SUMMARY
Launch a Matlab GUI for estimating source depth and mechanism.
SYNTAX
depmec
(An extended description of how to use this tool is included below)
DESCRIPTION
depmec allows the user to estimate source depth and mechanism using both grid search and simplex methods to fit ray theory synthetics to data. Results from this analysis are stored in the EVDP, USER0(strike), USER1(dip), USER2(rake) header variables. A detailed description of the technique can be found in:
Goldstein, P., and D. Dodge, Fast and accurate depth and mechanism estimation using P-waveform modeling: a tool for special event analysis, event screening and regional calibration, Geophys. Res. Lett., 26, 2569-2572, 1999
LATEST REVISION
June 2000 (Version 00.59)
************************************************************************ ** DEPMEC: P-wave Modeling Based Depth and Mechanism Estimation Tool ** ************************************************************************ * * Authors: Peter Goldstein and Doug Dodge * * Lawrence Livermore National Laboratory * L-205 * P.O. Box 808 * Livermore, CA 94550 * USA * * peterg@llnl.gov * (415) 423-1231 * * * Please cite the following reference when you use this tool: * * Goldstein, P., and D. Dodge, Fast and accurate depth and * mechanism estimation using P-waveform modeling: a tool for special * event analysis, event screening and regional calibration, * Geophys. Res. Lett., 26, 2569-2572, 1999 * ************************************************************************ The following is a hands on description of how to use DepMec V1.0. It assumes that you have version 59 or later of sac2000 and have the appropriate matlab licences and libraries. *** System Preparation: Make sure your SACAUX and MATLABPATH environments are set. *** Data preparation: Depmec will work with any vertical component velocity or displacement data that can be read into sac, provided the appropriate SAC data headers are set. You may find it useful to store your data in event based directories. The following SAC headers should be set: reference time (nzyear nzjday nzhour nzmin nzsec nzmsec) origin time (o) should be zero. reference time type (iztype=io) (The origin time should be the reference time) * If necessary, you can use the following to set a reference origin time. ch o GMT year jday hour min sec msec (only if origin isn't already set) ch allt (0. - &1,o ) event latitude (evla) event longitude (evlo) event depth (evdp) event magnitude (mag) data type, velocity or displacement (idep) station lat (stla) station lon (stlo) station name (kstnm) component name (kcmpnm) (Optional) Pick first arrivals and store in T0 sac header. If no picks are defined, depmec will set the first arrivals using the default velocity model (currently AK135). The user can also repick these interactively in depmec. *** Depth and Mechanism estimation *** At this point it is assumed that you are running SAC2000 and have read in data with properly loaded headers (see above). 1) To start depmec type: depmec (You should see the matlab insignia flash by, a waitbar saying loading initialization file will popup and disappear and then the main depmec window will appear with menus across the top.) Before analysing the data there are a number of preferences you may want to set under the edit and preferences menus. However, if you are just getting started, I'd suggest using the defaults. Further discussion of the preferences is given below. 2) Load the data using the file menu. You can use either displacements or velocity. Displacements may be better if the SNR is large at low frequencies because there are fewer concerns about cycle skipping when doing the cross-correlation to estimate the misfit. However, velocities have potentially higher resolution and work better when the low-frequency SNR is poor. Click file>Load SAC Velocity If your data have the necessary headers loaded and are within an appropriate distance range 15 to 90 degrees they will plotted along with synthetics and a mechanism beachball in the main window. * Most of the following involves using tools started by * clicking on Icons at the top of the depmec main window. * If you can't figure out what a particular Icon is for * place your cursor over the icon and an short * explanation should appear. 3) Picking and Filtering: If you haven't already done so, you may want to filter and pick or repick some of the traces. The easiest way to do this is to open both the single trace pick window (Icon with single trace and pick mark), and the filtering window (Icon with spectrum and F in upper right). It is usuall best to begin by applying some nominal filtering to the data and then remove traces with very poor SNR. While the main window and the picking and filter window are all visible, adjust filter parameters by selecting the the filter type and corners then selecting assign to all. Then apply this filter to all the data with the button to the left of the close box in the upper right corner. If these filter parameters are unacceptable, filter selected traces by stepping through the data with the red arrows at the top of the pick and filter windows. After a preliminary filtering of the data, traces with poor SNR can be deleted using the wastebasket icon on in the pick window. Alternatively, you can deselect a trace by clicking the check box to the left of a trace or the check mark icon in the pick tool window. This changes the color of the trace to gray and leaves it visible but it will not be used to estimate the depth or mechanism. To adjust the picks point the cursor at the vertical black line labeled P and use the middle mouse button to drag it to the desired pick location. The size and location of the pick window can be adjusted using the grey box enclosing the current trace in the main window. Point to the edge and use the middle mouse button to drag the whole window or point to the bowtie handle on the right hand side and drag to reduce or enlarge the lengthe of the window. You can also drag this gray box to another trace and the pick and filter windows will be updated appropriately. When the picks and filters are satisfactory close the pick and filter windows and click the floppy disk icon at the top of the main menu to store these parameters in the SAC data headers. *** Note that these pick and filter parameters are not *** permanently written in the SAC data files unless you *** save them with write or writeheader after closing depmec. *** To close depmec you use the closebox icon at the upper right *** of the main window. 4) (optional) Estimate magnitude: You can estimate the magnitude by turning the Auto-calculate Magnitude on under preferences. However, the magnitude returned when using velocities is still currently under development. *** Event parameter estimation 5) Perform a fast grid search data for depth and mechanism by clicking the main window icon with the lightening bolt over a grid. Even with a lot of data this only takes a few minutes on an Ultra-Sparc with a ~300 MHz processor. If you have lots of data or are impatient, you can speed things up by deselecting all but the best SNR traces. This will often give a reasonable preliminary fit to the data which can be iteratively improved upon by reselecting the rest of the traces and using the simplex method (diamond or kite shaped Icon) to refine the solution. Estimate errors by clicking on the bootstraping icon (the boot with a strap next to the simplex tool). This can take 10 or more minutes so it may be better to run this in batch mode if you have many events to look at. 6) Storing/Saving: Filters, Picks, Source Parameters and filelist: If you select Save SAC Data under the file menu, or click on the floppy disk Icon with the S on it the filter, pick, and estimated parameters will be stored in the SAC file headers. This will also create a new de2RestrictFile file with a list of the files that are currently visible. This file can be used as a SAC macro to read in selected data for future analysis. You can also save the estimated source parameters to a text file using the "add current solution to log file" Icon or the option on the file menu. Preferences: When you exit depmec it will write a file called de2.ini that contains your preferences. The next time you start depmec in this directory it will read this file. You can copy this file to other directories if you want to use it with other data. 7) Setting Preferences: If necessary, modify the default parameters using the edit and preferences menus. If you're running depmec for the first time in the current directory or you've removed any previously generated de2.ini files you should find that most default preferences such as automatic magnitude calculation, and filtering are off. Under the edit menu the key parameters are a) max correlation lag This determines how much the first arrival can be shifted. If you think your picks are really good set this to zero. b) max isotropic percent This determines how big an isotropic component depmec will allow for in the moment tensor. For events deeper than a few km (e.g., earthquakes) this is best left as zero or small since it is another free parameter. However, can be useful to test hypotheses when looking at shallow events. c) min synthetic length (T_sP - T_P): Initially this should be long enough to encompass any of the modeled arrivals over the depth range of interest. For events shallower than 70 km 25-30 s should be enough. * Be careful! If this is too short you can force depmec to find a shallow depth source. However, This could be useful for hypothesis testing with a limited number of noisy stations or stations with large site effects where you only want to fit the initial part of the waveform. Under the prefereneces menu the key parameters are: a) S/N cutoff: I usually start with this at .1 and remove traces interactively. b) Data selection: - file mask: This determines what data files get read in. - pre pick length: - post pick length: - min delta: Current default is 15 deg but beware of triplication before 30 deg - max delta: current default is 90 deg c) Auto filter traces: will automatically filter data when read in using filter parameters stored in SAC headers. d) Auto calculate Magnitude: Will calculate magnitude from the data based on an estimated corner frequency, the assumed stress drop (see edit menu), an assumed circular crack model, and the Kanamori and Anderson Moment-Magnitude relation. This works well with good snr displacement data but still needs work in the velocity domain e) Use Restriction File: If selected, depmec will read from the list of files specified in the de2RestrictFile file in the current data directory.