Project Abstract
|
The composite source model (CSM) implementation on the broadband platform initially used the layered velocity model as provided to the broadband modelers, described by Goulet et al. (2015) [13]. However, this model introduced some artifacts in the synthetics, and was not successful in reproducing the distance dependence of ground motions. This study starts with the observation of Christensen and Mooney (1998) that models for crustal structure generally find that the crustal waveguide is on average best modeled with a gradient from the depth of 5 km to the Moho. The results of this study suggest that a crust matching the gradient they found for the global average crust is likely to be more successful. The gradient is sampled at 1 km intervals for the CSM input. The shallow layers also are approximated by a smooth functional form, resampled at small depth intervals. These thin layers minimize impedance contrasts, resulting in fewer resonances in the generated spectra.
This study also considers how to improve the coda model. In early versions of the CSM, a coda transfer function is convolved with each Green’s function. The initial Green’s functions can be more polarized than usual observations. The new model partially addresses this problem by adding the coda contribution for each component in varying fractions to the other components. At this time, the revised model is ready to be submitted for implementation in the next available version of the broadband platform for additional calibration and testing. |
