Patricia M.'s blog

Finally, success!  I've got the time search program working, and I've been able to find some windows with LFE looking things.  The reason I wasn't getting any data was simply that I hadn't been thorough enough about changing dates from my old program - I was telling Matlab at one point to cut a data set between times A and B when time B was actually before A..... yeah.  So no wonder it wasn't working.  I also fixed a minor timing issue, or rather it fixed itself, because I don't really know what I did, but it stopped being an issue.  For some reason when Matlab cuts the windows of data out, the time stamps that mark the beginning of the seismic record for that window get put in the format 2009:08:25:00:60, rather than 2009:08:25:01:00.  But when you use the datenum function on both those time vectors you still get the same number, so it all works out.  I've made the program pretty user friendly - the program starts by asking you to enter a time in minutes and seconds, and then it pops up the window corresponding to the time you entered (you have to enter specific times 2:30 apart, it won't do just any time).  You are then given four options: 1) advance to next window 2) go back to previous window 3) enter a new time and 4) quit.  It's simplistic, but it serves my purposes, and so I am well satisfied.

I have to say, one of the things that drives me absolutely bonkers about geophysics is how many hoops you have to jump through just to look at the data!  It's enough to make one wish for the days when seismic intstruments were an ink pen scribbling on paper.....

Anyway, I determined after finally getting the correct inputs for the ReviewDataP.m function that the function was too cumbersome for what I wanted.  It ran very slow on my computer, and I couldn't get it to zoom in.  So I've decided instead that I'll adapt some code I used to look at .SAC files from earlier in the summer for my present purposes.  Back then I didn't know about the input function and such, so I'm making it more user friendly.  I got the first part working (at least I think), the part that reads in the .SAC files, stacks the seismograms, and cuts the stacked seismogram into 5 minute overlapping windows.  However, the second part, the part where I actually get to LOOK at the wiggles, is refusing to cooperate.  The program works by converting a time the user gives it to a number, and then searches among the cut-up stacked seismogram for a piece that had a start time the same as the time the user enters.  I got it to work with the old data, so I know it worked for that.  The problem is it can't match a time from a seismogram piece with the time the user gave.  I'm not sure if this is because the times I'm giving it are off, or if there's some deeper problem.  I have checked the start times of the pieces, and tried those start times, and it still doesn't work. 

......maybe SOMEDAY I'll get to look at the data.....

On a less frustrated note, I was kind of surprised, well, before I got stuck, how easily the Matlab was going.  I came out of the summer feeling like I knew nothing, yet now when I go back to it I find I know a lot more than I thought.  So that, at least, is encouraging.

So my IRIS internship is sort of being resurrected from the dead right now, as I'm finishing up my work as my project for an Intro Matlab course at my school.  I'm going to look for LFEs in the Slumgullion data, and then if I find some, compare them to LFEs from the Juan de Fuca plate.

This here is from the blog I have from the class, pretty much copy-pasted because I'm being lazy, if we can be frank.

"All right, updates.  I'm uploading to my folder copies of the programs I'm working with, as they stand now.  Both are programs written by someone else that I'm bending to my own arcane purposes.

xCorr.m is a program that, given a sample window that contains the sample LFE (low frequency earthquake), takes the data from that sample window and compares it to wave forms across other data sets from hours around the event.  Most of my work this summer was just getting the code to function.  My first goal is to get it to read from the new landslide data I just got my hands on at GSA, and do cross correlations across that data.  Theoretically, it should be fairly straightforward, since it's the same data type, but it will probably end up being far more complicated than just adjusting for different directories.  At present the code doesn't save any work, so eventually I'd at least like to to print to a file how many events it found for each event in each hour.  It'd be even cooler if I could get the individual times of those events, which at present the code doesn't do.  This would be even more useful, because then it would enable me to go back and look at the likely events by themselves.

Well, due to my later start, while most everyone else is packing up to return to their respective places of origin, I still have a week to go.  Which is not long, and won't feel long, considering I am now doing fieldwork on the Slumgullion Slide, which has got to be the coolest place I've ever seen, geologically speaking.  Either side of the slide is bounded by very defined transverse faults, which basically look like the Carrizo Plain in miniature.  There's also some normal faulting on parts of the slide which look like chasms opening up.  And selenite crystals everywhere!  I now have way too many than is good for me.  I'm enjoying running around the weird topography of the slide changing Texans and helping with survey points whilst trying not to get struck by lightning.  Life is good in Colorado.

As far as my project goes, that's on a back burner for now – due to some unfortunate circumstances I wasn't able to get very far at all while in Seattle, so since I won't have anything in time for AGU I'm planning on presenting at SSA in April with the help of one of my professors at the University of Oregon.  I'll miss seeing everyone, but I would much rather do a good job on my project than show up at AGU with something slapdash and half baked.

 I'll save summing up my final experience for when my experience is really over – it's difficut right now to accurately convey my sense of what I have taken away from this internship.