## Global anisotropic models

Much of my research has focused on the development of global scale models of radially anistropic shear wave velocity in the Earth's mantle. Two models have been published based on inversion of a database of body and surface waveforms inverted using Non-linear Asymptotic Coupling Theory (NACT). SAW642ANb is the most recent and preferred model. Further details and downloads of the models are available on the individual pages.

### References

Panning, M.P., and B.A. Romanowicz, "A three dimensional radially anisotropic
model of shear velocity in the whole mantle," *Geophys. J. Int*,
**167**, 361-379, 2006. pdf

Panning, M.P., V. Lekic, and B.A. Romanowicz,
"The importance of crustal corrections in the development of a new global
model of radial anisotropy"
*J. Geophys. Res.*, **115**, B12325, doi:10.1029/2010JB007520, 2010.

## Regional tomography of Southeast Asia using modified Born kernels

### Summary (adapted from paper abstract)

Southeast (SE) Asia is a tectonically complex region surrounded by many active source regions, thus an ideal test bed for developments in seismic tomography. Much recent development in tomography has been based on 3D sensitivity kernels based on the first-order Born approximation, but there are potential problems with this approach when applied to waveform data. In this study, we develop a radially anisotropic model of SE Asia using long period multimode waveforms. We use a theoretical "cascade" approach, starting with a large-scale Eurasian model developed using 2D NACT sensitivity kernels, and then using a modified Born approximation (nBorn), shown to be more accurate at modeling waveforms, in order to invert a subset of the data for structure in a subregion (longitude 75° to 150° and latitude 0° to 45°). In this subregion, the model is parameterized at a spherical spline level 6 (~200 km). The dataset is also inverted using NACT and purely linear 3D Born kernels. All 3 final models fit the data well, with just under 80% variance reduction as calculated using the corresponding theory, but the nBorn model shows more detailed structure than the NACT model throughout and has much better resolution at depths greater than 250 km. Based on variance analysis, the purely linear Born kernels do not provide as good a fit to the data due to deviations from linearity for the waveform dataset used in this modeling. The nBorn isotropic model shows a stronger fast velocity anomaly beneath the Tibetan plateau in the depth range of 150 km to 250 km, which disappears at greater depth, consistent with other studies. It also indicates moderate thinning of the high velocity plate in the middle of Tibet, consistent with a model where Tibet is underplated by Indian lithosphere from the south and Eurasian lithosphere from the north, in contrast to a model with continuous underplating by Indian lithosphere across the entire plateau. The nBorn anisotropic model detects negative ξ anomalies suggestive of vertical deformation associated with subducted slabs and convergent zones at the Himalayan front and Tien Shan at depths near 150 km.

For the simple joy of it, here's a word cloud of the published paper.

### Model files available for download:

gzipped tarfile (~3.1 MB)

The distribution includes all necessary model files, as well as some tools for obtaining model values at specified locations in the mantle. To use these tools, you will need C and Fortran90 compilers. Try the GNU compiler collection if you do not already have these available on your system. Some sample plotting scripts for plotting depth slices and cross sections are also included. These require GMT and the ability to run csh scripts (standard in Unix, linux, and OS X systems).

### Reference

M.P. Panning, A. Cao, A. Kim, and B.A. Romanowicz

"Non-linear 3D Born shear waveform tomography in Southeastern Asia"

*Geophys. J. Int.*, **190**, 463-475, doi: 10.1111/j.1365-246X.2012.05489.x, 2012.

pdf of accepted submission

This work was supported by NSF grants EAR-0738284 and EAR-0911414. The linear Born kernels were calculated using a code written by Y. Capdeville.