Seismic Explorationists First View of the Miyagi Earthquake

Bob Clapp and Jon Claerbout

Japan has 600 high quality (each down a bore hole) seismograph stations on a dense, somewhat-uniform mesh. Here we exploration seismologists take our first look, looking at the magnitude 7 earthquake at Miyagi, Japan in 2003

Time vs Distance from epicenter

Here are the seismograms sorted according to distance from the epicenter. Distance is in degrees. Time in seconds.

We are shocked to see no coherency from trace to trace on the first arrival. At later times we do see coherent stepout that is approximately straight lines projecting back towards an origin time of about t=640 seconds. Energy arrives in two trends, presumeably a p-wave trend and an s-wave trend. Theoretically the s-wave trend contains a first arriving s-wave and is a curve concave upward. Theoretically the Rayleigh wave is on a linear trend shortly following the s-wave.

There should be 600 traces here but there are fewer. The reason there are fewer than 600 is that we first binned the data to a regular cartesian mesh in the (x,y)-plane. If we want to see any spatial coherence, we are going to need to do some low-pass filtering. We probably better skip the binning too.

Time-slice movie

Now we put the data in bins on a regular mesh on the earth surface. Successive frames are successive times. Eventually the wave field is observable everywhere in Japan. North is to the right. It looks like random noise filling the expanding ring of the first arrival.


Changing numbers in the lower left corner of the frame are time in seconds.

Time vs Distance around desired core-mantle reflection

Greg was interested in looking for a core-mantle reflection. It should be lying flat in the middle here. We don't see much. The steeply dipping thing in the upper right is the Rayleigh wave.

Maybe there is a little something flat at t=1070 and degrees=12. To see any horizontal events we should do some high cut filtering and some spatial averaging. There do seem to be some unexplained dipping events that might turn out to be interesting.

Jump to filtered data