INTRODUCTION

This special issue focuses on the integration of different disciplines, ranging from seismic sequence stratigraphy and seismic interpretation to stratigraphic and basin modelling. Integration has been a recurring topic for several decades in the industry, but it is a topic that requires constant attention to bring to new levels of maturity in the workplace. The question is why – especially since the industry has widely acknowledged that the integration of disciplines is vital. Looking at possible factors, integration is often difficult due to lack of knowledge of the other disciplines, a perennial need for well-documented examples, and obstacles to communication. With this special issue of Petroleum Geoscience, and the EAGE Research Workshop in September 2006 from which these articles derive, we are contributing to discipline integration by communicating results of real-life case studies. Improving communication and exchanging results generated by these different disciplines will enable us to achieve a better overall understanding of the geological setting in basins.

For example, basin modellers are aware that their models can be improved by additional information, such as porosity values, lithology variation within formations (Hantschel 2006), fluid migration pathways (Ligtenberg & Thomsen 2003) and detailed structural data, available from application of sequence stratigraphy, stratigraphic modelling, interpretation and inversion of seismic data (Klarner et al. 2008). Similarly, sequence stratigraphers should be able to derive useful additional information from seismic data and could be assisted by information from stratigraphic and basin modelling. However, such integration of different methods is often hampered due to unstructured communication between the different disciplines involved and insufficient case studies to establish best practice.

The current practice of building reservoir models often patches together separate models for the different data, which can lead to inconsistent results without truly balancing scales, resolution and coverage. For example, to achieve integration, what type of seismic resolution is required to image certain geological targets (Rabineau & Nouze 2006)? How accurate does time-to-depth conversion need to be? Are we aware of uncertainties created in time-to-depth conversion, which may be carried forward automatically by reservoir model building software (Jardin et al. 2006)?

SEISMIC SEQUENCE STRATIGRAPHY

This discipline analyses, from seismic images, the sedimentary response to changes in base sea level and the depositional trends that follow from the interplay of accommodation and sedimentation. It has vast potential to improve predictive models for economic hydrocarbon exploration and production.

Seismic sequence stratigraphy originated in the 1960s with facies models. However, it really came alive in 1977 with AAPG Memoir 26, Seismic Stratigraphic Applications to Hydrocarbon Exploration, with significant contributions from Exxon (Vail et al. 1977). Ever since this landmark publication, seismic stratigraphic interpreters have derived stratigraphic information from seismic data instead of restricting themselves to the more traditional interpretation of structural features. Such stratigraphic interpretation relied on the identification of reflection terminations on sparse, 2D time sections. Traditionally, the seismic stratigraphic interpretation is carried out by:

• identifying the termination of seismic reflectors at discontinuity surfaces;

• establishing which of these surfaces form the sequence boundaries;

• numbering the sequence boundaries into stratigraphic order;

• classifying and labelling the stratal geometries of the sequences and the associated system tracts, validated and calibrated to well-log data through modelling.

The major impact stirred by the publication by Vail et al. (1977) was followed by later improvements, published in AAPG Memoir 39 (Berg & Woolverton 1985) and in SEPM Special Publication 42 (Wilgus et al. 1988). However, the discipline became less popular in the late 1990s.

Currently there is a re-awakening of interest in this discipline. One reason is the steadily increasing importance of understanding and finding subtle stratigraphic traps (e.g. Santos and Campos Basins offshore Brazil, Middle East and elsewhere). Another reason for this revival is the quality and accessibility of 3D seismic data, combined with the capabilities of recent software and powerful workstations that enable even 3D interpretation of chronostratigraphic domains. It started with automatic 2D Wheeler diagram generation and interpretation in the Sismage package developed by Total (Keskes 2002), followed by isochronous flattening methods by Stark Reality (Stark 2004) and, subsequently, creation of 3D Wheeler diagrams and system tract analysis in OpendTect SSIS by dGB (Ligtenberg et al. 2006; de Bruin et al. 2007). The workflows in these packages provide a better understanding of sediment deposition, erosion and geological timing, leading to an improved prediction of lithofacies and potential stratigraphic traps. These are tools that enable sequence stratigraphers to improve their work and speed up their workflow, but there remains room for improvement.

SEISMIC INTERPRETATION

Seismic interpretation is critical for seismic sequence stratigraphy, but there is more to seismic interpretation than picking horizons: the modern interpreter has access to a vast array of pre- and post-stack seismic attributes in multiple 3D volumes, within which technologies such as neural networks can help to detect trends, classify geobodies and populate interpreted models with properties.

Seismic attribute analysis, using seismic amplitude maps, spectral decomposition (Partyka et al. 1999), coherency and acoustic impedance inversion, offer a means of assisting the interpreter in revealing very subtle changes in the seismic response (Brown 2004; Chopra & Marfurt 2006). It is based on the relationship between lithofacies and acoustic/elastic impedance in layers delineated by seismic reflections. By analysing the local characteristics, such as waveform variability, it is possible to identify changes that may reflect seismic facies variation and to associate those facies with the stratigraphic development as recognized from well-logs. It is imperative to correlate the observed seismic signature with that from synthetic traces computed from impedance logs. These results will provide a better insight into the depositional setting and distribution, and can assist in finding stratigraphic traps.

The computation of geometrical (coherence and curvature) and lithostratigraphic (acoustic and elastic impedance) seismic attributes, subsequently integrated, validated and calibrated by well-log data, is invaluable. This powerful data integration approach, which can be utilized in either a regional or a prospect context, is exemplified in this issue by Al-Zahrani & Neves (2008) using onshore data and by Klarner et al. (2008) using deep-water offshore data.

A further development is the combination of seismic attributes by means of neural network technology (e.g. Aminzadeh & de Groot 2006). Clustering of seismic waveforms and/or clustering of seismic attributes assists in interpreting individual sedimentary elements or geobodies. Seismic geomorphology is the latest outcome of our improving ability to handle 3D volumes effectively (e.g. Posamentier 2004), in contrast to the traditional 2D approach to seismic sequence stratigraphy as originally conveyed in 1977.

Nevertheless, some theoretical and practical challenges for a more efficient and robust stratigraphic interpretation still remain, such as the old dilemma of what the seismic reflections are really showing, i.e. whether seismic horizons are chronostratigraphical or lithostratigraphical surfaces. This topic is addressed in this issue via forward stratigraphic modelling by Cacas et al. and Teles et al. in a carbonate depositional setting. Most current seismic interpretation packages require seismic horizons to be picked in a phase-consistent manner; that is, a seismic event is picked as a peak or a trough throughout the entire seismic volume. However, seismic phase polarity reversal can be associated with change in depositional setting along a chronostratigraphic horizon. Furthermore, some important geological time-lines do not give rise to large impedance contrasts and therefore lack seismic expression, due to their gradational character, e.g. when changing from offshore transition zone mudstones to shoreface sandstones.

Seismic interpretation and basin modelling are also strongly linked. Basin or petroleum system modelling receives direct structural input, such as horizons and faults, from seismic interpretation. In addition, seismic attribute and neural network analysis results can populate the model with rock properties (Hantschel 2006). Interpretation of fluid migration pathways may either be input to basin models or may allow validation of the kitchens and hydrocarbon flow paths postulated by petroleum system modelling (Ligtenberg & Thomsen 2003). Frequency-dependent seismic attributes, from spectral decomposition of seismic data in an interval of interest, and supported by geochemical analysis, were shown to identify coal beds and channels in the Jurassic section in the Danish North Sea. A direct correlation exists between these coal beds and gas migration pathways and is presented in a Grenoble paper by Schmidt (2006).

Okui et al. (2008) observed anomalies on seismic data, interpreted as hydrocarbon indicators. However, it turned out that volcanic material played an important role behind these seismic anomalies, demonstrating that both lithology and fluids can affect seismic data. Understanding the regional geology, with its tectonic and sequence stratigraphic history, is crucial. Similar observations were made in the Santos Basin, offshore Brazil, by Klarner et al. (2008). They carried out a detailed study on the different responses of lithologies, especially a mixture of sandstones and volcaniclastics, by means of AVO, fluid inclusion and petrophysical property analysis to understand better the seismic character of these complex rock types.

STRATIGRAPHIC MODELLING

Stratigraphic modelling has been recognized as a discipline since the 1980s, but it is still quite new on the exploration stage. Stratigraphic modelling simulates sedimentary processes at basin scale and geological time-scale, incorporating sea-level changes, uplift, subsidence, changes in accommodation space, sediment flux and erosion rates. Stratigraphic modelling may provide a new and improved insight into the distribution of different sequence stratigraphic deposits and should assist the seismic interpreter in improved identification of potential reservoirs.

Most of the stratigraphic modelling contributions at the Grenoble workshop were presented as validation tools for seismic interpretation (Cacas et al. 2008; Teles et al. 2008). However, the goal for stratigraphic modelling is to develop a predictive tool that could lead to new potential prospects and plays. A number of interesting predictive examples were presented by Waltham et al. (2008) and by Griffiths et al. (2006). Waltham et al. (2008) present a workflow modelling turbidity currents to obtain a good understanding of ancient turbidite sediment distributions and they apply the workflow to a North Sea prospect to reduce the exploration risk.

It is important to understand the scaling, principles and limitations of stratigraphic modelling. For example, how to constrain stratigraphic models with seismic data? Conversely, how to constrain seismic interpretation with stratigraphic modelling? Cacas et al. (2008) address the question of integrating stratigraphic modelling and seismic interpretation. They introduce a new and simplified stratigraphic modelling method which approximates the sedimentary history by a succession of sequences laid down during steady-state sedimentation. Their approach allows sedimentary settings and time-interpreted seismic sections to be considered as input data to the stratigraphic modelling schemes. Furthermore, they show the value of seismic data as a validation tool for different sedimentary models, but caution that the importance of the former depends both on the seismic modelling methods as well as the input acoustic parameters.

Basin modellers are also interested in incorporating stratigraphic modelling results into their basin models. Upscaling is required, but how? It is still an unsolved problem requiring further study. As with any modelling work, many parameters are involved, which also introduce their own uncertainties. Knowledge of these uncertainties and their impact on the modelled results is crucial. None the less, stratigraphic modelling is developing into an important discipline – not to be overlooked – especially when considering the growing need for finding increasingly subtle stratigraphic traps.

BASIN MODELLING

Petroleum systems modelling, or basin modelling, has already become a platform for integration of different disciplines (Hantschel 2006). We observe direct connections with, for example, structural modelling, fault seal analysis, geochemistry, sequence stratigraphy, seismic attribute and parameter inversions, and other analysis workflows. Okui et al. (2008) clearly illustrate the value of discipline integration to unravel the petroleum system in the deep-water area, offshore Japan. Seismic interpretation, sequence stratigraphy concepts, fluid inclusion analysis and limited well data were used as input. Step by step, they explain how these different sources of information were combined in a basin model to shed light on the many unknowns of this petroleum system.

As with stratigraphic modelling, many parameters and factors are involved, with their associated uncertainties. How to manage these uncertainties? Sylta (2008) shows the importance of a tight coupling of uncertainty from seismic data into hydrocarbon migration modelling. Iteration, working back and forth with seismic interpretation, is valuable, leading to the strong request from basin modellers to be involved at very early stages of exploration seismic interpretation, providing relevant input to interpreters, and vice versa. A requirement is good and clear communication on uncertainties involved in each set of results before handing over the results to another discipline. Furthermore, a better understanding of the other disciplines assists in dealing with uncertainties.

Basin modellers would clearly benefit from obtaining as much information as possible from seismic interpretation, sequence stratigraphy and stratigraphic modelling. These disciplines could help in improving the understanding of source-rock distribution, trap generation, migration pathways and the quality of the top seal. Tømmerås & Mann (2008) present a workflow to model the deposition and preservation of source-rock material in basins that will lead to a better modelling of hydrocarbon generation and expulsion in petroleum systems.

CONCLUSION

The integration of geophysical and geological workflows, facilitated and made economically viable by ever more powerful computer hardware and software packages, is instrumental to a wider adoption and integration of seismic interpretation, sequence stratigraphy, stratigraphic modelling and basin modelling in the oil and gas industry. How should one improve integration of disciplines? By exchanging knowledge, workflows and expertise at meetings such as the Grenoble research workshop in 2006 and by documenting experience and best practice. We hope this issue may stimulate new ideas and contributions for a future research workshop on the integration of disciplines.

Finally, does integration of disciplines reduce uncertainty of interpretation, or are we adding complexity and increasing uncertainty? At the EAGE Grenoble workshop it was agreed that we are definitely reducing risk, but there is a price: complexity. A sound understanding of different disciplines and their uncertainties is therefore critical. Please judge for yourself by reading the papers of this special issue. We believe you will find them interesting and informative, fostering the integration of both subsurface practitioners and their data.