Effects of 1-D versus 3-D velocity models on moment tensor inversion in the Dobra Voda area in the Little Carpathians region, Slovakia
Abstract
Retrieving the parameters of a seismic source from seismograms involves deconvolving the response of the medium from seismic records. Thus, in general, source parameters are determined from both seismograms and the Green functions describing the properties of the medium in which the earthquake focus is buried.
The quality of each of these two datasets is equally significant for the successful determination of source characteristics. As a rule, both sets are subject to contamination by effects that decrease the resolution of the source parameters.
Seismic records are generally contaminated by noise that appears as a spurious signal unrelated to the source. Since an improper model of the medium is quite often employed, due to poor knowledge of the seismic velocity of the area under study, and since the hypocentre may be mislocated, the Green functions are not without fault.
Thus, structures not modelled by Green functions are assigned to the source, distorting the source mechanism. To demonstrate these effects, we performed a synthetic case study by simulating seismic observations in the Dobra Voda area of the Little Carpathians region of Slovakia.
Simplified 1-D and 3-D laterally inhomogeneous structural models were constructed, and synthetic data were calculated using the 3-D model. Both models were employed during a moment tensor inversion.
The synthetic data were contaminated by random noise up to 10 and 20 % of the maximum signal amplitude. We compared the influence of these two effects on retrieving moment tensors, and determined that a poor structural model can be compensated for by high-quality data; and that, in a similar manner, a lack of data can be compensated for by a detailed model of the medium.
For examples, five local events from the Dobra Voda area were processed.