In these
images, one can immediately identify which anomalies are trustworthy
and which ones are caused by noise. For example, in 2006 for the odd
subset along the left (south) side of the cross section we observe low
Vp/Vs, while for the even subset, the Vp/Vs ratio is high. Same discrepancies
are observed in other years mostly in marginal areas. Thus, they
should be ignored or interpreted with prudence. At the same time, the
anomalies in the central part of the study area seem to be consistent.
Note that halving of anomalies in the odd/even test lead to the resolution
decrease in some parts of the models. For example, the shallow
anomaly beneath Kluchevskoy in 2004 is less prominent in the odd/
even tests than in the main model. All these facts should be taken into
account when interpreting the inversion results.