Oklahoma Geological Survey annotated 24 Hour Low frequency seismograms.


LONG EXPLANATION: hardware and software details.

Revised and updated, 2003 AUG 14, JL.

These seismograms represent vertical earth velocity sensed by a Guralp
CMG-1TBD seismometer at a depth of 850 meters in a borehole, A digitizer
inside the seismometer case digitizes the output of a vertical and two
horizontal sensors. The 2000 per second is repeatedly anti-alias filtered
and resampled to produce three output streams: 200, 20, and one sample
per second.

This data is compressed and assembled into one kilobyte packets. Each 
packet is time tagged with signals from a Global Positioning System
sattelite receiver at the surface. 

The data packets are sent by infared pulses in an optical fiber running
850 meters up the borehole. and 400 meters overland. The packets are
received by Guralp SCREAM40b12_linux software running on a PC in the
main Observatory building. The PC uses a SuSE 8.0 Linux operating system.

Each packet is retained on a ring buffer in the seismometer until that
packet is acknowledged by the SCREAM process on the PC.

Each data steram is recorded on hard disk, displayed in one or more
displsy windows, and broadcast on our Local Area Network to six other
computers (four Linux, two windows98) running SCREAM. Packets are also
sent by internet to the Ominplex in Oklahoma City, and other locations.

For these Low Frequency seismograms, only the vertical motion one sample
per second data is used. It is decimated by Lawrence Livermore Seismic
Analysis SAC (Seismic Analysis Code) software to one sample per two seconds.

 SAC bandpass filters the one sample per two second data to pass
mainly frequencies between 0.0143 and 0.0625 Hertz. This eliminates all but
surface waves from earthquakes, and some body waves from larger earthquakes,
and some background earth noise. At these frequencies, the background noise
consists mainly of the "single frequency" ocean generated microseisms with
frequencies around 0.05 Hertz. The better known, and larger "double
frequency" oceanic microseisms with frequencies around 0.14 Hertz are
not seen in the passband used on these seismograms.

Drum recorders with the trace going around while moving sideways,
make a helical trace. When the paper is removed, it has a series of lines,
each representing a certain time period, with each trace being one time
period later than the one above it. The SAC software was programmed to
make 24 one hour lines, which appear as if they had been made on a drum
recorder. Each line begins on a Greenwich Mean Time/Universal Time 
Coordinated (GMT or UTC) hour. To convert to CST subtract six hours.
To convert to CDT subtract five hours. Sometimes this will make the
CST/CDT date different from the GMT/UTC date.

The SAC image is converted by Image Magick open source to an approximately
779 by 424 pixel png image file. Image Magick is also used to label the
earthquakes.

If there are very large surface waves, they may go beyond the top and
bottom of the seismogram, and obscure other earthquakes. If the magnification
is reduced to bring them on scale, much of the seismogram may be reduced to
straight lines, preventing smaller earthquakes from being seen. For this
reason, most seismograms are SIGNED ROOT COMPRESSED.


On most of these 24 hour seismograms, one or more earthquakes is identified
by region name and magnitude, eg. "Irian Jaya 5.7Ms". Where practical the
lettering is below the recorded earthquake.

In other cases, the lettering is labeled with something like (below), or
is color keyed. Starting 2000 AUG27 the great circle distance from the
epicenter to Leonard in great circle degrees is labeled (eg: D=146).
Deep earthquakes have small surface waves. If an earthquake hypocenter is
deeper than ~50 km., it may be labeled (eg. h=280km) to explain the smaller
surface waves.

If earthquake surface waves overlap (not unusual due to the length of
the surface wave trains), or are close together on the seismogram, colored
type and rectangles may be used to help sort them out.

The epicenters used in annotation are from these three sources:
1. SED = Swiss Seismological Service/Schweizerischer Erdbebendienst
   seismo.ethz.ch/redpuma/redpuma_ami_list.html
2. USGS = U.S. Geological Survey, National Earthquake Information Center.
   neic.usgs.gov/neis/bulletin/
3. KNET = Kyrgyzstan Real-time Broadband Array
   epicenter.ucsd.edu/~quake_KN/bulletin.txt

SED epicenters include USGS and KNET epicenters, as well as epicenters from
several other sources. 

These web pages are processed by a 
NAWK program which calculates the distance to Leonard, Oklahoma U.S., and
the expected arrival times of P/Pdif/PKP and LR. Sources 1. and 3. tend to
have lower magnitudes than source 2.

Beginning 2001 FEB05 some magnitudes are annotated as "M", because the USGS
stopped distinguishing the magnitudes of their locations as Mw, Ms, mb, etc.

Before 2002 JAN 01 the seismograms are indexed by individual dates.
Beginning Jan 01 2002, the seismograms are presented in a scroll
of the four most recent days, and in individual scrolls of all
seismograms for each calendar month.

Back to seismogram index