Verification of bright spots in the presence of thin beds by AVO and spectral analysis in Miocene sediments of Carpathian Foredeep
- 67 Downloads
Twin structural traps that lie within the Miocene strata of the Carpathian Foredeep that are localized above Cierpisz and Mrowla-Bratkowice highs exhibit identical bright-spot seismic anomalies, but only those associated with the Cierpisz high are profitable gas reservoirs. Bright spots can be a result of weak gas or water saturation, but also seismic interference known as tuning effect. For these reasons, it is crucial to differentiate between seismic anomalies. In this article, we present the possibilities of verification of seismic anomalies that occur within the siliciclastic Miocene sediments of the Carpathian Foredeep with the application of AVO analysis and spectral decomposition. AVO methodology enabled to limit the number of anomalies that are present in the post-stack seismic data. These anomalies, however, may also be a result of tuning which is common for the heterolithic sequences in the Miocene sediments of the Carpathian Foredeep. For classification of anomalies in the view of the above, spectral decomposition based on the Basis Pursuit algorithm was applied. Spectral decomposition enabled to divide AVO anomalies in those that are the result of gas saturation and the tuning effect. Gas-saturated zones are characterized by higher spectral amplitudes of the lower frequency range, whereas tuning effect yields higher spectral amplitudes for the higher frequency content. This relation is visible for the data set and enables qualitative differentiation for the set of seismic anomalies.
KeywordsAVO analysis Spectral decomposition Thin beds Miocene sediments Carpathian Foredeep
Natural gas accumulations within the Carpathian Foredeep are located mostly in small structural–stratigraphic traps that are present in its whole depth profile (Myśliwiec et al. 2006 and references therein). Interpretation of such objects is difficult due to their small sizes, finely layered inner structures and the fact that the Miocene sediments that fill the Foredeep exhibit strong facial changes in all directions. These factors result in strong horizontal and lateral variations of physical parameters of both reservoir and cap rock.
Many reservoirs in the Miocene sediments of the Carpathian Foredeep are within finely layered (up to 5 cm) intervals of loosely consolidated sandstones. These kinds of sediments are built of thin layers of changing sandstone–mudstone, which are known as heterolithic sequences (Reineck and Wunderlich 1968). In favourable structural conditions, such an interval can accumulate hydrocarbons. Interestingly, in such a case thick sandstones within heterolithic sequences usually have only insignificant or almost no gas saturation. In the Miocene basin, gas derives nearly totally from the microbial reduction of carbon dioxide generated by breakdown of type III (humic) kerogen (Kotarba 1998; Kotarba et al. 2005). The presence of the methane within the heterolithic sequences of low porosity and permeability suggests that the source rock is also a reservoir rock. Hence, such gas accumulations can be understood as reservoirs that exist at the border between conventional and unconventional targets (Paszkowski et al. 2009).
The focal method for reservoir characterization is an analysis of direct hydrocarbon indicators (DHI). Gas accumulations within the Carpathian Foredeep exhibit strong negative anomaly of a bright spot type. Such anomalies, unfortunately, are very sensitive to the presence of any gas saturation (Biot 1956; Bała and Cichy 2007), which often causes seismically visible bright-spot anomalies to be non-perspective, which have resulted in several exploration failures (e.g. Allen et al. 1993; Myśliwiec 2004a, b). Many anomalies have completely different origins, for example, caused by lithofacial changes, thin bedding or uneven compaction in the Miocene sediments. For this reason, it is crucial to calibrate such an anomaly by other seismic tools, such as AVO analysis (AVO—Amplitude Versus Offset) or spectral decomposition. AVO analysis helps to classify anomalies more precisely, whereas spectral decomposition enables qualitatively to describe tuning effect (Kwietniak 2016).
Application of the frequency analysis for hydrocarbon exploration is widely described (Partyka et al. 1999; Li and Zheng 2008; Shixin et al. 2011; Ahmad et al. 2017) and applied by many authors. It is agreed that gas saturation is the main reason to generate frequency anomalies (e.g. Korneev et al. 2004). Low-frequency shadow, that is often used in such scenarios, exists but its strength and frequency drop, in reality, are very low (Barnes 1991). Nevertheless, practice shows that frequency analysis can be of some use as a DHI factor. The focal problem with the application of frequency analysis is that seismic data from a given area are never uniformly processed and that the frequency characteristics vary from one seismic survey to the other. There exists an endless list of reasons for that, for example acquisition parameters, type of source, source signatures, processing steps, etc. In reality, it is rarely possible to cover the study area with unbiased seismic data that are needed for frequency analysis. For these reasons, diagnostic frequency anomalies can have a very different origin. This is also the case in the Carpathian Foredeep—seismic data cover the majority of the basin, but these data are of very different qualities. The contractors changed, processing path and methodology have developed, and the data cannot be unified and treated with the same fidelity. For that, it is crucial to treat the frequency analysis quantitatively, rather than to search for low-frequency anomalies of a given value. Our approach is to define regions of higher-than-average and lower-than-average frequency anomalies within each separate data set. The frequency analysis applied uses basis pursuit spectral decomposition technique that enables the analysis of amplitude and phase behaviour for a given frequency range.
The described methods and implementations applied mostly for model data (Chopra and Castagna 2014 and references therein). For synthetic models, AVO analysis and spectral decomposition give unique solutions. For real data examples, results are not that ubiquitous and other effects can be observed. For these reasons, we perform the analysis on real data examples, trying to evaluate the quality of their performance.
The article aims to a present scenario, where in similar structural conditions heterolithic sequences generate DHI anomalies but not all of them exhibiting anomalies are gas-bearings traps. Such an example is the Cierpisz and Mrowla-Bratkowice zones. For similarly elevated zones of the similar heterolithic sequences, at the distance of less than 8 km one exhibits commercial gas accumulation (Cierpisz reservoir, see Syrek-Moryc 2007) and the other gives only weak, non-commercial gas flows.
Study area and geological setting
Data set and methodology
Input data for this study come from the 3D-volume ‘Trzciana–Cierpisz–Zaczernie’, which was realized in 2006 by Geofizyka Kraków SA for Polish Oil and Gas Company (PGNiG). The volume covers the area of 324 km2 (see Fig. 1b). The bin size was 20 × 20 m with offsets ranging up to 3 km, which gives the average fold of 42. Seismic sweep signal was generated with the length of 14 s and with bandwidth 8–100 Hz. Such parameters resulted in the average frequency of 46 Hz in the Miocene strata. The data set had a sampling rate of 2 ms. The processing scheme included trace edition, spherical divergence correction, coherent noise removal, first break mute, surface consistent amplitude scaling and deconvolution, static correction, NMO correction, DMO correction, spectral whitening, CMP sort, stack, post-stack signal enchantment, FX migration (Cichostępski 2016). The data were processed with relative amplitude preservation; (Cichostępski et al. 2019a, b). In general, these data have a record length of 4 s two-way time (TWT) and provide good-quality imaging down to the Palaeozoic levels, except the topmost 0.35 s.
The standard procedure used to visualize seismic anomalies caused by gas saturation is the application of seismic attributes. They are additional information obtained from seismic data, allowing visually reinforcing or quantifying the interesting features of the record. For reservoir interpretation, we used the post- and pre-stack seismic attributes. From the post-stack seismic attributes, we choose sweetness, which joins instantaneous amplitude and frequency. Areas characterized by high values of the instantaneous amplitude and low values of instantaneous frequency will give high values of sweetness. Therefore, this attribute highlights gas saturation zones but can also be an indicator of contact between clay and sand (high impedance contrast) (Yushun et al. 2011).
For the evaluation of the observed anomalies on post-stack data, we applied AVO analysis. AVO analysis can be defined as the variation in the amplitude of a seismic reflection with the angle of incidence or source–offset distance (Ostander 1984; Castagna and Backus 1993; Chopra and Castagna 2014). The method is based on Zoeppritz equations (Zoeppritz 1919) which describe seismic wave partitioning at an interface. The equations relate the amplitude of incident P-wave to reflected and transmitted P- and S-waves at a plane interface for a given angle of incidence. The changes in amplitude depend on velocity, density and Poisson’s ratio changes at the interface. The AVO method is often used in unconsolidated sediments as a hydrocarbon gas indicator, because gas generally decreases Poisson’s ratio which results in an increase in amplitude with incident angle/offset.
Class I comprises high-impedance-contrast reservoirs described by high intercept and negative gradient.
Class II is made up of near-zero-impedance reservoirs. They can be described as both weak positive (Class IIp) or negative (Class II) intercept and negative gradient.
Class III are low-impedance reservoirs characterized by strong negative intercept and negative gradient.
Class IV are low-impedance reservoirs characterized by strong negative intercept but a positive gradient.
In unconsolidated Tertiary clastic sediments, class I reservoirs often yields dim spots, class III and IV bright spots and class II reservoirs can be difficult to see unless they have considerable increase in gradient.
From the group of AVO pre-stack attributes, we have chosen the Product attribute which is simply a multiplication of intercept and gradient. This attribute works very well with class III responses (Rutherford and Williams 1989) which commonly occur in the Miocene formation of the Carpathian Foredeep. Unconsolidated sandstones with hydrocarbons will have a strong negative intercept and a strong negative gradient at the top and opposite on the bottom, so that they will show a positive response in the Product attribute from both top and bottom. Non-hydrocarbon-bearing sandstones will be weak or have negative Product (Avseth et al. 2010; Chopra and Castagna 2014).
Basis Pursuit (BP) decomposition is similar to matching pursuit algorithm that decomposes a signal in the individual functions (decomposition atoms) of the predefined dictionary (Tary et al. 2014). The theory behind matching pursuit spectral decomposition is given by Mallat and Zhanh (1993) for which the set of the predefined functions consisting of elements that differ in time and frequency is used. These functions, however, are not sine and cosine functions, as is used in the Fourier transform. The algorithm includes the step of recognition of these functions to find the optimal set of elements, which are different in time and frequency and if integrated give the original signal. In the process, the choice of a wavelet dictionary is of high importance since it predefines the accuracy of decomposition (Chen et al. 2001). Basis pursuit approach is a modified version of matching pursuit—the BP algorithm simultaneously identifies atoms and subtracts them from the signal (Zhang and Castagna 2011). Also, BP uses a minimal number of atoms of lower amplitudes, which results in a sparse representation (Chen et al. 2001).
BP algorithm applied to the data set starts with the initial atom data set that is iteratively adjusted by changing atoms in the dictionary to obtain the optimal solution. The criteria used by the algorithm are based on the L1 norm.
For the decomposition, we used a dictionary consisting of Ricker wavelet with regularization parameter 0.5 (default). Regularization parameter controls the sparsity of the time–frequency maps; so the larger the parameter, the sparser the frequency map. The maximum iteration to complete the decomposition was set to 50—this, at one seismic line, took app. 15 min (on a workstation with Intel Core i7-4770 CPU @ 3.40 GHz, 32 GB RAM). For the analysis, we used frequency slices and average frequency computed based on the decomposition.
Additionally, the interpretation of AVO Product results enabled us to indicate another interesting zone that lies above the M3 horizon—flat spot at time 650 ms drilled by B-4 well. On this level in well B-4, gas flow was obtained. At the same level above M3 horizon, similar anomalies occur (near M-1 well and C-2), but there were no gas flow tests completed.
C-B anomaly (Fig. 10) is a seismic image of the multi-horizon gas reservoir which is currently exploited. In the seismic image (Figs. 2, 4), this zone manifests itself between times 920–1110 ms as bright spots and flat spots that correspond to the subsequent gas-bearing horizons that also exhibit class III AVO. These intervals are built of heterolithic deposits, within which both thin sandstone layers, of porosity reaching 17%, and mudstone intervals, of very low porosity, are gas-saturated.
The anomalous zone MB-B (Fig. 11) lies between wells M-1 and B-4. It gives clear, bright spots at times 1030–1170 ms. These anomalies also give class III AVO response. These zones, due to their localization, cannot be tied to wells. They can be associated with the top of the heterolithic sequence which has higher mudstone content with thin layers of higher sandstone additives. Similar to the previous cases, gas saturation is higher in mudstone of lower porosity, than in sandstone.
Zones associated with the Mrowla-Bratkowice region (MB-A and MB-B) show a slightly different response. MB-A zone does not show increased amplitudes for 20 Hz frequency slice (Fig. 12), but it gives very high amplitudes for the 40 and 60 Hz ones (Figs. 13, 14, respectively). AVO and seismic attribute analysis enable the interpretation of gas reservoir, but saturation had not been proven due to the lack of a test for gas flow in this interval. The lack of low-frequency component could be associated with the existence of thin beds—in comparison with other zones, MB-A is built by the most finely layered interval.
Zone MB-B, although showing preferable characteristic for gas saturation on AVO and attribute analysis, in spectral decomposition shows no differentiation for lower and higher frequencies. This gives arguments to reason that the zone cannot be treated as a possible prospecting target.
The last zone which is worth mentioning is the shallowest zone near well B-4 (marked by black arrow on Figs. 12, 13, 14). For this, the test for a gas flow gave positive results. This anomaly manifests also high amplitudes for lower frequencies, similar to zone C-A. Lithological characteristics for these two are also similar—both zones are built by heterolithic sequences. Such characteristics classify the B-4 anomaly as a potential exploration site.
The analysed anomalous zones C-A and MB-A are localized within one seismic facies that belongs to the deltaic sediments (Marzec et al. 2018). Both these zones are composed out of heterolithic sequences that are built out of several-dozen metres of intervals of mudstone and sandstone sequences of various ratios. Anomalous zones C-B and MB-B are localized within deeper deltaic sediments. These heterolithic intervals manifest more mudstone additives. Within Cierpisz high, gas saturation is generally present within heterolithic sequences of high sand additives and better petrophysical quality.
The zone associated with Mrowla-Bratkowice high is characterized by higher amplitudes and more prominent anomalies than Cierpisz high, which may indicate better reservoir properties. Nevertheless, after drilling of well M-1, it turn out that the interval exhibits very strong heterolithic character, with weak gas saturation, flows from almost all intervals within the Miocene sediments that are linked to very thin mudstone layers. Sandstone layers that have far better petrophysical parameters have no gas saturation. This scenario may be attributed to the thin bedded mudstone source rock from which the generated gas migrated directly to a very thin sandstones within the heterolithic sequence or stayed within source rock.
AVO anomalies correlate well with the gas-saturated intervals. All zones exhibit class III AVO anomaly—response characteristic for gas-bearing sandstones of good reservoir properties, though they are produced by heterolithic sequences that consist of thinly layered mudstones and sandstones. Such anomalies are also indicated outside the gas-bearing zones. The existence of the false anomalous zones may be a result of the non-optimal processing sequence and the specification of land seismic and acquisition parameters. Non-regular offset distribution results in the weakening of amplitudes for near offsets, low S/N ratio or strong surface wave that is not reduced from the final data may cause amplitude changes, for which AVO analysis is very sensitive (Castagna and Backus 1993; Downton et al. 2000; Chopra and Castagna 2014; Cichostępski et al. 2019a, b). The existence of AVO anomalies may also be a result of geological features such as uneven compaction of clastic sediments, rapid change in VP/VS ratio that can be related to lithofacial changes rather than gas saturation and thin beds that can cause tuning effect. Tuning may result in amplification of AVO results which can influence reservoir reasoning (Castoro et al., 2001; Yang 2003). AVO anomalies can be produced by mudstones or sandstones, with no difference for thin or thick sequences. For this reason, it is crucial to use other methods that can differentiate between the thicknesses of the reservoir, for example spectral decomposition.
Spectral decomposition can be a good indicator of thin bed thickness (Partyka et al. 1999), but its quantitative ability is mostly useful for modelling purposes, i.e. give accurate results for a simple wedge model (Kwietniak 2016). Nevertheless, the selective attenuation caused by gas-saturated zones sometimes exhibits higher spectral amplitudes for lower frequencies. At the same time, frequency analysis is a reliable tool for an indication of tuning effect that limits AVO analysis. With the application of spectral decomposition, we can distinguish between the frequency characteristics of the seismic anomalies that give the same bright-spot signature on the post-stacked seismic data. Anomalies that exhibit higher amplitudes for lower frequencies are classified as gas-bearing zones. Those anomalies that are associated with higher frequencies are interpreted as consisting of more finely layered heterolithic sequences. For this reason, the later are treated with lower reliability when it comes to AVO analysis. Anomalies that lie above the Mrowla-Bratkowice high are identified as caused by tuning effect and hence should be excluded from reservoir reasoning. The application of frequency analyses enabled us to differentiate anomalies in terms of thickness relation of layers that lie within them.
The performed analysis included two structural highs of the Miocene sediments: Cierpisz and Mrowla-Bratkowice. Within the Cierpisz high, there is a multi-layered gas reservoir, Mrowla-Bratkowice high and anomalies that lie within this area are not classified as possible gas reservoirs even though the geological setting is similar to the Cierpisz high.
AVO analysis indicates possible gas-saturated zones that are verified by well logs and gas flow tests. They occur for both sandstones and mudstones as well as for the heterolithic sequence that complicates indication of the source and reservoir rock.
AVO anomalies can be also a result of wave interference; hence, it is crucial to perform a tuning analysis for the local geological setting. Spectral decomposition analysis shows that anomalies that are associated with the Mrowla-Bratkowice high are characterized by relatively higher frequencies than those of the Cierpisz high. This pattern may suggest that anomalies above the Mrowla-Bratkowice high are more likely linked to the tuning effect, than to the gas saturation. Moreover, those anomalies do not exhibit substantial low-frequency characteristics, which is one of the diagnostic indicators for gas saturation.
Based on the obtained results, authors suggest comprehensive seismic interpretation for gas exploration by application of post- and pre-stack attributes together with spectral analysis of indicated anomalous zones within thin beds. Altogether with the seismic analysis, it is crucial to perform integrated geological analysis that enables to recover basin morphology and sedimentological history. Only such an approach can explain the distribution of gas saturation in a given sedimentary basin. Our current work is focused on seismic stratigraphy within the area of studies and its proximity, altogether with the seismic analysis. For the Miocene strata of the Carpathian Foredeep, only such a methodology will give reliable results that will enable reservoir reasoning.
The research was supported by the National Science Centre (NCN) via Grant N525 254040 awarded to Porębski SJ and AGH-UST University of Science and Technology in Krakow, Poland, Grant No. 188.8.131.525. We thank Polish Oil and Gas Company (PGNiG) for providing access to the seismic and well-log data set and CGG for their provision of seismic interpretation software through the University Software Grant Program. We would also like to thank Mariusz Majdański and other anonymous reviewer for providing valuable and insightful feedback on the manuscript.
Compliance with ethical standards
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
- Avseth P, Mukerji T, Mavko G (2010) Quantitative seismic interpretation: applying rock physics tools to reduce interpretation risk. Cambridge University Press, CambridgeGoogle Scholar
- Barnes AE (1991) Comments on instantaneous frequency. SEG Technical ProgramGoogle Scholar
- Cichostępski K (2016) Analiza zmian zapisu sejsmicznego z offsetem jako narzędzie do identyfikacji stref akumulacji gazu zmiemnego w cienkowarstwowych utworach zapadliska przedkarpackiego. Dissertation, AGH Univesity of Science and Technology in Krakow. http://winntbg.bg.agh.edu.pl/rozprawy2/11172/full11172.pdf. Accessed 17 June 2019
- Downton JE, Russell HB, Lines LR (2000) AVO for managers: pitfall and solutions. CREWES Research Report Volume 12Google Scholar
- Kotarba M (1998) Composition and origin of gaseous hydrocarbons in the Miocene strata of the Polish part of the Carpathian Foredeep. Przegląd Geol 46:751–758Google Scholar
- Kotarba MJ, Więcłąw D, Kosakowski P, Kowalski A (2005) Potencjał węglowodorowy skał macierzystych i geneza gazu ziemnego akumulowanego w utworach miocenu zapadliska przedkarpackiego w strefie Rzeszowa. Przegląd Geol 53:67–76Google Scholar
- Kwietniak A (2016) Spectral decomposition of a seismic data—thin bed thickness estimation and analysis of attenuating zones, Doctoral Thesis, Hard copy in Library of AGH-UST, Access www.bg.agh.edu.plDissertation, AGH-UST University of Science and Technology
- Myśliwiec M (2004a) Poszukiwania złóż gazu ziemnego w osadach miocenu zapadliska przedkarpackiego na podstawie interpretacji anomalii sejsmicznych—podstawy fizyczne i dotychczasowe wyniki. Przegląd Geol 52:299–306Google Scholar
- Myśliwiec M (2004b) Typy pułapek gazu ziemnego i strefowość występowania ich złóż w osadach wschodniej części zapadliska przedkapackiego. Przegląd Geol 52:657–664Google Scholar
- Myśliwiec M, Borys Z, Bosak B, Liszka B, Madej K, Maksym A, Oleszkiewicz AK, Pietrusiak M, Plezia B, Staryszak G, Świętnicka G, Zielińska C, Zychowicz K, Gliniak P, Florek R, Zacharski J, Urbaniec A, Górka A, Karnkowski P, Karnkowski PH (2006) Hydrocarbon resources of the Polish Carpathian Foredeep: Reservoirs, traps, and selected hydrocarbon fields. In Golonka J, Picha FJ (eds) The Carpathians and their foreland: geology and hydrocarbon resources. AAPG Memoir, vol 84, pp 351–393Google Scholar
- Oszczypko N (2006) Powstanie i rozwój polskiej części zapadliska przedkarpackiego. Przegląd Geol 54:396–403Google Scholar
- Paszkowski M, Porębski SJ, Warchoł M (2009) Koncepcja projektu otworu kierunkowego w mioceńskich utworach zapadliska przedkarpackiego. Wiadomości Naftowe i Gazownicze 3:4–13Google Scholar
- Syrek-Moryc C (2007) Złoże gazu ziemnego Cierpisz ważnym argumentem w problematyce przyszłych prac poszukiwawczych w cienkowarstwowych utworach miocenu zapadliska przedkarpackiego i związanych z nim potencjalnych zasobów gazu ziemnego. Nafta-Gaz 2:95–99Google Scholar
- Wiggins R, Kenny GS, McClure CD (1983) A method for determining and displaying the shear-velocity reflectivities of a geologic formation. European patent application 0113944Google Scholar
- Yang M (2003) Monochromatic AVO: an indicator that sees through wave interference. In: 73rd annual international meeting SEG, Expanded Abstracts, pp 201–208Google Scholar
- Yushun D, Zhaoquan P, Lingbang Z, Mingbo B (2011) Carbonate reservoir and gas‐bearing property detection using sweetness. In: SEG annual meetingGoogle Scholar
- Zoeppritz K (1919) Erdbebenwellen VIII B, Uber Reflexion und durchgang seismischer wellen duch unstetigkeitsflachen. Gottinger Nachr 1:66–84Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.