Abstract
While exploring for hydrocarbons in rift-related basins, volcanic and volcaniclastic rocks and their erosional products are commonly encountered lithologies. At the reservoir level, the presence of rock types derived from volcanism or affected by post-volcanic re-deposition leads to highly varied complex lithologies with wide ranges of inherent rock properties.
In the Santos Basin, a number of exploration wells have encountered basalts emplaced along the basin hinge line that have controlled the sedimentation and basin geometry. The main reservoir units of the basin consist of complex Upper Cretaceous sandstones deposited during periods of active volcanism. For one of the main exploration targets, the Senonian Ilhabela Formation, integrated geological and geophysical interpretation of recently acquired seismic and well data has led to a refined sedimentological model, an enhanced AVO model as well as a deeper understanding of the petroleum system. The encountered sandstones are of shallow-marine and deltaic origin, with local occurrence of mafic volcanic and subvolcanic rocks. Dependent on the presence of volcanic rocks in the vicinity of the particular reservoir, two different types of sandstones with different AVO characteristics have been identified. The presence of volcanic rocks may be interpreted from seismic data by their distinct geometry, usually high amplitudes and unique AVO behaviour, expressed as a strengthening of positive amplitudes with increasing offset.
INTRODUCTION
Volcanic rocks can have a significant impact on reservoir rocks in rift-related basins. Partial or complete reservoir substitution, complex diagenetic alteration by circulating hot fluids and addition of mineral components have led to a number of economically unsuccessful wells, both in exploration and field development projects (Avseth et al. 2001; De Vincenzi & Klarner 2002). Furthermore, volcanic rocks may form lateral seals or migration barriers, providing both positive and negative impact on the petroleum system. Non-permeable volcanic layers can seal the top of the reservoir, preventing it from breaching (O'Halloran & Johnstone 2001), or they can build a migration barrier for the fluid on its way from the source rock into the trap. The possible impact of volcanism in time and space on the maturation of source rocks is still a subject of intense discussion (see, amongst others, Scotchman et al. 2002). It is therefore essential to understand the distribution of volcanic rocks in the vicinity of the reservoir.
A number of exploration wells in the Santos Basin offshore Brazil have penetrated the Late Cretaceous Ilhabela shallow-marine deposits, a part of the Senonian Itajai Açu Formation (Fig. 1). Most of the sandstones have a complex and immature composition, and have been altered via multiple diagenetic processes. In addition to complex reservoir rocks, the bounding lithologies include shale, siltstones and volcanics, making the interpretation of pre- and post-stack seismic reflection data both difficult and problematic.
Schematic stratigraphic column, Santos Basin, offshore Brazil. The main exploration targets are the Upper Cretaceous Ilhabela sandstones.
Close co-operation of specialists from different disciplines was required in order to solve unique problems related to seismic interpretation and reservoir prediction within this complex environment. Sedimentology and fieldwork studies of recent analogues (e.g. Porebski & Steel 2003) provided the basis for recognition of the depositional environment used for seismic interpretation. Special seismic processing was utilized to calculate amplitude versus offset (AVO) attributes such as ‘Far minus Near times Far’, scaled Poisson's ratio, Fluid Factor (see Ross 2002) from selected offset stacks and pre-stack data. These data were interpreted using results from rock physics analysis – elastic parameters of different lithology types and seismic forward modelling. The AVO analysis supported the lithology identification from seismic geometries. The input for the rock physics was delivered by the petrophysicist – log-derived properties and interpretations. Petrography was used for provenance analysis, as well as analysis of diagenesis and fabric. In fact, petrography was the key to explain the observed anomalous AVO behaviour of some of the sandstones. Finally, seismic sequence stratigraphy related the different levels of sandstones to periods with and without volcanic activity.
The understanding of the petrophysical and especially the elastic properties of the Ilhabela sandstones, related to their genesis in conjunction with a sound regional geological model, is progressing. The results of currently ongoing exploration will contribute to the generation of refined, predictive depositional and seismic models, which will help to lower future exploration risk.
REGIONAL GEOLOGY
The Santos Basin offshore Brazil is a typical passive margin rift basin that developed following the separation of South America from Africa in the Early Cretaceous. It contains volcanism as a product of the several rifting phases, which have led to the opening of the Atlantic. Initial rift sediments include organic-rich lacustrine shales of Barremian and Aptian age grading into coarse clastic rocks along the rift edges, and carbonate rocks. These shales are thought to constitute the main source-rock interval within the basin (Mello et al. 1994, 1995).
The post-rift subsidence and thermal sag phase of the Santos Basin has been accommodated largely by basement flexure across a hinge line, coupled with normal faulting (Mohriak et al. 2002). This feature, also referred to as the Santos hinge line, marks the western flank of the Santos Basin and is defined by the uplifted rift margin (Figs 2, 10⇓). Successive marine flooding of the rift, following the opening of the Atlantic, resulted in the deposition of thick evaporite successions during the Aptian. The continued subsidence resulted in the establishment of a passive margin sequence and carbonate platform during the Albian (one of the secondary exploration targets). The onset of the Late Cretaceous is marked by the collapse of the carbonate platform, following salt mobilization and salt withdrawal. Widespread marine flooding and continued salt-related structuring established a marine shelf to upper-slope environment across the Santos Basin. Late Cretaceous sedimentation is characterized by gravity flow deposits and associated channel and fan systems in intra-slope and inter-salt basins, while deltaic deposits dominate the more proximal areas. Repeatedbase-level changes starting with the Campanian eustatic fall characterize the drift-fill sequence for the remainder of its basin-fill history. Where the subsequent salt movement has created salt windows between the diapirs, the pre-salt hydrocarbons could migrate into the post-salt fluvial-deltaic and deeper-marine Upper Cretaceous to Tertiary clastic reservoirs.
Seismic line and interpretation from the Santos Basin, traversing the Upper Cretaceous basin hinge. The strong amplitudes within the Ilhabela member (arrow) are basalt-related. Horizon slices through this feature are shown in Figure 3.
Deeply seated faults, intersecting the stretched continental crust, served as magma conduits during multiple phases of mafic volcanic activity at several stages of the rifting and basin development. A correct stratigraphic indexing is difficult due to the lack of reliable biomarkers in the section, but the age of the volcanic rocks ranges between 80 Ma and 65 Ma (Mohriak et al. 2002). During the deposition of one of the main reservoir units, the Upper Cretaceous Ilhabela sandstones, multiple extrusions and intrusions were emplaced through faults and fissures along the basin hinge line (Fig. 3). This gave rise to the assumption that this fault trend was very active and an ideal conduit for the magma. Interestingly, the occurrence of volcanic rocks does not strictly follow the main fault, but occurs also along smaller fractures associated with it. Within the area of interest, at least four stages of Upper Cretaceous volcanic activity have been recognized. They exhibit distinct geometries comparable to those of recent surface volcanoes. Massive subaqueous basalt flows, indicative of sheet volcanoes with a thickness of several tens to hundreds of metres and covering an area of up to 400 km2, massive subaerial basalt flows with single lava flow geometries, smaller individual cones and lava flows as well as single dykes and dyke complexes can be identified on seismic data. Furthermore, due to their low shear velocities, the volcanic-related rock types show an anomalous AVO behaviour. As a general rule they exhibit a strengthening of their positive amplitudes with increasing offset (Klarner et al. 2006a, b).
Horizon slice (a) at the top of a massive 400 km2 basalt flow (see Fig. 2) contemporary with the Ilhabela sandstones, indicative of a small mainly subaqueous shield volcano (dark colours). The dashed line marks the basin hinge line. The arrow indicates a large crater, terminating a deep-reaching chimney. In between the seismic layers, representing the volcanic unit, on one of the horizon slices (b), a single channel-like feature can be seen (dashed arrow). As there were no intercalated sediments penetrated, this must be interpreted as a subaerial low-viscosity lava channel, as can be observed on dipping flanks of recent volcanoes. Both images show a number of circular features, interpreted as vents and craters.
Sediments coming from outboard of the basin and crossing the volcanic units were mixed locally with erosional products of those mafics and deposited downdip. The high percentage of unstable components subsequently led to a very complex diagenetic history, completely altering the reservoir properties of the sandstones. Early cementation by chlorite rims led to porosity preservation (Anjos et al. 2003), while later alteration and cementation led to the building of a pseudo-matrix, significantly influencing its permeability and elastic properties. The recognition of these sandstones is crucial to the correct prediction of the seismic properties of the reservoir.
UPPER CRETACEOUS DEPOSITIONAL MODEL
One of the main reservoirs in the Santos Basin is the Senonian Ilhabela Member of the Itajai Formation. It is composed of proximal deltaic deposits grading into shelf margin deltas and gravity flow deposits towards the deeper parts of the basin. The general basin trend is complicated locally by deposition in intra-slope basins formed by salt withdrawal. The sediment geometries are interpreted as local gravity flow deposit reservoirs forming stacked channel complexes and fans located east of the carbonate platform margin and deltaic reservoirs towards the west, above the carbonate levels (Fig. 4).
Semi-transparent seismic amplitude volume (approx. 40 km × 40 km). Bright amplitudes in the central part show sand geometries (channels and beach-parallel sand ridges on the fluvial-deltaic plain background). In the lower right corner the first shelf-edge sandstones appear. Concentric bright features in the northern half exhibit volcanic vents and craters.
ORIGINAL AVO MODEL AND SEISMIC OBSERVATIONS
The analysis of the variation of seismic amplitudes with increasing offset or incidence angle (AVO) is a standard tool in exploration, used for lithology identification, hydrocarbon detection and fluid parameter prediction. AVO forward modelling using log data or generalized models for p- and s-wave velocities and density is adopted to calibrate the observed seismic amplitudes (Castagna 1993).
In the Santos Basin, until now, shear information had been obtained in only a few drilled wells, and even this information was scarcely available. Therefore, for the generation of AVO models prior to drilling, the shear part had to be derived synthetically. One well, which had encountered the target formation at 4200 m (Ilhabela sandstones, water bearing, with approximately 12–15% total porosity), was chosen to be representative for the target area of a planned well. This sandstone exhibits a higher p-wave velocity than the bounding shale, but a significantly lower density. The empiric Greenberg–Castagna relationships for Vp/Vs ratio in sandstones and shales (Greenberg & Castagna 1992) were used for the model. This is an approach normally used when there is no shear log information available, as in the Santos Basin case, and it can show good results in predicting sandstone properties (Dillon et al. 2004). The fluid substitution modelling shows that, in the brine and oil case, no observable AVO effects should occur, whereas for the gas case a polarity reversal and a Class II–III AVO (weak negative intercept, and negative AVO gradient, Rutherford & Williams 1989) should be expected. Modelling also predicts that a decrease in shale volume and corresponding increase in porosity enhances the AVO effect by increasing both the negative intercept and the negative gradient. The analysis of partial offset stacks and pre-stack seismic data around the prospect area showed clear sandstone geometries (channels and stacked terminal lobes), and a subtle Class III AVO anomaly. This anomaly appeared to match the gas model for quartz sandstone with 12–15% total porosity.
WELL RESULTS
The exploration well encountered a sandstone unit several tens of metres thick, as predicted from seismic data, embedded in relatively high impedance shale (Fig. 5). The sandstone is poorly sorted, and consists of a complex mixture of quartz, feldspar, lithic and volcanic rock fragments, clays and diagenetic minerals. Beneath this main sandstone, the well encountered a high impedance basaltic unit of 50 m thickness. Underlying this volcanic unit, an additional sandstone layer was encountered, composed predominantly of quartz, with lesser amounts of feldspar, chlorite and rock fragments. Due to its proximity to the basalt, the brine pore filling and its high average impedance, this lower sandstone is nearly invisible on seismic data.
Seismic section and schematic cross-section through the discussed well location. Red indicates an impedance increase.
In an effort to understand better the local rock properties and AVO response(s) in the area, a comprehensive suite of data was acquired across all the major lithologies (Fig. 6). This included a conventional logging suite, as well as dipole sonic data, a nuclear-magnetic resonance tool, borehole image (FMI), sidewall cores and zero-offset vertical seismic profiling (VSP). The analysis of the dipole sonic data shows that the post-volcanic sandstone (Fig. 7, orange dots) has low acoustic impedance but surprisingly high Vp/Vs ratios of around 1.8. The Vp/Vs ratios are much higher than those predicted using the Greenberg–Castagna ‘sand’ coefficients, and are also higher than the Vp/Vs ratios of the bounding shales. The cross-plots reveal that this is an effect of very low shear velocities (2200 m s−1) rather than anomalous high p-wave velocity (4000–4500 m s−1). Indeed, Vp/Vs ratios across the post-volcanic sandstone interval are similar to those of carbonate rocks. Somewhat surprisingly, the density-neutron response for this sandstone interval is also similar to that of limestone (Fig. 6). Various post-drill AVO models, using the measured log data, suggest that these sandstones should give rise to a Class IV AVO response when filled with gas, which matches their low impedance. This is clearly different from the interpreted Class III AVO anomaly from the pre-stack seismic data.
Selected logs showing Upper Cretaceous lithology types in a well from the Santos Basin. The high amount of basic rock fragments in the post-volcanic sandstone causes an anomalous neutron porosity behaviour and high Vp/Vs ratio. The 25 Hz synthetic gather (red for positive amplitudes) demonstrates the AVO effects at the top and base of the basalt (see text).
Vp versus Vs ((a) high water saturation, (c) gas case) and acoustic impedance versus Vp/Vs ratio ((b) high water saturation, (d) gas case) cross-plots in a well in the Santos Basin, having encountered Upper Cretaceous volcanics. Purple: basalts; green: various shales; brown: quartz-rich sandstones; orange: sandstones containing volcanic rock fragments. The volcanic-related lithologies exhibit very low shear velocities and, even for the gas case, an anomalously high Vp/Vs ratio.
Shear velocities for shales and quartz sands throughout the rest of the well can be predicted accurately using the Greenberg–Castagna coefficients for sandstone and shale. The lower sandstone unit, as discovered by the well in the pre-volcanic section, is characterized by higher impedance and lower Vp/Vs ratios than the bounding lithologies (shales and basalts). This sandstone unit has Vp/Vs ratios typical of quartz sands, and also exhibits the expected density-neutron response (crossover of density and neutron-porosity logs calibrated for limestone porosity, Fig. 6). AVO models show that this sandstone should produce a Class II AVO response in the brine case, turning into a Class III AVO anomaly when filled with gas.
DATA QC DISCUSSION
As the Vp/Vs response in the post-volcanic sandstone does not match the Greenberg–Castagna model commonly used for the Santos Basin, many intensive discussions have taken place concerning the quality of the shear log data and possible sources of error. The monopole and dipole sonic data were determined to be of high quality. This determination was based on careful analysis of the raw waveforms and evaluation of other petrophysical data. Formation damage due to mud filtrate invasion was excluded using log and sidewall core data. Laboratory data were of limited help because sidewall core sampling did not cover all lithologies. The interpretation of the resistivity images from the FMI showed no indications of fractures. Therefore, the anomalous Vp/Vs ratio was accepted as real.
PETROGRAPHY RESULTS
Throughout the well a number of sidewall cores were taken, including pre-volcanic sandstone, volcanic section and post-volcanic sandstone. Thin section images are shown in Figure 8.
Thin section images of different lithology types from the Santos Basin: (a) pre-volcanic sandstone, grain-supported arkose (representing the basin trend); (b) locally occurring post-volcanic sandstone, pseudo-matrix-supported lithic arkose; (c) conglomerate with fragments of volcanic and non-volcanic rocks, cemented by calcite; (d) olivine basalt with intergranular texture of reddish altered olivine crystals. The microcrystalline matrix of volcanic-related rock types (b, c, d) is the presumed reason for the observed shear weakness.
Both sandstone types exhibit a rich mineralogy and a very complex diagenetic history. The pre-volcanic sandstone consists of about 50% quartz grains, 20% feldspars and their diagenetic products, minor amounts of rock fragments, about 3% chlorite, some calcite and other minerals and 12–15% total porosity. According to Folk (1968), this is an arkose with scarce or no volcanic rock fragments and similar to other sandstones described in the basin (compare De Ros et al. 2003). The volcanic section contains mainly brecciated basalts and conglomerates of mafic character containing hyaloclastic particles (shattered material formed when lava encounters water). The FMI image suggests also the presence of pillow lava, indicating a subaqueous environment.
The detrital composition analysis of the post-volcanic sandstone shows a picture completely different from the knownbasin trend. The quartz content is reduced strongly to less than a third, the amount of mafic rock fragment and chlorite increased to about 15% respectively and total porosity reaches values up to 20% at a depth of 4600 m. The samples are classified as lithic arkoses and feldspathic litharenite, dominantly medium-grained, rich in volcanic rock fragments (compare De Ros et al. 2003).
Both sandstone types have a complex diagenetic history. The pre-volcanic sandstone is quartz and calcite cemented, with intense chemical compaction. For the post-volcanic sandstones, intensive early chloritization has prevented quartz overgrowth and preserved porosity during later stages of burial (Anjos et al. 2003).
In addition, there is another interesting detail distinguishing both sandstone types: the texture. While the pre-volcanic sandstones show grains in direct contact with each other (Fig. 8a), the grains of the post-volcanic sandstone are covered fully by chlorite rims; in parts they even seem to float in a chlorite-calcite pseudo-matrix (Fig. 8b). This texture seems to affect the elastic properties of the sandstone strongly (compare Vernik 1994).
PETROPHYSICAL PROPERTIES AND ENHANCED AVO MODEL
A possible explanation of the anomalously high Vp/Vs ratio of the post-volcanic sandstone, as described above, is suggested by petrography. Along with the high percentage of chlorite and mafic rock fragments, thin section analysis over large intervals shows this sandstone to be pseudo-matrix supported (Fig. 8b). The intergranular space is filled with a more compliant chlorite-calcite cement. The chlorite plates do not show parallel orientation as in normal compacted shales. Together with the partly even poikilotopic calcite crystals (oversized crystals enclosing smaller crystals of other minerals), they fill the intergranular space, forming a pseudo-matrix, which is believed to cause the obvious shear weakness, expressed in low shear velocities and a very low shear modulus.
Fluid inclusion studies, as well as log interpretation from resistivity data, suggest that the upper part of the sandstone is partially gas saturated. To estimate the effect of the fluids on the elastic properties, Gassmann (1951) fluid substitution models were calculated. The cross-plot of Vp/Vs ratio versus acoustic impedance (Fig. 7d) shows that one can observe the expected drop in Vp/Vs ratio for full gas saturation. It is still high but the points are separated clearly from the initial highly water-saturated reservoir (Fig. 7b). Fluid substitution modelling shows that the post-volcanic sandstone should yield a Class IV AVO response when filled fully with gas (negative intercept and positive gradient). Nevertheless, the difference between the gas and the brine case is subtle at the seismic bandwidth, making fluid prediction for this sandstone type difficult (Klarner et al. 2005).
Looking at the base of the upper reservoir, the clastic–volcanic rock interface shows a distinct brightening with offset (Fig. 6), caused by the extremely low shear velocity of the basalts (Fig. 7). Petrography implicates that again the reason might be the microcrystalline texture of the rocks (Figs 8c, d). As a result, volcanic rocks show anomalous behaviour in certain intercept/gradient-dependent AVO attributes, for example strong positive Fluid Factor anomalies (Fig. 9b). These observations open the way for use of amplitude analysis for the recognition and mapping of volcanic rocks in the area.
Comparison of the Fluid Factor attribute: (a) negative anomaly value (hot colours), imaging the top of a gas-filled channel feature; (b) positive anomalies (purple colours), imaging the tops of a volcanic layer.
The current interpretation is that the Class III AVO anomaly observed on the seismic data is an artefact from the interface between the sandstone and the basalt. In the case where the sandstone is directly above the basalt the interface between the sandstone and the volcanic rock will produce an increase in amplitude with offset for the main seismic loop, but also for the pre- and post-cursor.
Contrary to the post-volcanic sandstone, the thin sections show the pre-volcanic sandstone as clearly grain-supported (Fig. 8a). The empirical Greenberg–Castagna sandstone and shale coefficients were found to match the measured data in this sandstone as well as in the shale. The AVO response to gas for the grain-supported quartz sandstone units is expected to be a Class II, with a polarity reversal (weak positive intercepts and negative gradients). Gas-filled higher porosity would result in a Class III anomaly. Cross-plot techniques and special AVO attributes, such as Fluid Factor (Gidlow et al. 1992), enhance these effects (Fig. 9a). Therefore, conventional AVO analysis will be one of the key methods to reduce the exploration risk for this sandstone type prevailing in the Santos Basin.
ENHANCED SEDIMENTOLOGICAL MODEL
In this case, the primary target was a stratigraphic pinchout of Ilhabela gravity flow deposits in a structurally high position. The encountered sandstones, however, were deposited in a much more proximal and shallower-water position than expected (compare Porebski & Steel 2003). They are poorly sorted and very angular, indicating a rather proximal source area. Hummocky cross-stratification, coarsening-upward cyclicity and the occurrence of carbonaceous material indicate a nearshore depositional environment.
The results of the interpretation revealed a structural development in the area that has not only controlled the distribution of the reservoir formations, but also the formation of the traps and hydrocarbon migration pathways. The major bounding growth faults control the shelf edge and led to the development of rollover structures. Continued movement along the bounding fault and ongoing burial resulted in the formation of salt pillows and diapirs. Not only did the salt ridges control the geometry of the carbonates and the distribution of sand flowing into the basin, continued halokinesis led to salt withdrawal and depletion along the main fault plane. These salt windows are major areas of charge for the overlying Lower Cretaceous reservoirs. The reservoir consists of deltaic sands of Campanian–Santonian age. These deltaic sands are thought to have been deposited on the shelf edge during a lowstand event. The reservoir is defined by a series of reflections showing evidence of prograding sandbody geometries (compare Porebski & Steel 2006). Mapping isopach thickness and amplitude anomaly geometries revealed a number of feeder channels and prograding sand packages constituting beach sand ridges in front of the fluvial-deltaic plain sediments (Fig. 4).
The same holds for the shales in the section: they are not clean marine shales but partly contain significant amounts of silty particles. Finally, the restricted spatial distribution of marine shales in a shallow-marine deltaic environment causes serious problems for the definition of an effective seal.
CONCLUSIONS
As a result of the integrated team approach, an enhanced sedimentological model for the Santos Basin was developed (Fig. 10). The fluvial-deltaic and shelf-edge deposits reach far deeper into the basin than initially presumed. The updated model includes the definition of two main types of Ilhabela sandstones, which obtain different seismic properties. The basin-wide distributed arkosic sandstone (the pre-volcanic sandstone) is an acoustically hard, grain-supported sediment; the Vp/Vs ratio follows the Greenberg–Castagna equation. Hydrocarbons in this sand would display a Class II–III-type AVO anomaly. The second sandstone type (the post-volcanic sandstone) is present only locally. It contains significant amounts of redeposited volcanic rock fragments – being the time-equivalent of the main phases of volcanism in the area and acquiring completely different elastic properties. The thin section images show quartz and feldspar grains as well as volcanic lithoclasts, floating in a microcrystalline pseudo-matrix of diagenetic chlorite and calcite. This causes a distinct shear weakness. The resulting abnormally high Vp/Vs ratio, best simulated by limestone rock properties, must be considered in AVO analysis (Class IV-type anomalies). For fluid substitution modelling, this sandstone type can be identified in wells lacking shear information by its anomalously high neutron-porosity readings.
3D seismic image illustrating the depositional model over the area of interest. (1) Faulted terraces providing sediment input; (2) shallow-marine shoreface sandstones (reservoir target) and volcanics; (3) shelf-edge sandstones, containing redeposited volcaniclastics (reservoir target, well location). Figure 2represents a cross-section from the NW corner to the SE corner of the cube.
A number of wells in the Santos Basin have encountered basalts along the basin hinge line, and more volcanic rocks are identified in seismic data. In their vicinity the seismic interpreter should be aware of sandstones exhibiting anomalous elastic properties.
Acknowledgements
The authors would like to thank all their colleagues (see references) for discussion of the results and Wintershall and Petrobras for their co-operation in providing data and for the permission to publish this work. Thanks to Fernando Barbosa (Petrobras) for the AVO aspects text revision.
- 2008 EAGE/Geological Society of London
References
Enhanced depositional and AVO models for lithologically complex sandstones in the Santos Basin, offshore Brazil
