Resloving Thickness Changes at Vertical Incidence

Here we perform a series of tests in which we:

• modify the vostok model slightly by adding a layer of rock on the bottom and, possibly, a layer of till bewteen ice and rock (call this the unperturbed model);
• form a series of perturbed models by increasing the thickness of the bottom layer by 1, 2, ... 5 cm;
• compute the vertical incidence response of the model (up to 3-rd order multiples);
• convolve the response with a radar wavelet (f0=100 MHz, df=5 Mhz);
• gate out a section of the timeseries centered on the primary reflection from the rock (1 ns sampling, 512 samples long);
• taper the first and last eighth using a cosine taper;
• compute the lag with respect to the unperturbed model (using ah_timediff3);
• compare the result to the exact answer.

Lag when there is no till layer. Circles) Calculated, line) exact. The thickness change is well resolved. (Postscript version).

Lag when there is a 100 m thick till layer. Circles) Calculated, line) exact. The thickness change is well resolved, since the reflection from the top of the till is well-separated from the reflection from the rock. (Postscript version).

Lag when there is a 1 m thick till layer. Circles) Calculated, line) exact. The thickness change is not well-recovered, because there is too much interference between reflections. (Postscript version).

Lag when there is a 0.1 m thick till layer. Circles) Calculated, line) exact. The derived lags follow a linear trend, but not the right one, due to the effect of having two nearly-overlapping pulses (one that moves out, and one that doesn't). (Postscript version).