With all the new discoveries that seem to occur almost monthly, it’s sometimes hard to remember that finding exoplanets is still a relatively new and difficult science.  As part of those continual discoveries, back in 2018 a team announced they had found a planet candidate around Barnard’s Star, one of the closest to our own.  Now, a different team has re-analyzed the data, collected some new data, and found that the planet detection was likely a false positive.  

A variety of factors were suggested as a cause of the false positive.  But it is easiest to think through it by considering the detection method, the star’s rotational period, and a signal processing artifact.

The team that originally detected the planet was known as CARMENES, which is a consortium of Spanish and German astronomers.  They did their due diligence, observing the star over the course of 23 years using 7 different instruments to collect data.  In fact, they were not the first group to suggest that the second closest star system to our own might have a planet.  In 1963 an astronomer named Peter van de Kamp claimed to have found a planet around the star in what would then have been the first discovery of an exoplanet.  That finding was later debunked by a variety of researchers who also used the astrometry method but were not able to reproduce the “wobble” that van de Kamp found.

Discussion of the astrometry method of exoplanet hunting.
Credit: ESA YouTube Channel

Fast forward to 2018 and the CARMENES team announce their new planet around Barnard’s star that they found using the radial velocity (RV) method.  This method relies on the slight blue-shifting and red-shifting that takes place as a star approaches or recedes from us that is caused by the pull of a planet it is orbiting around.

Some factors confound these RV measurements.  One that seems to have caused the false positive around Barnard’s star results from the star’s rotational period.  Barnard’s star has an extremely slow rotation – around 145 days, which is almost 6 times longer than the Sun’s 25 day rotational period.  

Barnard’s star is close enough that it can actually be seen moving against the stellar background over the decades. It has the fastest movement rate of any star
Credit: Steve Quirk

That rotational period is important for a few reasons.  First, sunspots are tricky to account for in RV measurements, as they can easily be interpreted as a shift.  To account for this, scientists usually collect data on more than one rotation sequentially to see if they can track the sunspot as it rotates around the star.  Unfortunately, the entire observational time of Barnard’s star is only around 270 days, meaning that scientists cannot collect data on a full two rotations at one time.  Sunspots can also exist on M dwarf stars (of which Barnard’s is one) for more than 10 rotations, so even spread over multiple observational periods, sun spots could affect the RV readings of the star.

According to a team at the Habitable Planet Finder (HPF), that is exactly what happened. What first clued them into that possibility was the orbital period of the proposed planet. At 233 days, it could be detected during a single observational window.  However, there is a quirk of signal processing, known as aliasing, that makes that period unlikely.

Example of aliasing of a periodic signal.  If samples are only taken at points 1,5, and 9 then either of the two signals would fit the data.
Example of aliasing of a periodic signal. If samples are only taken at points 1,5, and 9 then either of the two signals would fit the data.
Credit: Andrew Jarvis, Wikimedia Commons

Periods of 145 days (the star’s rotational period) and 233 days (the proposed planet’s orbital period) are aliases of each other.  Aliasing happens when data is only intermittently sampled.  That is exactly what happens when ground based telescopes are no longer able to observe the star, either because of the Earth’s rotation or because of its position around the Sun – hence the 270 days observational window mentioned above.

If a signal is “sparsely” sampled, the resulting data could be fit by multiple sinusoidal patterns.  Those sparsely sampled data points can also be affected by the sunspots mentioned above.  In fact, they can do so in a way that can create a false positive, which the HPF team believes happened in the case of
Did you miss our previous article…