75th EAGE Conference & Exhibition incorporating SPE EUROPEC 2013 London, UK, 10-13 June 2013

Th-07-15Using Extended Correlation Method in RegionalReflection Surveys - A Case Study from PolandM. Malinowski* (Institute of Geophysics PAS) & P. Brettwood (IONGeophysical)

SUMMARYIn the effort to provide constraints on the deep crustal structure we have applied extended correlationtechniqueto the ION GXTechnology PolandSPAN seismic reflection data. It allows to extend nominal record lengthof thesurvey (12 s in this case) to much longer times (18 s and 22 s tested here), given that raw uncorrelated dataare storedand the up-sweep is used. The technique is not novel and has been successfully used, e.g. in Canada,during theLITHOPROBE project to save the time spend on single VP. For the times greater than the nominal recordlength,data are correlated using self-truncating sweep resulting in the original sweep spectrum kept for thenominal recordlength and the higher frequencies cut off for the greater times. Given the broad sweep spectrum (2-150 Hz)used in the survey, the high-end frequency at 22 s is 57.5 Hz, which is way above the expected frequencyof deep crustal arrivals (usually below 30 Hz). The correlation was performed both using the pilot sweepsignal andthe mean of the measured ground force recorded for each separate vibrator and VP. Processing of theground-forcecorrelated data produced clearer reflectivity in the deeper section.

75th EAGE Conference & Exhibition incorporating SPE EUROPEC 2013 London, UK, 10-13 June 2013

Introduction

The majority of the land seismic data is acquired using the Vibroseis technique and recent advancements in the technique both in terms of the actual acquisition and processing are mostly focused on operational efficiencies in large 3D surveys (i.e. simultaneous sources, e.g. Bouska, 2010) and generation of the low frequencies (Wei and Phillips, 2011). Here we demonstrate that recording the raw uncorrelated data and applying the extended correlation method during processing, together with the measurement of the ground-force signature provides both operational efficiency (through reducing of the listening time) and improved imaging of the deep crustal structure along a regional 2D seismic survey.

Data

Here we use data from a large (2200 km) regional 2D seismic reflection program recently completed in Poland by ION Geophysical (PolandSPANTM project). It bridges the gap between the excellent coverage of the deep wide-angle reflection/refraction profiles (see e.g. Guterch et al. 2010) and numerous shallow exploration seismic projects and provides a unique opportunity to link geological processes from the near-surface to the deep crust. Basic acquisition parameters are as follows:

Receiver/shot interval: 25 m

Maximum offset: 12 km

Nominal fold: 480

Nominal record length: 12 s

Sweep length: 16 s

Sweep frequency: 2-150 Hz

Data were acquired using INOVA AHV-IV Commander (62,000lb) Vibroseis trucks, set up to use a custom broadband sweep starting as low as 2Hz (Wei and Phillips, 2011). Raw unstacked, uncorrelated data were recorded during production together with auxiliary data regarding Vibroseis performance (reaction mass and base plate acceleration and weighted sum estimate, i.e. the Vibrator ground force).

Method

The idea of extending the record length of the Vibroseis data during post-acquisition processing is not novel. It has been used since the mid-70s in deep reflection surveys (e.g. COCORP), but the method was largely restricted to academic research. The necessary requisite is to record raw uncorrelated data. In normal Vibroseis acquisition, a record length of duration tprofile is formed by crosscorrelation of the data recorded for a duration record (listening time) with the sweep of duration sweep. Therefore tprofile=record-sweep. In order to increase tprofile post-production, we need to reduce the sweep length. If a typical up-sweep characterized by a frequency band f0-f1 is used, then truncating the sweep means cutting off high frequencies. Correlation with such a truncated sweep (called fixed-bandwidth extended correlation) will result in loose of higher frequencies throughout all the times after correlation (also for t < tprofile). Alternatively, the so-called self-truncating extended correlation (Okaya and Jarchow, 1989) can be used. In this mode of operation, the correlation is made with the sweep that is gradually truncated past the tprofile. Therefore, the original sweep frequency bandwidth is maintained after correlation down to tprofile. The highest frequency fmax at a given time can be described by the following equation (Okaya and Jarchow, 1989):

(1)

75th EAGE Conference & Exhibition incorporating SPE EUROPEC 2013 London, UK, 10-13 June 2013

For the parameters of the PolandSPANTM survey: sweep = 16 s, f0=2 Hz, f1=150 Hz and the nominal record length tprofile=12 s, the fmax for a new extended record length of 18 s is 94.5 Hz and 57.5 Hz at 22 s. Such a frequency is too high for typical signals recorded from the deep crust and mantle (25-35 Hz). Results We applied the extended correlation technique using a self-truncating sweep for a 68-km long part of line 5100 of the PolandSPANTM project located in SE Poland. The correlation was performed either with the pilot sweep or the mean ground force signature (see Figure 1 for a plot of signal autocorrelation) calculated as the average of the weighted-sum estimates (Ziolkowski, 2010) for every single shot position. Initially the data were correlated down to 18 s; after selection of the preferred correlation signal (ground force), it was further extended to 22 s. Data after correlation were stacked using diversity stacking. The following minimalistic processing flow was applied to the data (true-amplitude preserving):

Spherical divergence (down to 10 s)

bandpass filtering (2-6-38-48 Hz)

f-k filtering (fan)

muting

surface-consistent amplitude balancing

elevation statics

NMO with a regional velocity field

Normal CDP stacking

semblance-based coherency filtering

finite-difference migration using regional velocity field

trace balancing

trace scaling for display

Figure 2 compares migrated time sections down to 18 s obtained from the data correlated with a pilot sweep (Figure 2a) and ground force (Figure 2b). There is a lot of reflectivity occurring in the deeper crust below the 12 s which was the nominal record length. We note that the section obtained using ground force-correlated data is characterized by the lower frequencies. It is related to the fact that the frequency characteristics of the ground force is shifted towards low frequencies as the higher frequencies are more rapidly attenuated by the near-surface (compare Figure 1a and 1b). This is however preferential for deep crustal imaging and therefore correlation with the ground force was considered superior for this purpose. The reflectivity present at around 15-16 s TWT likely represents the Moho discontinuity (i.e. the crust-mantle boundary) as confirmed by the coincident deep refraction profile (Grad et al. 2006). Another interesting feature is the rich reflectivity observed in the lower crust – a feature often found in the lower crust in an extensional tectonic setting. This observation was confirmed along a deep reflection profile located further towards SE (Malinowski et al. submitted).

Conclusions

We have applied the extended correlation technique to the PolandSPANTM seismic reflection data. It allows the extension of the nominal record length of the survey (12 s in this case) to much longer times (18 s and 22 s tested here), given that raw uncorrelated data are recorded and an up-sweep is used. Processing of the ground-force correlated data as compared to pilot sweep correlated one produced clearer reflectivity in the deeper section, which can be attributed to the lower-frequency characteristics of the ground-force signal.

75th EAGE Conference & Exhibition incorporating SPE EUROPEC 2013 London, UK, 10-13 June 2013

Acknowledgements

ION Geophysical is acknowledged for the permission to use and show PolandSPANTM data. Processing was done using GlobeClaritasTM software under the license from GNS Science, New Zealand.

References

Bouska, J. [2010] Distance separated simultaneous sweeping, for fast, clean, vibroseis acquisition. Geophysical Prospecting, 58, 123-153. Guterch, A., Wybraniec, S., Grad, M., Chadwick, R.A., Krawczyk, C.M., Ziegler, P.A., Thybo, H. and De Vos, W. [2010] Crustal structure and structural framework. In: Doornenbal, J.C. and Stevenson, A.G. (Eds.) Petroleum Geological Atlas of the Southern Permian Basin Area, 11-23, EAGE Publications B.V. Grad, M., Guterch, A., Keller, G.R., Janik, T., Hegedus, E., Vozár, J., Ślaczka, A., Tiira, T. and Yliniemi, J. [2006] Lithospheric structure beneath trans-Carpathian transect from Precambrian platform to Pannonian basin: CELEBRATION 2000 seismic profile CEL05. J. Geophys. Res., 111, B03301. Okaya, D.A. and Jarchow, C.M. [1989] Extraction of deep crustal reflections from shallow Vibroseis data using extended correlation. Geophysics, 54, 555-562. Wei, Z. and Phillips, F. [2011] Analysis of vibrator performance at low frequencies. First Break, 29, 55-61. Ziolkowski, A. [2010] Review of vibroseis data acquisition and processing for better amplitudes: adjusting the sweep and deconvolving for the time-derivative of the true groundforce. Geophysical Prospecting, 58, 41-53.

Figure 1 Amplitude and phase spectrum of the sweep autocorrelation function: a) pilot sweep; b) mean of the weighted-sum estimate (ground force) for all VPs.

75th EAGE Conference & Exhibition incorporating SPE EUROPEC 2013 London, UK, 10-13 June 2013

Figure 2 Migrated stacks from the processing using extended correlation technique: a) correlation using pilot signal; b) correlation using weighted-sum estimate (i.e. the ground force). Note that the nominal record length of the data was 12 s. Some structural elements are marked in b).