b'Seismic window Seismic windowMichael Micenko Associate Editor for Petroleum mick@freogeos.com.auFigure 1.Selected assortment of wavelets showing the differences are essentially in the side-lobe Wavelets, spectra and transforms amplitude and extentI know Ive done this before (Preview December 2017) but posts on variousreflections from the top and base users groups suggest there is still someof a bed result in a high amplitude misunderstanding of the applicationanomaly when the bed thickness is a of spectral decomposition. Spectralquarter of a wavelength (the primary decomposition is the process oftop reflection and side-lobe peaks breaking down seismic data into discretefrom the base reflection are a half cycle frequency components and its mainapart in TWT and align). Obviously the use by interpreters is to aid stratigraphicwavelength depends on frequency so interpretation. The method relies ondifferent wavelengths give information tuning curve properties to isolate thinon different bed thicknesses. Figure 2 beds and estimate their thickness. Hereis a handy table that shows the tuningFigure 2.Table of tuning thicknesses associated are some common user questions andthickness for some given frequency andwith various frequency and velocities. A table like answers with illustrations based onvelocity. this is handy to have on your desk.simple models.One way to analyse the seismic data What wavelet should I use in modelling? and determine bed thickness is to use a Frequency vs TWT plot at a specific I prefer one that matches the seismicplace such as a proposed well location data but extracted wavelets are difficult(Figure3). The colour depicts the to describe so a mathematical wavelet isamplitude of each frequency. On this more useful. There are many theoreticaltype of display a single bed will appear wavelets that could be used and quiteas an anomalously high amplitude often a Ricker wavelet is chosen but in(caused by tuning) at the tuning practice there is little difference betweenfrequency and TWT of the bed. Figure3 wavelets once the frequency range isshows the spectral decomposition known. Figure 1 has some exampleresults for a 10 m thick bed. The 45H z wavelets. The Ricker wavelet is thepeak frequency estimates the bed to be neatest with only a single but higher10.1 m.amplitude side-lobe while the Ormsby and Butterworth wavelets have broader side-lobes with more reverberation.What is the wavelet effect and how can it Something to remember is that allbe removed?seismic sources are impulsive and start at time zero and it is only in processing thatFailing to take account of the wavelet symmetric wavelets are created. spectrum is a pitfall for the unwary. The wavelet spectrum can often affect How can bed thickness be determined?the thickness calculation because it is What are these plots? embedded in the data and is commonly the dominant part of the spectrum. OneFigure 3.Continuous wavelet transform of a Spectral Decomposition relies onway to remove this overprint or reducesingle bed 10 m thick at 90 ms. The CWT predicts the tuning effect where for thin bedsthe wavelet effect is to normalise eachthe correct thickness and location of the bed. constructive interference betweenfrequency. This will flatten the spectrum(Velocity used is 2000 m/s).DECEMBER 2021 PREVIEW 50'