# Why is the TESS spacecraft finding fewer exoplanets than expected?

Judging from the Wikipedia page, I got the impression that TESS is underperforming, being expected to detect about 20000 candidate exoplanets in its primary two-year mission, but with only a few weeks left, it is now clear that it won't come close to that figure. I am just curious to know why that is.

From Astrobio.net's April 2018 article The Astrobiology Magazine Guide to TESS:

“The number of Earth-sized and super-earth planets that TESS should find over the course of its two-year primary mission will be in the range of 500 to 1,000 new planets, and overall the number of planets that will be established is likely to be in excess of 20,000 all together,” enthuses MIT’s George Ricker, who is TESS’ Principal Investigator.

That is roughly 10 times greater than the updated total as of March 20, 2020 being in the 1700-1800 range according to NASA's status tracker.

• You can hardly say that the satellite is 'underperforming', when nature just doesn't put enough planets around red dwarves. If anything, it reflects our flawed understanding of planet formation and planet statistics that gave rise to the overly optimistic predictions. Also I remember more of a figure of 2000 predicted exoplanet yield, are you sure you didn't add a zero? Mar 20 '20 at 18:28
• @AtmosphericPrisonEscape I'm going by the figure in the wikipedia page which is 20,000 as of today at least. The source of that claim is this article, also claiming 20,000 in a quote attributed to a scientist in the TESS project. Mar 20 '20 at 19:46
• "it is now clear that it won't come close to that figure" is a premise of your question, and in Stack Exchange we should support our premises. I've taken the link from your comment, moved it to your question and quoted the relevant numbers. I wonder if "...is likely to be..." and "enthuses" in association with the missions PI are in any way suggestive of a potential answer?
– uhoh
Mar 21 '20 at 2:25

I am not using TESS to find planets, but I'm reasonably sure that the answer is that nobody has looked very carefully at all the data yet.

The TESS data comes in two flavours. There were a subset of about 200,000 stars that were observed with a 2 minute cadence. These targets were chosen to be a combination of bright and small, heavily weighted towards nearby K and M dwarfs - to optimise the chances of finding a few thousand small planets. The vast majority of the reported planet discoveries so far will have come from this list.

https://tess.mit.edu/science/data/

In addition, the most of the sky was mapped repeatedly such that there is "full frame images" of patches taken every 30 minutes over at least 27 days.

In principle, every star in the sky is present in the full frame data. In practice, only stars brighter than about 15th magnitude have reasonable data that might allow the detection of larger planets; the so-called hot Jupiters.

There are expected to be about 20 million stars that are bright enough to facilitate hunting for hot Jupiters. The majority of these will be solar-type stars or more luminous. The occurrence rate of such objects is known to be about 1%, suggesting 200,000 stars will have them. But then only a fraction of about 10% of those happen to transit in front of the star, which is where the estimate of 20,000 comes from.

Extracting and analysing 20 million light curves from the full frame images is not a straightforward task and it simply hasn't been completed yet.

EDIT:

Chapter and verse on this can be found in Barclay et al. (2018). They study the likely numbers of planets to be found by TESS. They split the study into three samples. The 2-minute cadence sample of 200,000 stars, a 3.2 million candidate target list (CTL), which are generally fainter (but still with $$V<13$$) main sequence stars of solar-type and cooler, where there is a good chance of finding small planets, then a final list of about 20 million main sequence stars with $$V<14.7$$ around which giant exoplanets might be found; 16 million of these have $$T_{\rm eff} >5500$$ K and only 4 million have $$T_{\rm eff}<5500$$ K.

The estimated yields from these three samples are 1250 in the 2 minute cadence sample; about 4400 in the CTL sample and of order 10,000 in the faint full field image sample. Basically in agreement with what I wrote above.

A further important point is that the large full field image sample of giant exolanets will be heavily contaminated by false positives (i.e. not planets). It is understandable therefore that most of the effort so far has been on investigating and verifying the smaller sample of planets found around brighter stars.

• Ah, if that estimate is coming from Hot Jupiters, then the flaw in the 20.000 is clear: 1% Hot Jupiters are known to exist around G-type stars (Mayor+2011), while TESS is mostly observing M-type stars, due to their higher occurrence rates (and the habitability hype). If that numbers is 1% around M-type stars is currently unknown (as you said in your answer) but also planet formation models would expect much less than that. Mar 21 '20 at 8:05
• @AtmosphericPrisonEscape your assumptions are flawed. TESS is observing ALL stars (above a brightness limit), as explained in my answer. If you take a magnitude limited sample, then a minority are M dwarfs. Mar 21 '20 at 8:50
• My assumptions hold, as most stars are redder than G. Mar 21 '20 at 9:11
• @AtmosphericPrisonEscape Your correct statement that most stars are redder than G is not relevant to what kinds of stars planets will be found around in the TESS data. The primary factor determining detectabilitiy is brightness. In a brightness limited survey, most of the stars are intrinsically more luminous, despite the preponderance of intrinsically less luminous stars. Of the 20M main sequence stars with $V<15$ about 80% are spectral type G or hotter. I have added an edit, which includes a paper where you can see these numbers. Mar 21 '20 at 10:13
• Thank you for the clarification! Mar 21 '20 at 10:47