Abstract
The paper is a follow-up of Sokos and Zahradník (2008), in which the basic version of the ISOLA software was described. The present paper describes main recent innovations that allow the assessment of solution quality.
Different stations may be used in the moment tensor (MT) inversion with a different frequency range. The low-frequency range suitable for the MT inversion is determined by analyzing the signal and noise spectra.
The solution quality is measured based on the signal-to-noise ratio, variance reduction, condition number, and two new indices, FMVAR and STVAR. FMVAR (the focal-mechanism variability index) quantifies the stability of the focal mechanism in a space-time grid search within a prescribed correlation threshold.
STVAR (space–time variability index) measures the stability of the source position and time in a grid search within the same correlation threshold. The uncertainty analysis is also provided by means of a 6D theoretical error ellipsoid, suitable mainly when designing a seismic network, when real waveforms are not available.
Automated jackknifing is a possible complement of the standard MT inversion, in which real data and modeling errors are partly taken into account. The main innovations are illustrated by five examples.
All of them concern the same (Mw 4.3) earthquake, processed in ten different ways. Various scenarios have been simulated, in which just a few stations were available and the goal was to assess the reliability of the solution, while keeping the reference solution blind.