#TITLE_ALTERNATIVE#

• Main Author: LUTFI HERNANDI (NIM : 10212021), MUHAMMAD
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/23284
• Institution: Institut Teknologi Bandung

Deconvolution is a filtering process which aims to reconstruct reflectivity function by reversing the process of convolution. Deconvolution compresses the wavelet in the recorded seismogram, attenuates reverberations and short-period multiples, thus increases temporal resolution and improves interpretation result of seismic sectioin. The commonest way in performing deconvolution is by designing Wiener Filter in a least-squares sense. Seismic signal is non-stationary which means that the shape and band-width vary with travel time. Nonstationarity is caused by spherical divergence and high frequency attenuation. By the seismic waves travel deeper, the amplitude will be decreased by spherical divergence effects. On the other side, high frequency attenuation will be increased because of frequency absorption. As the result, some parameters like mean, variance, correlation function, and other parameters of seismic signal vary with time. <br /> <br /> <br /> Ordinary Wiener Filter deconvolution method is a time-invariant deconvolution where the source wavelet is assumed to be stationary. In fact, the source wavelet is non-stationary, so that to overcome the nonstationarity problem, time-varying deconvolution is needed to be perfromed. One of time-varying deconvolution methods is Gated Wiener Deconvolution. Time-varying deconvolution operators are derived by dividing seismic signal into some sections and treating each section as time-invariant process. <br /> <br /> <br /> Three different deconvolution methods were evaluated by their error energy in this work such as Wiener time invariant deconvolution, time invariant deconvolution with linear inversion, and Gated Wiener time varying deconvolution where T=0.05 s. The error energy for Wiener time invariant deconvolution, time invariant deconvolution with linear inversion, and Gated Wiener time varying deconvolution are 0.4461, 0.3948, and 0.1882, respectively. Gated Wiener time varying deconvolution produces better deconvolution result compared to time invariant deconvolution. Gated Wiener time varying deconvolution produces best result when the length of each gate T is 0.05 s over 0.25 s, with error energy value 0.1882.