One objective of seismic data processing is to reduce both coherent and random noise contaminating the reflection data, so that the physical properties of the rocks and their contained fluids may be more readily estimated and/or imaged from trace amplitudes. It can be important to preserve the relative amplitudes of reflection events during processing, especially laterally, since various interpretive techniques are based on trace-to-trace amplitude comparisons. Most prominently, AVO and AVA methods compare amplitudes from the same reflection or conversion event at different source-receiver offsets (AVO), or at different raypath angles (AVA). We describe here an empirical study in which physical model data were subjected to three commonly used processing techniques. The amplitudes along single reflections in those data were compared at each stage of processing. We found that, within the limits of measurement error, subtractive noise attenuation (radial trace filtering) and non-stationary deconvolution (Gabor deconvolution) have no significant effect on relative event amplitudes. As we might expect, however, a multi-trace process like F-X deconvolution, does affect relative event amplitudes and should be avoided in processing data for AVO.
View full article as PDF (0.90 Mb)