The signal

The campaign is up and running! As of February 9th 2016, we have only lost two nights on HARPS due to weather, which means we have 15 good qualityspectra ready for processing. On the photometric follow-up side, the LCOGT telescopes have been obtaining good data (photometry in UBV bands), and the ASH2 telescope has already accumulated 19 nights of good photometry as well (visual and red colours). The BOOTES station is suffering technical difficulties that we hope to sort out soon. Fortunately, the ASH2 telescope (the last one to join the effort, but the most successful photometer so far!) offered a degree of redundancy that saved the day! We will introduce all the observatories involved in forthcoming posts.

Feb8
Status update of the observations. We recently reached 25% of the HARPS data, and two photometric follow-up observatories have been operating nicely over the same period. We expect more downtime due to weather, but the survey goals will probably be achieved if we can hit 80% of the planned observations.

Now that we also had the opportunity to read about the Doppler method and how stellar activity can mimic the presence of a planet, let’s talk about what we are trying to achieve here. Analysis of previous campaigns show that a possible smooth signal was observable when monitoring the star at moderately high cadence; but we must remain cautious because stellar activity can produce the same kind of variability. As described in the article by Paul Gilster, Proxima has been monitored for small planets before. The most exhaustive works include the UVES/ESO survey for rocky planets around M-dwarfs, conducted between 2000 and 2009; the searches with HARPS by the Geneva team; and recently obtained data from our own high cadence program with HARPS, called Cool Tiny Beats(2013-2014). Here are some of the technical details for those of you who are interested in them…

The VLT/UVES Doppler data, and possible signals in it

The UVES Doppler measurements were published in Kuester & Endl 2008. In previous posts (eg. see Figure 1), we have seen that if we have a planet we should see an oscillatory motion over time. These measurements didn’t look much like that (Figure 2). Still, the velocities of UVES seemed to be not completely random.

Example of measurements (in red) overplotted on the expected Doppler signal caused by an exoplanet on the Star. Changes in the velocity of the Sun-like star 51 Peg used by M. Mayor and D. Queloz to infer the presence of a gas-giant planet in a short period orbit around the star.
Figure 1. Example of measurements (in red) overplotted on the expected Doppler signal caused by an exoplanet orbiting a star. Changes in the velocity of the Sun-like star 51 Peg, used by M. Mayor and D. Queloz were used to  infer the presence of a gas-giant planet in a short period orbit around the star.
Doppler measurements of Proxima from UVES. No clear sinusoid can be spotted by eye, which already rules out the presence of long period gas giants around the star.
Figure 2. Doppler measurements of Proxima from UVES. No clear sinusoid can be spotted by eye, which already rules out the presence of long period gas giants around the star.

Stars are only visible for a few months of a year, so that could be the smoking gun of a planet with a period similar to Earth that we happen to be sampling at more or less random moments of the orbit. Kuester & Endl 2008 had reasons to suspect that this variability was indeed caused by activity, or even some unknown instrumental effect. Once that possible signal was removed by fitting a sinusoid to it, very little remained in the residuals besides apparently random noise at the 2-3 m/s level. The Doppler signal of a planet is stronger if the planet is closer to the star (as in the Solar System where Mercury takes less than three months to circle the Sun, the motion of close-in planets is faster). So, while no clear signal could be extracted from these measurements, the data did tell the researchers that no large planets were orbiting the star with periods shorter than few hundred days.

Limits to the minimum mass of planets orbiting Proxima. Concerning the 'Habitable Zone' (here marked in green between 4 and 15 days, but new models suggest it extends to periods as long as 27 days), planets down to 3 Earth masses (minimum mass) were ruled out by the data.
Figure 3. Limits to the minimum mass of planets orbiting Proxima. The ‘Habitable Zone’  is marked in green between 4 and 15 days, but new models suggest it extends to periods as long as 27 days. Planets down to 3 Earth masses (minimum mass) were ruled out by the data. Extracted from Endl, M.; Kürster, M. 2008 A&A

The minimum masses of planets ruled out by UVES are illustrated in Figure 3. Let us note that we say ‘minimum mass’ of the planet because the Doppler method only measures the motion along our line of sight. Even in that case, statistical arguments indicate that it is highly unlikely to find any planet less than ~5 times the mass of the Earth in its habitable zone, with other techniques. With this upper limit set, the UVES program stopped observing Proxima and another handful of M-dwarf stars at the end of 2008.

The HARPS/Geneva team observations of Proxima prior 2012

During the same years, Proxima was observed about 25 times with HARPS. While the star did show variability at the 2-3 m/s level, it also showed evidence of activity in occasional flaring events, and some excess of radiation coming from its chromosphere. In any case, the measurements were consistent with those of the UVES survey in the sense that no obvious signal was detectable above ~2 m/s. In 2013, the star was again observed in the extended HARPS survey for M-dwarfs led by the ex-Geneva astronomer X. Bonfils, now based in Grenoble, but no report has appeared on significant period variability so far. So these campaigns led to no convincing evidence of a signal.

Doppler velocity measurements by X.Bonfils and his team taken between 2002 and 2009 with HARPS. Source : Bonfils et al. 2013 A&A, available via arXiv.
Doppler velocity measurements by X.Bonfils and his team taken between 2002 and 2009 with HARPS. Source : Bonfils et al. 2013 A&A, available via arXiv.

The ‘HARPS – Cool Tiny Beats’ observations (2013-2014)

In 2013, the same team as  the Pale Red Dot campaign started a programme to measure radial velocities at high cadence (focused on a small sample of very nearby M-dwarfs) to hunt for short period planets, pulsations and understand the connection of stellar activity with apparent Doppler signals. Proxima was a natural target for the survey, which was executed in two runs of 12 nights each (May 2013-Jan 2014). As opposed to the rest of the stars in the sample, the radial velocity measurements of Proxima were smoothly varying over both observing runs. Unfortunately, given the length of both runs, the strict periodicity of the variability could not be verified. Worse than this, the long term Doppler variability found by the UVES survey was still present but seems rather unpredictable, meaning that combining the data from years ago did not help much in confirming it. This is when Pale Red Dot was conceived…

Doppler measurements of Proxima obtained in 12 consecutive nights in May 2013, suggestive of smooth variability on the timescale between 10 and 20 days. The origin of this 'signal' is what the Pale Red Dot campaign is trying to figure out. Image credits : G.Anglada-Escude.
The top panel contains Doppler measurements of Proxima obtained in 12 consecutive nights in May 2013 which are suggestive of smooth variability on the timescale between 10 and 20 days. The lower panel is what we call a ‘periodogram’, which is a mathematical tool to identify possible periodicities in the data. Because the observing run was limited to ~12 days, we cannot really constrain the putative period with this data only. The Pale Red Dot campaign is trying to figure out if this is a strictly periodic signal feature by observing the star ~60 nights in a row, thus covering several cycles of the putative signal, and comparing the variability with simultaneous photometry. Image credits: G.Anglada-Escude.

So while we are convinced there is a signal in the Doppler measurements of Proxima, previous data do not allow to confirm its presence and clarify its origin. The long term variability of Proxima spoiled our attempts to combine data from previous observations so we needed a dedicated campaign.

In summary

Combination of UVES and HARPS data at different cadences suggest that the star is showing a smoothly varying Doppler signal. Since the UVES survey set an upper limit between 2-3 Earth masses and if the signal is not activity induced, it must correspond to a planet smaller than that (between 1-2 Earth masses>). The signal might well be caused by stellar activity, which should be quasi-periodic as opposed to the strict periodicity of the orbital motion of a putative planet. So this is what we want to figure out! If you really want to learn more, feel free to any member of the Pale Red Dot team!

What is the exact plan?

We are following Proxima Centauri for about two months. If the planet is there we should see the velocity of the star going up and down between 3 and 5 times depending on the precise period. Simultaneously, we are monitoring Proxima with telescopes from LCOGT.net, the ASH2 Atacama telescope and the BOOTES network. The contiguous and regular sampling of the observations together with quasi-simultaneous photometry should allow us to model its long term variability better and, hopefully, confirm whether the Doppler signal is caused by a planet or not. If it isn’t…  we will move on and keep searching for #palereddots around other nearby stars…

Thanks for following!

A search for Earth-like planets around Proxima Centauri