kwinkunks
commented
10 years ago I think simple is fine for now -- add 'thermal' noise (ideally from a spectrum, allowing for different colours of noise) to the seismic after convolution, consistent with the simple convolutional model, S = R * W + N. However, it must be per seismic sample (in time and space), not per image pixel (I am not certain if these are different, I think they are?). In which case the best way to do it might be to add the noise to the reflection coefficients, not the final image.
As per our MO, it would be fun to be able to vary noise systematically across a plot -- adding another dimension to our hypercube. But just adding in some percentage of some colour of noise to any plot would be OK for now.
+++++++++++++++++++ Matt Hall PhD PGeo Nova Scotia, Canada
- 1 902 980 0130 G+ Tw FB @kwinkunks agilegeoscience.com thehubsouthshore.ca
On Wed, Mar 12, 2014 at 10:13 AM, Evan Bianco notifications@github.comwrote:
Generally speaking, there are potentially several ways of incorporating noise into the seismic experiment. Noise at receiver, noise in the earth model, noise as a function of source-receiver offset, etc.
easiest: add (gaussian) random noise level by setting signal : noise parameter for each trace, or for each ensemble of traces
more difficult: add "noise" (heterogeneity) to earth model? This might be a way to vary rock property variations within a single layer. It would make more spikes, and would wreck the economy of sparsity that we are using.
Reply to this email directly or view it on GitHubhttps://github.com/kwinkunks/modelr_app/issues/47 .
ben-bougher
Generally speaking, there are potentially several ways of incorporating noise into the seismic experiment. Noise at receiver, noise in the earth model, noise as a function of source-receiver offset, etc.
easiest: add (gaussian) random noise level by setting signal : noise parameter for each trace, or for each ensemble of traces
more difficult: add "noise" (heterogeneity) to earth model? This might be a way to vary rock property variations within a single layer. It would make more spikes, and would wreck the economy of sparsity that we are using.