMS) amplitudes measured within time windows from the seismic data. The NRMS (normalized RMS) value measured in 4D processing is the RMS amplitude of the difference, normalized by the average of the RMS amplitudes.
Predictability measures, such as signal to distortion (SDR) ratios and signal to noise analysis, are also useful gauges of repeatability. Attributes such as amplitude, phase and time shifts between the baseline and monitor data are also checked using a series of horizons above and at the target reservoir levels. These are repeated at every key stage of the 4D workflow. The statistical distribution of these quantities should converge through the 4D sequence, centering on zero amplitude phase and time shift differences; for all horizons except those affected by reservoir production.
Comparison of the NRMS amplitude difference at a reference horizon for an early and later stage of 4D processing. As expected, the histogram of values (plotted on the right of each panel) is converging to a narrower distribution with lower mean value.
Seismic Methods Vary in the Real World
Baseline and monitor surveys come in many varieties, and processing workflows must be capable of dealing with that reality. In 2018 alone, PGS has processed 4D differences from eight different acquisition scenarios:
- GeoStreamer on GeoStreamer
- multisensor streamer on GeoStreamer
- GeoStreamer on single-sensor flat streamer
- GeoStreamer on single-sensor slanted streamer
- Single-sensor streamer on single-sensor streamer
- GeoStreamer on OBN (ocean bottom node)
- OBN on OBN
- PRM (permanent reservoir monitoring)
Ideally, both monitor and baseline surveys are acquired with reliable and repeatable multisensor data. In the absence of a permanent reservoir monitoring installation, GeoStreamer is the best option. It enables optimal towed streamer repeatability, with P-UP on P-UP 4D processing. Access to the separated wavefields -- the up-going pressure wavefield (P-UP) and the down-going pressure wavefield (P-DWN) -- allows the dynamic and non-repeatable imprint of changes in the sea-surface height (exclusively contained in P-DWN) to be removed from P-UP.
Illustration of GeoStreamer multisensor wavefield separation. P-DWN contains the unwanted dynamic effects of changes in sea-state, inescapably embedded in hydrophone-only streamer data (recorded as P-TOT)
Three 4D Processing Essentials
Dealing with free-surface effects
Dynamic forces during acquisition introduce non-repeatability in 4D processing. Changes in sea-surface height from shot-to-shot affect the local depth of each receiver for each shot and the depth of the air guns for each shot. This contributes to shot-by-shot variations in the emitted source wavefield. After removal of the receiver-side ghost, shot-by-shot source wavefield variations are removed. This is followed by source-side ghost removal, and a correction for source-side time shifts caused by the dynamic changes in source depth.
Accommodating long-period effects and non-repeatable noise contamination
Long-period dynamic effects occur from tidal variations and changes in salinity and water temperature in the water column. These affect primaries and multiples in different ways. 4D processing solutions include both trace-by-trace and spatially-variant corrections such as ‘4D warping’. A variety of man-made and environmental noises impact each survey in different ways. Every cause of non-repeatability is isolated and addressed during 4D processing.
Processing simultaneously for optimal products
Full integrity 4D processing uses ‘simultaneous processing’ of both datasets. Also known as ‘co-processing’, the baseline and monitor datasets are processed together with workflows that are as equal as possible. This ensures the latest technology is deployed to the processing of all data. Additionally, simultaneous 4D processing uses only the traces in each dataset that will contribute to optimal repeatability, an example being ‘4D trace binning and regularization’. For every common midpoint location, the baseline and monitor datasets are analyzed so that every source-receiver trace pair for every offset class have azimuths that are as comparable as possible.
Sometimes More Sophisticated 4D Processing is Required
As for any seismic project, the baseline and monitor data should have optimal image resolution and clarity. While critical, satisfying a set of 4D repeatability metrics is not enough, the 4D difference should be in the right place. It should also compensate for such things as variation in acquisition-related illumination.
The PGS portfolio of imaging solutions has been developed with 4D in mind. 4D reflectivity inversion has been shown to compensate for illumination differences (see figure below) and to recover the full signal bandwidth, when different survey geometries such as towed streamer and ocean bottom seismic are used for 4D.
Illumination for towed streamer (left), OBN (middle), and the cross survey illumination computed for 4D survey matching of towed streamer and OBN data (right).
Conclusion
In short, 4D seismic processing must allow a comparison of data acquired by different methods, or under different conditions. Because, while you can’t compare apples and oranges, sometimes you are dealt tangerines and mandarins. Effective 4D co-processing should be timely, with the staged delivery of 4D products to guide planned interventions. The 4D attributes asset teams rely upon to make production decisions must reflect the reality of the reservoir. The results must be reliable, measured by tailored 4D QC that highlights the healing of non-repeatability issues from the acquisition.
Next time you are faced with a critical decision on an intervention deadline, challenge our highly skilled and fully integrated processing team to see what the innovative PGS 4D toolkit can do to help your field development.
Contact a PGS expert
If you have a question related to our Imaging & Characterization services or would like to request a quotation, please get in touch.