Noise suppression

So far the reflection coefficients have been very high (0.5) and still the images have been relatively noisy. If this technique is to be used for looking at real targets, reflection coefficient of an order of magnitude smaller will have to be imaged.

Tests showed that the signal to noise ratio decreased as the $\sqrt{n}$, where n is the number of time samples, and by increasing the length of the time series used I was able to clearly image reflection coefficients of 0.05, as shown in Figure 7.

 

many
Figure 7 Common shot gathers for a model with a single horizontal layer with reflection coefficient 0.05. Left panel used 65,000 points in the time series, the center panel about 130,000 and the right panel about 260,000.

A similar study, but this time comparing signal to noise ratio with the number of incident plane waves, was also conducted. Interestingly, it showed that more plane waves did not increase the amplitude of the observed signal to noise; however, more plane waves do improve the shape and definition of the hyperbola.