Design a filter to remove the bubble effect and enhance the signal
The main objective of the Designature is to get the optimum source wavelet what we designed at the time of the acquisition. Before going to the Designature let us discuss about
Why do we need Designature and what is the procedure and how does it work?
In the marine data acquisition, we use air guns as source. The air gun arrays are designed in such a fashion that we get the desired signal and maximum peak to bubble ratio. Each air gun have different pressure volumes and based on the size of the air gun arrays it will create an impulse. Simultaneously it creates bubbles. Bubbles are created by the expansion and contraction of the air. The period of the bubbles based on the size of the gun array volume. The bigger the size of the gun array volume, more the bubble period. These bubbles appear as low frequency content in the data and we need to attenuate it otherwise they will compromise the resolution of the seismic data. When all the guns (array) fire at the same time, all the peaks will align and add together but the bubbles won’t. By this way we can minimize the bubbles in the data set however there is always some residual bubble left over in the data. To attenuate this we need to design an operator. Designature is the process where we design the operator and attenuate the bubble effect.
How do we design the operator?
During the time of acquisition or start of the acquisition, survey companies test various gun array combinations to get the desired source wavelet. They provide the source signature or we can call it as far field source signature. We take this far field source signature and create an operator (wavelet) to get the desired or close to the modeled wavelet. By doing so we attenuate the bubble and we convert the wavelet to minimum phase. Predictive Deconvolution expects the data should be in minimum phase. Deconvolution enhances the signal to noise ratio and attenuates the water reverberations.
There are different ways to get the source signature. If the acquisition company provides the far field signature that is great, if not then we can use the near field hydrophone information and design the desired source wavelet. Also if we have the acquisition report and the source (air gun) configuration we can design the source wavelet by using some software like Nucleus. There is also another way of doing this. If we have the signature on a paper section we can digitize it and use that also.
There are various kinds of source wavelet to generate. Single source wavelet which is also known as far field signature. Source signature for each individual shot gather. This one is especially available with the modern acquisition. It is called as CMS (Calibrated Marine Signature). g-Platform currently supports single signature wavelet. But in the future we will add more reading options of the source wavelet.
Far field signature is basically having source ghost only. During the design of the output source wavelet we can add receiver ghost to design the operator. Once we have the minimum phased far field signature with the receiver ghost added we need to create the operator by adding any minimum phase equivalent pass band filter like Butterworth.
User Recommendation
To run Designature in g-Platform first we need to read the far field signature by using the module “Read Wavelet from file”. Currently this module accepts only one source signature file. Under this you have to provide the desired sample of the output wavelet.
After inputting the input file name, and desired output wavelet parameters i.e., by truncating the input wavelet gives us the desired output wavelet from the "Read wavelet from file" module.
These two wavelets i.e. input and output wavelets (Input desired wavelet) are used in the Designature module as inputs.
Input gather - Connect/reference Input gather from the seismic loop. If the designature module is within the Seismic loop, it'll automatically connects the input gather of Seismic loop to Designature module. In this case, the user interference is not required.
The pre-stack seismic gather to be designatured. This is typically the raw or minimally conditioned shot gather or common-midpoint gather coming from your seismic reading loop. Both input wavelets are automatically resampled to match the sample interval of this gather before the operator is applied, so the wavelet and data sample rates do not need to match in advance.
Input wavelet - use the Input wavelet (far field signature) from the "Read wavelet from file" module
The actual source signature recorded during acquisition, typically the far-field signature provided by the survey company or derived from near-field hydrophone measurements. This wavelet describes the actual shape of the energy emitted by the air-gun array, including the primary pulse and bubble oscillations. Connect the Input wavelet output of the Read wavelet from file module here.
Input desired wavelet - Use the Output wavelet from the "Read wavelet from file" module.
The target wavelet that you want the data to look like after designature. This is the idealized or modeled source wavelet representing the desired acquisition output — typically the truncated or band-limited version of the far-field signature, produced by the Read wavelet from file module as its Output wavelet. The shaping filter computed by this module transforms the Input wavelet into this desired shape. If Output signal min phase is enabled, the module converts this wavelet to minimum phase internally before designing the filter.
Operator type { Time, Frequency, D^(Z) } - There are 3 types of operators available to choose from the drop down menu.The operator performs the auto-correlation in the time and frequency domain. D^(Z) - is a Wiener Designature operator works in Frequency domain.
Selects the mathematical domain in which the shaping filter is designed and applied. Default is Frequency. The three modes are:
Time — Designs a time-domain shaping filter using the Wiener-Levinson (Toeplitz) least-squares approach. The filter length is controlled by the Operator Length parameter. The autocorrelation of the input wavelet is used to build the normal equations, and the cross-correlation with the desired wavelet provides the right-hand side. The resulting filter is applied via convolution to every trace in the gather. Use this mode when you want direct control over the operator length in milliseconds.
Frequency — Designs the operator in the frequency domain using a Wiener-type spectral division: each frequency component of the input data is multiplied by the complex conjugate of the input wavelet spectrum and by the desired wavelet spectrum, then divided by the power spectrum of the input wavelet plus a noise stabilisation term. This approach is efficient and avoids the operator-length constraint of the Time mode. This is the recommended default for most designature workflows.
D^(Z) — An advanced Wiener designature operator formulated in the z-transform domain. Like Frequency mode it operates in the frequency domain, but uses a different mathematical formulation that can yield improved phase stability in some datasets. Use this mode when Frequency mode produces residual phase artefacts or the bubble attenuation is incomplete.
Operator Length - In case the Operator type is Time, provide the auto correlation operator length. By default 400 ms.
This parameter is only active when Operator type is set to Time. It controls the length of the time-domain shaping filter in seconds (default 0.4 s, equivalent to 400 ms). A longer operator allows the filter to model more complex wavelet-to-wavelet transformations, including multi-period bubble effects, but increases computation time. A shorter operator produces a more compact filter suitable for simpler wavelet shapes. As a starting point, set this value to at least twice the expected bubble period. For typical marine air-gun sources with bubble periods of 100–200 ms, a value of 0.4–0.6 s is usually appropriate.
Prewhitening factor - This is the amount of percentage noise added to the operator to stabilize the output. By default 10%
A percentage (default 10%, stored as 0.1) that is added as a noise floor to the wavelet power spectrum before computing the inverse filter. This prevents division by near-zero values at frequencies where the input wavelet has little or no energy, which would otherwise produce numerically unstable amplification of noise. In the Time domain mode this value whitens the autocorrelation diagonal; in the Frequency and D^(Z) modes it scales the peak spectral amplitude of the input wavelet before adding to the denominator. A higher value (e.g. 15–20%) provides greater stability but reduces the sharpness of the designature correction, leaving more residual bubble energy. A lower value (e.g. 1–5%) produces a more aggressive correction but can amplify noise at weak spectral frequencies. Start with the default of 10% and adjust based on QC of the output wavelet shape and the data signal-to-noise ratio.
Output signal min phase - By default Checked (Yes). In case the user wants to get the final output in Zero phase, uncheck this option.
When enabled (default), the desired output wavelet is converted to its minimum-phase equivalent before the shaping filter is computed. This ensures that the designatured data has a minimum-phase character, which is a prerequisite for predictive deconvolution to work correctly. Keep this option enabled if Designature will be followed by Predictive Deconvolution in the processing workflow. Disable this option only if you intend to produce zero-phase output for direct interpretation use, for example when comparing with well ties that use a zero-phase synthetic. Note that enabling this option applies a minimum-phase spectral factorisation to the desired wavelet inside the module; the shape of the actual output wavelet can be inspected using the Designature operator output.
Auto-connection - By default Yes (Checked)
Bad data values option { Fix, Notify, Continue } - This is applicable whenever there is a bad value or NaN (Not a Number) in the data. By default Notify. While testing, it is good to opt as Notify option. Once we understand the root cause of it,
the user can either choose the option Fix or Continue. In this way, the job won't stop/fail during the production
Calculate difference - This option creates the difference display gather between input and output gathers. By default Unchecked. To create a difference, check the option.
Number of threads - One less than total no of nodes/threads to execute a job in multi-thread mode.
Skip - By default, No (Unchecked). This option helps to bypass the module from the workflow.
Output gather - This module outputs the designatured output gather
The main output gather containing the seismic data after designature processing. The bubble effect has been attenuated and the source wavelet has been shaped toward the desired target wavelet. The output has the same trace count and time length as the input gather. Connect this output to the next module in your processing workflow, typically Predictive Deconvolution when working with minimum-phase output, or directly to stacking/migration for zero-phase output.
Gather of difference - Generates the difference gather of input vs output gather
Available when the Calculate difference option is enabled in the Settings section. Displays the sample-by-sample difference between the input gather and the output gather (input minus output). This QC gather is useful for verifying that the designature is removing bubble energy without distorting primary reflections. A good result shows residual bubble energy in the difference and clean reflections in the output.
Designature operator - Generates the designature operator as an output.
The computed shaping filter in the time domain, output as a single-trace gather for inspection. This is the actual operator that is convolved with the input data to produce the designatured output. Examining this gather allows you to verify the operator shape and length: it should resemble a compact wavelet that transforms the input signature toward the desired output shape. The operator is recomputed automatically in the GUI whenever any parameter or input wavelet is changed, so you can interactively optimise the parameters by watching the operator change in the vista display before running the full job.
Here is an example designature workflow using far field signature as an input wavelet by using "Read wavelet from file" module.
From the Designature module, we can get Input wavelet, Input desired wavelet and Designature operator, besides input and output gathers after designature.
YouTube video lesson, click here to open [VIDEO IN PROCESS...]
Yilmaz. O., 1987, Seismic data processing: Society of Exploration Geophysicist
* * * If you have any questions, please send an e-mail to: support@geomage.com * * *