Purpose

This writeup begins with a general description and links to explanations of procedures. I've annotated the links so you will know what parts of the procedures you need to do. The next part of the writeup lists your assignment, and contains the links to the data sets you will use.

More information on the use of refraction equipment to estimate S-wave velocity structure is available in the following draft reports:

• ``Obtaining Surface-Wave Velocity Spectra from Seismic Refraction Recording Equipment,'' http://www.seismo.unr.edu/ftp/pub/louie/papers/disper/refr.html

• ``Geophysical Surveys of Basin Properties and Geometry: Examples of Seismic Hazard Evaluation From Nevada and New Zealand Basins,'' http://www.seismo.unr.edu/ftp/pub/louie/talks/vuw-ses/vuw-ses.html

Explanations of Procedures

1. Convert the data from the seismograph (Sun only). You will not have to do this for the lab exercise, since I have you download data examples from the assignment section.

2. Obtain the Java software (any computer). You should have already done this for the previous lab. The page does offer some helpful links.

3. Run the velocity-spectrum analysis (any computer). This you will do for the lab exercise. Here you will be opening both SEG-Y and raw float data sets. The link here goes in after the instructions for opening the data file, but you will want to scroll up to those instructions when I give you a raw float data set. You will be instructed in the assignment to also apply geometry information.

4. Model structure from the results (Sun only). Where this page talks about labeling and tracing axes to make picks, you should realize that you can now use Viewmat's Pick Window to get picks exactly. The class can work on the Sun-dependent modeling together after lecture on Thurs. Feb. 10. For each data set you interpret, make a text file of your picks from the velocity spectrum, and email it to me before the Feb. 10 class period. Then we will all be able to compute an S-velocity profile from your picks. Also, you are welcome to log in to a Sun workstation in LME 320. Use the login ``guest''; I will give you the password in class. On a Seismology workstation you will not have to download the disper program; just start the program with the command ``disper''.

Assignment

1. Analysis of surface waves on refraction records: First we will identify and analyze surface waves recorded in Dixie Valley during the same survey we analyzed for refraction picks last week. Rob Abbott and Christine Mann both helped collect this data set in March 1998. We collected it as a seismic reflection survey; you can see the many different purposes a good data set can be used for.
   • Download two separate records in binary SEG-Y format: DIXI0106.sgy and DIXI0114.sgy. Record 114 is from a playa in the center of Dixie Valley while record 106 is from a shallow granite pediment (the Piedmont) at the edge of the basin.
   • Open each record separately in Viewmat. These SEG-Y files do have the 3600-byte reel header block present (so leave the box checked), but they still have IEEE/Java float and not IBM float trace sample data. Note that the first 24 channels are out of order. Apply the source-receiver geometry to each record, similar to how you applied geometries in the previous lab. For record 106, the single line of observer's report to use is ``0106 193 2 219 196 220 243''. For record 114, the observer's report is ``0114 53 2 76 53 77 100''. These observer's reports are not the same as what we used last week. They account for the trace ordering by defining two receiver patches, one reversed. For the station coordinates use the very same ``dixie-med-vp.txt'' file you used last week. Applying geometry ensures correct analysis but will not fix the reversed appearance of the plotted records. Note these records have 4 seconds total time.
   • In Viewmat use the Pick Window to identify the following phases on record 114: P-wave refraction, direct S wave, Rayleigh wave. Report their approximate velocities. For information on surface waves see Telford et al. p. 153-154, 182. Do you see an air wave? What measurement would confirm its existence? Is there an air wave on record 106?
   • Identify some spatial aliasing effects on record 114. Turn in a plot with the artifacts circled, or describe them.
   • Now run the velocity-spectrum analysis on each record separately, for two different velocity-spectral plots. When you run the vspect method, note that you should confirm the correct source-receiver offset range for the record. Use the default Fmax and Vmin values, but increase Np to 48 for improved velocity resolution.
   • Try several values for the clip levels of the velocity-spectral plots. Identify the Rayleigh-wave phase-velocity dispersion trend in each. Try to make a least a dozen picks along the center of the trend in each. If you see more than one trend, pick the lower-velocity trend, since the fundamental-mode Rayleigh waves we want to interpret have lower phase velocities than the higher-mode overtones. What is the lowest frequency at which you can get a good phase-velocity estimate for each record? Turn in the two picked, grayscale velocity-spectrum plots (or email color plots). Also email the saved velocity-spectrum picks to louie@seismo.unr.edu, hopefully as plain text in the body of your email, and not as attachments; send your picks before class on Feb. 10.
   • Discuss briefly how well your Rayleigh-wave phase-velocity picks agree with the velocities of the waves you could observe directly in the record 114.
   • Compare the phase velocities from the two records. How do they differ? Can you justify the difference from the information above about their recording locations?

2. Comparison with last week's records:
   • Load the data set used for the previous lab into Viewmat and apply geometry, just as you did before, using the observer's report lines from that lab.
   • Run the velocity-spectrum analysis on all four records together, for one composite velocity-spectral plot. Note that the Methods menu has an On Each Plane sub-menu that contains the vspect command, since there are multiple records in this file.
   • Compare the result to the velocity-spectral result from record 106 you got above. Versions of record 106 were used in both these analyses, yet they look very different. Explain what happened. Hint: frequency resolution is proportional to the duration of time recorded. It is not dependent on the time sampling interval.

3. Analysis of microtremor noise. Here we look at a passive seismic record, taken without any active source. A 24-channel refraction layout was made along Mill St. in Reno, extending west from the corner with Rock Blvd. It was afternoon, and there was plenty of traffic. We can again thank Rob Abbott and Christine Mann for helping get the data.
   • Download the rno-noise.flt 1.1 Mbyte data file, setting raw binary FTP mode if needed.
   • Import the data into Viewmat. This file has the Raw Float format (just a stream of IEEE/Java floating-point binary trace amplitude values, no headers or labels) and is not SEG-Y. In the Open Binary File... dialog, set the Binary File Type to Raw Float. As the instructions describe, you need to know the number of bytes to skip from the beginning of the file (0), the number of samples per trace (elements per vector = 3000), the number of traces per record (vectors per plane = 24), and the number of records in the data file (number of planes = 4).
   • Plot the record. The further instructions show how to apply an existing plot parameter file. Download the plrno-noise.par file (it is plain text). Now use the Edit menu on the data plot that came up, and select Apply Parameter File....
   • Now you should be able to Animate and interpret the passive records. How many seconds long is each record? What is the aperture of the 24-channel spread? Since the receivers were in a straight line and evenly spaced, the plot parameter file actually defines the geometry well enough for this analysis.
   • Examine the records to decide where their energy comes from. You'll notice a thick, fuzzy group of high-energy waves running almost the length of each record; some have more than one group. Estimate the apparent velocity of such a group - it will be very low. Make a few picks in Viewmat along the center of the group, and you will be able to compute the velocity easily. This is not a seismic wave-propagation velocity. Convert the velocity to miles/hour and you will be able to say what is propagating.
   • Radiating from each group of fuzzy energy in these records are noticeable seismic waves that have reasonable velocities - 100 to several hundred meters/sec. Turn in a plot of one record (email is OK) where you have made a couple of picks of one such wave, and computed its velocity.
     (If you cannot plot the record in Viewmat, look at this PDF plot.)
   • Now run the velocity-spectrum analysis on all four records together, for one composite velocity-spectral plot. When you run the vspect method, note that you should confirm the correct dt=0.016 s and dx=15.24 m. Use the default Fmax and Vmin values, but increase Np to 48 for improved velocity resolution.
   • Adjust the clip level of the velocity-spectral plot, identify the surface-wave dispersion trend, and try to make a least a dozen picks along the trend. Pick the lowest velocity reasonable, since a passive array records energy from all directions, and any energy not propagating in the array direction will have an artificially high apparent velocity. What is the lowest frequency you can get a good phase-velocity estimate for? Turn in a picked, grayscale plot (or email a color plot).
     (If you cannot plot the vspect result in Viewmat, look at this PDF plot.)
   • Discuss briefly how well your Rayleigh-wave phase-velocity picks agree with the velocities of the waves you could observe directly in the noise record.
   • Estimate the wavelength of a Rayleigh wave that samples 0-30 m depth S-wave velocities best (Hint: it is more than 30 m). Justify your estimate briefly with one of: surface-wave or skin-depth theory (e.g. Aki and Richards p. 259-267); references from the engineering literature (the Stokoe papers are useful); or simple principles of wave propagation (e.g. Telford et al. p. 149, 153-154). Identify from your velocity-spectral picks a frequency and phase-velocity combination that matches that wavelength, and report your estimate for the 0-30 m depth average S-wave velocity at Rock and Mill.
   • Examine the soil-type classifications for earthquake ground-shaking hazard mapping found at:
     • Seekins, Linda C. , Boatwright, Jack, and Fumal, Tom, 1999, Soil type and shaking hazard in the San Francisco Bay Area: U.S. Geological Survey web site at http://quake.wr.usgs.gov/hazprep/soil_type/
     Using your 0-30 m velocity estimate, report the ground-shaking hazard classification for Rock and Mill (don't call any newspapers; we already know). Now you have boiled 1.1 megabytes of data down to one letter.

4. Modeling S-velocity structure.
   • If you have emailed your velocity-spectral picks from Dixie Valley records 106 and 114 to louie@seismo.unr.edu in advance of the Feb. 10 lecture, we will be able to do the velocity modeling together, and I will email the resulting structures to you.
   • The picks you email to me are saved in this FTP directory, as files ending in ``.pck''. Using a C program I converted the picks to files ending in ``.dis'', just pairs of period values in seconds and velocity values in km/s, ready for disper. The models we estimated are in files ending in ``.mod''. If you can't list the directory try a URL like: http://www.seismo.unr.edu/ftp/pub/louie/class/453/surf/boskie114.mod
   • Alternatively you can follow these instructions to do the modeling yourself on an available Sun workstation.
   • Report the maximum depth your dispersion data constrain velocities to.
   • Report any attempts to evaluate velocity-depth trade-offs.
   • Briefly compare the two velocity structures from records 106 and 114, and justify these differences in terms of the local geology I described above.