The goal? To test the sensitivity of the WAVE fiber-optic network, a cutting-edge distributed acoustic sensing system designed to detect seismic vibrations with exceptional precision. We wanted to see if the combined jumping of a small group of people could generate vibrations detectable of WAVE in the PETRA ring, located close by, or even beyond?

In the following Figure we show the waterfall plot from the WAVE network, thus time on the vertical and position along the fiber on the vertical axis. The signal (50 second jumping: Thank you to all the hungry people in the line) we generated is clearly visible after applying a low pass filter to the data. The results offer a glimpse into how even small-scale human activity can be measured and analyzed using this advanced technology.

In the following spectrogram you can see that the main frequencies, that we were observing in the data, are around 2Hz. The is the jumping frequency of the people, so the beat of the song.

We’ll be repeating this experiment at the ET German Community Meeting in Hamburg this Friday January 31st between 11:35AM and 12:00PM again at CFEL. Then, also using the correct low pass filter in the livestream to make the signal more clear. But maybe you can find the signal also in the clip of the last Livestream?

A video of the demolition was streamed live by Hamburg Hafen Live on Youtube:

The station lies around 10 km from the DESY/Campus Bahrenfeld, where the WAVE Network is located. Considering both the urban noise of Hamburg and the river Elbe in between, this is quite a significant distance. Nevertheless, we are able to see the towers falling onto the ground in the data of the WAVE Network.

The Waterfall plot above shows how the fibre experiences increased strain which travels in ripples thourgh the campus. In the waterfall plot, the time is one the y-axis, while the section of the fibre in the WAVE Network is the x-axis. Every line denotes 1 m of fibre. Through this, we can see how the signal and in which direction it travels. As the fibre is mirrored, the picture appears twice in the plot.

This plot is filtered, meaning only certain frequency components of the data are taken into account. This is done to lower the influences of surrounding noise. The process is similar to peeling back layers of dirt to reveal a picture underneath, only it is done with data.

Let’s delve deeper into the world of gravitational wave detection and its connection to the WAVE project!

Gravitational Waves

When massive celestial bodies like black holes merge, they spiral around each other, drawing closer and spinning faster. This motion releases gravitational waves—ripples in spacetime itself. These waves stretch and compress spacetime, though by incredibly tiny amounts. For instance, if a gravitational wave passed between Earth and the Sun, the 150-million-kilometer distance would change by less than an angstrom, smaller than an atom!

Despite these tiny distortions, gravitational waves reveal details about the universe’s most energetic events. The first detection, GW150914, in 2015 by Laser Interferometer Gravitational-Wave Observatory LIGO, captured a black hole merger 1.4 billion light-years away, confirming Einstein’s theory and marking a new era in astronomy. Gravitational waves also come from neutron star collisions, like the 2017 GW170817 event, where scientists observed both gravitational waves and light, launching multi-messenger astronomy. Nearly 400 gravitational wave events have been detected so far.

Unlike electromagnetic telescopes that see the universe, gravitational wave detectors let us hear it. But detecting these waves is challenging because they are really really tiny, requiring incredibly sensitive instruments.

Credit: K.-S. Isleif

Gravitational wave detectors, like LIGO, measure tiny distance changes between two suspended mirrors (which we call test masses), 4 km apart, using a laser interferometer. These mirrors are suspended on ultra-thin strings to simulate free-fall and isolate them from ground vibrations, reducing seismic noise interference. The mirrors, acting as test masses for gravitational waves, perfectly reflect laser light, allowing precise measurement of light travel time in two perpendicular directions. When a gravitational wave passes, it stretches one interferometer arm while compressing the other, causing a shift in optical power that’s detected by a photodiode, revealing the gravitational wave signal.

We need to detect very small position changes down to 1e-18 meters, so the test masses must stay almost perfectly still. However, many noise sources, from ground vibrations to local gravitational fluctuations (“Newtonian noise”), can disturb the masses. Even nearby footsteps can create slight gravitational pulls! Reducing this Newtonian noise is key to enhancing gravitational wave detection.

Seismic noise

Seismic waves can result from tectonic shifts, human activity, ocean waves, and many other sources. They can produce noise in the gravitational wave detector by shaking it and moving the test masses. You can learn more about seismic waves here.

Credit: K.-S. Isleif

To suppress seismic noise, we make the suspensions of the mirrors super long, and combine several stages, so we actually have a multistage-pendulum, on which shiny mirrors are hanging. This is what we call passive isolation system. But this is not enough, and, additionally, the suspensions produce resonance frequencies (you know these frequencies when bridges can be destroyed), which needs to be damped, actively. Active noise cancellation involves measuring the noise from the ground or the system, and then either adjusting components in the detector (such as the suspensions) to counteract this noise during operation, or to remove the noise during the data analysis. For this we use, again, laser interferometers! Yes, we have small laser interferometer to improve the big laser interferometer gravitational wave detector.. But we also use seismic instruments like geophones and seismometers.

And this is where WAVE comes in!

Could Distributed Acoustic Sensing (DAS) help? Is a large seismic sensor network like WAVE beneficial for gravitational wave detectors? And can we even address the challenging issue of non-shieldable Newtonian noise, a problem yet to be solved in the future Einstein Telescope? These are the exciting questions that WAVE scientists are exploring!

Newtonian noise

As we aim to improve the sensitivity of gravitational wave detectors, such as the Einstein Telescope (ET), Newtonian noise, caused, for example, by seismic noise, will limit the sensitivity and thus the detection of cool cosmological events which are even heavier or further away than those detected so far. Seismic waves cause tiny fluctuations in ground density, effectively mass changes, which in turn alter the local gravitational force. These fluctuations near a sensitive detector can interfere with its ability to detect gravitational waves, making Newtonian noise a significant challenge. Credit: K.-S. Isleif

Suspensions can’t block Newtonian noise because it’s a gravitational force, and gravity can’t be shielded. To suppress it, we need new strategies. One idea is to measure Newtonian noise and subtract it from the gravitational wave signal. But Newtonian noise is incredibly faint, making it difficult to detect. While some scientists are developing Newtonian noise sensors, these are as complex to build as gravitational wave detectors. In fact, a gravitational wave detector is itself a Newtonian noise sensor!

Instead of measuring Newtonian noise directly, we can track its source — seismic waves!

That’s one reason for building the WAVE network!

The goal is to capture the full seismic wave field—often quite complex—and use this data to predict Newtonian noise through a smart algorithm or machine learning. Since the density of the ground affects how seismic waves create Newtonian noise, precise, site-specific measurements are crucial. This approach needs extensive data to accurately capture the complex seismic wave field and predict Newtonian noise, especially for sensitive detectors like the Einstein Telescope. To gather this data, Distributed Acoustic Sensing (DAS) can be used, providing thousands of sensors to create a detailed seismic picture.

WAVE

Since DAS uses optical fibers, it functions like an array of thousands of seismic sensors spaced every meter along kilometers of fiber. By wrapping the fiber around the vacuum chambers and beam path of the detector, it can precisely map environmental vibration, creating detailed profiles that show noise levels throughout the detector. This detailed mapping enables noise hunting, allowing us to improve noise cancellation not only for seismic and Newtonian noise but maybe even for unexpected noise sources we may not yet fully understand. DAS could reveal unknown noise origins, offering new insights and enhancing the overall sensitivity of the detector.

Credit: K.-S. Isleif

While DAS has not been implemented in a gravitational wave detector to date, there are plans to implement it especially in future generation detectors such as the Einstein Telescope, which is a planned detector in Europe.

Some additional information

Einstein Telescope (ET)

The Einstein Telescope ET will be Europe’s next-generation gravitational wave detector, designed to improve sensitivity by 1 order of magnitude compared to current detectors like LIGO and reach lower frequencies (1-10 Hz), where seismic and Newtonian noise must be suppressed by even one million! Its development will involve a major multinational effort over the next decade, with members of the WAVE team contributing strategies and sensors for seismic and Newtonian noise cancellation.

Credit: K.-S. Isleif

ET will be larger, with 10 km long interferometer arms, increasing sensitivity to detect smaller gravitational waves. It will also feature three overlapping interferometers arranged in an isosceles triangle, with each side containing two arms, thus multiple detectors in one experiment.

Additionally, the Einstein Telescope ET will be built 200-300 meters underground to reduce seismic and ambient noise, crucial for low-frequency detection. However, large seismic sensor arrays, like those achieved through Distributed Acoustic Sensing (DAS) by simply deploying fibers, are being investigated by WAVE. We are exploring how to optimally use this wealth of data to map and cancel both seismic and Newtonian noise in the ET.

More Gravitational Wave Detectors - current and planned

There are multiple gravitational wave detectors which are already in operation, while even more are being planned. The detectors currently in operation are the Laser Interferometer Gravitational-Wave Observatory LIGO in the USA, KAGRA in Japan and VIRGO in Italy. For a future generation of detectors, there are plans to build the Einstein Telescope ET in Europe and Cosmic Explorer CE in the USA. While they all rely on the Michelson interferometer topology, the detectors vary both in size and sensitivity. LIGO has armlengths of about 4 km, while Virgo has 3 km.

In addition to the earthbound detectors, there will be a gravitational wave detector which will operate in space, the Laser Interferometer Space Antenna LISA. This detector will have three satellites and have a different interferometry scheme and will have arm lengths of 2.5 Million kilometers! The satelites will trail behind earth while sending laser beams inbetween them in a giant triangle. This will enable the detection of low frequency gravitational waves. These waves come from older parts of the universe. The planned launch date for the LISA mission is in 2035.

Key Observations

This is a spectrogram of the five hours covering the football match Hamburger SV vs. Jahn Regensburg. The stadium is located at a distance of ca. 1.5 km from the part of the WAVE fiber network we have used for this analysis. The horizontal axis of the plots is time and the vertical axis beats per minute, which can be converted to a frequency in Hertz by dividing by 60.

The spectrogram clearly shows the signature of the two halftimes. Also, before 13:30 h and after 15:30 a larger amount of noise indicates the people walking to and leaving the stadium, which is why this noise is not visible during the halftime break. The game ended 5:0 and also without having watched the game five distinct signals at around 144 bpm (beats per minute) can be seen in the spectrogram. This frequency coincides with the beat of the goal jingle “Always Hamburg” by Scooter, which is played when HSV scores. However, it is likely that the measured signals are not only caused by the music itself, but rather by people jumping to the beat and thereby exciting waves. We have generated spectrograms of the two halftimes and marked the goals and a few other distinct events during the game.

Apart from the two goals scored by HSV there was also a goal scored by the opponent Regensburg that was revoked by the video assistant referee (VAR) two minutes later. Since no goal jingle is played for opponent goals, there is no distinct signal visible. However, it can be seen that the support of the fans dies down for a moment before recommencing a few moments later. After the goal was revoked, very light signals in the range from 100-150 bpm might be caused by the cheering of the fans, but the signals are too weak to clearly confirm this. Two very strong signals during the first half starting in minutes 16 and 37 were caused by the supporters jumping synchronously. We could correlate a third one starting in minute 19 with synchronous clapping and drums.

Also in the second half, the three goals scored by HSV in minutes 76, 89 and 93 can be clearly identified by their distinct signature. In comparison to the first half, the amplitudes seem slightly lower, which may be caused by the fact that the game was already decided and the explosion of joy of the fans is often more prominent when a game is tight. In general, in accordance with the game itself, the activities of the supporters were more calm during the first part of the second half. Two strong signals that we could correlate to jumping by watching the TV broadcast are visible starting in minutes 64 and 77. Also at around 15:30 h there is a very strong signal, which is caused by the supporters jumping synchronously while celebrating the victory together with the HSV players.

The songs can be roughly categorized according to their beat. Most of the songs like Wutboy, Hört ihr die Signale or Remmidemmi show a clear beat throughout the whole song with about 2.2 beats per second. The fans were jumping on this beat generating the specific observed frequencies. The more jumping, the larger the observed amplitudes in the sprectrogram. In addition there are some songs with a slower beat like 99 Bierkanister and Keine Party. Kids in meinem Alter shows the fastest beat during the concert with about 3 beats per second.

Wutboy Spectrogram

Below the spectrogram to the song Wutboy is shown. The signals were particularly strong as indicated by the high amplitudes. The increasing and decreasing amplitudes throughout the song may show the repeated chorus of the song where people were jumping to the beat.

Remmidemmi Correlation to YouTube Video

For Remmidemmi we were even able to correlate the recorded spectrogram to a video of the song available on youtube. The higher amplitudes clearly correlate to the chorus and the extensive jumping of the fans while smaller amplitudes correlate to times inbetween.

Seismic Waves: Hört ihr die Signale

Here you see, the seismic waves caused by the song Hört ihr die Signale (Do You Hear The Signals). After a few seconds you can see how the signal strength clearly increases as the song starts and the waves coming from Trabrennbahn travel westward through our fiber in the PETRA ring and all the way through the European XFEL. Thus – do we hear the signals? Yes, we do!

At 20:58 the HSV supporters could be seen jumping on the TV signal. Also in our fiber optic network, a strong signal was visible at that time.

A spectrogram of the measured vibrations during 5 hours around the match between 19 h and 0 h clearly shows the signature of the match starting at 20:30 h. Even the people arriving at the stadium before the match and leaving it afterwards seem to be visible. However, the strongest signals seem to be caused by the drums of the supporters and synchronized jumping during the match.

A closer look at the first half shows the strong support in the first minutes, after the goal and around minute 27 when jumping supporters were visible in the TV signal. It can also be seen that the signature of the goal scored by HSV looks different than other signals, which might be related to the goal jingle. The end of the first half is clearly visible as the signals die down.

During the first part of the second half, the signals in general seem to have slightly less amplitude. This might be related to the fact that Hertha BSC was getting stronger. The most important scenes of the second half, a free kick by HSV that hit the post and the equalizer by Hertha BSC just afterwards do not seem to have a clear signature. Another strong signal is visible shortly after the final whistle, which is probably the moment in which the HSV players thanked the fans for their support.

For more details on our earthquake tracking and previous findings, check our past analysis of Earthquake signals of the M7.5 Earthquake in Turkey 2023 and M7.4 Earthquake in China 2021 .

Further insights for the M7.1 Earthquake in Japan will follow soon.

This is when we decided to set up a livestream. A livestream was also required for the Science City Day on the Bahrenfeld campus, where we wanted to set up a so-called WAVE field so visitors could explore the DAS technology. Here, we discovered that it is actually not easy to live stream data, as the software provided by the DAS interrogator’s company did not support this at the time. So we started sharing the screen of the software running on the DAS interrogator.

EM soccer match: DAS Software Streaming

Screen sharing of the DAS software was used for our first livestream. It was for the Euro Cup soccer game, the last one that took place in Hamburg’s Volksparkstadion.

The downside of this visualization was that we could not filter the data. Since we expect vibrations mainly in the low-frequency range of about 5 Hz, this representation was unfortunately not well-suited for clearly displaying the vibrations, but it was all we had at the time, as live data filtering is not simple to implement. Nevertheless, listeners found it fascinating, asked many questions on Twitch, and the Reddit post generated a lot of discussion.

Why 5 Hz signals? Because these are the seismic waves that travel long distances through the ground to the campus.

Taylor Swift: Our Realtime Viewer

Erik, our Ph.D. student who deals extensively with data analysis and computing on servers, managed to display the data from the DAS interrogator with relatively low latency (about 5 seconds) in a browser. He got a first version running for the Taylor Swift concert. On the first evening, July 23, the filter didn’t work. But the Swiftquakes, especially during “Shake It Off,” were easily visible in the viewer. Discussions on Twitch were intense, and Katharina and Oliver were busy answering all the questions, even though both were slightly feverish with COVID-19 and working from home.

On the second evening, Erik got the live filtering working, and the signals were even clearer to see.

In this livestream, on the right side, you also see a bubble plot with a map of the Bahrenfeld campus. We determined the positions of most of the fibers (the seismic sensors) through georeferencing, meaning that the individual channels visible in the waterfall diagram can be assigned GPS positions. The fiber laid through the PETRA Ring was displayed in this plot on the right, showing the signals. We divided the signals into different frequency windows and represented them as bubbles of different colors. The larger a bubble, the stronger the signal in that frequency band. For the Taylor Swift concert, this meant that the red bubbles were consistently active and lit up the entire PETRA Ring, especially during “Shake It Off.”

HSV Soccer match August 2024: Next Generation Realtime Viewer

Erik stepped up and designed a real-time viewer with improved display and filtering. Even during the soccer match, where we hadn’t seen anything in previous viewers, the vibrations were so visible that we will be busy for the next few weeks figuring out whether these signals truly all came from the Volksparkstadion, 2 km away.

Alexander watched the soccer game live and could at least associate some of the signals with goals or drums. It seems real! The Deichkind concert is next, on August 30, this time right next door at the Trabrennbahn. We are excited.

Spectrogram of the Taylor Swift concert, with some songs annotated. The brighter yellow colors indicate stronger vibrations. The ‘ladder structures’ shift up and down with the beats per second of the song – this is probably a result of the fans jumping to the beat, and with that, they generate specific frequencies.