The localization isn't performed through parallax, it's done through triangulation and time-of-arrival. For a pair of observatories, a difference in the arrival time of a signal (moving at the speed of light) defines a circle on the sky of possible source locations. With three observatories, there are three circles on the sky, which intersect at a single point.
The Virgo detector, outside of Pisa, is in the final stages of installing its advanced instrumentation. With LIGO+Virgo a source like the first detection, from September (a very loud event), could be localized to a patch of sky about 10 square degrees in area.
> So basically we can only determine the general direction?
With current-generation detectors installed at three stations which is all we can hope to achieve at the current level of funding in the very near future.
More money, more detections at lower energies, better idea where they come from. There are plenty of instruments in cosmology, astronomy, and physics that have already been designed and planned out, but which do not have the funding to be built; Wait 20 years and five percent of them might come to fruition. You could pour trillions of dollars on these problems without running out of novel questions that we've already proposed ways of answering, and novel results from exploratory instruments we've already proposed building. We spend about 30 billion a year on basic science research according to the NSF, spread over all fields. For comparison, the military gets upwards of 600 billion.
We have good ideas about how to make gravitational wave detectors much, much better, but not how to make them much, much cheaper.
Not so bad as that, but it is a large area, and it's a challenge for optical astronomers to detect a faint source in such a large patch of sky. It's about the size of your two palms held at arms length. (Or, 40 times the size of the full moon, but that sounds less optimistic.)
Well, the goal of LIGO is to detect if gravity is a wave. So we had to build sensitive sensors, over large distances to be able to pull off this feet. It was (to my knowledge) only build to do this. The fact you can use it for triangulation is a bonus. If they wanted triangulation, I suspect they would have made it with 3 sensors instead of 2.
It sounds like you've already read about how we estimate the false-alarm rate for signals, so I'll just add that the estimated rate of BBH mergers was highly uncertain before this observation; the error bars spanned three orders of magnitude. See, for example, Fig 5 of http://arxiv.org/abs/1111.7314, which compares the previous LIGO-Virgo upper limits on similar events to the expected rate from population synthesis models and observations of high-mass X-ray binaries, known BNS systems, etc.
The rate inferred from GW150914 is on the high end of the rate estimates from astronomers, but it's completely consistent with prior observations. Certainly, if we had seen ten events in the first 16 days of data, it would not have made sense! But one event is well within expectations.
A lot of this information seems to be available, for example figure 4 here[1]. According to that, the horizon distance for "a binary black hole system with the same observed spin and mass parameters as GW150914 for optimal sky location and source orientation and detected with an SNR of 8" was 1.5-2 Gpc. Now we only need an estimate of the rate at which events with these properties occur, not using the data from GW150914. I presume this is somewhat (a few orders of magnitude) less than the rate at which the mergers occur in general.
Thanks, I'll have to look more closely at that paper since I only see discussion of the upper bounds. The reason for my concern is that I noted elsewhere that the prior lower bounds on the merger rates got pretty low: ~0.1 Gpc^-3 yr^-1. For only 16 days of observing the expected number of events drops to 4.38e-3 Gpc^-3.
Then if the horizon distance is somewhat less than than 1 Gpc we need to scale this further. Say it was 0.5 Gpc, we scale by 0.5^3 to get ~5.475e-4 expected events. For 0.2 Gpc horizon we get ~3.5e-5! These values are getting dangerously close to the estimated background rate, at least using this crude calculation at the lower ends of the prior estimates.
In Figs 6 and 7 of the second paper you can see the constraints from GW150914 on what are known as the "post-Newtonian" expansion terms of Newtonian gravity. Previously, terms beyond first order were only loosed bound, mostly from observations of the Double Pulsar system J0737-3039.
Just from this one event, we can also constrain the mass of the graviton to an order of magnitude less than the previous best measurement.
In addition to the two LIGO sites, there is the Virgo instrument outside of Pisa, Italy, which will come online later this year. The KAGRA detector is currently being assembled underneath a mountain in Japan. And, mentioned in another reply, LIGO has the equipment for a third detector. This is currently in storage in the hopes that the Indian government will build a facility. By 2023 there should be five widely-spaced detectors worldwide.
The resolution of time-of-flight between the two LIGO sites for the signal we just detected was about half a millisecond. This resolution is somewhat dependent on the signal strength and the location of the source relative to the detectors. With three sites we can localize most sources to tens of square degrees on the sky. This is still very large; the moon is a quarter of a square degree.
The odds are pretty good to observe only one event in 16 days of data, and the likelihood of seeing the event on the first day is the same as the likelihood as seeing it on the last day. The analysis of the remaining data from the first observing run (ended Jan 12th) will probably take a couple of months.
Another LIGO scientist here. It takes an overwhelming amount of energy to generate gravitational waves, and detecting them from terrestrial sources is about 20 orders of magnitude more difficult than the measurement we just made. Space, is extremely stiff; bending it enough to be detectable requires a huge amount of mass-energy.
LIGO scientist here. The way this is presented can be a little deceptive - the isolation is very frequency dependent. At high frequencies (>10Hz), the pendulums and blade springs in the suspension isolate the mirrors very well, so they are moving by only these small amplitudes (10^-19 meters). But at low frequencies (<1Hz) the isolation ratio is essentially 1, so the amplitude of the mirror motion is roughly the same as that of the ground (about 10^-6 meters).
Yes, which makes sense physically. If the earth's rotation slowed by a constant velocity, in order to damp that motion the mirrors would have to displace themselves in the opposite velocity. Of course, there's no room for them to do so, so damping such low frequencies is not possible.
