|
<< Click to Display Table of Contents >> Navigation: Modelling > Noiser |
Noiser adds synthetic noise to seismic gathers for algorithm testing, processing parameter sensitivity studies, and synthetic dataset generation. The module generates a noise wavelet of the user-specified type and frequency, scales it by the noise amplitude factor, and adds it to each trace sample. When Muted area only is enabled, noise is injected only in the muted (zeroed) zone of the gather, which is useful for filling the pre-first-break zone with realistic noise to test demuting or first-break picking algorithms. When disabled, noise is added uniformly to all samples. The noise amplitude at each sample position is randomised, with optional seed randomisation for reproducible or uniquely random results.
Container for the standard input data connections. Connect this to the output of the preceding module in the processing sequence.
The input seismic gather to which synthetic noise is added. This may be real or synthetic data; the module adds noise on top of the existing trace amplitudes.
The amplitude scaling factor of the added noise. The default is 10. This value is multiplied by a random amplitude drawn from a uniform distribution to produce the noise sample added at each trace position. Increase this value to add stronger noise; decrease it for subtler noise injection. The appropriate value depends on the amplitude range of the input gather.
Controls whether the random number generator seed is varied between gathers (default: false). When disabled, the same seed is used for each gather, producing repeatable (reproducible) noise realisations across repeated runs. When enabled, the seed changes with each gather, producing a different noise realisation for every gather processed — useful for testing algorithm robustness against varying noise patterns.
Container defining the shape of the synthetic noise wavelet. The noise is generated by convolving the selected wavelet with random amplitude spikes before adding it to the traces.
The wavelet shape used as the noise kernel. Options include Ricker1, Ricker2, AKB, Berlage, Gaussian, GaussianDeriv, MinPhase, Klauder, Ormsby, Spike, Zero, and Unit. The default is Ricker1. Choose a wavelet type that matches the expected spectral character of the noise you want to simulate.
The dominant frequency of the noise wavelet in Hz. Set this to match the frequency of the noise type being simulated — for example, use a low value (5–20 Hz) to simulate ground roll and a higher value (50–80 Hz) to simulate high-frequency random noise.
The total length of the noise wavelet window in seconds. This should be at least three times the period of the dominant frequency (1/Frequency) to avoid truncation of the wavelet.
When enabled (default: true), noise is injected only in the muted (zero-amplitude) region of the gather — typically the pre-first-break zone above the top mute. This is useful for producing realistic test gathers where the muted zone contains noise rather than zeros. When disabled, noise is added to all trace samples, including the live data zone.