3D Seismic in Areas of Poor Data and Complex Geology of South Oman.
This paper describes the problems of 3D seismic data in the complex geology of South Oman fields. Weak reflectors, strong multiples and ground roll result in poor seismic. However, 3D seismic has opened a wealth of hidden information. Amplitude maps (Marmul field) have revealed areas of the Gharif reservoir not previously recognized. In the more complex Al Khlata reservoir, seismic modeling indicates that sand/diamictite interfaces can locally give a reflector. This was successfully tested by a horizontal sidetrack and followed by three horizontal wells. In addition, gross seismic character (SW Marmul area) could be related to fluid movements in the reservoir.
The geology of South Oman fields varies from the more predictable Gharif and Haima reservoirs to the more complex periglacial Al Khalta reservoirs. A simplified sketch (figure 1) shows the 3 reservoirs from a typical South Oman Field, sealed by the Nahr Umr shale and overlain by a thick layer of the cretaceous Natih carbonate.
Interpretation of seismic data has been hampered by the generally poor data at reservoir level. Strong interbed multiples (figure 2) obscure primary events. Scattered ground roll confuses the interpretation of steeply dipping reflectors. The old 2D seismic was used with difficulty for structural interpretation at reservoir levels and maps were mainly based on well control. The coming of 3D seismic was a big step forward not so much in data quality but in the ability to view data in different directions. Weak events submerged in noisy data are effectively captured by steering through 3D data. 3D seismic surveys in South Oman started in 1989 and now cover nearly all the fields.
Good continuous events define the Natih carbonate interval above the Nahr Umr shale. However, seismic data quality of the Gharif/Al Khlata reservoirs is variable. Figure 3 shows 3 data vantages across 3 South Oman fields (Thuleilat Marmul and Umr). The Thuleilat data (figure 2a) is poor partly because of the low acquisition fold (5-6 traces/ bin at the reservoir level).
Acquisition at higher fold is not necessarily the solution. Imaging of Al Khlata paleovalleys illustrate this point. The Marmul paleovalley (figure 3b) is seismically visible (32 traces lbm), in contrast to the Nimr paleovalley (96 traces/bin - figure 3c). Thick layers of Al Khlata sands and diamictites fill the Marmul paleovalley reflecting strong seismic events. A geological model of the Marmul paleovalley lined with a hard (high acoustic impedance) basal diamictite (figure 4a) generates a seismic response (figure 4b) which matches the observed seismic profile (figure 3b). Absence of the basal diamictite (possibly a Nimr type paleovalley) shows a poor match with the seismic (figure 4c).
Near surface problems, low fold of shallow data. shooting skips over surface facilities can result in 3D data artifacts. However, the main source of interpretation pitfalls has been strong interbed multiples interfering with primary events. Normal stacking of traces will not remove interbed multiples. The velocity function (figure 2a) between the intra-Natih reflectors (i.e. the multiple generators) and the reservoir intervals, is nearly constant.
Image processing has however improved multiple suppression. The method depends on pattern recognition, where multiple generating horizons (the intra-Natih reflectors) are flattened together with the rest of the data. Multiples appear as parallel events and a filter is designed to remove them. However, the method is not effective when multiple generators are divergent. Over the eastern half of the section (figure 5) multiples are successfully suppressed. However, over the western half of the section, the intra-Natih reflectors cascade divergent multiples which are not removed. A multiple was interpreted as the Top Haima event leading to an incorrect prognosis of MM-251.
Capturing steep reflectors can be a problem where scattered ground roll occurs. Source and geophone arrays will not effectively attenuate ground roll scattered from shallow wadi deposits. It is also not possible to optimize source and geophone arrays under variable near surface conditions.