Seismic Tomography Models

This repository contains several seismic tomography models as netCDF4 files, saved in a consistent and useful radius-lat-lon format for plotting cross-sections and basemaps with Matplotlib.

Format

The format used in this repository is loosely based on the format used by the Seismological Facility for the Advancement of Geoscience (SAGE):

• Dimensions of the netCDF file are radial, latitude, longitude.
• The geographic coordinates have units of degrees North and East respectively.
• The longitudinal coordinates use a bipolar convention (with a twist, see below).

However, there are also some key differences:

• The radial dimension is actually the radius (in meters), and depth in kilometers is an additional dimension.
• Where it is reasonable¹, tomography models have been radially extrapolated to the surface and core-mantle boundary.
• Longitude ranges from $-180$ up to but not including $+180$ degrees, primarily because this resolves many issues when converting from bipolar to positive (from $0$ up to but not including $360$ degrees).
• The longitude dimension contains a 'convention' attribute, which tracks whether the longitude is in bipolar or positive format².
• Data variables have more meaningful names (e.g. if P-wave velocity perturbation is in %, it's called dVp_percent)
• Significantly less information is stored in the top level attributes of the netCDF4 files—some have been dropped entirely (e.g. full author lists) while others have been moved to the appropriate secondary level (e.g. units for dimensions are now attributes of those dimensions).

Processing

The processing folder contains the Python scripts that convert the source files into the format used in this repository. These scripts may require some combination of the following libraries: numpy, xarray, and pandas.

Plotting

The plotting folder contains an example script³ for plotting a radial cross-section, given two arbitrary geographic coordinate pairs. This depends on the spherical.py file, containing functions which are useful for computing things in spherical coordinate systems.

Obviously you can use any language/libraries you wish, but the example plotting script in this repository is for Python 3 and requires the numpy, scipy, xarray, cmcrameri, cartopy, and, of course, matplotlib libraries, as well as a working $\LaTeX$ installation. (If you don't already use $\LaTeX$ for typesetting your plots, you should!)

A big motivation for me to develop this workflow was my stubbornness to avoid using GMT or PyGMT. Since I was already very familiar with Matplotlib and using it for ordinary scientific plots, once I started working with geoscientific data I would either have to transition all my plotting to GMT, or learn how to do geoscientific plots with Matplotlib if I wanted my plots to look consistent (which I did). Unfortunately, a lot of things that are easy to do with GMT require a bit more effort to achieve in practice with Matplotlib, most notably, cross-sections. Matplotlib was just never designed with plotting heatmaps on circular sectors in mind, and it's not something many people are doing (they're mostly doing it with GMT).

Data sources

Every tomography model has its respective journal article(s) and download link saved as an attribute in its netCDF4 file, which in Python may be accessed using xarray (SciPy does not support netCDF4).

Footnotes

1. I.e., in the cases where the the shallowest depth is a few tens of kilometers from the surface or the deepest depth is a couple of hundred kilometers from the core-mantle boundary. ↩

2. In a future version, which will be Soon™, this will change from 'convention' to 'antimeridian', which will track the location of the antimeridian such that a global basemap can be plot centred on any meridian. ↩

3. Soon™ it will also have one for a spherical quadrilateral (or basemap), given four geographic coordinate pairs which are joined by great circles. ↩