Directed Follow-Up of Gaia Photometry in Search of Transiting Planets

Shay ZuckerYifat Dzigan

Tel-Aviv University

• Dzigan & Zucker, MNRAS, 415, 2513, 2011• Dzigan & Zucker, ApJL, 753, L1, 2012• Dzigan & Zucker,, MNRAS, accepted

Aposteriori Detection of HD209458b in Hipparcos Data

Söderhjelm 1999Robichon & Arenou 2000

Castellano et al. 2000

Why not a real detection?

• No one believed it was possible• Photometry not precise enough• Low cadence

Can we use the data for detection somehow?

Directed Follow-Up – General Idea

• Actually sampled transits provide some knowledge about the period, phase, duration and depth of a hypothetical transit, if it exists.

• Using MCMC we can present this scarce information as a Probability Distribution Function over (P,Tc,w,d) (assuming a BLS-like model).

• For each future time – t, we can use this PDF to calculate an instantaneous transit probability – ITP(t)

• Each new follow-up observataion is added and the process is repeated.

Could HD209458b have been detected in 2004?

Could HD209458b be detected in 2004?

Could HD209458b be detected in 2004?

Lessons from Hipparcos HD209458 Exercise

• With five sampled transits, low cadence data, augmented by DFU, allow detection.

• Data become useless after a few years.• For completeness – we should mention HD189733b.

Testing DFU on Gaia

• Simulated Gaia light curves inspired by known transiting planets (period and duration, coordinates)

• Gaia scanning law• Phase chosen to produce required number of

sampling transits• Photon noise level – 1 mmag.• We defined three scenarios of detection.

First Scenario – Gaia data alone

• Gaia samples enough transits to allow detection of the transit using only Gaia data.

• 1-mmag precision makes this scenario possible.• Simulation inspired by CoRoT-1b (P=1.51 d, w=0.1 d)

• Assuming five sampled transits (Ntot=64)• d > 0.005 mag

First Scenario – Gaia data alone

Second Scenario – one DFU observation

• Gaia samples enough transits to allow several possible solutions. One DFU observation is enough to constrain the period.

• Simulation inspired by CoRoT-4b (P=9.20 d, w=0.16 d)

• Assuming three sampled transits (Ntot=63)• d > 0.001 mag

Second Scenario – one DFU observation

Third Scenario – Few DFU Observations

• Gaia samples enough transits to allow several possible solutions. A few DFU observations are needed to constrain the period.

• Simulation inspired by WASP-4b (P=1.338 d, w=0.104 d)

• Assuming four sampled transits (Ntot=83)• d = 0.005 mag

Third Scenario – Few DFU Observations

Third Scenario – Few DFU Observations

Only Gaia data

Gaia data + 1 DFU observation

Gaia data + 2 DFU observations

We shouldn’t wait till the last minute

• ITP deteriorates over time• Therefore – we should start observing even before the

end of the mission• Tres-1b, (P=3.03 d, w=0.104 d)• Three sampled tranists (Ntot=48) (mid-life of mission)• Assume d=0.008 mag

Is it worth it?

Observational Window Function

70 Gaia Measurements 130 Gaia Measurements

Is it worth it?• Assuming 2 hr transit, Galactic model, transiting planet statistics from complete surveys (OGLE)

• Down to 14th G magnitude:minimum 7 transits: ~70 transiting HJs and VHJsminimum 5 transits: ~200 minimum 3 transits: ~600

• Down to 16th G magnitude:minimum 7 transits: ~300 transiting HJs and VJHsminimum 5 transits: ~900 minimum 3 transits: ~2600

To do list:• Get organized

– Dedicated observatory network?– CU7 follow-up network?– Science Alert team?

• Prescreening scheme (metallicity, brightness, activity etc.)• Develop a more efficient computational scheme than MCMC.• Objective criteria for applying DFU

– Wald statistics– ITP values– ITP skewness

• Smaller planets (Neptunian and below?)• Other low-cadence surveys

Additional slides Contingencies for potential questions

HD189733 – Hipparcos and DFU observations

HD189733 – Hipparcos and DFU observations

HD189733 – Hipparcos and DFU observations

HD189733 – Hipparcos and DFU observations

Sanity checks

Gaia WASP-4b – degradation of ITP over 10 years