The Aug. 7, 2000 Alliance Ohio Earthquake
This earthquake occurred in northeastern Ohio, yet not too close to Ashtabula. Thus it provides a valuable test of whether the "LLg" phase is only generated and propagated along the Lake Erie coast. This event was well-recorded across the region.
Lets look at some seismograms of the Alliance earthquake, here is COWO at a distance of about 75 km.
We can see the large amplitude LLg wave above, hence the LLg phase is not exclusive to the Lake Erie shore.
OSLO is about two hundred km epicentral distance.
We see that LLg is now a well-dispersed wavetrain with amplitude similar to the S wave.
SSUO is at about three hundred km epicentral distance.
LLg is still present at seismogram right edge as a well-dispersed wavetrain with amplitude still similar to the S wave. Note that the "official Lg" is now beginning to emerge just 5 to 10 seconds behind the S wave.
Even further, UOCO is about 360 km epicentral distance.
LLg is still present at the seismogram's right edge as a well-dispersed wavetrain with amplitude still similar to the S wave. Hence, LLg amplitude is as easy/difficult to measure as S or Lg at these far distances.
How do the OhioSeis seismograms compare to those from the USNSN stations? Overall, they look quite similar except that the USNSN system records higher frequencies, as allowed by their 40 samp/sec sampling rate. While this sampling rate includes much more high-frequency noise, it also captures the occasional larger high-frequency peak in the P and S waves. Below, I show the MCWV seismogram at an epicentral distance of 178 km, between the COWO and OSLO distances.
First of all, note that MCWV also records the LLg waves with U=1.85 km/sec.
The high-frequency peaks in the P and S wave have an effective period of about 0.25 sec, hence are filtered out by the 10 samp/sec OhioSeis system. However, it is well-accepted that magnitudes should not be based on a single high-frequency spike. For example, the details of the Hermann and Kijko (1983) formulation are that you should measure the third-highest peak — this is a simple device to effectively filter the seismogram. Application of this procedure to the MCWV seismogram reduces the P and S amplitudes to half their peak values.
Lets now plot the amplitudes for all phases and stations, including MCWV.
This graph shows amplitudes for P, S, Lg, and LLg waves recorded by OhioSeis and MichSeis stations (plus MCWV, see below) for the Alliance earthquake. Amplitudes are plotted in "digital units". The microseismic noise amplitudes for this day are around 60 du, hence most waves are less than the 6-sec microseismic amplitude. The SeismoView filter can extract these small amplitude waves since they have larger amplitude at periods less than 6 sec. Notice that LLg amplitude is greater than S amplitude for distances less than 100 km, but that S amplitudes are slightly larger at greater distances. For reference, a dash-dot line is drawn for log slope of -.90 and at a level appropriate for MN=2.4 for the ground velocities listed along right-axis. These ground velocities are converted from du at a period of 1.5 sec, the typical period of the LLg phase. The best-fit line for a power law between distance and LLg amplitude is plotted as the solid line. Notice that its log slope is -1.7, a much stronger spatial decay than the log slope of -.90 for the MN formula. The waves from USNSN station MCWV are plotted together with the OhioSeis data by matching microseismic amplitudes. The larger amplitudes for P and S waves are due to the fact that single large amplitude peaks occur with a period of about 0.2 sec in the MCWV seimogram. Also, application of the SeismoView filter would reduce the MCWV LLg amplitude.
| Back | To Ohio Waves & Magnitudes Main page | Forward |