Seismic records of the Swift Reservoir Power Canal collapse  on April 21, 2002

The Swift Reservoir Power Canal near Mount St. Helens collapsed early on Sunday, April 21st, with almost 800 million gallons of water spilling out in a few hours. A hydroelectric powerhouse was destroyed and part of a state highway was washed out.  Info from WA Dept. of Ecology The Columbian reports that a study by CH2M Hill suggested that the collapse was due to seepage into ancient lava tubes. Another Columbian article, from Oct 4, 2002.

Based on news reports the staff of the PNSN have looked in detail at the seismic recordings during the period of the canal collapse and have produced  the following preliminary interpretation based on the recorded seismograms.

The PNSN recorded a series of seismic signals starting at about 06:30:33 PDT Apr. 21.  See the seismograms for 2 1/2 hours around the event and a  50 minute seismogram/spectrogram from station, MTM  6km distant.  Several bursts of energy occurred in the ten to fifteen minutes following 06:30 superimposed on a increase in seismic background levels.  These background signals started slowly decaying after about 06:45 and had returned completely to normal background levels by about 07:15.  At 06:56:49 an impulsive seismic event occurred which was strong enough on several PNSN stations to trigger an automatic earthquake locator.  With careful manual analysis of this last event it was located near the head of Yale Reservoir at very shallow depth.  This is near the site of the failure and power plant (see map).  The size of this last event has an equivalent earthquake duration magnitude of 1.6.  Since this event is not an earthquake and occurs at the earth's surface this duration magnitude over estimates the energy release.  We estimate an amplitude magnitude for this event of 0.9 (equivalent seismic energy of 8x10^12 ergs).  Note that the frequency distribution of this last event is somewhat different from those preceeding it.  Based on waterlevel reports provided by the Cascade Volcano Observatory and eyewitness accounts we feel that the seismic signals are consistent with the following interpretation. The short transient signals reported here are  most likely related to the failure of the walls of the canal and the increased background levels are associated with the turbulent flood of water thus released.  Provisional data from water-level recorders operated by US Geological Survey for the Cowlitz County Public Utility District show that the water level in the canal started to decline slowly as early as 03:00. At about 06:00 the drop in level accelerated and by 06:45 the water had dropped below the bottom of the gage (about 12 ft below normal level). An eyewitness reported the sudden failure of the canal embankment and outflow of remaining water shortly before 07:00.  This final failure occurred after more than 12 ft of water had leaked out.   The seismic signals would indicate that canal integrity started failing arround 06:30 PDT.  It may have been progressive over the next 10-15 minutes and included quite a bit of water.  The event at 06:56 has a somewhat different character than the previous signals.  It is much richer in high frequency and very impulsive. It has the characteristics of an impact or small explosion.  If the final failure of the canal wall was very sudden causing a drop of a large amount of material onto a surface (not a sliding away) this would be the consistent with this signal.  The released energy causing the seismic event took place over only a few seconds. It's possible that this last event was initiated by accelerating seepage through the floor of the canal and the embankment.  Small slumps of material caused some sort  of void, into which  a large part of the embankment fell.

The map above shows the region around the canal. The red dot in the map shows the location of the source of the seismic signal, as determined by arrival times of signals at seismographs.  The errors in location could be as much as 2km.

Detailed seismograms of the final, impulsive event.  Click for either a "gif" or PostScript image.:
MTM: [gif] [postscript]    JUN: [gif] [postscript]    HSR: [gif] [postscript]    FL2: [gif] [postscript]   YEL: [gif] [postscript]   SOS: [gif] [postscript]
Quasi-record section: [gif] [ postscript]
This map: [gif] [ postscript]