If we’re confident that there is a real transit signal coming from a given star, we can apply for an allocation of time to use the Zorro speckle imager on the Gemini South telescope (which is a huge 8-metre telescope in Chile). This instrument takes lots of images of the star in quick succession, which allows us to “freeze out” the effects of the Earth’s atmosphere that causes light from stars to be distorted (similar to the distortion that causes stars to twinkle). This allows us to determine whether there are any nearby stars that might be contaminating our measurements. If a star were close enough that some of its light was “leaking” into our measurements of the candidate then we’d see a smaller transit depth than is actually the case. This would mean that we’d be underestimating the radius of the orbiting body which could mean that what we believe to be something the size of a planet could actually be the size of a star!

Below is a plot showing real data from the Zorro instrument on Gemini South for one of the most promising planet candidates from Planet Hunters NGTS (Subject 69654531 which you can read more about by clicking this link). The two different lines show data for two different wavelengths of light. The human eye typically sees light with any wavelength between 380 and 750 nanometres, but telescopes can focus on ranges centred around specific wavelengths with various filters. In this case, the telescope splits the beam of light (with a component called a dichroic) so that redder light (centred on 832 nm) goes to one camera while bluer light (centred on 562 nm) goes to the other. These two different measurements are shown in red and blue respectively on the plot below and allow us to obtain simultaneous two-colour imaging of the target.

The horizontal axis of this plot shows angular separation from the star in arcseconds, which is a unit used to measure the distance between points in the night sky (e.g. the Moon distance from one side of the moon to the other as you look at it from Earth is 1,900 arcseconds). The vertical axis is called “contrast” or “magnitude difference” and indicates how much dimmer the region around the star is compared to the star itself in units of apparent magnitude. What this plot shows is that between 0.1 arcsecs and 1.2 arcsecs, there aren’t any stars with an apparent magnitude within about 5 magnitudes of the target star’s brightness which lets us know that there probably isn’t a nearby star at these separations (although there could still be something really close in!). Using the results of these observations, we can be more confident that the transit we’re observing is “on target” (i.e. coming from the star we think it is) and we can have a more confidence in our estimates of the radius of our planet candidate as we know the levels of contamination aren’t significant. This gives us the necessary belief to push ahead with further follow-up observations which I will describe in the next part of this blog post series.

Once we have identified an object as a potential planet candidate from the NGTS data, the first step is to check whether we also see the transit signal in data from the Transiting Survey Exoplanet Satellite (TESS). Let’s walk through an example of one of the subjects that we initially thought was a planet candidate, Subject 69693101. The Planet Hunters NGTS light curve for this object looks promising. Also the odd/even and secondary eclipse (not shown) light curves look fine too and when we analysed the NGTS data we estimated a radius of 1.35 Jupiter radii for the orbiting body, which is big but not so big that we would immediately rule it out.

Therefore, we decided to check the TESS data for this star. Luckily TESS has observed this star in two “sectors” which means we have approximately 2 months worth of data to look at (TESS observes a region of the sky, called a sector, for a month at a time before moving on, but has is now revisiting regions to gather more data). The plot below shows the light curves for both sectors with the x-axis showing time in days since a fixed date. (Astronomers like to use a unit called Julian Days which is the number of days since January 1st 4713 BC, but this number is huge so we subtract standard values to make the numbers more manageable, in this case we’ve subtracted 2457000 days).

At first look, this star looks to be very variable, possibly due to some intrinsic stellar variability although periodic signals like this can be as a result of it being a binary star system. The first step is to try phase folding the data onto the period found by NGTS. This is a process where we take each section of the lightcurve that corresponds to (what we believe to be) one orbit of the candidate, centred on the transit, and plot the data on top of each other. This means that the transits should all be in the middle of the plot. The plot below shows the data folded onto the 2.51 day orbit that NGTS detected, with grey points showing the raw data and red points showing the data binned in 10-minute averages.

This looks like a clear transit signal, although the apparent variability of the star made me cautious when looking at this. Therefore, I tried folding the data to twice the detected period and, voila, we can clearly see two transits of differing depths which tells us that this is indeed an eclipsing binary system and not a planet candidate. While it’s unfortunate that this particular candidate isn’t a planet, it does mean we can focus our time and resources on more promising candidates that might really be planets!

You may be wondering how a subject goes from being one of thousands of subjects in the Planet Hunters NGTS project to becoming a candidate and then, hopefully, a bonafide planet! In this series of posts I hope to shed some light on the life cycle of a Planet Hunters NGTS subject.

Once a set of subjects are “retired” from the site, which occurs when all the subjects have received responses from at least 20 unique users, the data is collected from a huge “csv” file (that stands for comma-separated values). This contains every response by every volunteer for every subject included in the project and is effectively a giant spreadsheet, albeit one with millions of rows that would crash any standard spreadsheet viewer like Excel. This huge file is processed by a computer program and the most important data is pulled out and inputted into a large database that forms the hub of our operation.

From here, I can run another program that iteratively calculates scores for each subject for each possible response and computes weights for each user so that we can pay more attention to the classifications that identify transits more easily and therefore pick out subjects with high scores that might have been missed with simple vote counting. This is similar to the scoring and weighting schemes used in the Planet Hunters and Planet Hunters TESS projects, as well as many other citizen science projects. This lets me generate a list of the most promising candidates (e.g. subjects that are probably U-shaped) which are what you’ll see in the Secondary Eclipse & Odd/Even Transit Checks. I can then apply a similar program to classifications in these workflows to identify the subjects that are most likely to not have secondary eclipses or transit depth differences. These are the strongest candidates for being planetary transits instead of eclipsing binaries.

The next step is the most arduous as I look through all these candidates by eye (and there’s usually a lot of them!) and pick out the most promising ones to get a second opinion on. Filled with optimism and with candidates in hand, I show these to the NGTS team who evaluate them as a group to pick out the most promising planet candidates that we think are worthy of further investigation. At this point we would class these objects as potential planet candidates, although they can be ruled in or out through a series of tests using different data sources that I will go into in later blog posts.

We’ve spotted another transiting object around a star! With unprecedented image resolution we have captured the transit of this festive phenomena seen in the animation below.

The transit has a depth of ~6.5% around this small star that has a radius of only 0.4 Solar radii. That means that we can estimate the size of the transiting object (assuming a spherical Santa) to be R = 9.73 x 0.4 x sqrt(0.065) = 1 Jupiter radius! There were hopes that this would be an Earth ana-yule-logue with a Jingle Bell Rock-y composition but it seems to be a Ho-Ho-Hot Jupiter. The shape of the light curve is more similar to an exo-comet which means that it must be accompanied by an exo-cupid. Further observations will be required to identify additional exo-reindeer. Following its confirmation the festive object has been given the moniker HAT-P-Christmas.

Thank you to my fellow PhD students: Joe Murtagh, for drawing the silhouette of Santa and his reindeer; Lucy Dolan and David Jackson, for their festive-themed exoplanet puns. Merry Christmas and Happy Holidays to all!

This week, I was in Edinburgh for the UK Exoplanet Meeting 2022 (UKExom22, for short) where I presented details of the project and the results so far from Planet Hunters NGTS, including details on our 4 new planet candidates. This was an opportunity to showcase the brilliant work of the Planet Hunters NGTS volunteers to an audience of over 100 exoplanet astronomers based in the UK, as well as a chance to meet (in some cases for the first time!) and speak with members of the NGTS team who have been working hard to help us vet the best candidates from the project and coordinate follow-up observations.

The first day of talks included some exciting results from the new JWST mission as well as techniques being employed to try and detect Earth-analogues which are currently difficult to find due to stellar activity! Later in the afternoon, I gave a 12-minute presentation describing why citizen science is so great, how Planet Hunters NGTS works and what we know so far about our new planet candidates. There was plenty of interest in the project from the audience who were impressed by both the enthusiasm of the Planet Hunters NGTS volunteers and the fact we already have four new planet candidates from our very preliminary analysis.
Day two of the conference started with more interesting talks including hearing about the current status of the PLATO mission and the results of the ASTEP telescopes that are helping the hunt for exoplanets at an observatory out in Antarctica! The relaxed schedule of the afternoon provided the opportunity for me to meet with another member of the Planet Hunters NGTS team, Sam Gill, who can fit the NGTS data with better models than the initial search to provide us with more accurate estimates of the planetary radius. Below is an image of the output of Sam’s work, showing more plots of each transit and the phase folded light curve fitted with a more accurate model. I was also able to speak with Matt Burleigh and Alicia Kendall, collaborators from NGTS who obtained more photometry of some of our candidates discussed in the previous blog post and will be able to combine data from multiple data sources to provide us with a more accurate radius estimate for Subject 69654531.

It’s not all about science at a conference as Tuesday evening saw many make the short trek up Blackford Hill which is home to the Royal Observatory Edinburgh and provides excellent views of Edinburgh Castle, Arthur’s Seat and the sunshine on Leith. The conference dinner and a trip to a Karaoke bar that had a large number of astronomers attempting to sing Bohemian Rhapsody capped the second day of the conference.

The final day of the conference included talks on improved techniques for measuring the radial velocity of stars which will allow more accurate measurements of exoplanets and the first direct detection of planetary material being accreted by a white dwarf. This discovery was made by detecting 5 (yes, only five!) X-ray photons being emitted by a white dwarf which was then determined to be due to remnants of a planet falling into the atmosphere of the white dwarf.

That brought to an end my first in-person UK Exoplanet Meeting and I’m already looking forward to next year’s meeting that will be hosted by UCL in London! In the meantime, I’ll be vetting more candidates from the project and coordinating follow-up observations as well as considering applying for additional observing time that would potentially help us get closer to validating or confirming some of the candidates already mentioned or any others that we find in the coming months. Keep an eye on the blog to stay tuned!

You may have seen my previous post announcing that we have 4 new planet candidates discovered by Planet Hunters NGTS, if not you can find it here. We wanted to give you some more information on what we know about each candidate so far and the efforts we’ve made to gather follow-up data as we work towards ascertaining whether or not they’re real exoplanets! The first step once we decided that these candidates were worthy of further investigation was to fit the data using a more intricate model. This is typically what’s meant by phrases like “modelling estimates” or “fit results” that you may see below or in other pieces of science communication. Each of these candidates pose their own unique challenges in being able to confirm whether they’re really planets, however to already have 4 planet candidates is a promising sign for the future of the project and a real testament to the brilliant work put in by you, the Planet Hunters NGTS volunteers.

Subject 69695263: This is our most promising candidate so far as the transits look quite clear and our modelling estimates give encouraging results. We think this is a hot Jupiter orbiting a “K dwarf star,” which is a star slightly smaller and cooler than our Sun. We’ve estimated that the planet candidate has a radius equal to 1.09 times the radius of Jupiter, which we write as R_p=1.09 R_Jup. It orbits its host star every 1.74 days which means by the time you turn 18 Earth-years-old, you’d be almost 3800 years-old on this planet candidate (if it exists, and assuming you could live on it, which you can’t). We have received data from the Zorro team for this candidate (read more about the Zorro instrument and our follow-up efforts here), and it shows that the host star is isolated from other stars as far as we can tell, which means that we can be slightly more confident that this isn’t a false positive signal caused by an eclipsing binary! There’s still plenty of work to do before we can approach the possibility of confirming or validating this as a planet, but we’re cautiously optimistic about this candidate.

Subject 69593857:

This candidate has been causing something of a headache since we only managed to observe one full transit and a number of partial transits (i.e. we only see the start or end of the transit). We have estimated that the potential orbiting body has an orbital period of 3.97 days however our radius estimates range from Jupiter-sized up to a radius not much smaller than our Sun. This suggests we could have a planet candidate but it could also be what’s known as an eclipsing binary with low mass (EBLM). While these are interesting systems, they’re not the planets we’re looking for! We do have Zorro data for this candidate and we can’t see another star nearby however it’s possible that the signal could be a false positive caused by “systematics” which is a general term for any trend in the data that might not have been accounted for during processing.

Subject 69473954:

Much like the candidate above, this candidate is a troublesome one. The transits appear to be grazing (check out this post for a description of grazing transits) which makes it tricky to get a good estimate of the radius of the candidate. The radius of the planet candidate is estimated to be at least R_p=1.50 R_Jup which is already at the upper limit for what we believe can be a planet. Our upper estimate of the radius is one solar radius, meaning this could be a standard eclipsing binary. Sadly we don’t have Zorro data for this candidate to be able to check this just yet. This candidate, and the previous one, will require some more investigation to really understand what they actually are.

Subject 69654531:

Lastly, this candidate is very exciting because if our initial estimates are correct, this would be one of only a handful of discoveries to date of giant planets orbiting close to their small host star. The transit depth of ~12% is much deeper than we typically expect for exoplanet transits, however the host star has a radius equal to only 1/3rd the radius of our Sun. Therefore, we believe this could be a hot Jupiter, with a radius of R_p=1.13 R_Jup, orbiting an M dwarf star on a 2.10 day period. These types of systems are very rare and really surprising to find because our current understanding of how planets form suggests that such a large planet shouldn’t be able to form around such a small star because there simply shouldn’t be enough time for it to form or planet building material available. The Zorro data for this candidate is also encouraging as we see no companion stars. We hope to gather more data on this object in the coming months if possible, so stay tuned!

These four NEW candidates are really exciting! They weren’t spotted in the initial check of the data by the NGTS team which means that the Planet Hunters NGTS project has proven it really has the potential to be a complimentary search to help NGTS discover as many planets as possible. We couldn’t have done this without the help of everyone who’s classified subjects on Planet Hunters NGTS and we’re eager to see what else you can help us find in the future as there are still plenty of subjects to be classified! I’ll be presenting these candidates next week at the UK Exoplanet Meeting in Edinburgh before checking the latest results from the project for more possible planet candidates. We’ll also be starting to write up the first results from the project into a paper to be submitted to an academic journal!

Today, I presented the latest Planet Hunters NGTS results at the UK’s National Astronomy Meeting in the University of Warwick. Good news everyone! I am very excited to announce that we have four new planet candidates have been found by Planet Hunters NGTS. In addition, we have been able to get some observations of three of these new potential planet candidates with the Gemini South telescope in Chile!

I have spent the past several months developing a software pipeline to combine all of your assessments together for the various workflows that make up the website. This sifts through the candidates output from the NGTS algorithms to look for any new possible planets. In my preliminary search through the classifications available, I took the best candidates that were classified in the Secondary Eclipse Check and Odd/Even Transit Check and shared our findings with the rest of the NGTS team. Four of these candidates look like possible planet transiting planets and are shown in Figure 1.

There’s a lot of work that needs to be done to go from planet candidate to bonafide planet, so the four objects are still planet candidates. To confirm possible transit events requires using additional detection methods to get the mass of the orbiting body that confirm it has a mass less than a star or additional observations that can help statistically rule out the possible astrophysical false positives that can mimic planet transits (like eclipsing binaries). These candidates are around faint stars which will make validating these planets a tricky process.

Three of our planet candidates were observed in the past month to get follow-up observations. We worked with our collaborators in the US to apply for observing time on the Gemini South telescope. This involves: justifying why our candidates are interesting (there’s a lot of really interesting science that people want to do that we have to compete with!); justifying why the Gemini telescopes and Zorro instrument (see next paragraph) are the best tools for the job (in this case, Zorro is one of only a few instruments in the world that can carry out the kind of observation we need, another being ‘Alopeke on the Gemini North telescope in Hawaii); and calculating how much time we’d need to use the telescope for.

The instrument we are using is the ‘Zorro Speckle Imager.’ Zorro takes lots of images of the star in quick succession, which allows us to “freeze out” the effects of the Earth’s atmosphere that causes light from stars to be distorted (this effect is known as atmospheric seeing, see Figure 2). This allows us to spot whether there are any other stars so close to our targets that the NGTS telescopes couldn’t tell them apart. These background stars contaminate the light we measure for the main target star and dilute eclipsing binary light curves such that we can’t see the secondary transits, mimicking what we would observe for a true transiting planet. This isn’t a design flaw in NGTS but a reality of how different telescopes are built for different purposes. For example, Zorro isn’t designed to survey our targets for the long timespan, like NGTS has, in order to spot these transits in the first place. Using different telescopes for exoplanet follow-up and confirmation is much like a football (soccer) team: if the defenders don’t win the ball from the opposition (NGTS spotting transits), they can’t then pass it to the midfielders to move it up the pitch (Zorro checking for other stars).

Our observations were carried out by the excellent team of astronomers and support staff at Gemini and NASA a few weeks ago and we’re hoping to be sent the full final results soon.

What about the strikers in our analogy? If we find out that these targets are solo stars, that isn’t the final step in confirming an exoplanet (it’s also a big IF). We’d ideally take “radial velocity” measurements which allow us to measure the mass of the exoplanet. This technique works by detecting how much a star is “wobbling.” This wobbling is caused by the exoplanet orbiting the star and the amount of wobble relates to how much mass the exoplanet has. When we say the planets orbit the Sun, really we mean the planets AND the Sun orbit the entire Solar System’s common centre of mass. It just happens to be that this point is very close to the Sun since it’s so big. It’s the same story for exoplanets and their stars. The radial velocity measurements take the role of the striker in our analogy, although it’s important to say that this wouldn’t be the end of it and there’s still plenty other tests to do and data that we have to gather to confirm if any of these candidates are real exoplanets. If we’re unable to take radial velocity measurements then we can potentially use “multicolour photometry” to help towards validating the candidate. This involves checking whether the depth of the transit is the same when we observe the star with different filters on a telescope. These filters only let certain colours of light through, similar to how you’d mainly see pink if you wore Elton John’s famous tinted glasses. If there’s a difference in the depth then it suggests that there is a background eclipsing binary system that is mimicking the transit of an exoplanet. The difference in depth would be because stars have different colours depending on how hot they are, so if we see a shallower or deeper transit using a different filter it is because a background star isn’t as bright in that filter. For these four stars, getting radial velocity observations will be tough as they are very faint and would require lots of time on the world’s largest telescopes, but the first step is to see what the Zorro observations say. Once we can analyse and interpret the Zorro data, we will decide on the next steps.

 It’s very exciting to have candidates. Even if we can’t confirm these candidates as official planets, just finding these is an important step. We can still use these planet candidates to estimate the rate of exoplanets around the stars observed by NGTS. Thank you to everyone who has contributed to our project so far, whether it’s been through classifying light curves or getting involved with discussing potential candidates and weird subjects on the Talk boards. We couldn’t have done this without you. Also thank you to the extremely helpful team of instrument scientists at Gemini who helped us to setup our observations and the team at NASA for processing our data.

We also have many more subjects from the Exoplanet Transit Search still to sift through with the Secondary Eclipse and Odd/Even Transit checks. I performed an initial search, so there is much more I will be doing in terms of analysis of the classification data over the next many months. I am very hopeful that there will be even more candidates to find!  Stay tuned! We’ll keep everyone posted on the blog.

In this blog post, we explain the meaning of, and how to use, the different values and info given in the subject metadata on Planet Hunters NGTS.

On Planet Hunters NGTS, once you classify a subject you can choose ‘Done’ to move to the next subject and keep classifying, or you can choose ‘Done & Talk’ if you want to start a discussion with the Planet Hunters community about anything interesting you may have found. When you are looking at a subject image in the Talk discussion boards, you’ll be able to access the subject metadata by clicking the ‘i in a circle’ button. This metadata is additional info about the subject, some of which will be interesting to Zooniverse users wanting to delve deeper into their classifications.

Once you click the metadata icon, you’ll find a varying number of information fields, depending on whether the subject has linked images or is a known planet, but more on that later.

The first field is ‘sde’ which stands for ‘Signal Detection Efficiency.’ This value can be interpreted as how “strong” the signal is, according to the computer algorithm. The calculation of this value is described in Kovács et al. (2002) which describes the Box-fitting Least Squares (BLS) algorithm that forms the basis for the NGTS exoplanet transit searches.

As a demonstration, consider a full light curve (not phase-folded) such as the image below (which uses artificial data). I’ve injected a transit with a 3-day period into the light curve and we’re going to use a BLS algorithm to search for this.

I’ve used the AstroPy package’s BLS periodogram tool. This calculates the ‘power’ for a range of periods, this ‘power’ is equivalent to our SDE value mentioned above. The resulting ‘periodogram’ is plotted in the figure below. This shows the power value plotted against period with peaks in the periodogram corresponding to where the algorithm believes there is a strong periodic signal in the light curve.

As we can see, the algorithm has picked out the strongest signal at a period of 3 days, just as we expected as this is the true period of the signal I injected. We also see peaks (of decreasing power) around 1.5 days, 6 days, 9 days and 1 day. You may notice that these are all multiples or fractions of the true period. We refer to these as period aliases and it’s expected that they will produce peaks in the BLS periodogram. Sometimes these can even be the true period of a signal, which leads us to our next metadata field.

‘peak’ is the strength of the peak according to the BLS algorithm, ranked by SDE. 1 is the strongest peak (our 3-day period in the example above), down to 5 for the 5th strongest. Anything less significant than this doesn’t have a phase-folded image generated for Planet Hunters NGTS. Currently only Peak 1 objects are on the Planet Hunters NGTS site as these are most likely to correspond to the true period of a transiting object but we plan to include plots for other peaks in future.

‘prod_id’ and ‘obj_id’ are internal identifiers that are used as labels for the stars. These values therefore only have meaning to the NGTS science team.

‘plot_type’ simply refers to which workflow this subject is from:

• ‘primary_phased’ is for subjects in the Exoplanet Transit Search. These plots show where we think the primary eclipse of the phase-folded light curve is. See Figure 1.
• ‘secondary_phased’ is for subjects in the Secondary Eclipse Check. These plots show where we think the secondary eclipse could be, around phase=0.5. See Figure 5 below

• ‘odd_even’ is for subjects in the Odd Even Transit Check. These plots show the odd (1st, 3rd, 5th, etc.) transits in green and the even (2nd, 4th, 6th, etc.) transits in magenta. If the depths of the odd and even transits don’t match then that’s a clear indicator that we’re looking at an eclipsing binary rather than an exoplanet transit. See Figure 6 below for the Odd/Even image corresponding to the subject shown in Figure 1.

‘stellar_rad’ is the radius of the host star, expressed in units of Solar radii (i.e. a star with stellar_rad equal to 2.0 has a radius twice the size of our Sun). This can be used to estimate the radius of the transiting object. First, we estimate the depth of the transit. For the subject shown in Figure 1 (which is confirmed planet NGTS-5b), the transit depth is around 0.025 (or 2.5%), since the flux drops from 1.0 to 0.975 (typo corrected). We can then use this, along with the stellar radius of 0.75 Solar radii, in the equation:

to calculate an estimated planetary radius of 1.15 Jupiter radii. You can read more about how planet radius, stellar radius and depth are related here. The equation I use here is a quicker method as the multiplication factor of 9.73 is such that if you use R* in solar radii, as given in the metadata, then your answer will be in Jupiter radii. You can use this method to work out whether a transiting object is plausibly a planet, typically anything above 1.5 Jupiter radii is unlikely to be a planet.

(Note sometimes the stellar radius value will say ‘nan’ or ‘unavailable,’ this just means we don’t have a good measurement of the star’s radius.)

Extra fields

‘primary_phased,’ ‘secondary_phased’ and ‘odd_even’ provide links to the different plot types for subjects that have passed into the secondary and odd/even workflows. This makes it easier to vet candidates as you can view all the available data quickly and easily. 

Finally, we have the ‘known_planet’ field. This will only appear for previously known exoplanets (such as NGTS-5b above) and means that the subject image you are viewing corresponds to a planet that appears in the NASA Exoplanet Archive. Finding known exoplanets is a useful test of the detection efficiencies of the project as a whole, and it’s always exciting to know if you’re looking at data from a real exoplanet system too! Some subjects on the Planet Hunters NGTS site are still undergoing follow-up by the NGTS science team so might be strong planet candidates but won’t be confirmed and published exoplanets. This means that there will be subjects not marked as known planets in their metadata, but the NGTS team may already be aware of it and have put in the work to begin characterising the system further.

We hope this blog post can serve as a useful reference in future as you get involved with Planet Hunters NGTS.