Wavelet estimation and blind deconvolution of realistic synthetic seismic data by log spectral averaging
M. Tria, Mirko van der Baan, A. Larue, Jérôme Mars · 2007
Homomorphic wavelet estimation was a popular tool in the late 70s. It consists of taking the logarithm of traces transformed to the frequency domain. This maps a convolution in the time domain to an addition in the log(frequency domain). Wavelet estimation was however done on single traces. We improve on this concept by introducing a multi-trace approach. In current marine acquisition the source wavelet is very uniform across an entire survey. We assume that we have selected many traces, randomly sampled from a large area, that are all characterized by different reflectivity series. Averaging the log(spectra) then recovers the original source wavelet. The advantage of this approach is that it does not make any assumptions on the phase of the wavelet. After wavelet estimation, a standard deconvolution algorithm can be used to improve the resolution of the seismic data.