Signature
Signature by Paula Levine, earthquake interpretation software
Signature: technical details of an audio/visual simulation re-imaging the 1906 earthquake in California with an epicenter lying on the Roger’s Creek fault, Sonoma County
The simulation is based on data from two sets of seismographic measurements, taken in Sweden and Puerto Rico, respectively, during the 1906 earthquake. These were supplied by Paul Spudich of the USGS, who describes the method of digitizing the data as follows:
—–snip—–
from Paul Spudich, February 1 2006: “I received some 1906 seismograms from Seok-Goo Song, a graduate student at Stanford in the Department of Geophysics….
As you know, old seismograms are wiggly lines made by a moving pen on paper. The four files each contain a string of numbers. The numbers are the displacement of the wiggly line away from its undisturbed position, measured every second. Thus, a string like this:
1.234e-2
2.345e-1
4.321e+1
means that at the time of the first sample (let’s assume it was 6:00:00 a.m. on 4/18/06 the pen was displaced 0.001234 mm from its normal undisturbed position. One second later, at 6:00:01 the pen was displaced 0.02345 mm; at 06:00:02 the pen was displaced 43.21 mm. Etc.
The pen displacement is proportional to ground motion….
The seimsograms come from Uppsala, Sweden, and somewhere in Puerto Rico. There are two files from each place. One file is motion in the north-south direction, one is east-west.”
—-snip—–
The data sets are loaded into a buffer containing a number of discreet samples equivalent to the number of data in each set. Since the data contains no specific frequency information, frequency of the individual samples is approximated by performing a fast fourier transformation ( http://en.wikipedia.org/wiki/ Discrete_Fourier_transform ), in which small groups of sample data are analyzed to find a median frequency, which is multiplied by the given seismic amplitudes in order to bring the resulting sounds into an audible range.
The specific formula for generating the sounds of the quake follows a schema similar to the synthesis model for transmitting seismographic data as audio over an FM radio signal as suggested by Bob Smith of the Yellowstone Seismic Network and USGS at http://www.mines.utah.edu/~rbsmith/RESEARCH/ YellowstoneEarthquakes.html
Dr. Smith’s system follows this model:
“Eight FM center frequencies ranging from 680 to 3060 Hz are in use with a 340 Hz separation between center frequencies and an individual fixed bandwidth of 250 Hz.”
The video simulation is created through the modified use of a fluid dynamics visualization programmed by Joshua Clayton, Jeremy Bernstein, et al, in the Max/MSP/Jitter environment. The audio files created from the original seismic signatures are played back according to a pattern generated through the recording of GPS satellite data from outside the Sonoma County Museum. As the ’sounds’ of the earthquake are played back, they are analyzed in realtime for peak amplitude. This value is passed to the fluid dynamics simulation, where it is used to create displacements in the simulated fluid surface. As the amplitude of the signature rises, so does the amount of displacement which occurs in the video simulation.
This displacement is used as a means to mix together two video signals. The top image is a silhouette of the Santa Rosa city limits. The image below is a map displaying the Roger’s Creek fault and other seismic hazards within the Santa Rosa city limits. The displacement generated by the audio amplitude causes the sheering of the background ‘fault’ image, and a fading and feedback between the two images along the boundaries of the generated displacements, also proportional to the amplitude of the recorded audio signatures. The simulation runs in realtime, potentially in perpetuity, and repeats every 10000 iterations (10000 being the number of unique satellite signals recorded).