The Fireball and the Explosion
It occurred to me today that PNSN seismologists are, in a way, like nervous homeowners awakened in the middle of the night by a strange sound. One lays there trying to identify the cause and whether it needs immediate attention. Was it a burglar trying to open a window? Or a creak from the house settling, or perhaps just a passing car? You go over all the characteristics of the sound and throw out the hypothetical sources that don’t make sense. And then with any luck go back to sleep with peace of mind.
To illustrate this, in a recent PNSN blog post Steve Malone recounted the story of mysterious explosions on Orcas Island. Steve briefly discussed a high-altitude fireball that was observed in the area near the time of the explosion, noting that we discounted it as the source of the explosion. However, there’s a bit more to the story, and we’ll tell it here because it illustrates the processes and complexity of creative forensic seismology, and it showcases new regional monitoring capabilities (particularly at volcanoes) and our long-standing cooperation between volcano and earthquake monitoring in the region, and … well … because it is just too cool for school demonstrating Earth’s dynamic environment!
Background: over the past couple of years, our partners at the Cascades Volcano Observatory have been adding monitoring stations at the region’s dangerous volcanoes. Pertinent to this blog, CVO have added combined seismic and infrasound-sensing stations around Mount Rainier to help detect lahars (volcanic mudflows) and warn downstream communities of impending hazard. Lahars, by the way, may accompany magmatic eruptions on our glacier-clad volcanoes, but also may take place when a volcano is not erupting. The infrasound sensors are essentially microphones – they measure the atmospheric pressure; so, they’re good at observing explosions, fireballs, noisy lahars. Basically, anything that moves air in Earth’s atmosphere. When there is an array of multiple infrasound stations they can be used, by observing the pressure wave passing over the array, like an antenna to determine the direction from which a disturbance is arriving. In the parlance of array seismology (and atmospheric acoustics) this is called the “back azimuth”. In brief, the useful data provided by an infrasound array include the back azimuth, the time-of-arrival, the speed of propagation across the array, and the amplitude of the wave. The distance to the source of the disturbance and its characteristics (and hence its nature) must usually, unless known independently, be determined by that subtle interaction of deductive and inductive logic that is scientific reasoning.
Return to Orcas: Ok, so we initially determined that the signals we were looking at on our San Juan Islands seismic stations were basically atmospheric acoustic (i.e., pressure) waves that “coupled” into the solid Earth and shook the ground enough that they registered in our seismic data. We turned to our partner and colleague at CVO, Wes Thelen, to find out whether the infrasound arrays at Mt. Rainier detected any signals that could shed light on the Orcas explosion. Of course, at the time of our request to Wes we weren’t confident about the source’s location and size.
Wes responded that indeed there WAS an unusual signal detected at the Rainier array at about the time of the Orcas explosion, that it was large, low in frequency and long in duration (~ 1 minute), and it arrived from the North (Figure 1, supplied by W. Thelen). Subsequently, the video footage from social media of the fireball showed up online. Was THAT, perhaps, the source of the Orcas seismic data? The problem was, and is, the timing of the arrivals. The Rainier waves arrived several minutes later than one would expect from an Orcas Island source at the time indicated we saw on our closest Orcas seismic stations. Moreover, we soon determined that the Orcas signals were, as far as such explosion data go, quite small. It would be jaw-dropping for them to have propagated the >200 km distance to Mt. Rainier and appear so robust and large. Those of us with some experience interpreting infrasound signals rejected the Orcas explosion as the source of the Rainier signals. That said, still fresh in the back of our minds was the infrasound observations of HUGE atmospheric acoustic waves from the Hunga Tonga-Hunga Ha’apai submarine volcanic explosion in mid-January.
Figure 1: Data from the PR04 Infrasound Array near Mt. Rainier. Top panel: time series of atmospheric pressure (mean removed). Panel 2: mean cross-correlation maxima (MCCM) metric, a measure of signal correlation (similarity); times of MCCM above dashed line are determined to by “detections”. Panel 3: apparent velocity of acoustic wave propagating across array (at sea level atmospheric waves propagate at about 0.33 km/s). Bottom panel: Back azimuth (arrival direction). Note that the arrivals with MCCM > 0.6 (i.e., detections) are in warm colors.
The Fireball Problem: Now here’s where the fireball and the network data come back into the story. The photographic evidence suggested that it took place following the Orcas explosion (by 15 minutes or so). And that it didn’t appear on the San Juan seismic station data—at least not as a recognizable signal. And meanwhile a slew of further observations and citizen-science reports on social media made clear that the Orcas explosions were indeed explosions, somewhat small as such things go, and not high-atmospheric sources. So, Steve published his blog post.
But afterward, dissatisfied with this coincidence, we dug back into our seismic data looking for any evidence of a second source. Sure enough, Steve pulled out a set of signals from a wide area across the PNSN that clearly correlated to form a picture of a slowly propagating atmospheric pressure wave that fit with the Mt. Rainier infrasound data (Figure 2). A drop-in visitor from space like a bolide or a chunk of space junk can be really challenging to locate, however. One problem is that they’re moving as they stir up the atmosphere, so they don’t necessarily have a single “location”. Another problem is that there may be multiple explosions yielding loud high-frequency “pops” as they travel and break up. Getting a precise chronology of the fireball source takes a lot of work, skill, intuition, and (no doubt) luck. While we haven’t done that for the March 7 example shown in Figure 2, we did a rather crude estimate assuming a single source, which we estimate at a height somewhere between 5-15 km and near the town of Carnation (47˚56.95’ 122˚2.89’, Figure 3).
Figure 2. Seismograms from selected seismic stations in western Washington state. Top-to-bottom is arranged by distance to estimated source location covering an overall distance of about 250 km. The shaking related to the pressure wave from the fireball is seen as a clear impulsive arrival. Total time covered by the figure is about 6 minutes. The seismic data (not infrasound) from the PR04 site is the bottom trace in cyan background, and the very top trace is a “blow up” (pun intended) of the arrival at PR04 that’s highlighted in yellow background.
As to the relationship between the Orcas explosion and the Cascades fireball we are forced to favor a hypothesis that seismologists are trained to loathe and fear: two entirely independent and coincidental sources (near in time, and perhaps relatively close in location…). The ground-level explosion and resulting acoustic bang at about 04:12:30 local time was human-made and rattled windows and structures. The explosion woke up retired seismologist Tom Owens, who initially helped us with this study, and disturbed others. Being close to the source, the signals were comprised mostly by high frequencies, well-tuned to show up on our seismic stations built to look at those frequencies. Then a few minutes afterwards (at approximately 04:33 local time) some asteroid or piece of space junk streaked across the sky nearby. The fireball, being a moving source probably contained a mixture of frequencies, extended in duration, and high in Earth’s atmosphere. Such waves would probably propagate more efficiently the long distance to Mt. Rainier and show up well on the infrasound stations built for such a purpose. Further modeling and analysis would no doubt enable us to narrow the location down. But it’s not really our job!
Figure 3: Map showing our “best” estimate of the source of the March 7 fireball.
What is part of our job is to lay awake all night listening to the sounds in our neighborhood for threats. That is, to keep an eye on the data from whatever shakes our stations (or sometimes what DOESN’T shake our stations) and to make sure we understand as completely and accurately as possible what they’re trying to tell us.