3-D velocity model and traveltime
click on images to download a high res. pdf version
Creation the 3-D model
Each of the roughly 25 regions defined for Asia corresponds to a geologic province. The regions have been
selected so that the structure within any one region is relatively homogeneous. This allows a single
traveltime curve to predict trvaeltimes throughout the region. The traveltime curves have been been determined
empirically from decades of observations, in many cases by local seismic observatories. The structure implied
by the traveltime regionalization is one in which each region can be defined by a single one-dimensional
profile which changes abruptly at region boundaries. We attempt here to create a full three-dimensional model
with the same assumptions.
To achieve this, the regional 1-D velocity profiles are overlain on the prelimenary 3-D node structure. Each node on the surface of the velocity model is
tested to determine which region it falls within. The corresponding velocity profile is then applied,
including any including shifting nodes up or down in depth as required by the regional velocity models.
Meshing continuous boundaries with finite grid spacing would result in blocky borders. To avoid this, the
borders of each region were propagated inward (similar to a wave front) by roughly 100 km. Points falling in
between regions were assigned linearly interpolated velocities of the adjoining regions. This created a short
though smooth transtion between regions in the final 3-D model. (Not only is this better for ray tracing, it
is also more geologically plausible). The regions with transition zones are shown in the figure at right.
One initial result is that regional-scale (up to 20 degrees) raypaths throughout the region are essentially
straight lines. Deviations to raypaths may occur in the crust and during interaction with the Moho. However,
once in the mantle, the velocity variations (< 0.4 km/s) are not significant enough to bend rays significantly
from their azimuth. The example shown for station AAK sits near the edge of the Tibetan plateau, the thickest
crust in the model region. One would expect the path deviations to this station to represent the worst case
scenerio. (Rays were not traced at two azimuths.)
Model Files
| "C" class models | Description (files are Z compressed) |
|---|
| C: 20x25 nodes | Fast. Good for prelimenary modeling |
| C: 40x50 nodes | Suitable when errors of 0.05 s are acceptible |
| C: 80x100 nodes | Negligible model error but combersome and slow |
Traveltime residuals at central Asia stations
Raw data from ray tracing column format. See raytrace3d for details (files are compressed):
id x0 y0 z0 x1 y1 z1 T0 Ta TT T dT ratio obs_order pre_order amplitude
Bondar predicted traveltimes, provided by MRC (files are compressed):
lat lon tt_correction variance_of_tt_correction traveltime
