Wednesday, June 20, 2012

From the sensor to the concert hall

I am very excited to share the following with you! At 9:30 PM on the night of February 3, 2012 two rockfalls occurred into the Halema'uma'u Vent at Kilauea Volcano. These events were reported by HVO and recorded at our station (MENE) in Volcano, HI. This event has now been given new life as the inspiration for a percussion sextet. 

I am very happy to share with you "Volcano Music Mvt. II" by Jason Thompson for percussion sextet. Please go to his site and listen to the piece and read his program notes. 
UPDATE: unfortunately this piece is no longer available online. 

The following information and sound files are what was sent to Jason Thompson, and were used to create this wonderful artistic interpretation of a real event.



We generally try to clean our tracks up a bit, and the post processing procedure used here can be found in the following report. 

(This report is also available in pdf format on Jason Thompson's website and upon request. Just shoot me an email!)

Volcanic Infrasound Event VIE120203 Halemaumau Crater, Kilauea, Hawaii

Source: Halemaumau crater, Kilauea Volcano, Hawaii
Location: Halemaumau Crater
Origin time:  03 February 2012 09:30:00 PM HST, 03 February 2012 07:30:00 GMT. All times in narrative are GMT. 
Description: Two rock falls into the Halemaumau Crater were reported by the Hawaiian Volcano Observatory (Appendix A) on the night of 02 February 2012 (HST). Associated infrasonic signals were recorded at the local network.
Data Quality: Collapses occurred in the quiet of the night, with very little wind and surf noise.
Method Summary:  The initial signal characterization used automated analysis tools implemented at ISLA. These include array processing and spectral plots. Further characterization was completed with the aid of logarithmic scale spectrograms.

1. Instrumentation
The MENE infrasound array has a 24-bit Reftek 130 digitizer recording four Chaparral 2.2s and one Chaparral 50 infrasonic microphones at 40 samples/second. The Chaparral 2.2s are located at elements 1-4 and have a flat response down to 0.1Hz, and the Chaparral 50 is collated with MENE1 at element 5 and has a flat response down to 0.02Hz. Array processing returns the direction of arrival of infrasonic signals as a function of frequency, which allows a clear azimuthal and spectral separation of source regions.

2. Chronology
The sequence of events for the Signals of Interest (SOI1 and SOI2) can be seen in Figure 1, and are described below. Figure 2 shows the signals recorded by the array spanning the times for both signals of interest.  Figures 3 and 4 are logarithmic spectrograms, and figures 6 and 7 show a four minute window around the signals of interest.

1)  Between 1800 and 2200 GMT of 2 Feb 2012 we have a partition of energy between the Pu’u O’o Crater complex and the Halemaumau vent, with a clear frequency and azimuth separation.
2)  From around 2100 of 2 Feb 2012 to around 0315 of 3 Feb 2012 we see the emergence of several harmonics above 1 Hz. Array detections are predominantly from Halemaumau.
3)  From 0315 GMT to ~0615 there is a drop in the overtones, ~1.5 Hz signal stays, there is temporary drop in the total number of Halemauamau detections, then the ~1.5 Hz signal strengthens and detections pick back up
4)  After ~0615 GMT even the ~1.5Hz signal shuts off, and there is a transition to deeper frequencies that lasts around 10 minutes. The soundscape becomes ominously quiet above 1Hz, while there is an increase in observed detections from the Pu’u O’o azimuth.
5)  First collapse signal is at 07:29:20, and the second at 10:06:15. Both of these signals are broadband and extend down to .02Hz (Figure 3), there is a change in harmonic structure, with harmonics present below the ~1.5 Hz band and above the microbarom. Pu’u O’o detections continue and are joined by Halemaumau.
6)  There is a small event at 1500 which may be seismic or volcanic in origin.
7)  After 1600 to the end of the study period, Halemaumau quiets down while Pu’u O’o resumes radiation, but drops to a lower frequency.

3. Sound Editing
I select the 24h time period of 3 February, with Halemaumau as the initial prevailing volcanic source. We preserve all features in the original data by assigning it different sample rate of 8 kHz and 44kHz, essentially speeding up the raw data by a factor of 200 and 1102.5 times the original time. We use channel MENE5 because of its better performance down to 0.02 Hz (50 second periods). The resulting waveforms are shown in Figure 7.

A time compression of 200x maps 0.02 Hz to 4 Hz, which is below the response of most commercial speakers. It will map the 0.4 - 5 Hz volcanic bandpass typical of Kilauea to 80 - 1000 Hz. For this event, there are very interesting features beyond this bandpass that we would like to preserve.

A time compression of 1102.5 maps 0.02 Hz into 22 Hz, which good subwoofers can reproduce. It also maps 5 Hz to 5.512 kHz, well within the range of most speakers.

Master 8 kHz wave file, MENE5BDF2012020300_8khz.

3.1. Remove data gaps, clean attack and fade, save as esound_vie120203_8khz.wav.

3.2. ID neat signal at ~4:30 from origin, corresponding to small event at 1500 GMT. Sounds like a pop and a woosh.

3.3. Lowpass below 1500 Hz up to dusk (0-1:52) and leading up to the first small burst, and after 3:40 when helicopters start up again. Save as esound_vie120203_8khz_lp.wav.

3.4. Highpass above 80 Hz, which removes the microbarom band. Rescale by -12dB. Save as esound_vie120203_8khz_bp_sc.wav
Keeps only the volcanic signal and removed ocean and wind noise in the low register and aircraft noise in the high registers.

3.5. Now that I’ve deconstructed the sound, pick a dull noise segment and deconvolve it from the original esound_vie120203_8khz.wav. Rescale, clean up, save as esound_vie120203_8khz_decon.wav. Remove aircraft by lowpass filtering below 1500 Hz, and save as esound_vie120203_8khz_decon_lp.wav

Figure 1. Results from the automatic processing. Top panel is a linear-frequency spectrogram of MENE1, and lower panel is the array processing detections using PMCC3. Events 1-4 are illustrated with colored boxes. (1) Green: detections from Halemaumau, harmonics above 1Hz. (2) Red: decrease in harmonics above 1Hz besides the ~1.5Hz band and initial decrease in Halemaumau detections with a pickup of Pu’u O’o detections. (3) Yellow: drop in ~1.5Hz band and VLP/ULP event  over a 10minute span, Pu’u O’o detections and break in Halemaumau detections. (4) Blue: SOI-1 with associated coda, and harmonics, SOI-2 and stronger coda in both low and high frequency, detections from Pu’u O’o and Halemaumau.
Figure 2. Time series around the two signals of interest. Note the difference between the Chaparral 2.2s (1-4) and the Chaparral 50 (5). Next figures use only MENE1 and MENE5.
 Figure 3. Log-frequency spectrogram from MENE5 over .02Hz to 10hz. Note that SOI-1 and SOI-2 extend down to the lower bound at .02Hz. Also note the changes in harmonic structures over the time period.
Figure 4. Log-frequency spectrogram from MENE1 over .1 to 10Hz. Also note the change of harmonic structures.
Figure 5. Four minutes surrounding the first signal of interest with initial downward deflection and longer low-frequency coda.
Figure 6.  Four minutes surrounding the second signal of interest with a compressional onset and a higher frequency coda.
Figure 7. Original waveform and time-compressed .wav files.

Appendix A
Friday, February 3, 2012 7:02 AM HST (Friday, February 3, 2012 17:02 UTC)

19°25'16" N 155°17'13" W, Summit Elevation 4091 ft (1247 m)
Current Volcano Alert Level: WATCH
Current Aviation Color Code: ORANGE

Past 24 hours at Kilauea summit: The summit tilt network continued to record DI deflation punctuated with two abrupt positive offsets due to large rockfalls from the vent rim (Halema`uma`u Crater floor) into the lake at 9:30 pm and midnight; the first collapse involved a portion of the north rim while the second took a long sliver of the northeast rim; the first collapse apparently induced secondary collapses of the inner ledge and ejected hot spatter onto the nearby portions of the Halema`uma`u Crater floor; the second collapse deposited a large amount of debris into the northeast side of the lava lake. Both collapses severely disrupted the lava lake with the second significantly dropping the level which was slowly recovering lost elevation this morning. The most recent (preliminary) sulfur dioxide emission rate measurement was 600 tonnes/day on January 30, 2012.

Seismic tremor levels dropped when a small spattering source appeared on the north rim of the lake at 8:20 pm last night and remained low with the two large rockfall seismic signals superimposed. Ten earthquakes were strong enough to be located beneath Kilauea volcano: one north of and one beneath the summit caldera, one within the upper east rift zone, and seven on south flank faults.

Background: The summit lava lake is deep within a ~150 m (500 ft) diameter cylindrical vent with nearly vertical sides inset within the east wall and floor of Halema`uma`u Crater. Its level fluctuates from about 70 m to more than 150 m (out of sight) below the floor of Halema`uma`u Crater. The vent has been mostly active since opening with a small explosive event on March 19, 2008. Most recently, the lava level of the lake has remained below an inner ledge (75 m or 250 ft below the floor of Halema`uma`u Crater) and responded to summit tilt changes with the lake receding during deflation and rising during inflation.

No comments:

Post a Comment