Attribute analysis improvement by means of smart averaging
Seismic attribute maps can benefit from dedicated processing for suppression of noise and improvement in geological feature delineation. Traditional 2D windowed frequency filtering, or smoothing, can degrade the required resolution and hence impacts interpretability. We propose a ‘Smart Averaging’ (SA) technique, which is an optimization routine based on specific criteria, and we show that the root mean square (RMS) deviation minimization criterion is effective for both spikes and random noise, providing both visual and interpretation improvement. It out-performs conventional fixed window smoothing, or averaging. The dependency between interpretation reliability, data resolution and random noise level is demonstrated. When the attribute maps have very high levels of noise, we show that interpretation is still successful, even if the attribute map detail is noticeably compromised.