GW 170817 — Multi-messenger astronomy is born!

On October 16 2017, astronomers (including me) announced the first ever joint detection of an event by Gravitational Waves and light. This is a huge moment for astronomy. It opens up an entirely new way to study the cosmos. To try to explain why this matters so much, I use an analogy of a naturalist studying wild animals. Imagine a naturalist with only sight to use: they would learn a lot, watching how animals behave. Imagine then that someone found a way to detect sound from the animals: but the listening was independent of looking, and they couldn't tell where the sounds came from. Again, there would be a lot to learn. But then, one day, people managed to both see and hear the animals at the same time: this would revolutionise the way they do their science. This is a fair analogy for what we have now done in astronomy*.

I posted recently on the subject of this approach, which is called multi-messenger astronomy, so I'm not going to go into all the details of that here, instead, I'll focus on this first event. If you want to, you can go and check out a this video of myself and Prof Nial Tanvir discussing it.

Please note: this is a huge event and many, many astronomers were involved. This is a snapshot form my perspective. It is incomplete, and totally biased in that I've focussed on what I did and published. It is not meant to be a complete overview of the event. Maybe I'll write another blog later to do that, but there were nearly 80 papers on arXiv this morning on this subject, so it's going to take a while to get up to speed!

What actually happened?

GW skymap of GW 170817

The skymap of GW 170817. The image shows the entire sky, and the small black mark near the centre is the gravitational wave position: the waves came from somewhere in there. The blue shaded area is the gamma-ray burst localisation; the GRB came from somewhere in there.

At 12:41 UT on August 17th, two things happened. The LIGO and Virgo gravitational wave observatories detected a signal from what looked like two neutron stars merging; and the Fermi satellite detected a short gamma-ray burst or GRB. Given that we think short GRBs come from merging neutron stars we thought it likely the two things had detected the same event. For various reasons at first the gravitational waves were barely localised on the sky at all — they just came from ‘somewhere’, but pretty quickly scientists in the LIGO/Virgo collaboration pulled together all their data and gave us a really accurate position, by gravitational wave standards - an area about 31 square degrees on the sky, that's roughly 155 full moons (see the image on the right). But, they also told us it was nearby, so we could instead concentrate on nearby galaxies since, if two neutron stars had merged near(ish) to Earth, then they ought to be in or near to a known galaxy.

Once the alerts had been received, astronomers across the globe began searching the gravitational wave and gamma-ray burst error regions. We used ground- and space-based telescopes to search at all wavelengths, from gamma-rays and X-rays right down to infra-red and radio. Several teams independently found a new source in the nearby galaxy NGC 4993; about 130 million light years away (seriously, that's next door, cosmically speaking). The first to report this was the small ‘Swope’ telescope, but by the time they reported their finding, other teams had also detected it, including the infra-red VISTA telescope team which I am part of, led by my colleague Professor Nial Tanvir.

So, once the source was found you all jumped on it, right?

Yes, but it's a little more tricky than that. The problem is that space is dynamic. Lots of things flash and bang and fizzle all the time. So when someone says they may have found the counterpart to a gravitational wave trigger, you have to make a decision - fast. Do you stop your search and hit this source? If you do, and it was not the counterpart you've missed your chance of finding the real counterpart. Or do you ignore it, and carry on searching, and risk not getting any data on the the source if it's the real counterpart?

With Swift we did a bit of both. We interrupted our search to look at this new source (it has various names, I'm calling it ‘EM 170817’), and then resumed our search, interrupting it periodically to have another look at the source. One it became clear that EM 170817 really was the counterpart to the gravitational wave and GRB event, we abandoned the search. While I'm focussing on my own work for this blog, it's important to emphasize that large numbers of astronomers were doing the same thing with a wide range of observatories, at all wavelengths from very high energy gamma-rays right, through X-ray, ultra-violet, optical, infra-red and down to radio waves. The amount of data collected on this event is staggering, and I have not (yet) had time to read all of the results, which were only released while I was writing this blog. Since this is my Swift Twitter blog, I'm being unashamedly narrow-focussed, for now!

What were you expecting to see?

After a GRB, we normally see an ‘afterglow’: a glow of light at pretty much all wavelengths coming from the gas around the merging neutron stars, so, of course we were looking for this. But also, we were hoping to see something called a ‘kilonova’. This is light, most people suggest it's likely to be red or infrared, which comes from nuclear decay. The idea is that when the neutron stars merge, you should end up with a lot of very neutron-rich material around the aftermath, and this is a great site for something called ‘r-process nucleosynthesis’ which is where around half of the heaviest chemical elements in the universe — such as gold or platinum — are formed. These nuclear reations create radioactive isotopes which decay, giving off light, but also the same heavy chemicals are really good at absorbing light and trapping it. This ends up with the infrared glow I just mentioned.

And what did you actually see?

Vista detection

A Vista image of the counterpart to GW 170817. The bright object in the middle is the galaxy NGC 4993, and the small point to the upper left of the galaxy is the kilonova which followed GW 170817.

Telescopes like VISTA and Pan-STARRS saw the expected red/infra-red kilonova glow. This alone was really exciting because up to this point, the only kilonova detections so far consisted of maybe one discrepant datapoint. But of course, science (like life) is never straightforward. With Swift we expected to see X-rays from the afterglow, but we didn't. This is not unheard of after short GRBs, but given how close this one was, it should have been around 12,000 times brighter than a typical short GRB, so the lack of X-rays is really constraining. The NuSTAR telescope also found a total lack of X-rays, as did the Chandra telescope a few days later. Then after nearly 10 days, X-rays were found with Chandra, which watched them evolve over the next days to a week. But I'm getting ahead of myself. Going back to the day of the event: Swift observed the sourve about half a day after the gravitational wave trigger, and as well as seeing no X-rays, it saw bright ultra-violet and blue emission. This is not expected from a standard kilonova, althugh some models predict something like this may be seen. This bright pulse lasted only about a day, before fading away.

And what does it all mean?

There are a few parts to this answer. The infra-red data appear to be the expected kilonova. Fitting models to the data shows us that the emission matches what you expect if you rip of a suzable chunk of matter, and synthesize heavy chemicals like gold and platinum.

The ultra-violet data tell a different story though, and I've seen two explanations for this. The first, which is in my paper, is that there was a strong wind of neutrinos, either from a hyper-massive neutron star which formed after the merger (and, according to our results, probably collapsed to a black hole fairly quickly) or from a disc of material that formed around the this object. This wind drove some matter, about 3% of a solar mass, in a ball-like object (the red material is more like a doughnut shape). Because of the neutrinos, this material has fewer neutrons than the red doughnut, so instead of creating elements like gold, it only got as far as Zinc, Germanium, Gallium and similar. This produced the early ultra-violet pulse.

The second model for the ultra-violet emission (which is in a paper I'm a co-author on), is that jet which you get from a GRB heated up a ‘cocoon’ of material which was much wider than the pencil-thin jet. As this swept past the matter thrown off when the neutron stars spiralled in to each other, it created a mild shock, which glowed in the ultra-violet for a while.

Moving on to X-rays, this gets intriguing. If this event were a regular GRB, with the jet pointing towards Earth, the lack of X-rays suggests that there must have been hardly any gas in the environment of the object, which would be really surprising. But the fact that we detected X-rays about a week later is consistent with a different scenario: the jet was not pointed at us at all! Only when it's slowed down and starts to spread sideways do we see X-rays (or radio waves, which also were absent at first, and then rose up later). From this we infer that we were viewing the object at an angle, not straight down the jet. In fact, the angle we infer (around 25°) is very similar to the limit we worked out based on our ultra-violet data (≤30°) which is the same as the limit determined from the Gravitational Wave data. Which is fine, but if this is true, why did we see a gamma-ray burst? That's only supposed to happen when the jet is pointed towards us!

There are a few explanation for this floating around. Some people are suggesting that we maybe caught the very edge of the jet, for example. But the paper I was involved in looked into this mathematically and finds it very unlikely. Instead, in that work we suggest the cocoon that may explain the ultra-violet emission could also give rise to a GRB.

The reality is, we can't be certain about anything based on one object. We need to detect more of these with gravitational waves and light to start to build up a complete picture of what happened. But we've shown that we can do this. And we've shown that when we combine gravitational waves and light in this way, we find surprises!

What next?

Right now (October 2017) the advanced LIGO and Virgo facilities are offline, undergoing more upgrades. Meanwhile people like me are looking at what just happened, and what lessons we can learn to improve our response next time. In about a year, LIGO/Virgo should start up again, and hopefully we'll find more of these events, and so be able to build up a detailed, complete picture of what goes on when two neutron stars collide.

In the meantime, there are some specific things to consider. Until this event, we thought that a joint GRB and gravitational wave detection was unlikely. Many people were also predicting that the only real way to detected light from a binary neutron star merger would be with infra-red searches. This one event proved both of those predictions wrong, with a GRB, and with bright optical and ultra-violet emission. Of course, maybe we were just lucky, but this certainly suggests that our ideas about what sort of emission we should look for may need reviewing!

* I should point out that this is not the first ever joint light &non-light detection. In 1987 a handful of neutrinos were detected from the very nearby supernova SN 1987A in the Large Magellanic Cloud: a galaxy orbiting our Milky Way. However, we are expecting, or certainly hoping, that joint light & gravitational wave detections are going to become routine, so we believe this represents the beginning of multi-messenger astronomy as a regular part of our scientific toolkit.