Filtering in seismic data processing? How filtering help to suppress noises.
Download as DOCX, PDF11 likes14,603 views
The document discusses the importance of filtering in seismic data processing to suppress noise and enhance the signal-to-noise ratio. It outlines various filtering techniques, including frequency filtering, inverse filtering (deconvolution), and velocity filtering, each serving different purposes in improving seismic data quality. Additionally, it explains how different types of noise, both random and coherent, can be managed through these filtering methods to achieve clearer geologic feature representation.
More Related Content
Similar to Filtering in seismic data processing? How filtering help to suppress noises. (20)
Filtering in seismic data processing? How filtering help to suppress noises.
- 1. 1 Filtering in seismic data processing? How filtering help to suppress noises. The seismic trace is the combination of both signal and noise, the signal (wanted data) is the representation of the geologic feature but the presence of noise shows it different from real. Noises are further subdivided into Random (the natural noise with irregular pattern), and Coherent noise (generated by geophysical equipments and experiments) like surface wave generation by source. To enhance the signal-Noise ratio different techniques are used to remove the noises. Types of Seismic Filtering: 1- Frequency Filtering. 2- Inverse Filtering (Deconvolution). 3- Velocity Filtering. Frequency Filtering Basically seismic data processing is done in the frequency domain. The original time domain data is converted to frequency domain by applying forward Fourier transformation to make processing easy and filtering easy and cost effective. The frequency domain data filtering involves the product of amplitude spectrum of input trace with the filter operator (time-domain representation of wavelet). The processes of seismic filtering in time and frequency domain are based following concepts in time series. “Convolution in Time-domain is equivalent to multiplication in the frequency domain, same as convolution in frequency-domain is equivalent to multiplication in the time domain” (Yalmas, 2001). Frequency filters are designed in form of Band pass, Band reject, high pass or low cut or Low pass or high cut. All of them are working on same principle; simply generate the zero phase wavelet with an amplitude spectrum that satisfies the single specification from four.
- 2. 2 Original Seismic trace. Low Pass filtered trace. High Pass Filtered trace. Band Pass Filtered trace. Band Stop Filtered trace. Frequency filter are usually applied when the frequency of signal and noises are different, it is therefore can be separated on the function of frequency. Analogue frequency filter is still commonly used in which simply a range of frequency of desired signals is set in the filter. And it only passes the specific range of the frequency and blocks all the high and low frequencies traces. Band- Pass Filter is also working similar to analogue filter it commonly used and used at different stages of seismic data processing. It simply pass a certain bandwidth usually 10-70 HZ and block the lower and higher frequencies, typical seismic trace contains noises like ground rolls are example of low frequencies mainly of less than 10HZ. Some higher unwanted frequencies are produced by the ambient noises can be removed using band pass filter.
- 3. 3 Any other noises random and coherent are removed by frequency filtering in a way wind noise is removed by high cut filter/low pass filtering, it allow lower frequency to pass and removed higher frequencies, also known as smoothing filter.
- 4. 4 High Pass/Low cut filter is only allowed the higher frequencies and cut lower frequencies and it is widely used for edges detection. Notch Filter is the removal of removal of noises from electrical lines over geophones during recording; their frequency is mostly 60 Hz, just remove the effect on signal. Band Reject; if the range of noises is known then we simply mute or reject their band. It is totally opposite of band pass filter.
- 5. 5 Inverse Filtering (Deconvolution): Earth act as band pass filter for seismic waves that travel through it, higher frequencies are block and pass lower frequencies. For the better resolution of the seismic data we need to use greater bandwidth of the frequency so for that reason we apply inverse filtering. In which recover the higher frequencies, attenuate multiples, equalize amplitude and generate zero phase wavelet. Some of noises are left and are present with the same character of the reflected signal after frequency filtering. For this broad range of seismic inverse filters are used to remove the specific adverse effect of earth filtering (natural) along the transmission path such as absorption and multiples and (artificial) frequency filtering during the seismic processing.Examples from inverse filtering to remove the effect of frequency filtering; Reverberation: The ringing are removed that are associated with the multiple reflection in a water layers
- 6. 6 Deconvolution is best way to treat reverberation (reverberations) and multiples (deghostings). The short duration reverberation is exposed by auto-correlation function with series decaying of waveform, shown in waveform (a). The long reverberation is also exposed by auto-correlation as separate side lobes shown in waveform (b). The side lobes present with the time gap through which it align with the multiple reflection, then this lag of side lobes are periodically matched with the reverberations. De ghosting: Short path multiples are removed that directly travel from source to surface and reflect back to base of weathered layered to receiver. Whitening: In which amplitude of all the frequencies are equalized within the recorded frequency band. Below figure is showing the removal of apparent punch-out effect by recovering its amplitude and equalizing with its lateral extend.
- 7. 7 Velocity Filtering The velocity filter is use to remove the coherent noise from the seismic data on the base of angles at particular angles at which the event dips, this also known as fan filtering and pie slice filtering, (March & Bailey 1983). The angle of the dip event with it propagates across the spread of detector is determined by apparent velocity. Apparent velocity is calculated by; Va= v/sinα Va= apparent velocity V= seismic pulse traveling velocity α= angle with it propagate vertical across spread Along the direction of spread, every single sinusoidal component of the waveform having a apparent wave number ᶄa related to its frequency F. F= Va.ᶄa Hence plot of F and ᶄa is a straight line along the apparent wave number and apparent velocity. This filtering will apply to the F-K spectrum to split out a particular seismic event, which is shown as a sloping linear trend of peaks on the F-K spectrum, as Va = F/ᶄa. though, a typical
- 8. 8 shot data is in curved hyperbolic form on the original section containing different seismic event as shown in the below figure. The seismic event travel across spread way from source will plot in the negative wave number and the event travel towards the source will plot in the negative wave number like back scattered noises in the above figure. On the basis of difference in apparent velocity the unwanted noises are suppressed. The simple way to achieved the same results is by applying the f-k filtering, in f- k filtering the two dimensional data is converted from t-x to f-k domain then in the wedge shape model is easily classify the noises and signals.
- 9. 9 Other important function of the velocity filtering includes the removal of ground rolls from the short point gather. This will greatly help to better stacking and exact estimation of the stacking velocity. This is also used to remove the coherent noise from single shot data because of its anomalous dipping, for example the diffractions are removed by velocity filtering.