1. Adding detection codes to CSEP: An InSAR-entist’s tale GLRS (Salton Trough) Otherwise known as: What the heck do I do with this??

2. Adding codes to CSEP  Goals:  Have several operational detectors running by the end of SCEC III (AGU 2011 workshop)  Steps:  Lohman:  Provide a “simplistic” version of a detector that has the necessary parts, inputs, outputs  Make codes available to community and identify the key hurdles faced with out approach  Liukus:  Worked with Lohman on ensuring that the appropriate files and types of tests are available  Community:  Do we have any means of verification yet? Wait and see?

3. Basic steps  Get your code working with the provided input data (GPS, CMT solutions, etc)  GPS now in PBO format, as requested  Rowena’s scripts can download/sort into format that is the same as testing center  Identify transients  Time-delayed, retrospective detections are fine!  Detection format same as before  Location, timespan, spatial extent. Can interpret however you want but has to be correct format (csv)  Rewrite code so that some key paths, dates are in a single input file  Send to CSEP  They will install, change the input file to reflect their data structure, and test.  All programming/scripting languages discussed so far are fine with them.

4. Unexpected (by me) issues  Remove pop-up windows (figures, status bars)  Remember to reset flag when sending codes to Masha  Repeated testing  The center will repeatedly run your code using an earlier data set, and expects the code to produce the same results each time.  We have some say in what “same” means.  Issues with Matlab “now” command  Potential problems with codes that have supporting data that is updated each time. Might have to provide a copy of that, which is used for each retest.

5. Example algorithm and utility scripts  Will be made available at end of SCEC meeting http://collaborate.scec.org/transient/Main_Page  General use:  get_data_UNAVCO.pl  Downloads and untars PBO SNARF solution from UNAVCO  CSEP does same thing each day, only reruns algorithms if there is new data.  read_data_pbo.m  Extracts displacment time series, precision, etc  Okada, utm2ll, etc. scripts

6. My algorithm: get inputs/outputs right  Read in data  Id possible coseismic offsets  Cull sites  Fit rates/EQ for full series  Identify sites that diverge from long-term trend in “recent” time  Produce results file

7. My algorithm: get inputs/outputs right  Read in data  Id possible coseismic offsets  Cull sites  Fit rates/EQ for full series  Identify sites that diverge from long-term trend in “recent” time  Produce results file GLRS

8. My algorithm: get inputs/outputs right  Read in data  Id possible coseismic offsets  Cull sites  Fit rates/EQ for full series  Identify sites that diverge from long-term trend in “recent” time  Produce results file Assumes 100 bar stress drop, CMT Flag if forward model > 1/5 reported errors

9. My algorithm: get inputs/outputs right  Read in data  Id possible coseismic offsets  Cull sites  Fit rates/EQ for full series  Identify sites that diverge from long-term trend in “recent” time  Produce results file  <1 year of data  Too little data in recent testing period (~2 weeks)  Too few neighbors  Predefined # within # km

10. My algorithm: get inputs/outputs right  Read in data  Id possible coseismic offsets  Cull sites  Fit rates/EQ for full series  Identify sites that diverge from long-term trend in “recent” time  Produce results file  Fit for offset and 2 logarithmic decays  Problems with close inter-EQ times  Now: iterate over EQ  Next: do all simultaneously  Reduces detections at Sierra E. M sites, but does a bad job at Parkfield

11. My algorithm: get inputs/outputs right  Read in data  Id possible coseismic offsets  Cull sites  Fit rates/EQ for full series  Identify sites that diverge from long-term trend in “recent” time  Produce results file  Identify “triggered” sites  Only save ones that trigger at # of spatial neighbors

12. My algorithm: get inputs/outputs right  Read in data  Id possible coseismic offsets  Cull sites  Fit rates/EQ for full series  Identify sites that diverge from long-term trend in “recent” time  Produce results file

13. Conclusions  Postseismic transients continue to be an issue  Are of interest themselves  1: Crop out regions affected by postseismic deformation (bad!)  2: Model functional form of postseismic signal (one or more timescales) and remove before fitting  3: Forward model expected postseismic signal based on slip distribution/crustal constitutive laws  Manpower/human input into detections  Assessment?