The effect of uncertainties in seismic source parameters on (waveform) tomography

Abstract:

Earthquakes are the primary source of information used in seismic tomography in order to recover the 3-D structure of the Earth. In this work we investigate how errors in their assumed source parameters can affect tomographic results.

The usual tomographic workflow starts with a catalogue of suitable earthquakes, which are then used as input to the inverse problem that deals with determining Earth structure. The accuracy of the earthquake parameters used in the inverse problem is therefore a concern: errors in the source parameters can translate into errors in the recovered model. This may particularly be relevant in the case of waveform tomography, where subtle differences in the waveform affect misfit. Such errors may result from both the forward modelling approach and the Earth model used to invert for the sources, and may amount to significant differences in spatial location as well as differences in the moment tensor.

Here we investigate the effect of such likely errors on the tomographic problem. We make an overview of the typical errors that can arise in seismic source inversions, and assess the effect of such errors on the tomographic inverse problem in the particular case of waveform tomography. To this end, we compute both comparisons of forward simulations and sensitivity kernels. Horizontal spatial shifts are found to have the largest effect on data fit, and have the potential to locally affect the inverted model quite strongly. This may be a concern in locations where event coverage is poor.