Thursday, 11 May 2017

Frequency Decomposition Part 3 - HDFD (High Definition FD)

The previous two blog posts looked at 'standard' frequency decomposition techniques which applied convolution of the trace with bandpass filters in a traditional manner.  This post focuses on the High Definition Frequency Decomposition or HDFD.

Part 1 - Constant Bandwidth
Part 2 - Constant Q

Part 3 – High Definition Frequency Decomposition (HDFD)

Link to tutorial video here

The High Definition Frequency Decomposition (HDFD) algorithm uses a different approach to the ‘standard’ frequency decomposition filters. The application of a modified matching pursuit algorithm allows trace reconstruction with precise vertical localisation.

Matching pursuit is a trace based form of frequency analysis and decomposition. It uses a dictionary of Gabor wavelets which are correlated to each event in the seismic data individually. Once an event has been matched to a wavelet, it is extracted from the trace and the next event is matched. This iterative process continues until 99% of the trace energy has been matched and a synthetic trace has been generated.

Reconstruction of a particular frequency response is achieved by summation of the response of all wavelets that intersect the desired frequency. The relative proportion of the response included from each wavelet is determined by the degree of overlap of the bandwidth of each wavelet with the desired frequency. This is why the bandwidth of HDFD responses is so wide (and why the vertical resolution is so good).

First optimisation pass. a) Atoms matched during the first matching pass (red, blue and green) have been co-optimised to find the best combination of amplitudes to fit the seismic trace (black) over the region of the atoms’ overlap. b) The effect of co-optimising multiple atoms at once is to provide a better approximation (orange) to the seismic trace in regions of constructive or destructive interference between the atoms.

Two options of HDFD are available, one producing the best possible vertical resolution and one with an improved colour resolution.

When using the colour resolution option, the data is split into three band-limited versions of the input data using a modified FFT. Then the modified matching pursuit algorithm is applied separately to each of the three components and the results are combined to produce the final outputs.

Illustrative example of frequency splitting used in HDFD with colour resolution option

When using the vertical resolution option, no splitting is carried out and the modified matching pursuit algorithm is applied directly to the input data.

Monday, 24 April 2017

Frequency Decomposition Part 2 - Constant Q

This weeks blog post continues to look at frequency decomposition techniques available in GeoTeric. We previously looked at the Constant Bandwidth technique, we now look at another Standard Frequency Decomposition technique, the Constant Q.

Link to Part 1 - Constant Bandwidth

Part 2 – Constant Q

Tutorial video can be found here.

The Constant Q method utilises bandpass filters to carry out decomposition with properties similar to a Constant Wavelet Transform (CWT).  The main benefit of this technique is that due to variable filter lengths and bandwidths there is a reduced filter length at higher frequencies, therefore these bands provide an increased vertical resolution whilst the result can still be processed quickly over large volumes.

Comparison of Constant Q Blend Vs Seismic 

There are two Constant Q options available: Uniform Constant Q and Exponential Constant Q. One may be beneficial over the other to achieve a good frequency decomposition colour blend depending on the frequency spectrum of the data.

Friday, 14 April 2017

Frequency Decomposition Part 1 - Constant Bandwidth

The Standard Frequency Decomposition module uses bandpass filters to carry out decomposition with properties similar to either a Fast Fourier Transform (FFT) or to a Constant Wavelet Transform (CWT).  Due to the nature of the waveform transformation between the frequency and time domains there is resulting uncertainty, as defined by uncertainty principle. Therefore, the different frequency decomposition methods show differences between the frequency resolution and temporal resolution with the two being traded off against each other.

Over the next three weeks the blog will focus on the different frequency decomposition techniques available in GeoTeric with accompanying tutorial videos. In part 1 we will look at the Constant Bandwidth decomposition method.

Part 1 – Constant Bandwidth

Watch the Tutorial Video here

The Constant Bandwidth method utilises bandpass filters to carry out decomposition with properties similar to a Fast Fourier Transform (FFT). It is a good reconnaissance option since it provides excellent frequency resolution and is fast to process; however, it tends to lack the vertical resolution of other available techniques required for detailed reservoir scale analysis. Constant Bandwidth mode is generally used when the aim of the decomposition is to differentiate between different geological elements on the basis of their bulk properties, for example delineation of large channel systems, salt bodies or gas chimneys.

When using the Constant Bandwidth method the bandwidths and filter lengths of the individual frequency bands are the same, which allows a like-for-like comparison between the bands. The filter lengths are generally high and therefore there is a large amount of vertical smearing, however due to the relatively narrow bandwidths high frequency resolution can be achieved.

The filter length is controlled by the minimum frequency set; a low minimum frequency will require a longer filter length to sample, if this is increased a smaller filter length can be achieved. The bandwidth scale can also impact the filter length – a narrower bandwidth will require a larger filter length and vice-versa.

Thursday, 6 April 2017

The Stratigraphic Slicing Workflow

With the release of GeoTeric 2016.2.1, the user can now take advantage of the new Stratigraphic Slicing workflow. This workflow allows the user to rapidly create a series of stratigraphically conformant slices which can then be used to extract data from any of the volumes or colour blends available in their project. The Stratigraphic Slicing workflow consists of the following steps:

  • Creating the Interpretation HSV colour blend
  • Picking top and bottom surfaces
  • Creating intermediate slices using the Iso-Proportional Slicing (IPS) tool
  • Creating a Horizon Pack, which is new to GeoTeric 2016.2.1, and
  • Using the Horizon Pack to extract data from a volume or blend

Key benefits

The Stratigraphic Slicing workflow provides a rapid, novel approach to understanding both the structural and stratigraphic features of your dataset. Key benefits include:
  1. Interpretation based on detailed phase information
  2. Better visual separation of packages
Image 1. Picking on a particular phase angle and comparison of visual character

  1. Reduced cycle skipping in auto-tracking algorithms
Image 2. Explicitly encoded phase information reduces cycle skipping.

  1. The ability to see and honour fault information
Image 3. Picked horizon stops at faults.

  1. The ability to create meaningful surfaces through very challenging packages
Image 4. Challenging package can be imaged meaningfully and rapidly.

  1. A very short turn around time, allowing the user to quickly build up a full package of slices through their dataset.

Below is an example of the results that can be achieved in a few short hours.


Instructions for the the Stratigraphic Slicing workflow can be found in the following VIDEO

For further information please contact GeoTeric support on

Monday, 27 March 2017

GeoTeric 2D

GeoTeric 2D was released last week!

GeoTeric 2D is an Expression Tool optimised for attenuating noise in 2D data. The example driven, preview based interface allows you to rapidly optimise the filters for each 2D line. Multi-line batch processing means that you can easily apply the optimised parameters over your full 2D data set.

GeoTeric 2D Interface

GeoTeric 2D reads and writes the data directly from SEG-Y, avoiding any data loading/exporting issues.

Three filters are available:
  • SO FMH: It is a structurally oriented and edge-preserving filter which uses a combination of median and mean calculations, ensuring that both the coherent and random noise are attenuated. This filter ensures that details like edges, corners and sharp dips in the structure are preserved.
  • SO Mean: It is also a structurally oriented filter, using a more aggressive mean calculation, which increases the continuity of the reflectors and produces a smoother result.
  • Mean: It is a grid oriented filter. It is useful for situations of very low coherency and chaotic data, where structurally oriented filters will struggle.
The choice of filter to use will depend on the objectives of the noise cancellation: if the objective is to map a regional horizon, an aggressive filter like SO Mean will produce a smooth result and allow for an easier auto-tracking of the data. When looking at a reservoir scale, preserving subtle changes in the data is important, so SO FMH will produce a result which preserves the subtle breaks and amplitude changes in the data.

The interface shows 9 swatches, with three default filter sizes for each of the filters. Each filter can be further optimised by changing the filter size using the slider bar at the bottom of the main display.

A before/after comparison slider is available in the main panel, allowing for an easy QC of the results. The difference (result minus original) can also be visualised by clicking on the Difference button at the top.

The multi-line batch processing allows you to apply the same filter to multiple lines in a batch process, so a whole 2D survey can be conditioned in a matter of minutes.

This video shows how to use the GeoTeric 2D Noise Expression tool.

Subscriber Notice - We are in the process of launching our new website, as well as changing the location of our blog. If you have subscribed to the GeoTeric Blog via email, then your subscription will be taken across to the new blog and you will be notified as normal. If you follow by RSS feeds, you will need to update your own preference. Stay tuned for more information - posts will continue via Blogger for the time being.

Thursday, 16 March 2017

Attenuating Migration Smiles in Your Data

Written by Thomas Proença, Geoscientist, Brazil

A common issue observed in sub-salt imaging is migration artifacts such as "smiles" (Figure 1).  Migration smiles occur when the velocity survey used in the depth correction contains imperfections.  When re-processing is not an option, a specific data conditioning workflow can be applied within GeoTeric with the aim of removing/reducing migration noise and increasing the signal to noise ratio in the data; thus making it clearer and easier to interpret.

Figure 1 - Migration smiles

Noise Attenuation Workflow

The first step to removing the migration smiles is to create smooth steering volumes, with the aim of reducing the contamination of the steering dip and azimuth.  This is done by applying a grid oriented smoothing filter of the original seismic data. Migration smiles are the most problematic part in the Z direction of the seismic data, using a large filter in X and Y directions will attenuate its vertical response. A typical filter size for this process is 9x9x5. 

The smoothing attribute can be found at:

Reveal tab > Processes and Workflows > Processes > Attributes > Stratigraphic attributes

After the original data has been smoothed, the steering volumes can now be calculated.  The recommended filter size for dip and azimuth is above 15 (note that if the bin sizes are different between the X, Y and Z directions, the filter size for the steering volumes should be adjusted to compensate for the difference and make them square).

Once the steering volumes are generated, they are used along with the original seismic data as inputs to the SO Noise filter found under the Processes and Workflow menu. The recommended filter size for this process is between 5 and 9. If the data is extremely noise to get better result the SO Noise filter can be run twice.


Figure 2 - Original (left) x attenuated (right)

Friday, 3 March 2017

Interactive Facies Classification

By Tom Wooltorton, Senior Geoscientist

Why do Interactive Facies Classification (IFC+) in GeoTeric?

IFC+ offers the interpreter an advanced method of Seismic Facies classification in a rich multi-attribute environment. By utilizing the optimized blends, attributes and volumes created in GeoTeric, the IFC+ provides the optimal solution for translating the geology that you see in your data, into classified facies that can be embedded directly into the reservoir model.

The IFC+:

·    Provides a means to transform the geology revealed in colour blends, attributes and volumes into facies, and bridge the gap between visual interpretation and classification.

·    Affords the interpreter full control to delineate and tune the facies classes they see, or can operate in a partially supervised manner to discriminate subtle changes in the data that may be visually undetectable.

·    Allows the results to be clearly interrogated and validated against hard data.

·    Generates 3D results that can be directly imported into the reservoir model or volumetric estimates.

IFC+ tool interface.

Setting up your classification

To start, the interpreter chooses the optimized inputs they want to interpret facies classes from. These can include Frequency Decomposition RGB Blends, attributes, or imported results such as rock property inversions that highlight the features or geological bodies of interest. The best inputs are the ones that have been most effectively optimized to capture the facies variations, by carefully adjusting the frequency ranges or parameters in GeoTeric’s interactive tools. Up to ten volumes and one RGB blend can be used simultaneously in the tool.

By default the IFC+ will generate facies classes across the whole survey extent, however the classification can also be constrained to a 3D geobody that may delineate the reservoir or zone of interest. The geobody can be obtained by converting an Adaptive Geobody to a volume, and used in the classification by setting the opacity such that it does not obscure the other volumes. For information on how to do that, please see the video here.

Picking sample areas

Facies classes are interpreted by drawing small sample areas that capture the features of interest. This can be done in a localized manner, where the interpreter defines a single sample area to specify each individual facies, or pick sample areas crossing multiple features, which allows the IFC+ to detect the trends present in the data using a Gaussian Mixture Model and output a number of sub-facies, matching these trends. The interpreter then has the option to accept or adjust these detected sub-facies. To see how to pick sample areas, click here. Picking sample areas along wellbores is equally easy: simply by specifying two markers along a well trajectory, the IFC+ captures the seismic data values along the path between them and builds the sample area that way, also defining sub-facies according to the trends detected in the data. For picking sample areas along wellbores, see the video here.
Instead of using simple contouring to describe a complex range of data (left), the IFC+ uses a Gaussian Mixture Model to detect trends in the data and assign best-fit class centres (right).  This provides a more accurate reflection of the true facies being interpreted.

Facies tuning

Tuning these facies and sub-facies is easy and the interpreter can use the immediate, interactive feedback to ensure they capture the heterogeneity of the features of interest perfectly. Adjusting the acceptance controls how similar or dissimilar the data ranges to be classified are. Changing the opacity lets the interpreter inspect the facies distribution with respect to the underlying seismic. Watch a short video on facies tuning here.

Altering the acceptance level for a facies or set of sub-facies allows the interpreter to interactively decide what data ranges are appropriate for delineating the features of interest.

Evaluating facies using the Scatter Plot

To evaluate and ground truth the interpreted facies classes, the Scatter Plot tool is used to compare attribute values against each other and well logs. While the facies themselves are picked from specific volume combinations, they can be scatter plotted for any dataset available in the project, for example a sample area picked using the geomorphology shown in an RGB blend can be used to compare Near and Far amplitudes for AVO analysis, to relate the qualitative and the quantitative. Sample areas picked from well markers can be used to compare seismic values with log values, directly relating the geology to geophysics and providing vital calibration for the modelling of facies. To learn how to Scatter Plot facies, click here.

Finally, once the facies classes are interpreted, tuned, and calibrated, they can be generated in seismic volume form with each class given a discrete value. These are then ready to be integrated back into the GeoTeric workflow, exported using the software links, or embedded directly into the reservoir model and populated with reservoir properties.
Finished facies model, displayed in 3D render mode, this volume can be used as part of a geologic modelling workflow.

Friday, 17 February 2017

Hints & Tips for using GeoTeric’s Adaptive Geobodies

by Hugo Garcia, Senior Geoscientist.

Why should you do your geobody interpretation in GeoTeric?
GeoTeric’s Adaptive Geobodies changed the paradigm of geobody generation. The tool doesn’t rely on the standard threshold technique, where the user sets an opacity value in order to tell the software which attribute values to ignore.

The Adaptive Geobodies, in GeoTeric, uses the unique approach, of a set of interpreter guided Probability Density Functions (PDF), one for each attribute (up to ten), to interrogate the data. The user's experience is used to select the input data clusters, guide the extraction and when the limits of the data are reached (either due to data quality, data content or complexity of the structure) the user can manually manipulate any point of the surface.

The Adaptive Geobodies are in effect, a true 3D interpretation tool that uses a multi-Z surface and gives the interpreter the ability to extract any feature in the data.

What is the Acceptance Level doing? 
The Acceptance Level sets the fine scale control over the data variability to be included in the Geobody. It alters the acceptance cut-off level symmetrically from the centre of the PDF, hence sets the probability level at which a voxel is accepted as being part of the model. Because the acceptance cut-off levels are symmetrical, only the most dominant values that were selected in the Data Cluster are used, figure 1. This makes the Adaptive Geobodies resistant to noise in the data that may be included in the PDF (noise will be extreme values at the edge of the PDF and therefore not included in the selection criteria).

Figure 1 - Graphic display of the Acceptance Level parameter

Lowering the Acceptance Level will exclude more data while increasing the Acceptance Level will include more data, figure 2.

Figure 2 - Effect of changing the Acceptance Level on the growth of geobody

When only one attribute is used, the PDF’s and Acceptance Level are the same as defining a value range, when multiple volumes are used to define the Geobody, the data range is much more complex, and simple range definition would include a lot of non-desired values. PDF’s give more accuracy and control over the definition of the data range when multiple attributes are used, especially when multiple data clusters are used to describe the values. The Acceptance Level is the contour line around the data distribution, figure 3.

Figure 3 - Effects on data representation when using an increasing number of attributes/clusters

The difference between "Closest Cluster" and "All Cluster"?
With the Closest Cluster active, the Geobody will use the information from the cluster closes to the voxel that is being interrogated in order to determine if that voxel should be include in the Geobody or not.

This gives the user the ability to track a response in one portion of the geobody but ignore it in a different portion, figure 4. This is very useful when the attribute response of the structure and/or surrounding area changes laterally.

Figure 4 - Using multiple clusters with different cluster types (Internal & External)

With the All Clusters option active, the information in all of the clusters are used, independent of their proximity to the voxel that is being interrogated. 

How to edit your cluster?
The cluster can be edited in the Data Cluster tab. Data Clusters can be renamed, change the Type (Internal or External), change the colour, Auto-Seed on or off and add or delete cells.  For more on how to edit the Data Clusters, please click here.

How to make the geobodies grow faster?
The speed at which the Geobody grows is governed by two factors: the size of the Geobody (the bigger the Geobody the slower the growth) and the Mesh Granularity (the smaller the mesh the slower the growth).

In terms of the size, the Geobody can be extracted in different sections and later combined into one single object.  The quickest way to create a Geobody of a large salt body is to extract several smaller Geobodies and then combine them into one.  This is done by selecting two or more geobodies, in the project tree, and right click to get access to the menu, figure 5a.

Figure 5 - Combine geobodies using the Adaptive Geobodies right click menu.

In the menu chose the Combine Geobodies option. This will open a second window (figure 5b) where the user selects a name and choses from which original Geobody the new one will inherit the properties.
The Mesh Granularity specifies the maximum granularity that the mesh is allowed to be, but the adaptive nature of the algorithm means it will sometimes be finer than that if it thinks it needs to be to retain detail. It can never be coarser than you have specified. To increase the number of points representing the model, decrease the Mesh Granularity.

Although increasing the Mesh Granularity will make the Geobody growth faster, the thickness of the feature being extracted must be taken into account. If the feature is very thin the mesh must be kept low in order to track it. To see the effect of changing the Mesh Granularity please watch the video.

How to remove unwanted parts of the geobody?
Removing unwanted parts of the Geobody can be done in two ways:

1. The Geobody Splitting tool
Geobody Splitting allows the interpreter to manually interact with the Adaptive Geobodies to split a Geobody. On activating Splitting, an infinite cutting plane will be displayed in the screen. Pick where you want to split the model by positioning the cutting plane (the plane is define by drawing a line, one click (LMB) to define the start of the line, second click (LMB) to define the end of the line). Please watch the video for a more detailed explanation on how to use the Geobody Splitting tool.

If the option to retain both parts is selected (default option) on the Geobody menu option, Separate can then be selected to output the split Geobody sections as individual Geobodies, figure 6.

Figure 6 - Separating two geobodies

2. The Geobody Locking tool
This tool gives the user the ability to lock sections of the Geobody surface. If the section that need to be removed is left unlocked and the Acceptance Level is set to zero, the geobody will shrink and that portion will be removed. For a more detailed workflow for using the Locking tool to remove part of a Geobody, please click here.

How to make the geobody track more of the feature?
If on the volume, the colour in question is already represented in the internal cluster it might be just a question of increasing the Acceptance Level, this will allow the Geobody to take more of the represented colour in the PDF.

But if there is a portion of the volume you would like to extract as a Geobody and the colours on the volume are not represented in the original Data Cluster, you can either add those colours to the current Data Cluster by continuing to draw the Data Cluster or a new Data Cluster can be made from the drop down menu in the Adatptive Geobodies Toolbox. In the last case the Use all clusters must be active in the Adapt tab. 

How to make the geobody surface smoother?
By increasing the Mesh Granularity the surface of the Geobody becomes smoother. This is achieved, of course, at the expense of detail.  A courser Mesh Granularity will not be able to track in thinner areas and therefore portions of the Geobody may not be extracted, resulting in smaller volumes. 

How to edit the geobody surface?
The ability to edit any position of the Geobody surface is one of the most powerful features of the Adaptive Geobody tool. Geobody Manipulation allows for manual interaction with the Adaptive Geobodies to adjust the shape and position by dragging points of the mesh to the desired location.  Besides adjusting the desired seismic reflector, the interpreter is able to close gaps between parts of the same Geobody.  For a more detailed description on how to use the Geobody Manipulation tool, please click here. 

What to do next with your Adaptive Geobodies?
Once the Geobody is created and edited simply click with RMB to open the menu with the available options for your Geobody and options for export. A detailed description of the options can be found by watching the following video.