Predicting rock properties from seismic data is one of the pillars of dGB. In fact, back in 1995 we started out as a QI (Quantitative Interpretation) specialist company. Our first software system GDI focuses on integrating geology, logs and seismic data and supports stochastic pseudo-well modeling and neural networks for non-linear analysis. GDI is still a work horse for our in-house service projects that cover the entire spectrum of QI work: seismic inversion, fluid-replacement, pre- and post-stack modeling, AVO analysis, 4D etc.
With the development of OpendTect and plugins, we have broadened our technology but our passion for QI remains. More and more work flows originating from GDI are now supported in a more user-friendly way in OpendTect. For example the Neural Networks plugin supports waveform segmentation (visualizing seismic patterns along horizons) as well as inversion along well tracks. In the long run we intend to migrate all GDI functionalities to OpendTect, which should thus become the ultimate QI workbench.
Forward pseudo-well modeling and probabilistic inversion to achieve qualitative and quantitative seismic interpretation. It includes wedge models, stochastic models, pre- and post-stack synthetic seismograms and cross-matching (HitCube) inversion.
Sophisticated workflows in which model parameters are varied stochastically can be run to create a data base of pseudo-wells, representative of the expected geologic and seismic variations at target level. Such models are then used to predict rock properties with uncertainties from pre- and post-stack seismic volumes.
All kind of rock properties can be predicted with the use of the neural networks plugin. It is possible to relate any well log to seismic data. Examples are porosity, Vshale, lithology, permeability.
HorizonCube is a game-changing technology that impacts every level of seismic interpretation. A HorizonCube consists of a dense set of correlated 3D stratigraphic surfaces that are assigned a relative geological age. The technology was initially developed for sequence stratigraphic interpretation and as such was part of the SSIS plugin where it was called chronostratigraphy. Because the technology has a much wider application range the HorizonCube became a separate plugin in v.4.2. The HorizonCube can be utilized to build geo-models by interpolating well logs following the many horizons contained in the HorizonCube. Such geo-model can further be used to create low frequency models, necessary input for any model-based inversion.
Deterministic Inversion includes a 3D model builder for the construction of a priori impedance models using well log and seismic horizon data; a 2D error grid generation module to provide spatial inversion constraints and a model-based deterministic acoustic impedance inversion module.
Stochastic Inversion
Stochastic Inversion includes the MPSI UltraFast stochastic inversion module for the generation of multiple geostatistical realizations. This plugin allows the user to explore the impact of the geophysical uncertainties in the inversion process. Utilities for post-processing the multiple realizations are available to create expectation, standard deviation, and (joined) probability and trend volumes. This plugin requires a license for the deterministic inversion plugin to be in place.
ARK CLS Seismic Coloured Inversion enables rapid band-limited acoustic impedance inversion of seismic data. A single convolution inversion operator is derived which optimally inverts the data. The spectrum of the inverted data honours the available well data spectra in a global sense. Generally, traditional inversion methods (e.g. sparse-spike) are time consuming, expensive, require specialists and are not performed routinely by the Interpretation Geophysicist, whereas SCI is rapid, easy to use, inexpensive, robust and does not require expert users. SCI and unconstrained sparse-spike appear to give broadly equivalent results.
Understanding geology is the key to drilling successful wells. This means that one needs to gain understanding in the 3D distribution of lithologies, rock properties and fluid fill in the subsurface. Predicting these distributions from 3D seismic data with control points at well locations is the current industry standard. At dGB we take reservoir prediction a step further by looking beyond the numbers and use geological models to improve our prediction. Also, with the use of neural networks, we avoid using oversimplified linear models which can not accurately describe most rock property relations.