Shallow Transits

What is the effect of spot-covering events on stellar transits? Can we use out-of-transit photometry to correct(?) them, or do we need other kind of observations?

Previous studies have been for deep transits. 

For shallow transits where pl is comparable to the size of the spot, get bumpy lightcurve. How can we pick this up with detection algos? How will we deal with changing spots?

Can we use oot observations to help? Not sure; problem is that we don’t know much about spot properties (lots of degeneracies). Can be put any number of spots at any latitude. Would ground-based phot in multiple colours help? Is that the only way? What facilities exist to do this? 

  • What abou NIR phot? From space ltd opportunities, but what about ground? E.g. GTC, E-ELT? Will they have the precision we need? Likely not, so space but what to use – JWST? Would not be detection, but could we sell follow-up of a really interesting system?
  • Why not spec from e.g. ELT? Spectro-photometry obs? RM effect? Long-period, so long-duration – would need coordinated effort. Handful of targets so could justify? Validate approach on shorter-period systems?
  • Spectro-pol? How could we constrain from large-scale B the spot pattern? Is it possible?
  • Chromospheric survey?

FGS cameras might be able to help. But only 2 cameras. Not clear that this will be available for stars. Could we select stars from the guiding stars?

Are our most important transits all going to be shallow? How often will this be a problem? Will the transit duration help?

Quiet stars being targeted by PLATO, so will that help? Remember that the Sun is quiet! For PLATO, will have to prioritise them in order of activity.

Relate problem to detection / filtering / characterisation.

  • Flattening done before or simultaneously?
  • Flattening messes with transit depth.
  • Perhaps do two stages. 1) flatten then search 2) model simultaneous with flatten for characterisation.

TTVs? TDVs? 

Fit in time domain, and don’t phase fold. Maybe after search.

Is this really going to be a problem we can deal with? If have fidelity to pick it up, it will look really interesting and someone will pick up on it. If we don’t, it won’t be detected. (e.g. Kepler didn’t find anything, so should we be concerned at all?)

  • Will TESS help us with activity level of targets? Probably (though indicators will be different to RVs), though not cycles.

Are we interested in actual spot distribution, or just a distribution that models the transit?

  • The latter probably, as unlikely to get ‘real’ distribution.
  • Break problem down.

How well could phase-folding get rid of activity for shallow transits (and eclipses)?

Has phase folding had a role to play in lack of detections?

Many different sub-classes of detections, from multiple sys to circumbinaries. Need a prioritised list of which to tackle, and of problems associated with each one.

Smaller planets are more perturbed than larger ones. If unlucky with perturber and non-transiting, then it’s a problem.

Problem for TTVs AND TDVs.

Do we need to do similar simulation and modelling testing to check phase-folding / TTVs for shallow transits with modern characterisation code? Probably!

Binned phase folded data is dangerous – could lose lots of information potentially because of changes in activity over time. Potentially ok in time-domain.

  • Don’t gain anything except computational speed by binning / folding

Should SNR_min in a transit search change if a high-SN planet is detected (in that system)?

At which stage of the detection? Would find more signals, but also more false positives.

What if high-SN ‘planet’ tuns out to be false positive?

Could add a flag for ‘lower threshold detection’ in candidate list.

Have column for ‘number of transits’ in candidate list.

Target the secondary search using info on period ratios?



Incomplete sample – we don’t know what we missed! Sop injection tests are very important.


Also need to correct for detection biases.


With PIC, we know which stars we’ll be observing. So can we include this info somehow? 



Is 25s cadence useful for shallow transits? Would we better with longer cadence?

  • Earth-Sun system ingress is 7 minutes. If want 10 points to full characterise it, that’s 0.7min=42s.
  • Lose info if go longer.
  • Can always bin.


How will P5 deal with shallow transits with 600s cadence.



If have spots, most favourable case to characterise stellar variability because can use GPs.

  • Can’t use oot info from GPs to inform in-transit modelling.


Smaller the number of transits the worse the problem of spot occultations is. If miss a transit, can get wrong period.



Separate detection and characterisation.

Separate systems into ‘normal shallow’ i.e strictly periodic, & harder planets.


Problem of shallow transits is not the prime sample of PLATO, as won’t get the precision that we want.


Define our terms – what is a ‘shallow’ transit?

  • Anything that PLATO can’t get 3% radius precision because of depth (maybe 5%)
    • Better to do Rp/Rs because then don’t have effect from the star uncertainty.
    • ⇒ SNR>=10 (in theory) gives 100% completeness, but recent paper gives 90-95%

[at SNR=7, get 50% completeness]

  • V>10
  • G0 type
  • Define relative to noise lt? Why not an SNR definition? Or do it based on completeness?


Shallow mono-transits are the worst case. May have TESS data in some examples, but not all.


Shallow transits have fewer false positives, because fewer astrophysical mimics.




Different path in pipeline for active / inactive stars – do different detection, maybe not GPs if inactive.



When you detrend, how much can you affect the phase-curve?

Quite a lot.

Many independent data extraction tools have been developed for TESS; need to do systematic study to see which one is most suitable.


Time to start some H&H exercises.


WOTAN paper by Hippke+ (ref.)



Special case:

Detect two transits of a shallow, long period planet and predict 3rd. Apply Doppler-imaging if the star is appropriate. For this technique see:…321..496V