Most of the exoplanets we've discovered thus far have been found because they're easy to spot—Jupiter-sized giants orbiting close in to their host stars. But the Kepler mission has been providing a huge catalog of exoplanets and with it we've obtained a very different perspective, finding that planets in general are common and most of them are far smaller than the gas giants first identified. This new perspective has raised the prospect that we can identify some orbiting nearby stars, following identification with direct observations searching for signs that the planet's atmosphere is shaped by life.
More recently, astronomers have started making progress towards identifying planetary candidates that are close enough that we could eventually image them with an orbiting telescope. Just in October, astronomers announced there was a hint of a signal from an exoplanet in the light from one of the closest stars, Centauri B. Today the astronomers have released a paper that suggests there are several planets around the nearby star Tau Ceti, and one of them is likely to be within the star's habitable zone.
It's important to note the signals of the planets are buried deep in a variety of optical noise, both from the telescopes and instruments, and from the star itself. Further observations are going to be needed to confirm that the signals appear consistently. But this work certainly suggests those follow-up observations are going to be a high priority.
The Tau Ceti observations are based on what are called radial velocity measurements. As planets swing around a star during their orbits, they tug the star back and forth, accelerating it ever so slightly in various directions. If the planets' orbits are aligned so that the motion of the star occurs in the same plane as the Earth, this will cause a slight Doppler shift in the light that we receive from that star. The star's light will periodically appear somewhat bluer as it gets pulled towards Earth and redder when it's shifted away.
But these periodic changes can be very small. This is especially true when the plane of their orbits isn't close to edge-on from the perspective of Earth (which causes less of the star's motion to be in the same plane as Earth). But it will also be small when the planets are the sort of moderately sized, rocky bodies we'd be most interested in.
Adding to the challenge of detecting them, small motions in the instruments themselves can create a noise that obscures the signal. So can things like the rotation of the star (which brings brighter and darker regions to the fore) and long-term cycles in the star's activity. It takes a lot of effort to account for this noise (or, technically, "jitter") and find weak signals buried within it. So, the authors say they started out focusing on Tau Ceti simply because it's a very quiet, regular star. People had looked at it before and saw no hints of exoplanets, so it could provide a good test case for figuring out how to account for jitter.
In the past, one thing that researchers did was "bin" data, aggregating observations taken across several hours in order to average out short-term changes in the light we're seeing. The authors of the new paper suggested that this method, while getting rid of noise, could also be getting rid of some of the signal. So, they started with unbinned data and began trying to create a statistical model of the noise. Their procedure for doing this was to add signals for a number of planets to actual data from Tau Ceti. They then created various mathematical models of the noise and subtracted them from the data, trying to determine which models worked best (which turned out to be a combination of moving average and random white noise).
Once they had that in hand, the astronomers turned to the actual data from Tau Ceti without any added signals. And, this time, three signals did pop out, with each of them adding between 1010 and 1017 to the statistical fit with the real data. The authors concluded that there were three planets in this signal, orbiting with periods of 14, 35, and 94 days.
That's the most convincing part of their paper, but it wasn't the end of the paper. All of their initial data was obtained using the planet-finding HARPs instrument. But data on Tau Ceti had also been obtained using Hawaii's Keck telescope and the Anglo-Australian telescope in (you guessed it) Australia. These instruments had about three times the noise that the HARPS data did. When the authors checked the data from these instruments, they found... nothing, not even the planets they'd found in the earlier analysis.
Nevertheless, they went on and combined the data from all three instruments. A basic analysis didn't identify any signals, but they went ahead and started doing modeling, assuming their data contained noise and one or more planetary signals. The two planets at 13.9 and 94 days came out of the analysis with high significance, and the third also increased the fit of their model. But they kept going and found that adding two more planets to their model increased the fit, although not by nearly the same degree. One orbited with a period of 630 days, the second at 168 days. The latter one is the object that resides in Tau Ceti's habitable zone; it has a mass that's at least four times that of Earth's.
Even the authors admit that discussions about the planet, "remain merely speculative until the planetary origin of the signals can be verified by an independent detection." But there are a number of reasons to think that something is there. Tau Ceti has a bright debris disk that extends out to 55 Astronomical Units (55 times the average distance from the Earth to the Sun). These sorts of disks are thought to be the raw materials for planet formation. Tau Ceti is also a low-metal star (it doesn't have much beyond hydrogen and helium), which is in keeping with the fact that the signals suggest low-mass planets. And, finally, the specific configuration of planets detected in this study would be able to maintain stable orbits indefinitely.
Still, with planets this close to the limits of detecting, it's difficult not to remain a bit cautious for now (just as the authors have). Hopefully, a few more years of observations will help us know how much of this evidence is signal, and how much is noise.
没有评论:
发表评论