Incorporate propagation modeling for more accurate regional travel times

[liamtoney]

Currently, for regional applications of rtm (projected=False), only time_method='celerity' is supported — we assume a homogeneous wavespeed.

This can result is gross errors in travel time estimation for realistic atmospheres due to different infrasound phases as a function of range, as well as spatially (and temporally) heterogeneous atmospheric state.

We could compute more realistic travel times using e.g. InfraGA ray tracing, and store these in a lookup table. We could take the average vertical profile over the region defined by the source search grid + the station locations. The lookup table would not be valid through time, of course.

Users could supply their own atmospheric specification files, and/or [preferred!] we could use e.g. the ECMWF Python API to download relevant files for the user-specified spatial configuration and starttime, endtime, etc. automagically.

Ping @fkdannemann

[fkdannemann]

If you are still interested in this, we can discuss a collaboration/feature enhancement starting next Oct(ish). Writing my funding proposal for next FY and this came up as a possibility.

[davidfee5]

Thanks Fransiska. I know we are still interested in this but probably have too many other things going on right now! Let's chat about it soon though.