A step in seismicprocessing to determine the shape of the wavelet, also known as the embedded wavelet, that would be produced by a wave train impinging upon an interface with a positive reflection coefficient. Wavelets may also be extracted by using a model for the reflections in a seismic trace, such as a synthetic seismogram. A wavelet is generated by deconvolving the trace with the set of reflection coefficients of the synthetic seismogram, a process also known as deterministic wavelet extraction. Wavelets may be extracted without a model for the reflections by generating a power spectrum of the data. By making certain assumptions, such as that the power spectrum contains information about the wavelet (and not the geology) and that the wavelet is of a certain phase (minimum, zero), a wavelet may be generated. This is also called statistical wavelet extraction. A particular processing approach to establishing the embedded wavelet is to compare the processed seismic response with the response measured by a vertical seismic profile (VSP) or generated synthetically through a synthetic seismogram in which the embedded wavelet is known. The wavelet can also be extracted through the autocorrelation of the seismic trace, in which case the phase of the wavelet has to be assumed.