1

## What is the best way of selecting the parameters for Rock Solid Attributes (RSA) calculations?

I want to calculate some seismic attributes in Rock Solid Attributes (RSA) module. How should I select my parameters? Can you give me some background information?

Thanks

Before going into the Rock Solid Attributes (RSA), I would strongly suggest you to have a look at “Attributes Revisited” webpage by the Rock Solid Images (RSI) which is the developer of the RSA. I am sure you can find more in-depth materials on the internet if you know what you are searching for.

Since this is partly a question about the unique subsurface geology in your study area, I can’t tell you about what exact input parameters you should use, but instead, I can give you a general idea here. The rest will be up to you to make a series of tests with your data.

Basics:– RSA can be run both in time and in depth domains, thanks to the programmers at Kingdom

– RSA would be most valuable if you may squeeze information on the reservoir scale. So, as I mentioned above, make sure your sub-volume is a good representative data set for your reservoir

– Depending on the group of attributes, RSA calculates pretty much all attributes in the background to give you what is selected

– Always save your data at 32 bit. It would give you much better resolution to appreciate the RSA

Calculating 3D sub-volume:You should really have a sub-volume of your 3D seismic cube which represents your target depths. Even though you might get subtle differences at the end of RSA calculations, optimizing your input parameters rely on the geology at your zone of interest. You may refer to the below thread for this step:

http://ihskingdom.com/?p=1108

Selecting Attributes:Since there are many attributes are available to calculate you may select which ones you really need for geological uses by simply going through the help document.

– Geological Attributes

– Instantaneous Attributes

– Wavelet Attributes

Input Parameters:Depending on your RSA selection, you will be asked to set your input parameters grouped as follows:

– Bandpass: RSA uses bandpass filter (Butterworth), so you need to design it so that it covers your data spectrum. In doing so, be careful of not cutting too much of your data out in an attempt to get rid of the high frequency noise. Make sure to have a filter without steep slopes at each ends because if you do, you may end up getting Gibbs effects. Having smaller side-lobes is always better.

– Similarity compares the traces around the reference trace in sequential order. You need to have a narrow enough window to get higher resolution, but at the same time, wide enough window to run much better similarity calculation. Help document states “the recommended time window should be at least equal to 3 trace half-cycles” which means 1.5 wavelets. Any window selected smaller than that may not give you a good result because the similarity calculation may not have the optimum window length for cross-correlation with the adjacent wavelets.

– Dipscan handles corrections for any structural dips in Similarity calculation. The default is set to 3 which means 1 in-line and 1 cross-line in each direction from the reference trace in the center. Larger the number, more the traces would be scanned, and but slower to run the calculations.

Max Dip is the maximum time cap to look for similar event dips in the surrounding traces. If it is larger than the actual dips, less edge effects will be seen on the data. If it is smaller than the actual dips, than the dip calculation will not be correct. If it is set too wide, then you may end up correlation other events above and below.

Dip Scan Increment is the amount of time increments in cross-correlation in upward and downward directions. In the help documents, it is recommended to be “the half the sample interval of the data”. If a large number selected, then the scanning would loose the details; if a very small number selected, then the calculations may take a pretty long time.

Lateral Smoothing is optional. If it is checked, a simple structural smoothing algorithm is applied. It is recommended for consistency, especially for the noisy data and structurally complex geology.

– Variance is somewhat closer to the Similarity calculation. By default, it is 3 times larger the the Similarity window. It would be good if you keep that ratio when you play with the Similarity time window.

– Curvature calculation uses 9 traces by default with a reference trace in the center.

Correlation Window is the length of window used in 9-trace correlation. Similar to the correlation window discussed above, it is recommended to be at least 3-half wavelets to identify the wavelet shapes properly. Narrow correlation window may not give correct results.

Maximum Lag is the maximum time cap in running the correlation for both upward and downward directions. It is recommended to set it to at least about one wavelet length.

Smoothing (DPS) stands for Dip Preserving Smoothing. It applies a simple structural filter on the calculated dips. You should apply it in case you have noisy data.

Minimum Correlation Coefficient is the minimum acceptable value in the curvature calculation. Any calculated coefficient below this minimum value will be rejected. Higher values may give more accurate results.

– Spectral Decomposition: Please refer to the below thread for this topic:

http://ihskingdom.com/?p=1148

Was this answer helpful?

Like