We have also performed traditional checkerboard tests for different
time episodes. The alternating anomalies in the synthetic starting
model (not shown) are 5×5×5 km with 2 km of empty space between
anomalies. The anomalies of opposite signs for P and S-models have
magnitudes of 4% and −7%, respectively. These anomalies produced
high contrast Vp/Vs features of similar to the real data. In Fig. A1 the
results of the checkerboard reconstruction are presented in the same
section as used for showing the main results. The checkerboard test
results in most episodes show good vertical resolution beneath
Kluchevskoy volcano and allows robust retrieval of the negative–
positive–negative Vp/Vs series from 0 to 20 km depth. For the surrounding
areas the resolution changes with time: in 2004, 2007 and
2008 the resolved area seems to be larger than in other years. For the
deep part below 25 km depth, where we detected a prominent anomaly
related to the mantle channel, this checkerboard seems to be not well
resolved. Comparing the results of this test with those in Fig. 4 shows
that the mantle channel can only be resolved if the anomaly is much
larger than the size of the checkerboard patterns.