Expected precision on planet radii with

Transcription

1 Expected precision on planet radii with Adrien Deline 1, Didier Queloz 1,2 1 University of Geneva 2 University of Cambridge 24 th 26 th July 2017 CHEOPS Science Workshop 5 Schloss Seggau, Austria

2 Field of View 200 x200 (1 = 1 pixel) defocused PSF (Ø PSF 20 ) specific shape

3 Field of View 200 x200 (1 = 1 pixel) defocused PSF (Ø PSF 20 ) specific shape (extended halo) 3

4 Field of View 200 x200 (1 = 1 pixel) defocused PSF (Ø PSF 20 ) specific shape (extended halo) background stars field of view rotation (PSF not rotating!) jitter (about 1.32 rms) 4

5 CHEOPSim Developed by the Science Operations Center (SOC) main developer : David Futyan (david.futyan@unige.ch) 5

6 Photometric extraction tool Developed independently from the Data Reduction Pipeline (DRP) Purposes : Testing output data from CHEOPSim feedback during development Processing data from calibration campaign DRP not delivered specific field of view (on-ground calibration tracking system) Cross-checking with the Data Reduction Pipeline 6

7 Photometric extraction tool Raw data Corrected images Light curve Bias & RON estimation Background subtraction Detrending + fitting ADU to e conversion Dark current estimation PSF center estimation Aperture photometry Transit parameters Flat field correction 7

8 Photometric extraction tool σ 6h = 9.43 ppm Raw data Corrected images Light curve Bias & RON estimation ± 0.5 ADU Background subtraction ± 3 e Detrending + fitting ADU/e ADU to e conversion Dark current estimation ± e /s PSF center estimation Aperture photometry ± pixel Transit parameters Flat field correction Typical precision for - G5 star, m V = 9-6x 10s-exposures - nominal jitter - medium background 8

9 Main effects of new PSF PSF center estimation new method necessary : IWCOG (iterative weighted center of gravity) high CPU usage (not on-board -compatible) precision better than pixel rms (measured on datasets varying background and jitter) Background noise extended halo (more energy in the background) Original simulated PSF 9

10 Photometric precision (case 1) Target star G5, m V =9 K5, m V =12 Exposure time 6x 10s 1x 60s Jitter 2x nominal 2x nominal Thermal variations no no Background none none Flicker no no σ 6h = 8.95 ppm σ 6h = ppm 10

11 Photometric precision (case 2) Target star G5, m V =9 K5, m V =12 Exposure time 6x 10s 1x 60s Jitter nominal nominal Thermal variations yes yes Background medium medium Flicker no no σ 6h = 9.43 ppm σ 6h = ppm 11

12 Photometric precision (case 3) Target star G5, m V =9 K5, m V =12 Exposure time 6x 10s 1x 60s Jitter nominal nominal Thermal variations yes yes Background none none Flicker yes yes σ 6h = ppm σ 6h = ppm 12

13 Light curve fitting Example of a super-earth (2R earth ) transiting a G5 star (m V =9) Transit model batman (Kreidberg 2015b) used by CHEOPSim Gaussian processes george (Ambikasaran 2014a) to simultaneously fit correlated noise and transit Markov Chain Monte Carlo (MCMC) emcee (Foreman-Mackey 2013) to explore the parameter space 13

14 Details on Gaussian processes Kernels Two kernels considered for now (radial and periodic) Results presented for periodic kernel! x #, x % = ' ( exp 1 2. ( sin( 2 3 x # x % ' : amplitude scale. : roughness parameter T : oscillation period Transit fit Parameter Prior Orbital period Gaussian Eccentricity Fixed to 0 Limb-darkening coefficients Uniform (Kipping 2013) Semi-major axis Uniform (tight) Others Uniform 14

15 Precision on planet radii Target G5, mv=9 Duration 64 hours Exposures per image 6x 10 seconds Jitter none Background none Flicker no Planet radius 2 R Orbital period 10 days (1 transit) Impact parameter 0.2 (i = 89.4 ) RP truth ε ε = 2.6 % 3σ = 5.8 % 3σ Time [min] 15

16 Precision on planet radii Target K5, mv=12 Duration 48 hours Exposures per image 1x 60 seconds Jitter none Background none Flicker no Planet radius 1R Orbital period 14 hours (3 transits) Impact parameter 0.2 (i = 86.9 ) RP truth ε ε = -1.1 % 3σ = 4.0 % 3σ Time [min] 16

17 Precision on planet radii Cases - G5, m V =9, R P =2R - K5, m V =12, R P =R 7 scenarios increasing jitter intensity, background level or stellar noise Overall results - ε max = 2.6 % - 3σ max = 6.2 % Target G5, m V =9 K5, m V =12 Duration 64 hours 48 hours Exposures per image 6x 10 seconds 1x 60 seconds Jitter none none Background none none Flicker no no Planet radius 2 R 1 R Orbital period 10 days (1 transit) 14 hours (3 transits) Impact parameter 0.2 (i = 89.4 ) 0.2 (i = 86.9 ) Scenarios ε 3σ ε 3σ default 2.6 % 5.8 % -1.1 % 4.0 % Precision not affected significantly Variations of less than 1% Error varies randomly but always below 1.5 σ jitter x1 1.9 % 6.0 % 0.2 % 3.9 % jitter x2-0.6 % 5.6 % -1.7 % 4.2 % jitter x3 0.2 % 5.2 % -2.3 % 4.6 % bkg low -1.2 % 5.4 % 2.0 % 3.9 % bkg medium -1.0 % 5.3 % -1.2 % 4.6 % flicker 1.2 % 6.2 % 1.9 % 4.2 % 17

18 Precision on planet radii Number of data points in the light curve Single-transit case (G5 scenario) Reduce the number of out-of-transit points No significant effect Multiple-transit case (K5 scenario) Reduce the number of transits Significant effect on the precision South Atlantic Anomaly interruptions The SAA causes scattered interruptions that represent about 9% of the time. Significant effect, especially in the single-transit case Overall results : 3σ max = 8.3 % Duration ε Δt G5 = 3.9 h Δt K5 = 74 min G5, m V =9 3σ 64 hours (default) 1.9 % 6.0 % 42.7 hours (x2/3) 0.6 % 5.2 % 15.5 hours (4 Δt G5 ) 0.4 % 6.3 % SAA interruptions 0.5 % 8.3 % Duration ε K5, m V =12 3σ 48 hours (default) 0.2 % 3.9 % 32 hours (x2/3) * 1.0 % 4.8 % 5 hours (4 Δt K5 ) ** -0.1 % 7.0 % SAA interruptions 1.2 % 4.7 % * 2 transits ** 1 transit 18

19 Summary Precision on planet radii : - better than 10 % - error lower than 1.5 σ Jitter, background and stellar flicker seems to have limited effect on the radii precision Duration of the time series affects the precision mainly if the number of transits is reduced The SAA interruptions lower the precision, in particular in the single-transit case 19

20 Thank you! Questions?