Event Report, infrasound

Unknown Infrasonic Event – XIE140718

Unknown Infrasonic Event – XIE140718
Prepared by: ISLA
Distribution Date: 18 July 2014
Source: Unknown
Location: Keauhou, HI
Origin Time: ~4pm HST 07/17/2014 or ~2:00 UTC 07/18/2014
Description: A very large “boom” was reported in the Keauhou area of Kona, along with buildings shaking.
IS Array: I59US, KHLU, InfraSound LAborotory (ISLA) test rig “AACE”
Data Quality: Good, test rig at ISLA was undergoing high frequency noise test

Summary: A large booming noise was reported in Kona near the Keauhou and Kahalu’u area on 07/17/2014 at around 4pm local time. Witnesses say that buildings shook and car alarms went off. The boom was followed by the sound of jets (reported by several witnesses). This signal was clearly recorded at I59US and set off an automated large amplitude alert at ISLA.

Report

I59US is part of the Comprehensive Nuclear Test Ban Treaty (CTBT) International Monitoring System (IMS), and the data is open and available through the Incorporated Research Institutions for Seismology (IRIS). The station is located in the ahupua’a (Hawaiian traditional land division) of Kahalu’u on the Kona (leeward) coast of the Big Island of Hawaii. The site is located at an elevation of 3000’ in the Kahalu’u forest reserve. The site consists of four stations with Chaparral 5 infrasound sensors, housed in a vault with wind noise reduction systems. The array has an aperture of 2km.

KHLU is part of the University of Hawaii Infrasound Network, and is co-located with the I59US array. This site uses Chaparral 2.2 infrasound sensors, and is designed for higher frequency signals. The KHLU seismic station is part of the Pacific Tsunami Warning Center (PTWC) network, and is also co-located with I59US.

Data from I59US is sent from the site to ISLA on Keahole Point, where it is processed by an automated system and archived. Raw waveforms are displayed at near real time in the lab. The signal of interest (SOI) was observed on this display on sties I59US and the co-located UH network site KHLU, and the PTWC seismic site KHLU.

In addition to these sites, sensor noise tests were being conducted at ISLA during the time of the SOI. Six chaparral 2.2 sensors were hooked up to a common manifold to examine relative noise levels between the sensors. After going back and looking at the data, this system also recorded the SOI. However, these results are skewed to the low end of frequencies due to the manifold being sealed at the time of the signal.

1 I59US
1.1 Raw waveform data

The signal was recorded with good signal to noise on all the stations of I59US (figure 1). The signal is impulsive with one dominant peak and a much smaller secondary signal as seen in a more detailed view of the signal (figure 2). Due to the different sensors and configurations in the data set for this signal, the detailed waveforms are especially interesting.

ISLA_2014718_1islaRawWfPlotI59US
Figure 1: Figure showing the same 2 hour window as used in the I59US array processing analysis. The SOI has very high signal to noise and is impulsive in character.
I59US_wfdetail
Figure 2: Detailed view of the signal (amplitudes normalized for each station) for I59US. Note the shape of the signal for comparison to later detailed signals.

1.2 Spectrograms

Due to the high frequency nature of the SOI, a linear spectrogram was calculated (figure 3). This analysis further illustrates the high signal to noise, along with the large amplitude of the SOI. This analysis also revealed the broadband characteristic of the signal.

Figure 3: Linear spectrogram of the signal of interest recorded at I59US. The frequency range is 0.1 to 10 Hz. With this range the roll off of the sensor as it approaches nyquest can easily been seen above 8Hz.
Figure 3: Linear spectrogram of the signal of interest recorded at I59US. The frequency range is 0.1 to 10 Hz. With this range the roll off of the sensor as it approaches nyquest can easily been seen above 8Hz.

1.3 Array Processing

After the spectral analysis, the data were analyzed using array processing. This analysis is useful in determining not only the consistency of the signal across the array, but the signal approach direction from the array (back azimuth). The first plot (figure 4) depicts the results from the automated PMCC (Progressive Multichannel Cross Correlation) v 4.0 program running at ISLA. The results are also plotted with a log spectrogram calculated using the INFERNO algorithm.

The data were then re-processed using a manual rerun with PMCC 4.15 and those results can be seen in figures 5 and 6. Both analyses indicate a signal coming from the west (~260° back azimuth). The results have a tight time grouping and are consistent across the frequency range.

Figure 4: I59US results from the automated run of PMCC 4.0. The array processing results have been masked in correlation showing only results with a correlation of 0.7 and greater. The logarithmic spectrogram runs from 0.1 Hz to 8Hz.
Figure 4: I59US results from the automated run of PMCC 4.0. The array processing results have been masked in correlation showing only results with a correlation of 0.7 and greater. The logarithmic spectrogram runs from 0.1 Hz to 8Hz.
Figure 5: Manual re-run of I59US PMCC 4.15 results. Note again the tight grouping of detections.
Figure 5: Manual re-run of I59US PMCC 4.15 results. Note again the tight grouping of detections.
Figure 6: I59US radar plot created from the manual rerun showing the back azimuth of the signal.
Figure 6: I59US radar plot created from the manual rerun showing the back azimuth of the signal.

2 KHLU

KHLU is co-located with I59US and records at a higher sample rate. This is in part due to the fact that the site was designed to investigate higher frequency signals of interest than I59US. Due to this face, ISLA uses both sites when characterizing a signal.

The signal was also recorded with very high signal to noise at KHLU. Again, the signal is very impulsive (figure 7). The detailed waveform view (figure 8) also includes the vertical channel from the broadband seismic station located with I59US. In this channel, along with all the sites of KHLU; the signal of interest has a high frequency component that was not resolved in the I59US data.

Figure 7: The signal of interest as recorded at the UH infrasound station KHLU. Note KHLU is designed to pick up higher frequencies and sampled at a higher rate.
Figure 7: The signal of interest as recorded at the UH infrasound station KHLU. Note KHLU is designed to pick up higher frequencies and sampled at a higher rate.
Figure 8: Detailed view of the KHLU data. The top signal is a vertical channel from the KHLU seismic station run by PTWC. The signal coupling to the ground is consistent with witness reports (appendix 1) who reported buildings shaking and car alarms going off.
Figure 8: Detailed view of the KHLU data. The top signal is a vertical channel from the KHLU seismic station run by PTWC. The signal coupling to the ground is consistent with witness reports (appendix 1) who reported buildings shaking and car alarms going off.

2.2 Spectrograms

After the raw waveforms were analyzed, a linear spectrogram was also calculated. This spectrogram has a larger frequency range due to the higher sample rate at KHLU (figure 9).

Figure 9: Linear spectrogram for KHLU. Note the roll off for this sensor is at 18 Hz, and therefore more detail on the higher end can be seen.
Figure 9: Linear spectrogram for KHLU. Note the roll off for this sensor is at 18 Hz, and therefore more detail on the higher end can be seen.

2.3 Array Processing

Array processing for KHLU is also performed in the automated system. Like the I59US data, the KHLU array processing results are plotted with a logarithmic spectrogram calculated with the INFERNO algorithm (figure 10). Again the data were manual rerun with PMCC 4.15 (figure 11-12).

Figure 10: This figure is generated using the same settings as the I59US figure. KHLU has a higher number of detections due to its smaller aperture size and higher sample rate. The signal of interest PMCC results are not as obvious in this automated run.
Figure 10: This figure is generated using the same settings as the I59US figure. KHLU has a higher number of detections due to its smaller aperture size and higher sample rate. The signal of interest PMCC results are not as obvious in this automated run.
Figure 11: Results from the manual rerun of KHLU. Again note the increase in high frequency detections.
Figure 11: Results from the manual rerun of KHLU. Again note the increase in high frequency detections.
Figure 12: Radar plot of results from the manual rerun. While the grouping is not as tight the resulting azimuth is the same,
Figure 12: Radar plot of results from the manual rerun. While the grouping is not as tight the resulting azimuth is the same,

3 Test Rig “AACE”

3.1 Raw waveform data

At the time of the signal, ISLA was performing a test on several sensors. These sensors are in a huddle, connected to a manifold, and recording to the same digitizer. The manifold is sealed in order to examine the high frequency noise floor on the sensors. Due to this configuration, most high frequencies were filtered out. Despite the lab staff not observing or hearing the signal, the test data did recorded the signal (figure 13).

Figure 13: There are three signals in this plot. The first and last signals are when the front door of the lab was opened by an ISLA staff member taking a break. The signal between 2:00 and 2:02 is the signal of interest as recorded on the test rig.
Figure 13: There are three signals in this plot. The first and last signals are when the front door of the lab was opened by an ISLA staff member taking a break. The signal between 2:00 and 2:02 is the signal of interest as recorded on the test rig.

4 Interpretations

Through array processing, a back azimuth was drawn from the site locations (figure 14). Witness reports from the area can be seen in appendix A1. The results from both I59US and KHLU result in a back azimuth of ~ 260° which tracks over Keauhou and Kahalu’u. This is consistent with observations from Kahalu’u beach park area in which the noise was reported “right over head”.

The source of the signal is unknown, but due to several reports of hearing “jets” immediately after, and the signal characteristics, it is possible that this signal was due to an aircraft breaking the sound barrier. This hypothesis would need further testing, but is further supported by the larger than average military presence in the area due to exercise taking place offshore.

Figure 14: The calculated back azimuth of the signal plotted in GoogleEarth. Note that the calculated line passes directly over Kahalu’u.
Figure 14: The calculated back azimuth of the signal plotted in GoogleEarth. Note that the calculated line passes directly over Kahalu’u.

Appendix (A-1): Witness reports and News Articles

Witness reports:

1.”A big bang that shook the ground” – Kona
2.”HUGE BANG that scared all the animals followed by jet noise. It sounded like it was right overhead” – Donkey Mill Area
3.”It was super loud. Right over Kahalu’u. I think it was a jet. It was so loud. I talked to people at KTA and they said the whole building shook” – Kahalu’u beach park area
4.“Our whole house shook, a couple of the kiddos woke up and were scared! My mom said it sounded like an explosion and the neighbor was sure that her water heater had exploded!” – Holualoa
5.“Just about 4pm at Keauhou Shopping center. Really loud…set off car alarms!” – Keauhou

News:

World’s largest maritime exercises happening off Kawaihae – West Hawaii Today