Description:

"Effects of pre-processing on reverse time migration: a North Sea study Ian F. Jones* Abstract: Almost all conventional pre-processing is conceived with one-way wave propagation in mind. If we take into account the existence of events arising from two-way wave propagation, then many of the underlying assumptions of moveout behaviour implicit in some pre-processing techniques must be re-evaluated. Using 2D synthetic data, we demonstrate that the moveout behaviour of double bounce arrivals (a class of two-way propagating events) can be compromised by pre-processing designed to remove events exhibiting anomalous moveout behaviour. These observations are of interest to us because we are now beginning to employ two-way migration schemes to image complex structures. However, if we continue to use conventional pre-processing techniques, we run the risk of removing the very events we are trying to image. Introduction: The speed and cost effectiveness of contemporary computer systems now permit us to implement more general algorithmic solutions of the wave equation (Whitmore, 1983; Baysal et al., 1983; McMechan, 1984; Bednar et al., 2003; Yoon et al., 2003; Shan & Biondi, 2004; Zhou et al., 2006; Zhang et al., 2006). The restriction to one-way propagation can be lifted, and data migrated to take advantage of more esoteric propagation paths, such as turned rays, double bounce arrivals and, potentially, multiples (Mittet, 2006). However, in order to take advantage of these improved algorithms, we must ensure that the data input to migration have not been compromised in any way. Specifically, in this work we address the moveout behaviour of double bounce events (Hawkins et al., 1995; Bernitsas et al., 1997; Cavalca & Lailly, 2005), and note how many conventional pre-processing algorithms can damage these arrivals, thus rendering some aspects of any subsequent high-end migration superfluous. We commence our analysis by reviewing the conclusions of preliminary work on synthetic data (Jones, 2008, in prep) which discussed the moveout behaviour of turning waves (Hale et al., 1992) and simple double bounce events (also referred to as prism waves by some authors). For ease of demonstration, we firstly employ a ray-tracing package, with which we can model individual selected arrivals, and later create more complex synthetic data using an elastic finite difference (FD) package. Some brief details of these packages are given. After investigating the moveout behaviour of the simple models, we move on to a model representing a complex North Sea salt dome structure (Davison et al., 2000; Thomson, 2004; Farmer et al., 2006). We show the effect of various conventional pre-processing steps on double bounce arrivals, and carry these analyses through to migration with a 2D reverse time migration (RTM) algorithm capable of imaging the double bounce arrivals. We then extend this analysis and demonstration on 2D synthetic data to real data, where we see similar classes of event and the same degradation of double bounce arrivals shown in the synthetic trials. Work programme: Using a workstation-based 2D modelling system, we generated acoustic ray-traced CMP data as a control to identify various individual arrivals. The initial data creation and analysis were performed for simple geometries, and then repeated for a complex North Sea salt diapir model (2 ms sampling, peak-frequency ~35 Hz, shot interval 50 m, CMP interval 6.25 m, 6 km maximum offset). We then generated more realistic elastic FD shot gather data for the complex North Sea salt diapir model. These modelled data included attenuation with an absorbing boundary condition, using the same vertical interval velocity model as the ray-traced model for 1 ms data (resampled to 4 ms for processing), and a peak-frequency of ~17 Hz. In this study we have used an absorbing surface boundary; hence the FD data contain no free-surface multiples (whereas the ray-trace data do). The explicit 2D 3D elastic wave propagation code is 4th-order accurate in space and 2nd-order accurate in time, and is based on the elastodynamic formulation of the wave equation on a staggered grid (Madariaga, 1976; Virieux, 1986; Levander, 1988; Larsen & Grieger, 1998). We pre-processed the FD data using conventional data processing flows that are likely to damage double bounce events, including: * ION GX Technology EAME, 180 High Street, Egham, Surrey TW20 9DY, UK. E-mail: ian.jones@iongeo.com. n Tau-P mute for backscattered noise n Radon demultiple n CMP domain apex-shifted multiple attenuation (ASMA) After each pre-processing flow, we applied 2D RTM and assessed the preservation of double bounce (prism wave) arrivals in the resultant images. In all cases, the direct arrival was muted on the data input to the RTM process. The modelling: We commence by looking at three simple scenarios: n A simple right-angle corner reflector n An acute angle reflector (non-crossing rays) n An acute angle reflector (crossing rays) With this layout, there are no double bounce arrivals for an obtuse angle geometry: we would need extremely long offsets and large arrival times. These three scenarios are shown in Figure 1. It is clear that the moveout behaviour does not conform to what we expect for normal-incidence raypaths that can be migrated with a one-way algorithm, but more closely resembles events such as those resulting from scattered energy or diffracted multiples. We know that for simple quasi-1D cylindrical models all co-axially recorded events in a CMP gather will appear with their apex at zero offset. It is this observation that guides the design principle of various multiple suppression techniques and provides the justification for muting in Tau-P space to suppress backscattered energy. We now look at a full synthetic data set created along a 2D crestal line of the 3D production model representing our North Sea example, and show the effects of various pre-processing techniques on these data. For the geometry in Figure 2, we have: n A single bounce at the flat-lying part of a reflector n A single bounce at the dipping part of this reflector n A non-crossing double bounce involving the flat and dipping reflectors n A crossing double bounce involving the flat and dipping reflectors (not shown to avoid clutter) Figure 1: In each example, the model horizons are shown in green and arrows indicate the propagation direction on the sample ray paths. The resulting CMP gather is shown below each frame. (a) Right-angle corner reflector travel time is constant with offset: i.e., no moveout. (b) Acute angle, non-crossing events. Arrivals are only present on the near offsets, and arrival time decreases with offset. (c) Acute angle, crossing events. Arrivals are present on most offsets, and arrival time increases with offset. Figure 2: (a) Plot of a few sparse rays shown against the interval velocity model. The sediment velocity ranges from about 1900 m/s to 2200 m/s, with some shallow impedance contrast events. The absence of a strong sediment gradient precludes turning rays in the sediments, although a strong compaction velocity gradient below the Top Balder and Top Chalk does produce turning rays. The salt velocity (green) is 4500 m/s and the Chalk velocity is between 5500 m/s and 6000 m/s. (b) A single CMP gather from the surface location at 8 km. Figure 3: (a) The velocity depth model, with main horizons indicated. (b) The brute stack of the FD data. The direct arrivals have been muted. n Raypaths passing into the salt with an internal reflection at the steep salt wall and a second bounce outside the salt from the flat or steep reflectors (which are not considered here as they have relatively low amplitude due to the transmission coefficients at the salt wall) The velocity model is based on a 2D crestal line taken from an actual 3D North Sea example (Farmer et al., 2006). The production project in that case was anisotropic, using VTI 3D RTM code, but for simplicity here we are using 2D isotropic modelling and 2D isotropic RTM. From the ray-tracing exercises, we can clearly see which are the single and which are the double bounce events illuminating the salt flank. We can also identify a class of events passing through the salt body itself and illuminating the salt flank, but these are weak due to the impedance contrast at the salt boundary and will not be discussed here. Figure 3 shows the interval velocity model and associated elastic FD stack. No velocity analysis was carried out. The stack was produced using the RMS velocity function associated with the interval velocity model, and the direct arrival has been muted out. Figure 4 shows an enlargement of the diffraction tails on the right flank of the dome from Figure 3b. We see the ray-trace modelling with and without double bounce events, permitting us to id" Ключевые слова: break, compromised, muted, cmp, velocity, double bounce, no-salt model, simple, diffracted multiple, set, image, offset, break volume, workstation-based modelling, photo, data, cmp gather, scenario, flat-lying, identify, jones, rtm image, north sea, seg, based, mittet, ray-trace modelling, shin, two-way, conclusion, deleterious, sea, diffraction tail, hawkins, synthetic, behaviour, conventional pre-processing, near-vertical event, individual, cmp domain, eage, angle, moveout behaviour, successful project, isotropic rtm, elastic, technical article, backscattered noise, received, raw, domain, srme data, wave, rtm migration, package, volume june, salt flank, balder, event, damage, june, tau-p muting, demonstration, modelling, clear, performed, note, applied, north, no-salt, accurate, geophysics, farmer, synthetic data, conventional, propagation, arrival, firstbreak, bounce event, abstracts, tau-p, radon, flat, moveout, firstbreak org, removed, algorithm, pre-processing technique, assessed, dipping reflector, geology, process, volume, gather, modelled, reflector, pre-processing, bounce arrival, baysal, levander, complex, imaging condition, wave propagation, velocity model, interval, asma, edge, yoon, bounce, diffracted, project, wa, scattered energy, wave equation, mute, commence, application, pre processing, result, equation, flow, create, ensure, raw data, technical, location, crestal, discussed, salt, article, tau-p mute, bloor, vicinity, srme, lines, input, migration, imaging, madariaga, free-surface multiple, org, analysis, ray, salt model, order, real data, noise, one-way, apex, work, boundary, class, data input, model, case, designed, shifted apex, rtm, considered, fd, advantage, mcmechan, reverse time, time migration, davison, observation, maximum offset, zhang, time, multiple, hale, processing, pre, stack, double, attenuated, output, issue, flank, inset, apex-shifted, salt boundary, conventional processing, reverse, backscattered, energy