Geostatistical data conditioning
Checkshots, sonics, well tops, horizons, seismic velocities… A careful check of the available data must be conducted before using them as input for a velocity model. A proper data conditioning will prevent the propagation of artefacts in the model.
Currently existing commercial packages generally offer a limited set of conditioning tools for seismic velocities. An example with poor smoothing operators such as median or Gaussian filters: 1km x 1km smoothing, 2km x 2km smoothing, etc. Using these operators on most dataset can lead to a double negative outcome:
– the noise is spread over the dataset and not fully removed
– the data resolution is deteriorated as short scale geological structures are lost during the smoothing process.
A key step of our geostatistical QC & conditioning flowchart is the geostatistical filtering of seismic velocities. It is known that PSTM stacking velocities used for building a velocity model are often affected by different types of noise such as random noise, picking artefact, acquisition footprints, multiples, etc. Based on factorial kriging with local parameters, the ESTIMAGES M-GS® filters allow to remove any kind of noise from seismic velocities while preserving short scale geological structures.
M-GS® gridding
Gridding 2D seismic velocity lines or merging several seismic velocity datasets require the use of reliable spatialization algorithms. Dedicated workflows help to avoid artefacts such as line and edge effects and ensure the preservation of a maximum lateral and vertical resolution regarding data sampling and coverage.
Our M-GS® gridding solutions consist of 3D kriging algorithms with varying parameters. They take into account local structural characteristics of seismic velocities and their coverage. Guided by a couple of key horizons (when available), algorithms used by ESTIMAGES provide the best possible velocity estimation. If any, artefacts are detected in a spatial qualification phase and filtered out during the gridding process: an artefact structure is added to the M-GS® kriging model.
Solutions currently on the market tend to generate smoothed velocities volumes where resolution is degraded. Our gridding solutions give the best velocity estimates with a preserved resolution and consequently improve the quality of the final velocity model.
3D calibration (PSTM/PSDM)
Have you ever experienced data consistency issues when trying to reconciliate seismic velocities (PSTM/PSDM) with wells data? ESTIMAGES builds velocity models respecting both geological structuration and well data with a minimum depth error far from the wells thanks to a 3D geostatistical calibration solution.
A full 3D approach is required when building a 3D velocity model as it is essential to preserve 3D heterogeneities coming from seismic velocities. Our M-GS® calibration solution is based on a 3D structural analysis of the differences between wells and seismic velocities. This phase allows us to determine the spatial variations of parameters, such as the 3D range of influence of the wells. Optimal local parameters are then used into a 3D M-GS® bivariate kriging process in order to get an optimal scaling factor and to calibrate the seismic velocities to the wells velocities. The 3D process is performed on a stratigraphic grid built from the available key horizons, this ensures a geological consistency and the representativity of anisotropy ratios.
Building an accurate and consistent 3D velocity model is something complex. So it is a matter of choosing the right model and the right parameters. Our expertise and experience makes all the difference.
M-GS® surface model (PSTM/PSDM)
Many methods exist for building a velocity model based on the calibration of horizons to the wells tops. But unlike ours, none of them ensures an optimal compromise between 3D heterogeneities preservation, structural interpretations integrity and local fit to the wells data. For instance, a layer cake / multi-layer model cannot represent 3D heterogeneities coming from seismic velocities because it stacks them between surfaces. As a result, at your target horizons the estimation of the depth values between wells is less accurate.
Getting the best depth surfaces is made possible by our four steps workflow:
– conditioning seismic velocities with geostatistical filters,
– 3D calibration of seismic velocities to the wells velocities (checkhots, sonics calibrated to the checkshots, etc.),
– depth conversion (PSTM) or recomputation (PSDM) of the horizons with the 3D model,
– calibration of the surfaces to the well tops. In this last phase we use the M-GS® approach combined with a patented cross-validation method in order to get the optimal geostatistical parameters. The cross-validation allows lateral variations of the range of influence instead of using an arbitrary global range (5km? 10km? etc.).
This integrated solution values all sources of information, making sure that you get an optimal prediction of your key depth surfaces. Of course, you have the possibility to combine the M-GS® surface model with our uncertainties estimation solutions which are suitable for any scale (regional to reservoir).
Uncertainties (regional or reservoir)
What is the potential error associated with my depth estimations? From regional to reservoir scale, we offer refined uncertainty solutions based on advanced geostatistics. All sources of uncertainties – data, modeling, parameters – are integrated into one consistent uncertainty model.
We are the only oil and gas market player offering uncertainty volumes consistent with large scale velocity models (regional or licence scale). For any given interpretation on your area of interest, we provide you with the potential depth error associated to the use of the velocity model. For the estimation of the uncertainty volume we use an innovative data-driven geostatistical model where the uncertainty is related to the geological layers and their respective mean sea level positions. This is a unique and robust approach leading to a realistic and accurate uncertainty estimation.
For reservoir studies, our M-GS® workflow produces an optimal uncertainty estimation. Simulated depth surfaces rely on a non-stationary model integrating all possible sources of uncertainties, from processing phases to the data themselves (seismic velocities, horizons, well tops, time-depth functions). In some studies, we also integrate an uncertainty component for the lateral anisotropy which impacts seismic velocities. With our reservoir solutions, you can associate reliable uncertainties to your base case scenario and even go further, by asking us to conduct a volumetrics study for you.
Years of experience and high level of expertise are necessary to achieve high quality uncertainty outputs. We believe that a state-of-the-art data conditioning, the use of a tailored and relevant workflow and the tuning of a geostatistical model are key conditions for success that can not be fulfilled solely by clicking on some buttons in existing commercial software.