Physical properties, starspot activity, orbital obliquity, and transmission spectrum of the Qatar-2 planetary system from multi-colour photometry
L. Mancini, J. Southworth, S. Ciceri, J. Tregloan-Reed, I. Crossfield, N. Nikolov, I. Bruni, R. Zambelli, Th. Henning
aa r X i v : . [ a s t r o - ph . E P ] J un Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 15 August 2018 (MN L A TEX style file v2.2)
Physical properties, starspot activity, orbital obliquity, andtransmission spectrum of the Qatar-2 planetary systemfrom multi-colour photometry ⋆ L. Mancini † , J. Southworth , S. Ciceri , J. Tregloan-Reed , I. Crossfield , N.Nikolov , I. Bruni R. Zambelli and Th. Henning Max Planck Institute for Astronomy, K¨onigstuhl 17, 69117 Heidelberg, Germany Astrophysics Group, Keele University, Staffordshire, ST5 5BG, UK Astrophysics Group, University of Exeter, Stocker Road, EX4 4QL, Exeter, UK INAF – Osservatorio Astronomico di Bologna, Via Ranzani 1, 40127 – Bologna, Italy Societ`a Astronomica Lunae, 19030 Castelnuovo Magra (La Spezia), Italy
15 August 2018
ABSTRACT
We present seventeen high-precision light curves of five transits of the planet Qatar-2 b,obtained from four defocussed 2m-class telescopes. Three of the transits were observedsimultaneously in the SDSS g ′ r ′ i ′ z ′ passbands using the seven-beam GROND imageron the MPG/ESO 2.2-m telescope. A fourth was observed simultaneously in Gunn grz using the CAHA 2.2-m telescope with BUSCA, and in r using the Cassini 1.52-mtelescope. Every light curve shows small anomalies due to the passage of the planetaryshadow over a cool spot on the surface of the host star. We fit the light curves withthe prism+gemc model to obtain the photometric parameters of the system andthe position, size and contrast of each spot. We use these photometric parametersand published spectroscopic measurements to obtain the physical properties of thesystem to high precision, finding a larger radius and lower density for both star andplanet than previously thought. By tracking the change in position of one starspotbetween two transit observations we measure the orbital obliquity of Qatar-2 b to be λ = 4 . ◦ ± . ◦ , strongly indicating an alignment of the stellar spin with the orbit ofthe planet. We calculate the rotation period and velocity of the cool host star to be11 . ± . . ± .
13 km s − at a colatitude of 74 ◦ . We assemble the planet’stransmission spectrum over the 386–976 nm wavelength range and search for variationsof the measured radius of Qatar-2 b as a function of wavelength. Our analysis highlightsa possible H /He Rayleigh scattering in the blue. Key words: stars: planetary systems — stars: fundamental parameters — stars:individual: Qatar-2 — techniques: photometric
Transiting extrasolar planets (TEPs) are the most interest-ing exoplanets to study as it is possible to deduce theirphysical properties to high precision. High-quality photo-metric observations of TEPs are a vital component of suchwork, as they strongly constrain the density of the host star(Seager & Mall´en-Ornelas 2003). They also allow searches ⋆ Based on data collected with GROND at the MPG/ESO 2.2-mtelescope, BUSCA at the CAHA 2.2-m telescope, BFOSC at theCassini 1.52-m telescope, and DLR-MKIII camera at the CAHA1.23-m telescope. † E-mail: [email protected] for transit timing variations (TTVs; e.g. Holman et al.2010), which can be used to measure the masses of the tran-siting planets or show the presence of non-transiting objects(Nesvorn´y et al. 2013), and for variations of the planetaryradius with wavelength which trace opacity variations in theplanet’s atmosphere.Since 2008 we have been photometrically following upknown TEP systems from both hemispheres. The aim ofthis project is to obtain high-precision differential photom-etry of complete transit events, which can be used to refinethe measured physical properties of the planets and parentstars (e.g. Southworth et al. 2010, 2011, 2012a,b,c, 2013),search for opacity-induced planetary radius variations (e.g.Mancini et al. 2013a,b,c, 2014; Nikolov et al. 2013), and c (cid:13) L. Mancini et al. investigate starspot crossing events (Mohler-Fischer et al.2013; Ciceri et al. 2013). Our observations are performed us-ing medium-class defocussed telescopes, some of which areequipped with multi-band imaging instruments.In this work we present extensive new follow-up pho-tometry of Qatar-2, the second planetary system discov-ered by the Qatar Exoplanet Survey (QES) (Bryan et al.2012). This system comprises Qatar-2 A, a moderatelybright ( V = 13 . M Jup planet on a 1.34 d period . The late spectraltype of the host star means that the transits due to Qatar-2 b are deep and may contain starspot crossing events (e.g.Sanchis-Ojeda et al. 2011; Sanchis-Ojeda & Winn 2011;Tregloan-Reed et al. 2013).We report observations of three transits simultaneouslyobserved in four optical passbands using the “Gamma RayBurst Optical and Near-Infrared Detector” (GROND) at theMPG/ESO 2.2-m telescope, one transit simultaneously ob-served in three optical passbands with the “Bonn Univer-sity Simultaneous CAmera” (BUSCA) at the CAHA 2.2-mtelescope, one transit with the Cassini 1.5-m telescope, onetransit with the CAHA 1.23-m telescope and three furthertransits observed with a 25-cm telescope. We use these newlight curves to refine the physical properties of the systemand attempt to probe the atmospheric composition of Qatar-2 b at optical wavelengths (386–976 nm). In this section we describe the methodologies used to ob-tain accurate photometric observations of transiting-planetevents and get reliable physical information on the planetarysystem.
All the observations presented in this work were performedusing the telescope-defocussing technique (Alonso et al.2008; Southworth et al. 2009). In this method the telescopeis defocussed so point spread functions (PSFs) cover of order1000 pixels, and long exposure times (up to ∼
120 s) are usedto collect many photons in each PSF. This increases the ob-servational efficiency as the CCD is read out less often, thusminimising Poisson and scintillation noise. The large PSFsare also insensitive to focus or seeing changes, which mightotherwise cause systematic errors. The other main source ofsystematic error, flat-fielding, is decreased by two orders ofmagnitude as each PSF covers of order 10 pixels. Telescopepointing errors affect photometry via flat-fielding errors, sothese also average down to very low levels.The exposure time is chosen for a given observing se-quence from consideration of the brightness of the target The discovery paper also reported the possible existence of asecond planet, Qatar-2 c, in a ∼ and comparison stars, sky background, telescope size andfilter used. The amount of defocussing is then tuned so thepeak count rate in the PSFs of the target and comparisonstars is significantly below the onset of nonlinearity effects inthe CCD. Changes in seeing, airmass and sky transparencyaffect the count rate of the observations; this is accountedfor by changing the exposure times but not the focus settingduring an observing sequence. Time-series photometry of transit events can show anoma-lies due to the planet crossing over spots on the stellarsurface. The detection of starspots occulted by a tran-siting planet is becoming commonplace (e.g. Pont et al.2007; Rabus et al. 2009; Silva-Valio et al. 2010; D´esert2011; Sanchis-Ojeda et al. 2011; Sanchis-Ojeda & Winn2011; Tregloan-Reed et al. 2013; Mohler-Fischer et al. 2013;Mancini et al. 2013c). However, in the case of ground-basedobservations, similar signals could be caused by weather-related or instrumental effects. One method to sift the astro-physical from observational anomalies is to observe a tran-sit event from multiple telescopes at different observatories.Any feature present in all light curves is unambiguously in-trinsic to the target of the observations. We used this strat-egy to observe a transit of Qatar-2 using two telescopes atdifferent locations.This two-site observational strategy was successfullytested in the follow-up of HAT-P-8, where an anomaly wasdetected in both the light curves (Mancini et al. 2013a). Itwas also used for HAT-P-16 and WASP-21 (Ciceri et al.2013), although in these two cases no anomalies were de-tected. Conversely, Lendl et al. (2013) used this method toshow that a possible starspot anomaly in the WASP-19 sys-tem was of instrumental origin.
Precise photometric observations of planetary transits probethe chemical composition of the atmosphere of TEPs in away similar to transmission spectroscopy. A dependence ofopacity on wavelength causes variations in the radius of theplanet as found from transit observations. The effect can bebig enough to measure using medium-size telescopes withmulti-band imagers, assuming they have a good spectral res-olution. It is important to obtain the observations at multi-ple wavelengths simultaneously, to avoid variations in transitdepth due to unseen starspots rather than planetary radiusvariations, even if one should take into account that unoc-culted star spots may still cause wavelength dependence ofthe transit depth (Sing et al. 2011). In order to investigatethis effect, one should monitor the variability of the par-ent star for many years or, assuming that stellar activitydoes not change suddenly, repeatedly measure the transitdepth by observing several planetary-transit events a fewdays away from each other. As an example, in the case of the K-dwarf HD 189733 A,Pont et al. (2013) estimated 0 .
3% at 8 µ m (and then scaled atother wavelengths) as an additional uncertainty in the depth mea-surement of individual transits due to unidentified spot crossings.c (cid:13) , 000–000 hysical properties of Qatar-2b òòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòò òòòòòòòòòòòòòòòòòòòòòòòòòòò ææææææææææææææææææææææææææææææææææ æææææææææææææææææææææææææææàààààààààààààààààààààààààààààààààà ààààààààààààààààààààààààààà ôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôô ôôôôôôôôôôôôôôôôôôôôôôôôôôô N o r m a li s e d F l ux (cid:144) (cid:144) òæàô Sloan g ¢ Sloan r ¢ Sloan i ¢ Sloan z ¢ òòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòò æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààà ôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôô N o r m a li s e d F l ux (cid:144) (cid:144) òæàô Sloan g ¢ Sloan r ¢ Sloan i ¢ Sloan z ¢ òòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòò òòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòò æææææææææææææææææææææææææææææææææææææææææææææææææææææææ æææææææææææææææææææææææææææææææææææææææææææææææààààààààààààààààààààààààààààààààààààààààààààààààààààààà ààààààààààààààààààààààààààààààààààààààààààààààà ôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôô ôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôô BJD H TDB L - N o r m a li s e d F l ux (cid:144) (cid:144) òæàô Sloan g ¢ Sloan r ¢ Sloan i ¢ Sloan z ¢ Figure 1.
Light curves of three transits of Qatar-2 b observed simultaneously in four optical bands with GROND, ordered according todate.
Simultaneous multi-band observations also allow a de-tailed study of starspots which are occulted by the transitingplanet. For a single light curve the spot radius is stronglycorrelated with its temperature (e.g. Tregloan-Reed et al.2013). Multi-band light curves constrain the spot temper-ature relative to the effective temperature ( T eff ) of thepristine photosphere, thus providing additional informationwhich lifts this degeneracy.Simultaneous multi-band observations of plan-etary transits have been obtained for several TEP systems using the instruments BUSCA(Southworth et al. 2012b; Mancini et al. 2013a),GROND (de Mooij et al. 2012; Mancini et al. 2013b,c;Nikolov et al. 2013; Southworth et al. 2013; Penev et al.2013; Mohler-Fischer et al. 2013; Bayliss et al. 2013),ULTRACAM (Copperwheat et al. 2013; Bento et al. 2013)and SIRIUS (Narita et al. 2013). c (cid:13) , 000–000 L. Mancini et al. àààààààààààààààààààààààààààààààààà ààààààààààààààààààààààààààà æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ççççççççççççççççççççççççççççççççççççççççççççççççççççççç çççççççççççççççççççççççççççççççççççççççççç - - N o r m a li s e d F l ux GROND g ¢ filter à (cid:144) (cid:144) æ (cid:144) (cid:144) ç (cid:144) (cid:144) àààààààààààààààààààààààààààààààààà ààààààààààààààààààààààààààà æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ççççççççççççççççççççççççççççççççççççççççççççççççççççççç ççççççççççççççççççççççççççççççççççççççççççççççç - - N o r m a li s e d F l ux GROND r ¢ filter à (cid:144) (cid:144) æ (cid:144) (cid:144) ç (cid:144) (cid:144) àààààààààààààààààààààààààààààààààà ààààààààààààààààààààààààààà æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ççççççççççççççççççççççççççççççççççççççççççççççççççççççç ççççççççççççççççççççççççççççççççççççççççççççççç - - N o r m a li s e d F l ux GROND i ¢ filter à (cid:144) (cid:144) æ (cid:144) (cid:144) ç (cid:144) (cid:144) àààààààààààààààààààààààààààààààààà ààààààààààààààààààààààààààà æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ççççççççççççççççççççççççççççççççççççççççççççççççççççççç ççççççççççççççççççççççççççççççççççççççççççççççç - - Orbital phase N o r m a li s e d F l ux GROND z ¢ filter à (cid:144) (cid:144) æ (cid:144) (cid:144) ç (cid:144) (cid:144) Figure 2.
Phased light curves of the three transits of Qatar-2 b observed with GROND, ordered according to the filter used. Thishighlights the anomalies between the three transits in each colour. c (cid:13) , 000–000 hysical properties of Qatar-2b òòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòòò ææææææææææææææææææææææææææ æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ôôôô ôôôô ôôôôôôôôôôôôôôôôôô ôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôôô N o r m a li s e d F l ux (cid:144) (cid:144) òæô Gunn g Gunn r Gunn z ì ì ììììììììì ìììììì ìììììì ìììììììììì ì ììììì ìììììììììì ìììììììììì ììììì N o r m a li s e d F l ux ì Gunn g (cid:144) (cid:144) ææææææææææææææææææææææææææ æææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææææ ì ì ììììììììì ìììììì ìììììì ìììììììììì ì ììììì ìììììììììì ìììììììììì ììììì N o r m a li s e d F l ux (cid:144) (cid:144) ìæ Cassini Gunn r CAHA 2.2 m Gunn r CAHA 2.2 m + Cassini 1.52 m à ààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààààà
BJD H TDB L - N o r m a li s e d F l ux à Cousins I (cid:144) (cid:144) - MKIIICAHA 1.23 m
Figure 3.
Light curves of two transits of Qatar-2 b observed with three telescopes, shown in date order.
Top panel : light curves of onetransit of Qatar-2 b observed simultaneously in three optical bands with BUSCA at the CAHA 2.2 m telescope. The anomalies on thelight curves are interpreted as the occultation of a starspot by the planet.
Second panel : light curve obtained with the Cassini 1.52 mtelescope using a Gunn r filter. This is the same transit observed with BUSCA. Third panel : CAHA 2.2 m and Cassini 1.5 m Gunn- r light curves. Bottom panel : light curve obtained with the CAHA 1.23 m telescope using a Cousin I filter.c (cid:13) , 000–000 L. Mancini et al.
Table 1.
Details of the transit observations presented in this work. N obs is the number of observations, T exp is the exposure time, T obs is the observational cadence, and ‘Moon illum.’ is the fractional illumination of the Moon at the midpoint of the transit. The aperturesizes are the radii of the software apertures for the star, inner sky and outer sky, respectively. Scatter is the r.m.s. scatter of the dataversus a fitted model. β is the ratio between the noise levels due to Poisson noise and to combined Poisson and red noise. Telescope Date of Start time End time N obs T exp T obs Filter Airmass Moon Aperture Scatter β first obs. (UT) (UT) (s) (s) illum. radii (px) (mmag)ESO 2.2 m g ′ . → . → .
14 83% 23, 45, 75 1.22 1.26ESO 2.2 m r ′ . → . → .
14 83% 29, 55, 85 0.78 1.00ESO 2.2 m i ′ . → . → .
14 83% 29, 55, 85 0.88 1.06ESO 2.2 m z ′ . → . → .
14 83% 30, 55, 85 0.99 1.15ESO 2.2 m g ′ . → . → .
70 9% 29, 55, 85 0.96 1.00ESO 2.2 m r ′ . → . → .
70 9% 34, 55, 85 0.74 1.42ESO 2.2 m i ′ . → . → .
70 9% 33, 55, 85 0.84 1.55ESO 2.2 m z ′ . → . → .
70 9% 30, 55, 85 1.01 1.46ESO 2.2 m g ′ . → .
46 1% 28, 55, 85 3.26 1.60ESO 2.2 m r ′ . → .
46 1% 35, 60, 90 0.77 1.40ESO 2.2 m i ′ . → .
46 1% 32, 60, 90 0.93 1.32ESO 2.2 m z ′ . → .
46 1% 32, 60, 90 1.02 1.12CAHA 2.2 m 2012 05 09 21:25 01:43 89 120 200 Gunn g . → . → .
15 79% 15, 55, 80 1.65 1.18CAHA 2.2 m 2012 05 09 21:25 01:43 89 120 200 Gunn r . → . → .
15 79% 15, 50, 80 0.92 1.17CAHA 2.2 m 2012 05 09 21:25 01:43 89 120 200 Gunn z . → . → .
15 79% 20, 40, 55 1.70 1.01Cassini 1.52 m 2012 05 09 21:01 00:28 64 170 180 Gunn r . → . → .
11 80% 28, 55, 85 0.80 1.12CAHA 1.23 m 2013 02 16 01:53 05:54 128 140 180 Cousin I . → . → .
55 33% 20, 45, 60 1.36 1.26 æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ N o r m a li s e d F l ux (cid:144) (cid:144) æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ N o r m a li s e d F l ux (cid:144) (cid:144) æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ BJD H TDB L - N o r m a li s e d F l ux (cid:144) (cid:144) Figure 4.
Three transit events of Qatar-2 b observed in 2011 witha 25-cm MEADE LX200 telescope at the Canis Mayor observa-tory.
Five transits of Qatar-2 b were monitored at optical wave-lengths by five different telescopes in 2012 and 2013 (Ta-ble 1). One transit was followed simultaneously by two of thetelescopes, one was simultaneously observed through three filters, and the other three through four filters. All obser-vations were performed with autoguiding and defocussing.The light curves are given in Table 2 and shown in Figs. 1,2 and 3. We obtained observations of three more transits in2011 with a 25 cm telescope (Fig. 4).All observations were analysed with the defot pipeline(Southworth et al. 2009) written in idl . Debiasing and flat-fielding were done using master calibration frames obtainedby median-combining individual calibration images. Point-ing variations were corrected by cross-correlating each imageagainst a reference frame. No de-correlation with PSF loca-tion was necessary because, due to defocussing, the PSFsare much bigger than the pixel sizes of the CCDs and imagemotion as the telescopes tracked. Apertures were placed byhand on the target and comparison stars, and their radiiwere chosen based on the lowest scatter achieved when com-pared with a fitted model.The aper routine used to measure the differential pho-tometry commonly returns underestimated errorbars. Wetherefore enlarged the errorbars for each light curve to give areduced χ of χ ν = 1 versus a fitted model. We then furtherinflated the errorbars using the β approach (e.g. Gillon et al.2006; Winn et al. 2008; Gibson et al. 2008) to account forany correlated noise. We calculated β values for betweentwo and ten data points for each light curve, and adoptedthe largest β value. They are reported in Table 1.The twelve GROND light curves are plotted accordingto date (Fig. 1) and filter (Fig. 2) in order to highlight thestarspot anomalies found in each transit. Three transits of Qatar-2 b were monitored with theGROND instrument mounted on the MPG /ESO 2.2 m tele-scope at the ESO observatory in La Silla, Chile. The transitevents were observed on 2012 April 2, 17 and 21. GROND The acronym idl stands for Interactive Data Language and isa trademark of ITT Visual Information Solutions. aper is part of the astrolib subroutine library distributed byNASA. Max Planck Gesellschaft. c (cid:13) , 000–000 hysical properties of Qatar-2b Table 2.
Excerpts of the light curves of Qatar-2: this table will bemade available at the CDS. A portion is shown here for guidanceregarding its form and content.
Telescope Filter BJD(TDB) Diff. mag. UncertaintyESO 2.2 m g ′ g ′ r ′ r ′ i ′ i ′ z z r r I I is an imaging system capable of simultaneous photometricobservations in four fixed optical (similar to Sloan g ′ , r ′ , i ′ , z ′ ) and three fixed NIR ( J, H, K ) passbands (Greiner et al.2008). Each of the four optical channels is equipped with aback-illuminated 2048 × . ′ × . ′ at a scale of 0 . ′′ / pixel. The three NIRchannels use 1024 × ′ × ′ at 0 . ′′ / pixel. Unfortunately, due to alack of good reference stars in the FOV, we were not able toobtain usable light curves in the three NIR bands.The precision of the optical data are in agreement withthe statistical-uncertainty study performed by Pierini et al.(2012) with two exceptions. The z ′ transit observed on 2012April 2 has a lower depth compared to the other light curvesof the same transit (upper panel of Fig. 1) or the z ′ lightcurves of the other two transits (bottom panel of Fig. 2).This was caused by an unknown instrumental error, whichcould not be reliably corrected for during the data reduction.The g ′ data observed on 2012 April 21 were affected by ex-cess readout noise, caused by another unknown instrumentalproblem, so are very inaccurate compared to the other threelight curves of this transit (bottom panel of Fig. 1) or tothe g ′ light curves of the other two transits (upper panel ofFig. 2). We observed one full transit on the night of 2012 May 9,using the CAHA 2.2-m telescope and BUSCA imager atthe German-Spanish Astronomical Center at Calar Alto inSpain. BUSCA is designed for simultaneous four-colour pho-tometry and, unlike GROND, the user has a choice of fil-ters available for each arm. Each of the four optical chan-nels is equipped with a Loral CCD4855 4k ×
4k CCD with15 × µ m pixels, providing an astronomical FOV of nearly12 ×
12 arcmin.For our observations we selected a Str¨omgren u filter inthe bluest arm and standard Calar Alto Gunn g , r and z filters in the other three arms. This choice led to a reducedfield of view (from 12 ′ × ′ to a circle of 6 ′ in diameter), buthad two advantages. Firstly the g and r filters have a muchbetter throughput compared to the default Str¨omgren b and y filters. Secondly, the different filter thicknesses meant the u band, where the target star is comparatively faint, wasless defocussed. The CCDs were binned 2 × u band were too strongly affected by atmospheric extinction and poor signal-to-noise ratio to be useful. The g , r and z light curves areplotted in Fig. 3. The transit event of 2012 May 9 was also observed withthe BFOSC (Bologna Faint Object Spectrograph & Cam-era) imager mounted on the 1.52 m Cassini Telescope at theAstronomical Observatory of Bologna in Loiano, Italy. Thetransit was not fully covered due to the pointing limits ofthe telescope. The CCD was used unbinned, giving a platescale of 0 . ′′ / pixel, for a total FOV of 13 ′ × . ′ . A Gunn r filter was used. The CCD was windowed to decrease thereadout time and the telescope was autoguided and defo-cussed, allowing low scatter to be obtained even though theobservations were conducted at high airmass (see Table 1).The light curve is plotted in Fig. 3. The Cassini data areconsistent with the presence of the starspot anomaly in thefinal phase of the transit ingress. Another complete transit event was observed with theCAHA 1.23 m telescope, on the night of 2013 February 16.Mounted in the Cassergrain focus of this telescope is theDLR-MKIII camera, which has 4000 × − and a large FOV of 21 . ′ × . ′ .The transit was monitored through a Cousins- I filter, thetelescope was autoguided and defocussed, and the CCD waswindowed. The resulting light curve is plotted in Fig. 3. Ananomaly is also visible in this light curve, shortly after thetransit midpoint. Three complete transits were observed at the Canis MayorObservatory, located in Castelnuovo Magra, Italy. The in-strument used for the observations was a Meade LX200 GPS10 inch telescope, equipped with an f/6.3 focal reducer andan SBIG ST8 XME CCD camera. Science frames were takenthrough the Baader Yellow 495 Longpass filter and the ex-posure time was 300 s. The telescope was autoguided andslightly defocussed. The light curves are plotted in Fig. 4.
All of our high-precision light curves show possible starspotcrossing events, which must be analysed using a self-consistent and physically realistic model. We use the prism and gemc codes (Tregloan-Reed et al. 2013) for this.We have previously used these codes to model HATS-2(Mohler-Fischer et al. 2013) and WASP-19 (Mancini et al.2013c). prism models planetary transits with starspot cross-ings using a pixellation approach in Cartesian coordinates. gemc uses a Differential Evolution Markov Chain Monte Planetary Retrospective Integrated Star-spot Model. Genetic Evolution Markov Chain.c (cid:13) , 000–000
L. Mancini et al.
Carlo (DE-MCMC) approach to locate the parameters of the prism model which best fit the data, using a global search. prism uses the fractional radii, r A = R A a and r b = R b a ,where R A and R b are the true radii of the star and planet,and a is the orbital semimajor axis.The fitted parameters of prism are the sum and ratioof the fractional radii ( r A + r b and k = r b r A ), the orbital pe-riod and inclination ( P and i ), the time of transit midpoint( T ) and the two coefficients of the quadratic limb darkening(LD) law ( u A and v A ). Each starspot is represented by thelongitude and colatitude of its centre ( θ and φ ), its angularradius ( r spot ) and its contrast ( ρ spot ), the latter being theratio of the surface brightness of the starspot to that of thesurrounding photosphere.The datasets obtained with the 25-cm telescope weremodelled using the much faster jktebop code, as nostarspot anomalies are visible. The parameters used for jk-tebop were the same as for prism . We used the photometric data presented in Sect. 3 to re-fine the orbital period of Qatar-2 b. We excluded the third g ′ -band transit observed with GROND due to the large scat-ter of these data. The transit times and uncertainties for thehigh-precision datasets were obtained using prism + gemc .Those for the small telescope were calculated using jktebop and Monte Carlo simulations. To these timings we added onefrom the discovery paper (Bryan et al. 2012) and 14 mea-sured by amateur astronomers and available on the ETD website. The ETD light curves were included only if theyhad complete coverage of the transit and a Data Quality in-dex
3. All 34 timings were placed on the BJD(TDB) timesystem (Table 3).Possible unknown planets in the system could gravita-tionally perturb the orbit of Qatar-2 b and induce transittiming variations (TTVs). We therefore searched for pe-riodic variations in the transit times that might indicatesuch perturbations. We first performed a weighted linearleast-squares fit to compute a new system ephemeris of T T = T + P × E , where E is the number of orbital cy-cles after the reference epoch and T = BJD(TDB)2 455 624 . ± . ,P = 1 . ± . − . × − d . A plot of the residuals around the fit is shown inFig. 5. We then look for sinusoidal variations in the residualsby scanning through a wide range of periods (10–1000 orbitsof Qatar-2 b) and looking for the best-fitting sinusoid at eachperiod. Across this range of periods, the best-fitting sinusoidhas a period of 11.8 d, but there are a large range of localminima in the interval 15 −
200 orbits. All of these candidateTTV signals give χ ∼ jktebop is written in FORTRAN77 and is available at: The Exoplanet Transit Database (ETD) website can be foundat http://var2.astro.cz/ETD
Table 3.
Times of transit midpoint of Qatar-2 b and their resid-uals. TRESCA refer to the “TRansiting ExoplanetS and CAndi-dates”.
References: (1) Canis Major Observatory (this work);(2) Bryan et al. (2012); (3) Strajnic et al. (TRESCA); (4) ZibarM. (TRESCA); (5) Gonzales J. (TRESCA); (6) MPG/ESO 2.2-m g ′ (this work); (7) MPG/ESO 2.2-m r ′ (this work); (8) MPG/ESO2.2-m i ′ (this work); (9) MPG/ESO 2.2-m z ′ (this work); (10)Dax T. (TRESCA); (11) Masek M. (TRESCA); (12) Carreno A.(TRESCA); (13) Montigiani N., Manucci M. (TRESCA); (14)Cassini 1.52-m (this work); (15) CAHA 2.2-m g (this work); (16)CAHA 2.2-m r (this work); (17) CAHA 2.2-m z (this work);(18) Campbell J. (TRESCA); (19) CAHA 1.23-m (this work);(20) Ren´e R. (TRESCA); (21) Ayiomamitis A. (TRESCA);(22) Jacobsen J. (TRESCA); (23) Kehusmaa P., Harlingten C.(TRESCA); (24) Shadic S. (TRESCA); (25) Colazo C. et al.(TRESCA).Time of minimum Cycle Residual ReferenceBJD(TDB) − . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . of these best-fitting TTV models are all ∼
30 sec, so we quotethis value as the nominal upper limit of any TTV effects onthe orbit of Qatar-2b.
From this point we considered only the high-precision lightcurves (i.e. not the Canis-Mayor ones). These were indi-vidually modelled with prism + gemc , each time includingthe parameters for one starspot. We used GEMC to ran- c (cid:13)000
From this point we considered only the high-precision lightcurves (i.e. not the Canis-Mayor ones). These were indi-vidually modelled with prism + gemc , each time includingthe parameters for one starspot. We used GEMC to ran- c (cid:13)000 , 000–000 hysical properties of Qatar-2b ì ììáá ççç çççç çç ç ççç çìììì ììììììì ììììì ìì - - Cycle number R e s i du a l s H d a y s L Figure 5.
Plot of the residuals of the times of transit midpoint of Qatar-2 b versus a linear ephemeris. The timings plotted withdiamonds are from this work, while empty circles are from ETD and the box is from Bryan et al. (2012). The box has the same size ofthe corresponding error bar, which has been suppressed for clarity. domly generate parameters for 36 chains, within a rea-sonable initial parameter space, and then to simultane-ously evolve the chains for 50 000 successive generations; seeTregloan-Reed et al. (2013) for details. The light curves andtheir best-fitting models are shown in Figs. 6 and 7. The de-rived parameters of the planetary system are reported inTable 4, while those of the starspots in Table 5. We usedthe former to reanalyse the phsyical properties of the sys-tem (Sect. 5), and the latter for the characterisation of thestarspots and the planetary orbit (Sect. 6).The results concerning the first GROND z ′ data setwere not considered since these data were compromised byan instrumental error (Sect. 3.1). Due to its low quality,the starpsot parameters resulted from the fit of the thirdGROND g ′ light curve have very large error bars and werenot reported in Table 5.We compared the fitted LD coefficients with the ex-pected stellar atmosphere model values. For the g ′ , r ′ , i ′ , z ′ and I bands, we used the theoretical LD coefficients esti-mated by Claret (2004) with two different model atmospherecodes ( atlas and phoenix ). We also checked the values forother passbands ( V , R c , I c , R j , I j ), where predictions fromdifferent authors (van Hamme 1993; Diaz-Cordoves et al.1995; Claret 2000) are available. While there is a good agree-ment for most of the light curves, there are some, especiallythose related to the g ′ band, for which differences of ± . ± . prism + gemc .We discounted all the g -band data sets because the best-fitting models returned very large value for k compared tothose of the other redder bands. Considering the high ac-tivity level of Qatar-2 A, as can be seen from the numer-ous occulted starspots detected, we attribute this to the ef-fect of unocculted starspots (i.e. starspots in a region ofthe stellar disc is not crossed by the planet) which causethe transits monitored in the bluest bands to be deeper.This is easily seen in Fig. 8 in which, using Eq. (4) and (5)from Sing et al. (2011), we plot for the case of Qatar-2 thecorrection for unocculted spots for a total dimming of 1%at a reference wavelength of 600 nm for different starspottemperatures. Starspots are modelled with atlas9 stellaratmospheric models (Kurucz 1979) of different tempera- tures ranging from 4450 to 3700 K in 250 K intervals, and T eff = 4645 K for the stellar temperature. However, in or-der to apply the right correction for unocculted starspot,we need an estimate of the absolute level of the stellar fluxcorresponding to a spot-free surface, which can be obtainedthrough a continuing, accurate photometric monitoring ofQatar-2 A over several years. Since the QES discovery dataare not public and no other long photometric monitoringare available for this target, we decided to exclude the g -band data for the estimation of the physical parameters ofthe system. Instead, corrections on the other optical bands( λ >
540 nm) are expected to be of the order of . − ,adding only a small contribution to the uncertainties in the k values. We measured the physical properties of the Qatar-2 systemusing the
Homogeneous Studies approach (see Southworth2012 and references therein). This methodology makes useof the photometric parameters reported in Table 4, spec-troscopic parameters from the discovery paper (velocityamplitude K A = 558 . ± . − , effective temperature T eff = 4645 ±
50 K, metallicity [Fe/H] = − . ± .
08 dex,and eccentricity e = 0; Bryan et al. 2012) and theoreticalstellar models to estimate the properties of the system.A value was estimated for K b , the velocity amplitudeof the planet, and the full system properties were calculatedusing standard formulae. K b was then iteratively adjustedto find the best agreement between the observed r A and T eff , and the values of R A a and T eff predicted by a theoreti-cal model of the calculated mass. This was done for a grid ofages from zero to 5 Gyr, and for five different sets of theoreti-cal models, specified in Table 6). We imposed an upper limitof 5 Gyr because the strong spot activity of the host starimplies a young age. The formal best fits are found at thelargest possible ages (20 Gyr in this case), which implies thatthe spectroscopic properties of the host star are not a goodmatch for the stellar density implicitly but strongly con-strained by the transit duration (Seager & Mall´en-Ornelas2003). Given the large number of available light curves, thediscrepancy is best investigated by obtaining new spectro-scopic measurements of the atmospheric properties of Qatar-2 A. We found a reasonably good agreement between the re- c (cid:13) , 000–000 L. Mancini et al. - N o r m a li s e d F l ux g ¢ Μ m r ¢ Μ m i ¢ Μ m z ¢ Μ m2012 (cid:144) (cid:144) - g ¢ Μ m r ¢ Μ m i ¢ Μ m z ¢ Μ m2012 (cid:144) (cid:144) - N o r m a li s e d F l ux g ¢ Μ m r ¢ Μ m i ¢ Μ m z ¢ Μ m2012 (cid:144) (cid:144) - g ¢ Μ m r ¢ Μ m i ¢ Μ m z ¢ Μ m2012 (cid:144) (cid:144) - Orbital phase N o r m a li s e d F l ux g ¢ Μ m r ¢ Μ m i ¢ Μ m z ¢ Μ m2012 (cid:144) (cid:144) - Orbital phase g ¢ Μ m r ¢ Μ m i ¢ Μ m z ¢ Μ m2012 (cid:144) (cid:144) Figure 6.
Phased GROND light curves of Qatar-2 compared to the best prism + gemc fits. The light curves and the residuals are orderedaccording to the central wavelength of the filter used. The passbands are labelled on the left of the figure, and their central wavelengthsare given on the right. c (cid:13) , 000–000 hysical properties of Qatar-2b - - N o r m a li s e d F l ux g Μ m r Μ m z Μ m - - g Μ m r Μ m z Μ m - - Orbital phase N o r m a li s e d F l ux Cassini 1.52 - mCAHA 1.23 - m rI - - Orbital phase
Cassini 1.52 - mCAHA 1.23 - m rI Figure 7.
Top panels : phased BUSCA light curves of Qatar-2 compared to the best prism + gemc fits. The light curves are orderedaccording to the central wavelength of the filter used. The passbands are labelled on the left of the figure, and their central wavelengthsare given on the right. Bottom panels : as above, but for the Cassini 1.52-m and the CAHA 1.23-m telescopes. The residuals of each fitare plotted in the right panels. sults for the five different sets of theoretical stellar models(see Table 6). The Claret models are the most discrepant,but not by enough to reject these results. We also useda model-independent method to estimate the physical pa-rameters of the system, via a calibration based on detachedeclipsing binary stars of mass < M ⊙ (Enoch et al. 2010;Southworth 2011). These empirical results match thosefound by using the stellar models.The final set of physical properties was obtained by tak-ing the unweighted mean of the five sets of values obtainedusing stellar models, and are reported in Table 7. System-atic errors were calculated as the standard deviation of theresults from the five models for each output parameter. Ta-ble 7 also shows the physical properties found by Bryan et al. (2012), which are of lower precision. Our measurement ofthe stellar density is ∼ . σ smaller than that found byBryan et al. (2012). The effect of this is to yield larger ra-dius measurements for both star and planet, and a 30% lowerplanetary density. As explained in Sect. 4, the parameters of the starspots de-tected in the light curves were fitted together with those ofthe transit using prism and gemc codes. In this way, wewere able to establish the best-fitting position, size, spotcontrast and temperature for each of them. The results are c (cid:13) , 000–000 L. Mancini et al.
300 400 500 600 700 800 900 10000.00000.00050.00100.00150.00200.00250.0030 wavelength H nm L R b (cid:144) R A c o rr ec ti on Figure 8.
The effect of unocculted starspots on the transmission spectrum of Qatar-2 considering a 1% flux drop at 600 nm. A gridof atlas9 stellar atmospheric models at different temperature, ranging from 4450 to 3700 K in 250 K intervals, was used to model thestarspot coverage, while for the star a model with T eff = 4645 K was adopted. Table 4.
Parameters of the prism + gemc best fits of the light curves of Qatar-2 for the quadratic LD law with the coefficients includedas fitted parameters. The final parameters, given in bold, are the weighted means of the results for the individual datasets, but excludingthe g -band ones (see text). Results from the discovery paper are included at the base of the table for comparison. The orbital period foreach data set is reported in Table 3. The results concerning the GROND z ′ dataset of the first transit are not reported (see text).Telescope Filter r A + r b k i ◦ u A v A ESO 2.2-m g ′ . ± . . ± . . ± .
46 0 . ± .
039 0 . ± . r ′ . ± . . ± . . ± .
39 0 . ± .
048 0 . ± . i ′ . ± . . ± . . ± .
26 0 . ± .
044 0 . ± . g ′ . ± . . ± . . ± .
78 0 . ± .
106 0 . ± . r ′ . ± . . ± . . ± .
35 0 . ± .
060 0 . ± . i ′ . ± . . ± . . ± .
31 0 . ± .
044 0 . ± . z ′ . ± . . ± . . ± .
28 0 . ± .
042 0 . ± . g ′ . ± . . ± . . ± .
23 0 . ± .
218 0 . ± . r ′ . ± . . ± . . ± .
26 0 . ± .
073 0 . ± . i ′ . ± . . ± . . ± .
18 0 . ± .
144 0 . ± . z ′ . ± . . ± . . ± .
18 0 . ± .
038 0 . ± . g . ± . . ± . . ± .
58 0 . ± .
060 0 . ± . r . ± . . ± . . ± .
18 0 . ± .
043 0 . ± . z . ± . . ± . . ± .
55 0 . ± .
055 0 . ± . r . ± . . ± . . ± .
43 0 . ± .
146 0 . ± . I . ± . . ± . . ± .
35 0 . ± .
135 0 . ± . . ± . . ± . . ± . Bryan et al. (2012) 0 . ± . . ± . summarised in Table 5. The final values for the angular radiiof the spots detected in each transit come from the weightedmean of the results in each band. These are reported in Ta-ble 8 in km and in percent of the stellar disc.Current knowledge on starspot temperatures is basedon results coming from different techniques, such as simulta-neous modelling of brightness and colour variations, Dopplerimaging, modelling of molecular bands and atomic line-depth ratios. Planetary-transit events offer a more directway to investigate this topic, especially in the lucky casethat the parent star is active and that the planet occultsone or more starspots during the transit.For the current case, we observed starspots in everyone of the transits that we monitored with high precision. This suggests that Qatar-2 A could be in a peak of its stellaractivity, since no starspots were seen in the four light curvesobserved between February and March 2011 by Bryan et al.(2012), of which three covered the complete transit event.Taking advantage of our multi-band photometry, westudied how the starspot contrast changes with pass-band. Starspots are expected to be darker in the ultravi-olet (UV) than in the infrared (IR). From Table 5, it isclear that the starspots are brighter in the redder pass-bands than in the bluer passbands, for all four simulta-neous multi-band observations. Modelling both the pho-tosphere and the starspot as black bodies (Rabus et al.2009; Sanchis-Ojeda et al. 2011; Mohler-Fischer et al. 2013;Mancini et al. 2013c) and using Eq. 1 of Silva (2003) and c (cid:13)000
135 0 . ± . . ± . . ± . . ± . Bryan et al. (2012) 0 . ± . . ± . summarised in Table 5. The final values for the angular radiiof the spots detected in each transit come from the weightedmean of the results in each band. These are reported in Ta-ble 8 in km and in percent of the stellar disc.Current knowledge on starspot temperatures is basedon results coming from different techniques, such as simulta-neous modelling of brightness and colour variations, Dopplerimaging, modelling of molecular bands and atomic line-depth ratios. Planetary-transit events offer a more directway to investigate this topic, especially in the lucky casethat the parent star is active and that the planet occultsone or more starspots during the transit.For the current case, we observed starspots in everyone of the transits that we monitored with high precision. This suggests that Qatar-2 A could be in a peak of its stellaractivity, since no starspots were seen in the four light curvesobserved between February and March 2011 by Bryan et al.(2012), of which three covered the complete transit event.Taking advantage of our multi-band photometry, westudied how the starspot contrast changes with pass-band. Starspots are expected to be darker in the ultravi-olet (UV) than in the infrared (IR). From Table 5, it isclear that the starspots are brighter in the redder pass-bands than in the bluer passbands, for all four simulta-neous multi-band observations. Modelling both the pho-tosphere and the starspot as black bodies (Rabus et al.2009; Sanchis-Ojeda et al. 2011; Mohler-Fischer et al. 2013;Mancini et al. 2013c) and using Eq. 1 of Silva (2003) and c (cid:13)000 , 000–000 hysical properties of Qatar-2b Table 5.
Starspot parameters derived from the prism + gemc fitting of the transit light curves presented in this work. ( a ) The longitude of the centre of the spot is defined to be 0 ◦ at the centre of the stellar disc and can vary from − ◦ to 90 ◦ . ( b ) Thecolatitude of the centre of the spot is defined to be 0 ◦ at the north pole and 180 ◦ at the south pole. ( c ) Angular radius of the starspot;note that 90 ◦ degrees covers half of stellar surface. ( d ) Spot contrast; note that 1.0 equals the brightness of the surrounding photosphere. ( e ) The temperature of the starspots are obtained by considering the photosphere and the starspots as black bodies (see text in Sect. 6).The results concerning the GROND z ′ dataset of the first transit are not reported, because this dataset is affected by correlated noise.Due to very large uncertainties, the results concerning the GROND g ′ dataset of the third transit are not reported. The results for theCassini data set are also not reported due to the large uncertainties of the parameters due to the fact that the light-curve points are veryscattered during the transit time and the sampling is not so good.Telescope Filter θ ( ◦ ) a φ ( ◦ ) b r spot ( ◦ ) c ρ spot d Temperature (K) e ESO 2.2-m g ′ . ± .
65 73 . ± .
44 6 . ± .
64 0 . ± .
066 4068 ± r ′ . ± .
12 80 . ± .
27 6 . ± .
97 0 . ± .
091 4137 ± i ′ . ± .
53 73 . ± .
61 3 . ± .
26 0 . ± .
052 4096 ± g ′ − . ± .
38 66 . ± .
17 4 . ± .
51 0 . ± .
049 4166 ± r ′ − . ± .
26 70 . ± .
89 4 . ± .
48 0 . ± .
048 4201 ± i ′ − . ± .
70 69 . ± .
18 4 . ± .
67 0 . ± .
087 4167 ± z ′ − . ± .
66 67 . ± .
35 3 . ± .
43 0 . ± .
072 4143 ± r ′ . ± .
48 72 . ± .
23 4 . ± .
50 0 . ± .
091 4172 ± i ′ . ± .
39 72 . ± .
76 4 . ± .
76 0 . ± .
183 4061 ± z ′ . ± .
62 70 . ± .
74 3 . ± .
57 0 . ± .
116 4132 ± g − . ± .
13 67 . ± .
52 3 . ± .
93 0 . ± .
102 4078 ± r − . ± .
89 72 . ± .
73 3 . ± .
63 0 . ± .
095 4144 ± z − . ± .
83 84 . ± .
14 4 . ± .
99 0 . ± .
184 4217 ± I . ± .
22 68 . ± .
26 4 . ± .
16 0 . ± .
216 4120 ± Table 6.
Derived physical properties of the Qatar-2 planetary system.This work This work This work This work This work This work(dEB constraint) (
Claret models) ( Y models) ( Teramo models) (
VRSS models) (
DSEP models) K b ( km s − ) 175.8 ± ± ± ± ± ± M A ( M ⊙ ) 0.763 ± ± ± ± ± ± R A ( R ⊙ ) 0.783 ± ± ± ± ± ± g A (cgs) 4.534 ± ± ± ± ± ± M b ( M Jup ) 2.539 ± ± ± ± ± ± R b ( R Jup ) 1.265 ± ± ± ± ± ± ρ b ( ρ Jup ) 1.173 ± ± ± ± ± ± ± ± ± ± ± ± a (AU) 0.02172 ± ± ± ± ± ± . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . Table 7.
Final physical properties of the Qatar-2 planetary system, compared with results from Bryan et al. (2012). The first errorbar for each parameter is the statistical error, which stems from the measured spectroscopic and photometric parameters. The seconderror bar is the systematic error arising from the use of theoretical stellar models, and is given only for those parameters which have adependence on stellar theory.
This work (final)
Bryan et al. (2012)Stellar mass M A ( M ⊙ ) 0.743 ± ± . ± . R A ( R ⊙ ) 0.776 ± ± . ± . g A (cgs) 4.530 ± ± . ± . ρ A ( ρ ⊙ ) 1 . ± .
016 2 . ± . M b ( M Jup ) 2.494 ± ± . ± . R b ( R Jup ) 1.254 ± ± . ± . g b ( m s − ) 39 . ± .
52 43 . ± . ρ b ( ρ Jup ) 1.183 ± ± . ± . T ′ eq (K) 1344 ±
14 1292 ± ± ± . ± . a (AU) 0.02153 ± ± . ± . (cid:13) , 000–000 L. Mancini et al.
Table 8.
Starspot sizes and temperatures for each of the transits presented in this work.Telescope Spot radius (km) % of the stellar disc Temperature (K)ESO 2.2-m ± ∼ .
41% 4094 ± ± ∼ .
47% 4176 ± ± ∼ .
58% 4148 ± ± ∼ .
53% 4124 ± ± ∼ .
70% 4120 ± T eff = 4645 ±
50 (Bryan et al. 2012), we estimated the tem-perature of the starspots at different bands and reportedthem in the last column of Table 5. The values of thetemperature estimated for each transit are in good agree-ment between each other within the experimental uncer-tainties and point to starspots with temperature between4100 and 4200 K. This can be also noted in Fig. 9, wherewe compare the spot contrasts calculated by prism+gemc with those expected for a starspot at 4200 K over a stel-lar photosphere of 4645 K, both modelled with atlas9 atmospheric models (Kurucz 1979). The weighted meansare shown in Table 8, and are consistent with what hasbeen observed for other main-sequence stars (Berdyugina2005), and for the case of the TrES-1 (Rabus et al. 2009),CoRoT-2 (Silva-Valio et al. 2010), HD 189733 (Sing et al.2011), WASP-4 (Sanchis-Ojeda et al. 2011), HATS-2(Mohler-Fischer et al. 2013) and WASP-19 (Mancini et al.2013c; Huitson et al. 2013). All these measurements areshown in Fig. 10 versus the temperature of the photosphereof the corresponding star. The spectral class for most of thestars is also reported and allows to see that the temperaturedifference between photosphere and starspots is not stronglydependent on spectral type, as already noted by Strassmeier(2009).Another type of precious information that we can ob-tain from follow-up transit observations comes from observ-ing multiple planetary transits across the same starspotor starspot complex (Sanchis-Ojeda et al. 2011). In thesecases, there is a good alignment between the stellar spinaxis and planet’s orbital plane and, by measuring the shiftin position of the starspot between the transit events, onecan constrain the alignment between the orbital axis of theplanet and the spin axis of the star with higher precisionthan with the measurement of the Rossiter-McLaughlin ef-fect (e.g. Tregloan-Reed et al. 2013). On the other hand, ifthe starspots detected at each transit are different, then thelatitude difference of the starspots is fully degenerate withthe sky-projected spin-orbit angle λ .According to the orbital period of the transiting planet,the same starspot can be observed after consecutive transitsor after some orbital cycles, presuming that in the lattercase the star performs one or more complete revolutions.Therefore, in general, the distance D covered by the starspotin the time between two detections is D = ( n × πR lat ) + d, (1)where n is the number of revolutions completed by the star, R lat is the scaled stellar radius for the latitude at whichthe starspot have been observed and d is the arc lengthbetween the positions of two transits in which the starspotis detected.In the present case, we observed three close transits of Qatar-2 b, with GROND (Fig. 6). The transits θ, φ ) = (37 . ◦ ± . ◦ , . ◦ ± . ◦ )( θ, φ ) = ( − . ◦ ± . ◦ , . ◦ ± . ◦ )( θ, φ ) = (4 . ◦ ± . ◦ , . ◦ ± . ◦ ) (2)Could the starspots detected in the very close transits n = 0, i.e. v (72 ◦ ) = 0 .
31 km s − .For n = 1 we get v (72 ◦ ) = 9 . − , accomplishing a com-plete revolution in P rot = 3 . ± .
07 d at a colatitude of72 ◦ . For n = 2 we have v (72 ◦ ) = 18 . − , and for n > P rot ≈ πR A v sin i ⋆ = (14 . ± . i ⋆ , (3)where i ⋆ is the inclination of the stellar rotation axis with re-spect to the line of sight ( v sin i ⋆ = 2 . − ; Bryan et al.2012). From this we can exclude that the starspots detectedin transits n = 1 we find v (72 ◦ ) = 3 . ± .
13 km s − which corresponds to P rot = 11 . ± . σ of the equatorial value found above. Under the assumptionthat we have detected the same spot in these two transits,simple algebra gives the sky-projected angle between thestellar rotation and the planetary orbit to be λ = 4 . ◦ ± . ◦ . This is the first measurement of the orbital obliquityof Qatar-2 and is consistent with orbital alignment ( λ = 0).This result is also in agreement with the general idea thatcool stars have low obliquity (Winn et al. 2010).The fact that we observed spot crossing events in everyone of our high-precision transit light curves suggests thatthe host star has an active region underneath the transitcord. This idea was put forward by (Sanchis-Ojeda & Winn2011) for HAT-P-11, a rather different case where theplanet’s orbital axis is inclined by nearly 90 ◦ to the stel-lar rotational axis and the spot events cluster at two orbitalphases in the transit light curve (see fig. 24 in Southworth2011). c (cid:13) , 000–000 hysical properties of Qatar-2b ì ì ìì ì ì ìì ì ì ìì ì ììì
200 400 600 800 10000.00.20.40.60.81.0 wavelength H nm L S po t c on t r a s t ì GROND 2012.04.02 ì GROND 2012.04.17 ì GROND 2012.04.21 ì BUSCA 2012.05.09 ì CA 1.23m 2013.02.15
Figure 9.
Variation of the spot contrast with wavelength. All the points are from this work and are explained in the upper-left legend.The vertical bars represent the errors in the measurements and the horizontal bars show the FWHM transmission of the passbands used.Solid line represents the spot-contrast variation expected for a starspot at 4200 K over a stellar photosphere of 4645 K.
As discussed in Sect. 2.3, simultaneous multi-band transitobservations allow the chemical composition of the planet’satmosphere to be probed in a way similar to transmissionspectroscopy.In Sect. 5 we estimated an equilibrium temperature of1344 ±
14 K for Qatar-2 b, which suggests, in the terminol-ogy of Fortney et al. (2008), that this planet should belongto the pL class. Therefore, based on the theoretical per-spective of Fortney et al. (2008), it is not expected that theatmosphere of the planet should host a large amount of ab-sorbing molecules, such as gaseous titanium oxide (TiO) andvanadium oxide (VO). By using the data reported in Table 4,we investigated the variations of the radius of Qatar-2 b inthe wavelength ranges accessible to the instruments used, i.e.386–976 nm. In particular, we show in Fig. 11 the values of k (the planet/star radius ratio) determined from the analysisof each transit separately versus wavelength. The verticalbars represent the relative errors in the measurements andthe horizontal bars show the full-width at half-maximum(FWHM) transmission of the passbands used.The depths from the transit observations all agree witheach other within ∼ σ , even if the g -band data indicate alarger value of k at this wavelength region. As discussed inSect. 4, this is likely caused by unocculted starspots whichaffect the measure of the transit depth in the bluest opticalbands (see Fig. 8).In order to have a more homogeneous set of data, weconsider only the weighted-mean results coming from thethree transit observations performed with GROND over ì ì ìì ì ì ìì ì ì ìì ì ììììì
400 500 600 700 800 900 10000.1600.1650.1700.1750.180 wavelength H nm L R b (cid:144) R A ì GROND 2012.04.02 ì GROND 2012.04.17 ì GROND 2012.04.21 ì BUSCA 2012.05.09 ì Cassini 1.52m 2012.05.09 ì CA 1.23m 2013.02.15
Figure 11.
Variation of the planetary radius, in terms of theplanet/star radius ratio, with wavelength. All the points are fromthis work and are explained in the upper-right legend. The verticalbars represent the errors in the measurements and the horizontalbars show the FWHM transmission of the passbands used.
19 days. They are shown in Fig. 12 together with three one-dimensional model atmospheres developed by Fortney et al.(2010) for comparison. The green line has been calculated forJupiter’s gravity (25 m s − ) with a base radius of 1.25 R Jup at the 10 bar level and at 1250 K. The opacity of TiO andVO molecules is excluded from the model and the opticaltransmission spectrum is dominated by Rayleigh scatteringin the blue, and pressure-broadened neutral atomic lines ofsodium (Na) at 589 nm and potassium (K) at 770 nm. Theother two models are equal but with H /He Rayleigh scatter- c (cid:13) , 000–000 L. Mancini et al. ççç çççççççççççççç óó ææææææ T photosphere H K L T pho t o s ph e r e - T s po t H K L HD307938 H G2 L Sun H G2 L Sun H G2 L EK Dra H G2 L WASP - H G8 L WASP - H G8 L WASP - H G7 L HD189733 H K1 L TrES - H K0 L AB Dor H K0 L AB Dor H K0 L EV Lac H M4 L YY Gem H M0 L BY Dra H M0 L BY Dra H M0 L EQ Vir H K5 L V833 Tau H K4 L LQ Hya H K2 L AU Mic H M2 L HATS - H K L HATS - H K L CoRoT - H G7 L Qatar - H K L Figure 10.
Spot temperature contrast with respect to the photospheric temperature in several dwarf stars taken from Berdyugina(2005), shown with empty circles. The name and spectral type of the star are also reported for most of them. The values for TrES-1, CoRoT-2, HD 189733, WASP-4, HATS-2, WASP-19 are taken from Rabus et al. (2009), Silva-Valio et al. (2010), Sing et al. (2011),Sanchis-Ojeda et al. (2011), Mohler-Fischer et al. (2013), Mancini et al. (2013c) and Huitson et al. (2013), respectively. Their positionsin the diagram are marked with gray dots. The error bars have been suppressed for clarity. Note that some stars appear twice. The blackdots and the triangle refer to the values estimated with GROND and BUSCA, respectively, for the case of Qatar-2 (this work). ing increased by a factor of 10 (red dot line) and 1000 (bluedashed line). The GROND data clearly indicate a very largeradius of Qatar-2 b in the wavelength range 400–510 nm; thevalue of R b measured in the g ′ band differs by ∼ r ′ band, which equates to ∼ H , where H is the atmospheric pressure scale height. A reasonableexplanation for this nonphysical result is to advocate thepresence of unocculted starspots which strongly contami-nate the stellar flux in the g ′ band. Actually, if we correctthe value of k in the g ′ band by 0.0015 (cfr. Fig. 8), weobtain a planetary radius which is still large compared tothe other bands, but straightforwardly explicable within theerror bar by Rayleigh-scattering processes occurring in theatmosphere of Qatar-2 b (blue line in Fig. 12). A long pho-tometric monitoring of Qatar-2 A is mandatory to estimatethe right correction to make and new transit events obser-vations in the u or U bands are also suggested in order toconfirm the Rayleigh-scattering signature.The lower value of the radius in the i ′ band could sug-gest a lack of K, even if the spectral resolution of GRONDis not enough good to allow a clear determination. However,if true, this lack is attributable to a selective destructionprocess via photoionization, or its presence in a more con-densed state. Another possible explanation is that the planet formed very far from the star, on the boundaries of the pro-toplanetary disk, before migrating to its current position. We have reported new broad-band photometric observationsof five transit events in the Qatar-2 planetary system. Threeof them were simultaneously monitored through four opticalbands with GROND at the MPG/ESO 2.2-m telescope, andone through three optical bands with BUSCA at the CAHA2.2-m telescope and contemporaneously with the Cassini1.52-m telescope in one band. Another single-band obser-vation was obtained at the CAHA 1.23-m telescope. In totalwe have collected 17 new light curves, all showing anomaliesdue to the occultation of starspots by the planet. These werefitted using the prism + gemc codes which are designed tomodel transits with starspot anomalies. Three further tran-sits of Qatar-2 b were observed with a 25-cm telescope, andfitted with the jktebop code. Our principal results are asfollows.( i ) We used our new data and those collected from thediscovery paper and web archives to improve the precisionof the measured orbital ephemerides. We also investigated c (cid:13) , 000–000 hysical properties of Qatar-2b ì ì ì ìì ì ì ì R b (cid:144) R A á á á áá á á áá á á á
400 500 600 700 800 900 10000.20.40.60.81.0 wavelength H nm L T r a n s m i ss i on Figure 12.
Variation of the planetary radius, in terms ofplanet/star radius ratio, with wavelength. The black points arethe weighted-mean results coming from the three transit observa-tions performed with GROND. The vertical bars represent theerrors in the measurements and the horizontal bars show theFWHM transmission of the passbands used. The observationalpoints are compared with three synthetic spectra at 1250 K, whichdo not include TiO and VO opacity. With respect to the modelidentified with the green line, the other two have H /He Rayleighscattering increased by a factor of 10 (red dot line) and 1000 (bluedashed line). An offset is applied to all three models to provide thebest fit to our radius measurements. Coloured squares representband-averaged model radii over the bandpasses of the observa-tions. Transmission curves of the GROND filters are shown inthe bottom panel. possible transit timing variations generated by a putativeouter planet and affecting Qatar-2 b’s orbit. Unfortunately,the sampling of transit timings is not yet sufficient to detectany clear sinusoidal signal. A more prolonged monitoring ofthis system is mandatory in order to accurately characterisea possible TTV signal.( ii ) We have revised the physical parameters of Qatar-2 b and its host star, significantly improving their accu-racy. They are summarised in Table 7. In particular wefind that density of Qatar-2 b is lower than the estimateby Bryan et al. (2012). The theoretical radius calculated byFortney et al. (2007) for a core-free planet at age 4.5 Gyr anddistance 0.02 AU is 1.17 R Jup for a planet of mass 2.44 M Jup .These numbers are in good agreement with the parametersthat we have estimated and imply that Qatar-2 b is coreless.Fig. 11 shows the new position of Qatar-2 b in the mass-radius diagram (left-hand panel) and in the plot of orbitalperiod versus surface gravity (right-hand panel).( iii ) The detection of so many starspots in our lightcurves suggests an intense period of activity for Qatar-2 A.The extent of each starspot was estimated and found to bein agreement with those found for similar starspot detec-tions in other planetary systems during transit events. Theprojected positions ( θ, φ ) of each starspot was also deter-mined, and the colatitudes were found to be consistent with72 ◦ . For the four simultaneous multi-band observations, wedetected a variation of the starspot contrast as a function ofwavelength, as expected due to the different temperaturesof starspots and the surrounding photosphere. The multi- colour data allowed a precise measurement of the tempera-ture of each starspot. The values that we found are well inagreement with those found for other dwarf stars.( iv ) The starspots detected in the GROND transits P rot = 11 . ± . ◦ , andthe sky-projected spin orbit alignment, λ = 4 . ◦ ± . ◦ . Thelatter result implies that the orbital plane of Qatar-2 b iswell aligned with the rotational axis of its parent star.( v ) Thanks to the ability of GROND and BUSCA tomeasure stellar flux simultaneously through different fil-ters, covering quite a large range of optical wavelengths,we were able to search for a radius variation of Qatar-2 bas a function of wavelength. All of the measurements areconsistent with a larger value of the planet’s radius in the g -band when compared with the redder bands. This phe-nomenon is attributable to unocculted starspots which af-fect more strongly our measurements in the g band. By fo-cussing on the results coming from the three close transitsobserved with GROND, we reconstructed a more accuratetransmission spectrum of the planet’s atmosphere in termsof the planet/star radius ratio and compared it with threesynthetic spectra, based on model atmospheres in chemi-cal equilibrium in which the presence of strong absorberswere excluded, but with differing amounts of Rayleigh scat-tering (Fig. 12). If we correct the g ′ -band by the amountindicated by atmospheric models, the comparison betweenexperimental data and synthetic spectra suggests that theatmosphere of Qatar-2 b could be dominated by Rayleighscattering at bluer wavelengths. The low value of the radiusobserved between 700 and 800 nm should be explicable bya lack of potassium in the atmosphere of the planet (prob-ably caused by star-planet photoionization processes). Thishypothesis could be investigated with narrower-band filters. ACKNOWLEDGEMENTS
This paper is based on observations collected with theMPG/ESO 2.2-m located at ESO Observatory in La Silla,Chile; CAHA 2.2-m and 1.23-m telescopes at the Cen-tro Astron´omico Hispano Alem´an (CAHA) at Calar Alto,Spain; Cassini 1.52-m telescope at the OAB Loiano Observa-tory, Italy. Supplementary data were obtained at the Canis-Major Observatory in Italy. Operation of the MPG/ESO 2.2-m telescope is jointly performed by the Max PlanckGesellschaft and the European Southern Observatory. Op-erations at the Calar Alto telescopes are jointly performedby the Max-Planck Institut f¨ur Astronomie (MPIA) andthe Instituto de Astrof´ısica de Andaluc´ıa (CSIC). GRONDwas built by the high-energy group of MPE in collabora-tion with the LSW Tautenburg and ESO, and is operatedas a PI-instrument at the MPG/ESO 2. 2m telescope. Wethank Timo Anguita and R´egis Lachaume for technical as-sistance during the GROND observations. We thank DavidGalad´ı-Enr´ıquez and Roberto Gualandi for their technicalassistance at the CA 1.23 m telescope and Cassini telescope,respectively. L.M. thanks Antonino Lanza for useful discus- c (cid:13) , 000–000 L. Mancini et al. ææ ææ Planet mass H M Jup L P l a n e t r a d i u s H R J up L Ρ jup Ρ jup Ρ jup Ρ jup Ρ jup ææ Orbital period H days L P l a n e t s u rf ace g r a v it y H m s - L Figure 13.
Left-hand panel : plot of the masses and radii of the known TEPs. The orange symbols denote values from the HomogeneousStudies project and the blue symbols results for the other known TEPs. Qatar-2 b is shown in red (Bryan et al. 2012) and green (thiswork). Dotted lines show where density is 2.5, 1.0, 0.5, 0.25 and 0.1 ρ Jup . Right-hand panel sion. J.S. acknowledges financial support from STFC in theform of an Advanced Fellowship. The reduced light curvespresented in this work will be made available at the CDS(http://cdsweb.u-strasbg.fr/). We thank the anonymous ref-eree for their useful criticisms and suggestions that helpedus to improve the quality of the present paper. The fol-lowing internet-based resources were used in research forthis paper: the ESO Digitized Sky Survey; the NASA As-trophysics Data System; the SIMBAD data base operatedat CDS, Strasbourg, France; and the arXiv scientific paperpreprint service operated by Cornell University.
REFERENCES
Alonso R., Barbieri M., Rabus M., Deeg H. J., Belmonte,J. A., Almenara J. M., 2008, A&A, 487, L5Ballerini P., Micela G., Lanza, A. F., Pagano I., 2012, A&A,539, 140Bayliss D., et al., 2013, AJ, 146, 113Berdyugina S. V., 2005, Living Rev. Solar Phys., 2, 8Bento J., 2013, MNRAS in pressBryan, M. L., et al., 2012, ApJ, 750, 84Bryan, M. L., et al., 2014, ApJ, 782, 121Ciceri S., et al., 2013, A&A, 557, A30Claret A., 2000, A&A, 363, 1081Claret A., 2004, A&A, 428, 1001Copperwheat C. M., et al., 2013, MNRAS, 434, 661de Mooij E. J. W., et al., 2012, A&A, 538, A46D´esert J.-M., et al., 2011, ApJS, 197, 14de Wit J., Seager S., 2013, Science, 342, 147Diaz-Cordoves J., Claret A., Gimenez A., 1995, A&AS, 110,329Enoch B., Collier Cameron A., Parley N. R., Hebb L., 2010,A&A, 516, A33Fortney J. J., Marley M. S., Barneset J. W., 2007, ApJ,659, 1661Fortney J. J., Lodders K., Marley M. S., Freedman R. S.,2008, ApJ, 678, 1419 Fortney J. J., Shabram M., Showman A. P., et al., 2010,ApJ, 709, 1396Gibson N. P., Pollacco D., Simpson E. K., et al., 2008,A&A, 492, 603Gillon M., Pont F., Moutou C., et al., 2006, A&A, 459, 249Greiner J., et al., 2008, PASP, 120, 405Holman M. J., et al., 2010, Science, 330, 51Huitson C. M., et al., 2013, MNRAS, 434, 3252Kurucz R. L., 2013, ApJS, 40, 1Lendl M., Gillon M., Queloz D., Alonso R., Fumel A., JehinE., Naef D., 2013, A&A, 552, A2Maciejewski G., et al., 2013, A&A, 551, A108Mancini L., et al., 2013a, A&A, 551, A11Mancini L., et al., 2013b, MNRAS, 430, 2932Mancini L., et al., 2013c, MNRAS, 436, 2Mancini L., et al., 2014, A&A, 562, A126Mohler-Fischer M., et al., 2013, A&A, 558, A55Narita N., et al., 2013, ApJ, 773, 144Nesvorn´y D., Kipping D. M., Buchhave L. A., Bakos G. ´A.,Hartman J., Schmitt A. R., 2012,
Science , 336, 1133Nikolov, N., Chen, G., Fortney, J. J., Mancini L., South-worth J., van Boekel R., Henning Th., 2013, A&A, 553,A26Penev K., et al., 2013, AJ, 145, 5Pierini D., et al., 2012, A&A, 540, A45Pont F., et al., 2007, A&A, 476, 1347Pont F., Sing D. K. Gibson, N. P. Aigrain S., Henry G.,Husnoo N., 2013, MNRAS, 432, 2917Rabus M., et al., 2009, A&A, 494, 391Rothery D. A., McBride N., Gilmour I., 2011, An Intro-duction to the Solar System (Cambidge University Press)Sanchis-Ojeda R., Winn J. N., Holman M. J., 2011, ApJ,733, 127Sanchis-Ojeda R., Winn J. N., 2011, ApJ, 743, 61Seager S., Mall´en-Ornelas G., 2003, ApJ, 585, 1038Silva A. V. R., 2003, ApJL, 585, L147Silva-Valio A., Lanza A. F., Alonso R., Barge P. 2010,A&A, 510, A25 c (cid:13) , 000–000 hysical properties of Qatar-2b Sing D. K., et al., 2011, MNRAS, 416, 1443Seager S., & Mall´en-Ornelas G., 2003, ApJ, 585, 1038Southworth J., 2011, MNRAS, 417, 2166Southworth J., 2012, MNRAS, 426, 1291Southworth J., et al., 2009, MNRAS, 396, 1023Southworth J., et al., 2010, MNRAS, 408, 1680Southworth J., et al., 2011, A&A, 527, A8Southworth J., Bruni I., Mancini L., Gregorio J., 2012a,MNRAS, 420, 2580Southworth J., et al., 2012b, MNRAS, 422, 3099Southworth J., et al., 2012c, MNRAS, 426, 1338Southworth J., et al., 2013, MNRAS, 434, 1300Strassmeier K. G., 2009, Astron. Astrophys. Rev., 17, 251Tregloan-Reed J., Southworth J., Tappert C., 2013, MN-RAS, 428, 3671van Hamme W., 1993, AJ, 106, 2096Winn J. N., et al., 2008, ApJ, 683, 1076Winn J. N., Fabrycky D., Albrecht S., Johnson J. A., 2010,ApJ, 718, L145Zacharias N., Monet D. G., Levine, S. E., Urban S. E.,Gaume R., Wycoff G. L., 2004, American Astron. Soc.meeting, 205,
APPENDIX A: S/N ESTIMATIONS
In order to test the goodness of the measurements reportedin this paper, we present signal-to-noise ratio (S/N) expecta-tions for the simultaneous multi-band photometric observa-tions of Qatar-2 with GROND. For each of the four GRONDoptical bands, we can quantify the ratio of noise-to-signalper unit time by ρ = σ total p t exp + d readout , (A1)where t exp is the exposure time and d readout is the total deadtime per observation – the latter quantity is generally dom-inated by the CCD readout time, but for GROND we havealso to consider an extra dead time due to the synchroniza-tion of the optical observations with the NIR ones. σ total isthe total noise in a specific band for 1 mag measurement inone observation and takes into account five noise contribu-tions that are added in quadrature. They are the Poissonnoise from the target and background, readout noise, flat-fielding noise and scintillation, i.e. σ total = q σ + σ + σ + σ + σ . (A2)Following the procedure described by Southworth et al.(2009), we performed S/N calculations for the GROND ob-servations presented in this work. The count rates for thetarget (overall) and for the sky background (per pixel) weregathered from the SIGNAL code, scaling for the differencein telescope aperture and CCD plate scale. The magnitudesof Qatar-2 in Sloan g , r and i were taken from AAVSO archive. The magnitude in Sloan z was obtained by interpo-lating the previous ones with J , H and K magnitudes taken Information on the Isaac Newton Group’s
SIGNAL code canbe found at http://catserver.ing.iac.es/signal/. The American Association of Variable Star Observers(AAVSO) is a non-profit worldwide scientific and educational or-ganization of amateur and professional astronomers. from the NOMAD catalog (Zacharias et al. 2004). Other in-put parameters, specific for each band, were the readoutnoise of the CCD detector, the signal per pixel from the tar-get averaged over the PSF, the maximum total count in apixel (target plus sky) and the number of pixels inside theannulus of the target to apply flat-field noise. The resultingcurves of noise level per observation are plotted in millimag-nitudes as a function of t exp in Fig. A1 for dark time andhigh observational cadence (solid curves) and for bright timeand low cadence (dashed curves), cfr. Table 1. The scatter ofour measurements are a bit higher than the expected noise,but fully consistent. c (cid:13) , 000–000 L. Mancini et al. ììíí N o i s e H mm a g L GROND - g ¢ ììíí N o i s e H mm a g L GROND - g ¢ ììììíí N o i s e H mm a g L GROND - r ¢ ììììíí N o i s e H mm a g L GROND - r ¢ ììììíí N o i s e H mm a g L GROND - i ¢ ììììíí N o i s e H mm a g L GROND - i ¢ ììììíí Exposure time H sec L N o i s e H mm a g L GROND - z ¢ ììììíí Exposure time H sec L N o i s e H mm a g L GROND - z ¢ Figure A1.
Predicted noise levels for the GROND observations presented in this work, but as a function of exposure time. The solidcurves show the predicted noise level per observation for dark time and high observational cadence, whereas the dashed curves that perobservation in bright time and low cadence. Panels in the left-hand column are for each of the four GROND optical bands. Panels inthe right-hand column are the same of left-hand column, but zoomed to t exp < (cid:13)000