Oriol Falivene

Three-dimensional reconstruction of geological surfaces: An example of growth strata and turbidite systems from the Ainsa basin (Pyrenees, Spain)

The external and internal geometry of four turbidite systems outcropping around the Buil syncline (Ainsa basin, Spanish Pyrenees) has been reconstructed with reservoir-scale resolution in three dimensions (3-D). The irregular geometry of the syncline and the resolution required for the reconstruction cannot be resolved with cross sections. Therefore, reconstruction has been carried out with a new methodology that applies a 3-D dip-domain geometrical model and 3-D restoration techniques to achieve reservoir-scale resolution in kilometric-scale reconstructions. This methodology is aimed at resolving 3-D geometries in folded areas and regions with variable thickness stratigraphy.The 3-D reconstruction of the Buil syncline reveals the synsedimentary growth of an intrabasinal anticline and the foreland lithospheric flexure associated with tectonic loading north of the Ainsa basin.

Statistical grid-based facies reconstruction and modelling for sedimentary bodies. Alluvial-palustrine and turbiditic examples

The geological community is increasingly aware of the importance of geological heterogeneity for managing subsurface activities. In sedimentary bodies, facies distribution is an important factor constraining geological heterogeneity. Statistical grid-based sedimentary facies reconstruction and modelling methods (FRM methods) can be used to provide accurate representations (reconstructions or models) of facies distribution at a variety of scales, which can be conditioned to hard and soft data. These representations enable geological heterogeneity to be quantified; and therefore, serve as important inputs to manage projects involving subsurface activities. FRM methods are part of a general workflow comprising the construction of a surface-based framework, which is used to build the modelling grid in which these methods operate. This paper describes this workflow and provides an overview, classification, description and illustration of the most widely used FRM methods (deterministic and stochastic). Among others, two selected datasets comprising alluvial-palustrine and turbiditic deposits are used for illustration purposes. This exercise enables critical issues when using FRM methods to be highlighted and also provides some recommendations on their capabilities. For deterministic facies reconstruction, the main choice of the method to be used is between that employing a continuous or a categorical method. For stochastic facies modelling, choosing between the different techniques must be based on the scale of the problem, the type and density of available data, the objective of the model, and the conceptual depositional model to be reproduced. Realistic representations of facies distribution can be obtained if the available methods are applied appropriately.

Supervised identification and reconstruction of near-planar geological surfaces from terrestrial laser scanning

Terrestrial laser scanning is an effective method for digitally capturing outcrops, enabling them to be visualized, analyzed, and revisited in an office environment without the limitations of fieldwork (such as time constraints, weather conditions, outcrop accessibility, repeatability, and poor resolution of measurements). It is common practice in geological interpretation of digital outcrops to use visual identification and manual digitization of pointsets or polylines in order to characterise geological features using 3D CAD-like modules. Other recent and less generic approaches have focused on automated extraction of geological features by using segmentation methods, mostly based on geometric parameters derived from the point cloud, but also aided by attributes captured from the outcrop (intensity, RGB). This paper presents a workflow for the supervised and automated identification and reconstruction of near-planar geological surfaces that have a three-dimensional exposure in the outcrop (typically bedding, fractures, or faults enhanced by differential erosion). The original point cloud is used without modifications, and thus no decimation, smoothing, intermediate triangulation, or gridding are required. The workflow is based on planar regressions carried out for each point in the point cloud, enabling subsequent filtering and classification to be based on orientation, quality of fit, and relative locations of points. A coarse grid preprocessing strategy is implemented to speed up the search for neighboring points, permitting analysis of multimillion point clouds. The surfaces identified are organized into classes according to their orientations and regression quality parameters. These can then be used as seeds for building outcrop reconstructions or further analyzed to investigate their characteristics (geometry, morphology, spacing, dimensions, intersections, etc.). The workflow is illustrated here using a synthetic example and a natural example from a limestone outcrop, in which surfaces corresponding to bedding and three fault orientations were reconstructed.

Interpolation algorithm ranking using cross-validation and the role of smoothing effect. A coal zone example

For a property measured at several locations, interpolation algorithms provide a unique and smooth function yielding a locally realistic estimation at any point within the sampled region. Previous studies searching for optimal interpolation strategies by measuring cross-validation error have not found consistent rankings; this fact was traditionally explained by differences in the distribution, spatial variability and sampling patterns of the datasets. This article demonstrates that ranking differences are also related to interpolation smoothing, an important factor controlling cross-validation errors that was not considered previously. Indeed, smoothing in average-based interpolation algorithms depends on the number of neighbouring data points used to obtain each interpolated value, among other algorithm parameters. A 3D dataset of calorific value measurements from a coal zone is used to demonstrate that different algorithm rankings can be obtained solely by varying the number of neighbouring points considered (i.e. whilst maintaining the distribution, spatial variability and sampling pattern of the dataset). These results suggest that cross-validation error cannot be used as a unique criterion to compare the performance of interpolation algorithms, as has been done in the past, and indicate that smoothing should be also coupled to search for optimum and geologically realistic interpolation algorithms.

Synthetic seismic models from outcrop-derived reservoir-scale three-dimensional facies models: The Eocene Ainsa turbidite system (southern Pyrenees)

This study uses one-dimensional convolution seismic models to better understand which features of slope turbidite systems can (or cannot) be observed on real seismic data, aiming to improve subsurface seismic interpretation. Synthetic seismic sections and cubes were built from reservoir-scale three-dimensional facies models of the outcropping Ainsa turbidite system. This turbidite system developed in a foredeep and wedge-top depositional setting within a slope system. The turbidite system consists of laterally and vertically stacked sandstone- and debrite-dominated channel fills, grading into heterolithic and mudstone units, with intercalations of slump-deformed mudstone-rich units. Typical petrophysical values for subsurface Cenozoic sediments were assumed for the seismic models, which are presented at 25-, 50-, and 75-Hz resolution. Seismic models enabled the comparison between architectural and facies distributions observed in the outcrops and the geological models to their possible seismic expression in the subsurface. Comparisons show how seismic expression degrades when seismic resolution decreases. By using models at different geological scales, the effects of each heterogeneity scale are identified. Precise delineation of the internal architecture and facies distribution within channel complexes is beyond the reach of all seismic frequencies. The position of channel complex tops and margins is uncertain because of their gradual character. Differentiating between sandstone- and debrite-filled channels is not straightforward, and bed-scale heterogeneities within the sandstone-dominated channels are barely distinguishable in the seismic data. The net-to-gross predictive capability of root-mean-square amplitude extractions varies depending on the seismic frequency and unit thickness.

A geostatistical algorithm to reproduce lateral gradual facies transitions: Description and implementation

Valid representations of geological heterogeneity are fundamental inputs for quantitative models used in managing subsurface activities. Consequently, the simulation of realistic facies distributions is a significant aim. Realistic facies distributions are typically obtained by pixel-based, object-based or process-based methods. This work presents a pixel-based geostatistical algorithm suitable for reproducing lateral gradual facies transitions (LGFT) between two adjacent sedimentary bodies. Lateral contact (i.e. interfingering) between distinct depositional facies is a widespread geometric relationship that occurs at different scales in any depositional system. The algorithm is based on the truncation of the sum of a linear expectation trend and a random Gaussian field, and can be conditioned to well data. The implementation introduced herein also includes subroutines to clean and geometrically characterize the obtained LGFT. The cleaned sedimentary body transition provides a more appropriate and realistic facies distribution for some depositional settings. The geometric measures of the LGFT yield an intuitive measure of the morphology of the sedimentary body boundary, which can be compared to analogue data. An example of a LGFT obtained by the algorithm presented herein is also flow simulated, quantitatively demonstrating the importance of realistically reproducing them in subsurface models, if further flow-related accurate predictions are to be made.