Long-Offset Seismic Aquisition, Processing, and Imaging
Presiding: H Zhou, University of Houston; P L Stoffa, University of Texas at Austin
S52A-01 10:30h
Sub-Basalt Imaging Using Long-Offset Seismic Data
Basalt layers usually have a complex structure with thin layers and large velocity and density differences. They may be inhomogeneous in all directions and may display a positive velocity gradient. Such complex structures create quasi-anisotropy, high- frequency scattering, low-frequency absorption, and refraction. In general, it is not difficult to map the top of a basalt layer because it is a strong reflector. But multiples from the sea bottom and top of the basalt generally mask sub-basalt reflected waves. Multiple attenuation methods do not always provide the low-level of remnants of multiples in case of near-offset data. Sub-basalt reflected and refracted waves at far-offsets are free from the problems of multiples. In addition, reflected waves at far-offsets have high amplitudes because the reflection coefficient of the waves approaches unity, and the strong influence of basalt inhomogeneity is reduced. The long-offset (0-16 km) seismic data were used for sub-basalt imaging at 3-5 km depth. A three step methodology was utilized to construct the image of sub-basalt sediments: 1) getting of the image and velocity model of the low-velocity sediments above basalt (including its top) by processing of the reflected waves at near-offset (0-6 km); 2) construction of the basalt layer model by reflected waves observed at far-offset (6-16 km) with high apparent velocity (about 5 km/s); 3) recognition of the sub-basalt sediments and construction of the sub-basalt seismic image based on first two steps. Besides methodology the results of the sub-basalt imaging using long-offset data are discussed.
S52A-02 10:50h
Maintaining Frequency Content for Long-Offset Seismic Data
A serious drawback with long-offset seismic acquisition is the reduction of frequency content caused by the processing steps of migration and normal moveout. This frequency reduction leads to poor and/or unreliable structural images. With source-receiver offsets greater than three times the target depth, a frequency loss of 100 percent is not uncommon for reflection events after NMO corrections. A processing and interpretation approach has been developed that images all reflections but avoids normal moveout (NMO) stretch. That is, imaged reflections with long offsets have the same frequency content as traces with small offsets. We call this approach Seismic Wide-Angle Processing (SWAP). There are two main steps that are different from conventional processing. The first is a pre-stack Kirchhoff migration in the common-offset domain that maintains the offset separation between source and receiver during migration. That is, the process of NMO is removed from migration. Obliquity and aperture factors different from conventional Kirchhoff migration are introduced to maintain true amplitude. The second step is an NMO correction that is centered on the target horizon. Static shifts based on ray-trace traveltimes to the target horizon provide the NMO corrections; a process often referred to as block NMO. Even though anisotropy is applied in the processing stages, a final adjustment of traveltimes is required to handle lateral velocity changes. Only the target horizon is truly flat in the CMP gathers after applying the NMO static shifts. Depending upon the velocity gradient around the target horizon, interpretation is limited to a time window of approximately 50-200 ms around the target horizon. Initial tests indicate that both structural and amplitude interpretations are improved using data that have offsets greater than twice the depth of the target horizon.
S52A-03 11:10h
Long-offset detection of offshore airguns by onshore broadband seismographs
In aseismic areas such as the coastal regions of the Gulf of Mexico, using onshore seismographs to detect signals from airgun sources during oil industry offshore seismic surveys offers a great potential to yield a large amount of long-offset data. The feasibility of onshore seismographs to detect marine airgun sources has been demonstrated by the LARSE surveys in southern California and the SIGHT survey in New Zealand. These surveys used a 8470-inch3 airgun array of the R/V Maurice Ewing. However, a much smaller airgun array volume, about 3400 inch3 or less, is typically used in today's industry marine reflection surveys. It is also foreseeable that, to minimize the impact on marine life and environment, future airgun surveys will use even smaller volumes. Therefore, we have started a pilot project to assess the detectability of the small-volume airgun signals in the Gulf of Mexico using portable onshore broadband seismographs. This research project is collaboration between seismologists at the University of Houston and the Reservoir Geophysical Corporation, who is providing airgun source information during their field tests of a 2000-inch3 airgun array. The source area is between West Cameron and South Timbalier in the Gulf of Mexico, offshore of Louisiana. So far we have conducted a series of numerical analyses of the detectability of the airgun signal. The calculation suggests that the soft ocean-bottom sediments of our study area should be very effective to transmit the airgun energy through the ocean floor to the onshore stations. During the next several months, we will carry out several field experiments by placing broadband seismographs at different onshore locations during the airgun surveys. We hope to show the detectability of the airgun signal in terms of source-station offset distance, station site condition, signal frequency content, source water depth, and ocean bottom hardness. Because we will have all necessary information about source timing, location, site conditions and acquisition parameters, we should be able to conduct source stacks to improve the S/N ratio.
S52A-04 INVITED 11:30h
Advancements in Long-Offset Seismic Imaging: A Blind Test of Traveltime and Waveform Tomography
In 2003 a realistic long-offset synthetic seismic dataset was made available to the community for the purpose of testing modelling, inversion and imaging algorithms. Here we present the results of 2-D traveltime and 2-D waveform tomography applied by workers who, at the time, did not know what the true model was. The synthetic wide-angle dataset consisting of 51 shots was calculated for a realistic crustal model using a 2-D visco-elastic code; these data are still available at terra.rice.edu/department/faculty/zelt/ccss/. The model is 250 km long, and the shot and receiver spacings are 5 km and 90 m, respectively. The center frequency of the source is 5 Hz, with energy between 2-11 Hz. The true model contains large-scale features such as laterally-varying sediment thickness, a basement outcrop, a low-velocity zone, and regions where the crust-mantle boundary is sharp and smooth. Superimposed on this are non-stationary intermediate to wavelength-scale stochastic features. Both first arrival and simultaneous PmP/Pn traveltime tomography were applied to obtain a smooth velocity model with a sharp Moho. The traveltime model compares favorably with the large-scale features of the true model, although it does not capture the details of the low-velocity zone or the smooth crust-mantle transition zone. However, the model obtained from first-arrival traveltime tomography was essential as a starting model for the 2-D acoustic, frequency-domain waveform tomography method we have applied. Data windowing in time, re-weighting in offset, and model smoothing were applied, and a relatively low starting frequency was used, 0.8 Hz, progressing up to 7 Hz. The final model from waveform tomography predicts the input data to a high degree of accuracy for each chosen frequency, and a comparison of the original time domain data with time-domain forward modelling through the final model also reveals a good match. The final model from waveform tomography matches the large and intermediate-scale (down to ~1 km) features of the true model, including the recovery of the low-velocity zone and the structure of the crust-mantle transition. The combined results from traveltime and waveform tomography show the complementary nature of these approaches and the potential for the analysis of real wide-angle crustal data in the future.
http://terra.rice.edu/department/faculty/zelt/ccss/