Halemaˋumaˋu Rockfall May 3, 2015 | Infrasound Laboratory University of Hawaii

Volcano Infrasound Event – VIE150503

Prepared by: ISLA
Source: Rock fall into Halemaˋumaˋu lava lake
Location: 19.404365° N, 155.280515° W
Origin Time: Reported time ~23:21 05/03/2015 UTC
IS Arrays: AIND, AHUD, and MENE
REDVOX: iPhone6 with iTestMic, unit 1000000007.
Data Quality: Good at MENE, some wind noise at AHUD and AIND.

Part of the crater rim wall just below the Halemaˋumaˋu overlook failed and fell into the lava lake around 1:21 PM HT of 3 May 2015. The failure triggered a brief explosive event which was recorded on the HVO/UH infrasound network, as well as an iPhone6 with the REDVOX Infrasound Recorder. The explosion was also captured by several HVO webcams.

multimediaFile-1181 — Figure 1: In the top left image the small dust plume from the rockfall can be seen, followed by the small explosive event as recorded by the overlook webcam (HVO). Image provided by HVO.

A rockfall and associated explosive event occurred on 3 May 2015 at ~23:21 UTC in the Halemaˋumaˋu crater. This event was reported by HVO (appendix A) and was well recorded by webcams and infrasound stations around the crater (Figure 1). According to the HVO Daily Update on May 4, a portion of the wall of Halemaˋumaˋu crater collapsed and impacted the lava lake (which was at or near the rim of the overlook crater at the time). This rockfall then triggered a small explosive event.

The current Hawaii Island infrasound network consists of five arrays, two outside of Kailua-Kona, and three around Kilauea Volcano. MENE (UH) is located in the village of Volcano and is has a range of 7 km and heading of 250 degrees to Halemaˋumaˋu crater. AHUD (HVO) is located within Hawaii Volcanoes National park and has a range of 4 km and a heading of 335 degrees to Halemaˋumaˋu crater. AIND (HVO) is located on the flank of Mauna Loa with a range of 19 km and a heading of 78 degrees to Halemaˋumaˋu crater. All three infrasound arrays detected the event. In addition, a specially-configured iPhone6 at HVO also captured the infrasound signature at a range of 2 km and heading of 156 degrees.

islaPmccPlotMENE — Figure 2: Automated processing for the array MENE. The top panel is raw waveform data from the first element of the MENE array. In the center is a logarithmic spectrogram using the INFERNO framework for the element plotted above. In the lower panel is the array processing results. These results are colored by mean frequency of the detection set, and plotted as time vs. back azimuth of the signal. The size of the circles is scaled to the intensity of the detection.

The University of Hawaii Infrasound Lab (ISLA) runs several automated processes for real-time analyses of infrasound data streams. Figure 2 is one of the automatic spectral and array processing products. The raw infrasound data were reanalyzed to prepare this report.

ISLA uses Progressive Multi-Channel Correlation (PMCC) for array processing. By using the expected signal heading and estimated arrival time, the detections from this event can be isolated from other coherent detections at the array (Figures 3-5).

AHUD_PMCCresults_freqvstime — Figure 3: Filtered PMCC (array processing) results for the array AHUD, located south of the summit crater. The array processing results were filtered in time and back-azimuth. The results are colored by the back-azimuth, the size is the intensity of the detection and the results are plotted in time vs. frequency. This is a broad band signal that was detected between 0.6 to 5 Hz at the AHUD array.

AIND_PMCCresults_freqvstime — Figure 4: Filtered PMCC (array processing) results for the array AIND, located west of the summit crater. The array processing results were filtered in time and back-azimuth. The results are colored by the back-azimuth, the size is the intensity of the detection and the results are plotted in time vs. frequency. This is a broad band signal that was detected between 0.6 to 2 Hz at the AIND array.

MENE_PMCCresults_freqvstime — Figure 5: Filtered PMCC (array processing) results for the array MENE, located in Volcano Village. The array processing results were filtered in time and back-azimuth. The results are colored by the back-azimuth, the size is the intensity of the detection and the results are plotted in time vs. frequency. This is a broad band signal that was detected between 0.1 to 4 Hz at the MENE array.

ISLA uses the array processing results at each site to verify the source location. This method makes no assumptions about the time, location, or range of the signal, and only considers intersecting heading projections and the sound speed. The result of that association is shown in Figure 6.

AssociationResults — Figure 6: Event association and localization using the PMCC (array processing) results plotted in figures 3-5. This ana lysis uses intersecting backazimuths and then normalizes to the best solution.

The signal-to-noise ratios were substantially improved by beamforming along the heading angle and the application of the Infrasonic Energy, Nth Octave (INFERNO) algorithm (Figures 7-9).

islaIgnite_BEAM_AHUD — Figure 7: Signal-to-noise (S2N) INFERNO plot for AHUD beamformed data. The beam was generated using the validated back azimuth from the array to the source.

islaIgnite_BEAM_AIND — Figure 8: Signal-to-noise (S2N) INFERNO plot for AIND beamformed data. The beam was generated using the validated back azimuth from the array to the source.

islaIgnite_BEAM_MENE — Figure 9: Signal-to-noise (S2N) INFERNO plot for MENE beamformed data. The beam was generated using the validated back azimuth from the array to the source. Notice the signal has a higher signal to noise ratio than the HVO sites. MENE is located in a heavily vegated area that helps reduce the wind noise at the site

Spectral analysis was also performed on data collected by an iPhone6 running the REDVOX Recorder App designed to turn Apple iOS into infrasound recording systems. During the time of this event there was a single iPhone 6 recording at the HVO overlook. The data were recorded on the internal hard drive, and were recovered on 6 May. Spectral and INFERNO signal-to-noise analysis were performed on these data (Figures 10 and 11), as well as the barometer data recorded by the iPhone6 (Figure 12).

The iPhone6 was replaced with unit 1000000008 on 6 May 2015, and is streaming live infrasound data via wifi, with cell as a communication backup system. Prototype ubiquitous system data products can be viewed at ubq2s.isla.hawaii.edu.

1000000007 — Figure 10: Raw waveform data on the top and traditional spectral (FFT) analysis on the bottom panel. The signal of interest is visible at this sensor below 16 Hz. These data were recorded by an iPhone 6 located at the Jagger Museum overlook.

1000000007ignite — Figure 11: INFERNO analysis of the raw waveform data collected by the iPhone6. The signal timing, duration, and frequency band matches well the signatures recorded by traditional arrays. This is the first validated infrasound signature of a volcanic eruption captured by an iPhone.

islaIgnite_1000000007bar — Figure 12: INFERNO analysis of the bar barometer data collected by the iPhone 6.

-M. Garces, A. Perttu, & B. Williams

Acknowledgements

Mahalos to F. Trusdell for hosting MENE, to W. Thelen for his encouragement and constructive collaborations, and to L. Lee for comms support.