Home | deutsch  | Department | Legals | Data Protection | KIT

Applications of time-frequency transforms for seismic processing: Noise attenuation and inverse Q filtering

Applications of time-frequency transforms for seismic processing: Noise attenuation and inverse Q filtering



Dr. Tim Seher (Spectrum Geo Ltd./Great Britain)


The spectral content of a seismic trace is the basis for many seismic processing methods. However, this spectral content commonly varies with time and space due to the effects of seismic wave propagation and the non-stationary nature of seismic noise. Simple Fourier analysis is therefore insufficient for making full use of these time-varying properties. This issue is commonly addressed through the usage of time-frequency transforms that estimate the spectral content of a seismic trace over a time-window. In this presentation, I will first give an introduction to time-frequency transforms. I will then present two applications that make efficient use of these transforms: Time-frequency noise attenuation using a vector median filter and amplitude friendly inverse Q filtering.

Time-frequency vector median filtering is a novel technique for attenuating both coherent and incoherent noise in seismic data. It combines a time-frequency transform, an improved median filter that efficiently handles complex data, an efficient thresholding method and a fast dip scan. This novel noise attenuation method has proven particularly effective at low and intermediate frequency noise attenuation and has enabled the imaging of deep crustal reflectors.

Quantitative seismic interpretation commonly assumes the seismic data to satisfy the elastic wave equation. Preparing seismic data for interpretation thereby requires compensating for viscous effects present in the seismic data. One process that reduces the effects of attenuation and dispersion present in seismic observations is amplitude and phase inverse Q filtering. This process can again be implemented using time-frequency transforms. In this talk, I will show how to construct amplitude-friendly inverse Q filters, which reduce amplitude distortion during seismic processing. Finally, I will demonstrate the impact of inverse Q filtering on the estimated AVA gradients.