--> Abstract: Comprehensive Karst Delineation from 3D Seismic Data, by Gaynor Fisher, David Hunt, Arnout Colpaert, Brita G. Wall, and Jonathan Henderson; #90105 (2010)
[First Hit]

Datapages, Inc.Print this page

AAPG GEO 2010 Middle East
Geoscience Conference & Exhibition
Innovative Geoscience Solutions – Meeting Hydrocarbon Demand in Changing Times
March 7-10, 2010 – Manama, Bahrain

Comprehensive Karst Delineation from 3D Seismic Data

Gaynor Fisher1; David Hunt2; Arnout Colpaert2; Brita G. Wall2; Jonathan Henderson1

(1) ffA, Aberdeen, United Kingdom.

(2) StatoilHydro, Bergen, Norway.

Introduction

3D seismic data has the potential to provide an enormous amount of detailed information about karst and dissolution features. However, horizon based interpretation is not well suited to the analysis of such features and, in addition to being very time consuming, reveals little or no information on the paleokarst network and connectivity between karsts within a carbonate target. Without a good understanding of karst distribution it is difficult to compile a comprehensive geological model and appreciate the impact such structures will have on porosity evolution and reservoir quality (Budd et al., 1995; Neuhaus et al., 2004).

Although karst and dissolution features may be extensive, they can be hard to identify in Previous HitreflectivityNext Hit data due to their variable seismic character. The application of Image Processing and Analysis (IPA) workflows enables a rapid and detailed examination of Carbonate features, including karsts, as well as reefs / buildups and clinoforms. The IPA workflow employs attribute analysis to highlight the location and extent of these features, and then 3D geobody delineation techniques are applied to allow the 3D geometry of the highlighted features to be examined in detail. A major strength of these workflows is the built-in capability to detect karsts and dissolution features across a wide range of scales and with diverse morphologies.

This paper describes the IPA workflow and illustrates its application to the interpretation of upper Palaezoic carbonates in the Loppa High area of the Norwegian Barents Sea. Here it is estimated that some 300-500 m of uplift, erosion and karstification of a mixed carbonate-evaporite succession occurred during c. 20 million years of subaerial exposure (i.e. Roadian-Induan times). Major drainage systems can be traced across basement rocks and into and through the karstified carbonate succession. The carbonates are cut by steep km-scale canyons and penetrative sinkholes. The dataset shows a range of contrasting paleokarst features, so that some of the key seismic attributes and spectral decomposition methods used to delimit contrasting genetic elements of paleokarst systems can be illustrated. Results from the seismic data analysis have been quality-controlled against well data and horizon-based interpretations.

Methodology and Results

Application of data conditioning techniques to improve the signal to noise ratio is an important pre-processing stage in the workflow. With the large Previous HitvariationNext Hit in structure, lithology and continuity of stratigraphic intervals within the area of interest, different data conditioning filters were applied to different intervals in the data (Figure 1). This ensures that the appropriate level of noise suppression is applied to maximise the signal to noise ratio in each stratigraphic interval without removing information that is important for karst identification.

Figure 1. (a) Section through the original data (b) Data after application of interval dependent data conditioning. The green dotted lines define the boundaries of the three intervals to which separate noise attenuation workflows were applied.

Carbonate stratigraphies are represented in seismic data by a wide range of frequencies, so an investigation into the distribution of energy with frequency can provide a good overview of the carbonate features present. Frequency Decomposition generates a series of volumes that show the magnitude of the response for a range of frequency bands. It provides a more sensitive method of analyzing the seismic data than the full frequency amplitude response. Frequency decomposition volumes are most effective if the response in different frequency bands can be compared. This is done very effectively by blending three individual response volumes using a Red-Green-Blue (RGB) 3D colour space, which allows the interplay between the three frequencies to be easily visualised. Several different types of carbonate features can be highlighted using this RGB blending technique including karsts, mounds, clinoforms, and patch reefs as well as faults and fractures. It is commonly challenging to accurately visualise or analyse the extent of karst features using either Previous HitreflectivityNext Hit or full frequency magnitude data. However the RGB blend makes it simple to determine the location and extent of the karst features which can then be interrogated in a volumetric manner (Figure 2).

Figure 2a

Figure 2b
Figure 2. RGB blend of three frequency response volumes (Red = high, Green = Mid, Blue = Low frequency). (a) karsts highlighted as anomalous low frequency (blue) areas, surrounded by brown build-up and black inter-build-up areas (b) canyons feeding into a subterranean drainage system.

The third stage in the workflow aims to highlight structural variations associated with the karst features. This is achieved by independently identifying the edges of the karst boundaries and the zones of disturbance representing the karst infill. A Tensor Diffusion algorithm, which uses a combination of eigenvalues to identify discontinuities in the data, was the most consistent of the techniques investigated for identifying karst edges. The Tensor attribute is influenced by amplitude, so that karst features associated with abrupt amplitude changes are highlighted particularly effectively (Figure 3b).

Diagenetic features such as karsts or late dolomites can contain chaotic reflectors either in vertical cones, towers or in laterally-extensive cavern-like structures. This structural disturbance can be measured in several ways depending on the scale required. One of the most effective volumes created to highlight all zones of structural disturbance was the Conformance attribute. This attribute uses vector analysis to measure the stability of the reflector orientation within a defined footprint, and enables rapid identification of areas of disturbance (Figure 3c).

Figure 3a

Figure 3b

Figure 3c

Figure 3. Karst attributes. (a) The noise cancelled Previous HitreflectivityNext Hit data (b) The Tensor attribute highlighting the edge of the karst features. (c) The Conformance attribute highlighting zones of structural disturbance.

When combined the Tensor and Conformance attributes are very effective at isolating the karsts within a corrupted or structurally disturbed zone (Figure 4). Within the Tensor - Conformance attribute colour change represents a change in structural conformance; colour saturation represents an increased response in the Tensor attribute. Pink areas represent chaotic areas, dark pink / black areas represent karst boundaries. This enables accurate delineation of the boundary between the corroded zone and the overlying sediment. It also enables the karsts and pipes to be rapidly identified when viewing a time slice through the volume.

Figure 4. The Tensor - Conformance attribute combination.

Application of a connected components technique (Body Labelling) (Figure 5), to the attributes described above, is then used to delineate the karst features as 3D objects allowing their size and extent to be easily interrogated.

Figure 5. Body Labelling showing the 3D shape and connectivity of dissolution features.

Well Correlation

Initial well correlation has confirmed the ability of the IPA workflow to define geobodies that enable the distribution and inter-connectivity of the karst features to be more accurately defined and understood.

Summary

IPA techniques have been developed that can be used to successfully and robustly highlight and extract karst features, in an accurate and objective manner. These techniques are based on analysis of three different seismic characteristics, frequency, amplitude, and structural Previous HitvariationNext Hit. Each one analyses the data in a 3D volumetric manner independent of any interpreted horizons. The output volumes create a comprehensive suite of attribute and geobody volumes that are repeatable across data sets from different geographical areas, and stratigraphic epochs. The results have been correlated with well data and have enabled a better understanding of a variety of diagenetic features in multiple data sets. A significant finding of the study has been that different seismic properties/attributes are required to recognise and extract paleokarst features formed by different processes.

References

Budd, D.A., Saller, A.H., and Harris, P.M., (Eds.) (1995) Unconformities and porosity in carbonate strata. AAPG Memoir, 63, 313.

Neuhaus, D., Borgomano, J., Jauffred, J.-C., Mercadier, C., Olotu, S., and Grötsch, J. (2004) Quantitative seismic reservoir characterisation of an Oligocene - Miocene carbonate buildup: Malampaya field, Phillippines. In: Eberki, G.P., Masaferro, J.L., and Sarg, J.F. (Eds.) Seismic Imaging of carbonate reservoirs and systems. AAPG Memoir, 81, 169 - 920.

689592_A.jpgFigure 1. (a) Section through the original data (b) Data after application of interval dependent data conditioning. The green dotted lines define the boundaries of the three intervals to which separate noise attenuation workflows were applied.

689592_B.jpgFigure 2a

689592_C.jpgFigure 2b Figure 2. RGB blend of three frequency response volumes (Red = high, Green = Mid, Blue = Low frequency). (a) karsts highlighted as anomalous low frequency (blue) areas, surrounded by brown build-up and black inter-build-up areas (b) canyons feeding into a subterranean drainage system.

689592_D.jpgFigure 3. a,b,c Figure 3. Karst attributes. (a) The noise cancelled Previous HitreflectivityTop data (b) The Tensor attribute highlighting the edge of the karst features. (c) The Conformance attribute highlighting zones of structural disturbance.

689592_E.jpgFigure 4. The Tensor - Conformance attribute combination.

689592_F.jpgFigure 5. Body Labelling showing the 3D shape and connectivity of dissolution features.