March 21, 2012
Earthquakes, Impacts, and Seismic Imaging
Jeroen Tromp
Professor of Geosciences and Applied and Computational Mathematics; Blair Professor of Geology,
Princeton University
Earthquakes, Impacts, and Seismic Imaging
Jeroen Tromp
Professor of Geosciences and Applied and Computational Mathematics; Blair Professor of Geology,
Princeton University
Minutes of the 25th Meeting of the 70th Year
The 25th meeting of the 70th year of the Old Guard was held at the Princeton Public Library.
President Varrin gaveled the 79 members to order promptly at 10:15. Tom Fulmer let the assemblage in the invocation. Tony Glockler read his minutes of the preceding week’s meeting, describing the talk given by Ruth Mandel, Director of the Eagleton Institute at Rutgers. Ruth Miller introduced her guest, her husband, Bernie Miller.
Scott McVay introduced the speaker, Professor Jeroen Tromp. Professor Tromp graduated from the University of Utrecht and received his Doctorate from Princeton in 1992. From there he went first to Harvard and then to Cal Tech where he was the Director of the Seismology Laboratory from 2003 to 2008. In 2008 he succumbed to Princeton’s advances and joined the Department of Geosciences. He is Blair Professor of Geology and in addition is Director of the Princeton Institute for Computational Science and Engineering (so-called PICSciE). He is a joint author of a 1000 page book on Seismology and over 100 papers. He is the recipient of many awards and prizes and has spoken all over the world to scientific gatherings.
Professor Tromp began by explaining that Earthquakes were intimately connected with the plate tectonics of our earth. Earthquakes often occur on ridges or faults, but many occur where the plates come together. A seismic event there is generally due to one plate driving under another, a process called subduction. This was graphically illustrated by a world map showing how a large number of earthquakes exactly follow the common boundary of adjacent plates.
The data used for analyses by Professor Tromp’s research group comes from a network of hundreds of monitoring stations all over the world that freely share the time series data they record of displacements in three coordinate directions following a seismic event. The speaker said he would describe three areas of interest at Princeton: Earthquakes, Meteorite strikes, and seismographic mapping of the Mediterranean area.
It was pointed out that the Eastern United States is not immune from large quakes, for example one in 1811 in New Madrid Missouri rang church bells in Boston.
The tool at PICSciE used by the Princeton group is a bank of 384 Dell Central Processing Units. These provide the enormous computing power required for their mathematical models. Presently it is possible to execute a million billion floating point operations per second on one of these supercomputers. He would like to take advantage in the future of the much greater computing capability in Graphics Processing Units, developed for the computer games industry.
Before going into particulars Professor Tromp described a general approach to their analyses. The data observed is compared to a theoretical model based on: assumptions about the earth’s properties, a decomposition of the earth into many small cube-like pieces, and a solution to the equations governing the propagation of waves generated by a disturbance. The difference between the predicted waves and the observed waves yields information used to correct the assumptions of the model. A fascinating movie was shown of the modeling technique as applied to a drop of coffee landing in a half filled ceramic cup. The waves generated on the surface were represented in color and were seen to pass to the cup and its handle. More to the point, another movie showed how the disastrous 2011 Honshu earthquake, located at the confluence of three plates, generated shear and compressional waves which emanated from the epicenter in concentric circles, girdling the earth and 90 minutes later were focused on a point off the southeast coast of South America, antipodal to the original disturbance. In that quake the sea floor was displaced upwards by 15 meters, resulting in a huge tsunami.
The Princeton group’s model was applied in the case of a meteor impacting the earth at Chicxulub (“chicks-a-lube”) 65 million years ago. A movie of a physical model of perpendicular impact of an air puff upon a water sphere in zero gravity was shown and illustrated the focusing of the pressure wave at the antipode of the impact point. The analysis of the Chicxulube impact showed that under suitable conditions it could certainly cause seismic activity at the antipode, possibly volcanic activity and a tsunami, and perhaps even fracturing of rocks but not melting. In addition it probably did not cause the Deccan Traps volcanism in India.
Finally the model was utilized to investigate the geology of the earth’s upper mantle in the region surrounding the Mediterranean. There the model was able to utilize the relevant data from 338 stations’ records of 159 earthquakes to verify many known geologic features, and support current hypotheses of paleotectonics. The model took careful account for example of the differing speed of shear and compressional waves in different environments. It is hoped that the entire earth can someday be subjected to a similar analysis. This would require a thousand times as many computations.
Professor Tromp answered questions on rogue waves, (not due to seismic events), protection from earthquakes, (we can’t predict them but we can estimate effects from assumptions), the age of the tectonic plates (50-100 million years, but bits and pieces are 3.5 billion years old), what physical properties are revealed by analysis, (rock temperature and density), and prospects for prediction (with our current understanding, extremely poor except in terms of geologic time).
The meeting closed with a burst of applause for our speaker!
Respectfully submitted,
Lee Neuwirth
President Varrin gaveled the 79 members to order promptly at 10:15. Tom Fulmer let the assemblage in the invocation. Tony Glockler read his minutes of the preceding week’s meeting, describing the talk given by Ruth Mandel, Director of the Eagleton Institute at Rutgers. Ruth Miller introduced her guest, her husband, Bernie Miller.
Scott McVay introduced the speaker, Professor Jeroen Tromp. Professor Tromp graduated from the University of Utrecht and received his Doctorate from Princeton in 1992. From there he went first to Harvard and then to Cal Tech where he was the Director of the Seismology Laboratory from 2003 to 2008. In 2008 he succumbed to Princeton’s advances and joined the Department of Geosciences. He is Blair Professor of Geology and in addition is Director of the Princeton Institute for Computational Science and Engineering (so-called PICSciE). He is a joint author of a 1000 page book on Seismology and over 100 papers. He is the recipient of many awards and prizes and has spoken all over the world to scientific gatherings.
Professor Tromp began by explaining that Earthquakes were intimately connected with the plate tectonics of our earth. Earthquakes often occur on ridges or faults, but many occur where the plates come together. A seismic event there is generally due to one plate driving under another, a process called subduction. This was graphically illustrated by a world map showing how a large number of earthquakes exactly follow the common boundary of adjacent plates.
The data used for analyses by Professor Tromp’s research group comes from a network of hundreds of monitoring stations all over the world that freely share the time series data they record of displacements in three coordinate directions following a seismic event. The speaker said he would describe three areas of interest at Princeton: Earthquakes, Meteorite strikes, and seismographic mapping of the Mediterranean area.
It was pointed out that the Eastern United States is not immune from large quakes, for example one in 1811 in New Madrid Missouri rang church bells in Boston.
The tool at PICSciE used by the Princeton group is a bank of 384 Dell Central Processing Units. These provide the enormous computing power required for their mathematical models. Presently it is possible to execute a million billion floating point operations per second on one of these supercomputers. He would like to take advantage in the future of the much greater computing capability in Graphics Processing Units, developed for the computer games industry.
Before going into particulars Professor Tromp described a general approach to their analyses. The data observed is compared to a theoretical model based on: assumptions about the earth’s properties, a decomposition of the earth into many small cube-like pieces, and a solution to the equations governing the propagation of waves generated by a disturbance. The difference between the predicted waves and the observed waves yields information used to correct the assumptions of the model. A fascinating movie was shown of the modeling technique as applied to a drop of coffee landing in a half filled ceramic cup. The waves generated on the surface were represented in color and were seen to pass to the cup and its handle. More to the point, another movie showed how the disastrous 2011 Honshu earthquake, located at the confluence of three plates, generated shear and compressional waves which emanated from the epicenter in concentric circles, girdling the earth and 90 minutes later were focused on a point off the southeast coast of South America, antipodal to the original disturbance. In that quake the sea floor was displaced upwards by 15 meters, resulting in a huge tsunami.
The Princeton group’s model was applied in the case of a meteor impacting the earth at Chicxulub (“chicks-a-lube”) 65 million years ago. A movie of a physical model of perpendicular impact of an air puff upon a water sphere in zero gravity was shown and illustrated the focusing of the pressure wave at the antipode of the impact point. The analysis of the Chicxulube impact showed that under suitable conditions it could certainly cause seismic activity at the antipode, possibly volcanic activity and a tsunami, and perhaps even fracturing of rocks but not melting. In addition it probably did not cause the Deccan Traps volcanism in India.
Finally the model was utilized to investigate the geology of the earth’s upper mantle in the region surrounding the Mediterranean. There the model was able to utilize the relevant data from 338 stations’ records of 159 earthquakes to verify many known geologic features, and support current hypotheses of paleotectonics. The model took careful account for example of the differing speed of shear and compressional waves in different environments. It is hoped that the entire earth can someday be subjected to a similar analysis. This would require a thousand times as many computations.
Professor Tromp answered questions on rogue waves, (not due to seismic events), protection from earthquakes, (we can’t predict them but we can estimate effects from assumptions), the age of the tectonic plates (50-100 million years, but bits and pieces are 3.5 billion years old), what physical properties are revealed by analysis, (rock temperature and density), and prospects for prediction (with our current understanding, extremely poor except in terms of geologic time).
The meeting closed with a burst of applause for our speaker!
Respectfully submitted,
Lee Neuwirth