Tomography- the focus of my research

 August 1st 2009 -

Basically all the things I learned for the last six weeks in the lab were preparing me for doing a tomographic inversion of the data. A tomographic inversion is when you use a set of earthquake locations,  p&S wave arrivals and an initial velocity model to create a new and more accurate velocity model. Such a velocity model can help reveal show structure beneath the surface with higher velocity areas generally being composed of denser materials and slower areas being composed of less dense materials (including fluids).

The arrivals and locations used in the inversion will be those from the month I just picked as well as over a years worth of locations picked by Mike’s previous IRIS Intern, Kevin.  The goal of the tomography is to gain a better understanding of what the structure is at the plate interface in the Oaxaca region of Mexico. We have earthquake and tremor locations in this region, and it would be useful to have a better understanding of the structure to help explain what it going on.

The tomography code I am using is called tomoDD and was written by Haijiang Zhang from the University of Wisconsin and has already been used to work on tomography in Japan. The DD stands for Double Difference and means that the program looks for events that occurred near each other and would only have slightly different ray-paths. It then uses the information from these similar ray paths to solve for a velocity model in that region.  There are many different parameters that need to be tweaked before a decent image results and that’s where I come in. I basically play around with each of the different parameters to understand the affect they have on the inversion. The goal right now is just to have and image of the fast-velocity subducting slab as opposed to the images I’m getting now which look like someone left a three year old alone with a paint set.

So the steps to get to an accurate velocity model would be as follows:

1)      Collect Data from an array of seismometers

2)      Pick the P&S wave arrivals from hundreds of events (Thanks Kevin)

3)      Locate the earthquake using dbloc on Antelope (Again, thanks Kevin)

4)      Input the location into tomoDD and tweak the parameters

5)      Get colorful pictures (which hopefully make sense and don’t look like something Picaso made)

Now I understand why having large arrays of instruments is important. It takes at least 4 stations to pinpoint an earthquake. The more stations you have the better locations you can get and you get more picks. The better your locations and the more picks you have, the higher the resolution that is possible from an inversion.