OK so just watch a movie, get confused, think about it (and maybe even read more of my webpage), watch it again, have a revelation, then smile :)
The axes are latitude and longitude.
The numbers at the bottom are time in seconds, in each case the PDE catalog origin time is around 100 seconds.
The blue line is the peak amplitude for any grid point at that particular time.
Your attention span probably isnt large enough to read all the gory details from a website, so if you have any questions, please email me!
ahutko@pmc.ucsc.edu
Nov 14, 2007 M7.7 Chile earthquake using 124 US stations, linear stacking Nov 14, 2007 M7.7 Chile earthquake using 124 US stations, cube root stacking
If you look at the linear stacking movie it looks like 2 main stages of rupture with the second (much weaker) stage being 70 or so km to the south and about 26 seconds after the largest slip happened. The stars are the NEIC locations of the main event and aftershocks. The box is Paul Earles guess for the fault plane, which agrees with the spatial extent of bursts of high frequency energy imaged here. Tomography corrections were not applied, which probably would have helped with the static shift to the west.
All of these 2007 Indonesian events use whatever clean BHZ stations 30-90 degrees from the source that IRIS WILBUR gave me. There are many more BHZ stations north of Sumatra than south, hence the norhtely swimming direction.
Hi-net data were generously provided by NIED in Tsukuba, Japan http://www.bosai.go.jp/e/index.html
GSN, US Array and other North American data were provided by IRIS. http://www.iris.edu
Why do all the movies "swim" in one direction?!!!
Its always in the direction of the majority of stations.
Cumulative backprojection image summed over 600 seconds for the 2004 Sumatra earthquake
Its all about coherency. Higher frequencies = shorter periods = tougher phase alignment because the earth doesnt like to play nicely. Unaccounted for 3D velocity structure causes small advances and delays that inhibit the coherent arrival of energy. Here, static station corrections were made using a reference earthquake near the hypocenter. So gridpoints far away from the hypocenter (towards the North) have imperfect corrections for 3D velocity structure, hence our inability to image high frequency energy. This is a problem for 1000km long faults and not so big of a problem for smaller < 200km long faults. One (slow) solution: use many aftershocks to get independant station corrections for each gridpoint.
Fun with statistics
Backprojection is a pretty good quick and dirty estimate of a large earthquake's source time function.
This is for Sumatra using Hi-net data.
More STFs from the recent Kuril earthquakes
Note that while the October event is only M6.6, it appears to go on for almost as long as the January M8.1 event.
Is this true? No, but this a good illustration of how array geometry can dominate certain aspects of backprojection imaging.
Fun with color scales
Imaging rupture direction
Snapshots from backprojection movies. The top row is using European stations. The bottom uses US Array and other NAmerican stations.
How does backprojection work as a small earthquake and aftershock detector?
These Kuril earthquakes were detected using data from US Array. A M3.9 was clearly detected from >45 degrees away.
This is smaller than the 2006 North Korean nuclear test.
