################################################## # JWST Transit Planning Meeting ################################################## -------------------------------------------------- Introduction & Welcome Chas Beichmann & Jonathan Lunine 2014-03-11 08:17 -------------------------------------------------- Pay Ellen for food ($20/day, checks OK) 7.5x higher S/N vs. Spitzer 2.7x higher S/N vs. HST GOALS: --Define most important transit-science cases --Understand instrument capabilities --Inform project about unique requirements (duration, cadence, astrometry, etc.) --Learn from HST, Spitzer, Kepler --Plan for instrument validation --Identify necessary precursor activities - - Ephemerides precision? - - Knowledge of host stars? - - Model atmospheres? - - Early on-orbit tests --Inform community re. planning small/large/Legacy programs --Write a white paper GOAL: capture 25% of observing time. BE AMBITIOUS! Instrument tests are imminent: talk to Marcia Rieke or others. Observing templates are imminent: talk to J. Valenti or others. -------------------------------------------------- Spectroscopy of Giant Planets Jonathan Fortney 2014-03-11 08:46 -------------------------------------------------- We are 45 years behind solar system infrared science We are *NOT* 'tying up loose ends.' Spitzer leaves many questions unanswered. Big Science questions: --are giant planets metal-enriched? --how does this change w/panet mass, stellar type, migration history --what is impact of snow line condensation on abundance deviations? --how far out of radiative equilibrium are atmospheres? --what modulates surface brightness distributions (metallicity, rotation, incident flux) --Cloud opacity: what are they made of? How effect thermal emission? --Deviations from equilibrium chemistry --Vertical/horizontal mixing & chemical homogeneziation "BIG FIVE" molecules: NH4, CH4, H2O, CO, CO2 Photochemical products: HCN, C2H2, C2H4, C2H6 Beware of model-dependent inferences Need to TEST the 1D assumption of many models! Opacity problems: CH4 and CO2 are still quite uncertain. People need to know where the uncertainties come from: opacity lists, etc. -------------------------------------------------- Spectroscopy of Super-Earths Eliza Kempton 2014-03-11 09:13 -------------------------------------------------- Super-Earths: most common type of planet, but no Solar System examples Broad diversity of bulk compositions: what are they made of? GJ1214b: emission spectrum doesn't help much in discriminating between H2-dominated or H2O-dominated atmospheres! [Thought: if GJ1214b is covered in haze, presumably it has a temperature inversion! Tropopause at or above 0.1 mbar.] GJ 1214b: flat transmission. GJ 436b: flat transmission. (probably) HD 97658b: flat transmission. (probably) Need to choose between "low-cost" easy goals and "high-risk, high-reward" goals. Need to constrain cloud properties. --Is there a clear dividing line (or any discriminators) between H2-dominated and heavy molecule-dominated super-Earth atmospheres? Question: could we 'pre-screen' super-Earth candidates for clouds? Answer: maybe first do snapshot (HST or JWST); if not featureless, follow up with more time. To beat cloud opacity in transmission, go for long-wavelength (MIRI) thermal emission spectroscopy. Lunine: we will never understand clouds. How can we justify observing small planets whose masses we don't know? Doyon: this would probably be a waste of time. Bean: Listen to my M dwarf/RV talk tomorrow. Crossfield: are there any targets for which we will get good spectra but could not measure masses? Beichmann: maybe not...? Benneke: Whether or not, knowing the mass (from RV) helps your retrieval. -------------------------------------------------- Atmospheric Dynamics Heather Knutson 2014-03-11 09:45 -------------------------------------------------- Most planets in circular orbits: tidally locked. --> Broad jets, large vortices. Also: longitudinal variations in temperature abundances, etc. Atmospheric structure should be simpler than that of Jupiter. Habitable super-Earths around M dwarfs are likely also tidally locked! GCMs compromise between spatial scale & physics included. Choices that were appropriate for Earth or Jupiter might not be appropriate for exoplanets!!! Questions we can answer: --What is day-night temperature gradient? --WHat are locations of hottest and/or coldest regions? --How do the above 2 vary with altitude/pressure? --How do all these vary with location on planet? Eclipse mapping: mainly sensitive to first few modes (location & degree of concentration of hottest/coldest spot). Transits/eclipses: need stability on ~few hour timescales. Phase curves: need stability on 10-100 hour timescales. Spectroscopic phase curves. Thermal phase curves of temperate 2 R_Earth planets... if you can reach photon-limited precision on timescales of weeks-to-months. (!!!) Venus would have a low-amplitude phase curve (in optical or NIR). But hotter super-Earth phase curves should be feasible. Valenti: could you partially cancel systematics via differential phase curves? E.g., I could measure the temperature of the Sun from *normalized* spectra by examining individual line shapes. Question: could we switch between multiple instrument modes to extend wavelength coverage? Knutson: Spitzer and HST show us we need to keep things *stable* Mandell: does HST phase curve show slow, long-term drifts? Kreidberg: Yes. We had 3 visits, and need to remove a ~quadratic trend from each visit. Beichmann: What about very hot, rocky planets (Kepler-10b)? You could measure temperature contrasts. Crossfield: You could measure surface and/or exosphere composition, too. Question: Couldn't any phase curve be corrupted from thermal phase variations of other, non-transiting planets in the system? [This is a nightmare-question!] -------------------------------------------------- Photometry: transit validation, TTVs, high precision transits, etc. Drake Deming 2014-03-11 10:21 -------------------------------------------------- Infrared transits: essentially no limb darkening! Sharp ingress/egress. Measured radius is model-independent. [But there are also far fewer photons!] Validation of planet candidates: false positive "binary" systems would have very different optical-vs.-IR transit depths. This would be tough to justify w/JWST. TTV mass measurements (or ephemeris refinement). This would be tough to justify w/JWST, except for a very compelling target. Maybe the key case is: planets around M dwarfs and/or brown dwarfs (otherwise, why go to the IR for transits?) Deming's predictions: --Probably won't use JWST to reject TESS false-positives --Probably won't use JWST to measure photometric radii/transit depths. --Probably won't use JWST to measure TTVs --Might use it to find transits of other planets (e.g., Gillon et al. 2014) Deroo: Won't CHEOPS do transit followup? Deming: Maybe JWST would do better if the star were small & cool. T. Greene: How close to the photon noise do you come? Deming: Within 10-20% on short timescales, but worse on longer timescales. Carey: Binning multiple transits goes as sqrt(N), but data in a single transit improves more slowly. Carey: How will we discriminate between astrophysical and instrumental variations? We should do as *much* calibration on ground as we possibly can. Comment: It's tough to test the instruments at the level of precision you exoplanet people want! Beichmann: Heather, is it useful to look at eccentric planets like HD80606b? Knutson: Eccentric planets constrains radiative timescale. But how much can we extend these lessons to other (non-eccentric) planetary atmospheres? Spectroscopy could disentangle temperature from abundances. Valenti: Is there any big science gain from going to high (~3000) resolution? Benneke: Low-resolution probes features, not individual lines. Low-resolution captures shapes of features, so can measure mean molecular mass. Beichmann: one will have to trade off resolution vs. spectral grasp vs. S/N Howell: Answer depends on your science question. Mandell: Isn't it troubling that GJ1214b's transmission spectrum is flat out to Spitzer wavelengths? How far we might have to go? Fortney: You just need a wide distribution of particle sizes. Greene: JWST is not as cold as Spitzer: long-wavelengths will not be *so* easy. Christiansen: How do we prioritize our programs? Can we afford to look at just 2 super-Earths, or can we look at 20 slightly-less-"interesting" targets. Beichmann: We can do a large sample of Saturns and Jupiters [and Neptunes]. One could get ~50, easily. And phase curves for (at least) several. This is what I think of as "survey" science. Kempton: 2-band recon super-Earth photometry Benneke: No, take spectra instead. Doyon: Take 1 transit with NIRISS. -------------------------------------------------- Measurement Repeatability: HST/WFC3 Spatial Scan Mark Swain 2014-03-11 11:29 -------------------------------------------------- A detailed investigation of HST spatial-scan observations, down to the per-pixel level. Investigated: --Mean, std. dev., Gaussianity of individual 'frames' in each integrations --Properties of orbit-to-orbit measurements Small- and larger-scale pointing jitter always evident Finds various types of non-Gaussianity in the sampled measurements. Kreidberg: scan rate is not constanst, but if you add up all samples this effect shouldn't matter so much. Second you are comparing pre- and post-transit scans to look for reliability: this works at lower S/N, but at high S/N a more comprehensive modeling approach works better. Also: dispersion solution changes across detector. So you can't just assume that one column represents one wavelength Swain: I don't think this last point has a significant impact. Deming: I will never go back to 'staring mode' (unless there were infinite detector well depth). Many different studies show near photon-limited results on exoplanet spectra. So the effects you have discussed must not affect the final precision on the spectrum. Swain: I do not share your view. -------------------------------------------------- HST Best Performances and Best Practices David Sing & Avi Mandell 2014-03-11 11:50 -------------------------------------------------- HST: suffers from Thermal Breathing, focus & other changes First HST orbit is always different from the rest: schedule buffer time. First frame of each orbit is systematically low (with STIS) Orbital "breathing" trends seen in essentially all HST observations. Relative flux in PSF core vs. wings evolves over an HST orbit. Attain high S/N: with CCD, saturate detector wells (*not* A/D converter!) and bin up all photons. WFC3: systematics are not common-mode and not wholly repeatable when observing in 'staring' mode. WFC3 spectral resolution: results are near photon-noise limited for higher resolution... as you bin the data, error bars initialy shrink but then converge. Is this astrophysical ("forests" of low-amplitude spectral features) or instrumental (correlated systematics) ? R. Smith: when spatial-scanning, how do you flat-field with sufficient precision? Sing: we always come back to the same pixels, so we are making a relative measurement and absolute flat-fielding is unncecessary. Doyon: What determines how many pixels you scan over? Knutson: We use as many pixels as possible, while avoiding saturation. ** JWST will (probably) not do spatial scanning ** Knutson: initial JWST transit observations should be made public immediately. This lets everyone look at it and understand the data faster. -------------------------------------------------- Kepler Best Practices Jessie Christiansen 2014-03-11 11:49 -------------------------------------------------- Key point: always place target on the same pixels! Pointing: Spitzer goes back and forth over the same part of the pixel. JWST: Any pointing jitter should be on much longer (or shorter) timescales than timescale of interesting events! If pointing is good enough, you don't need to worry about flat-fielding. Heating cycles & temperature dependencies: these affect photometry! Rapid detector readout will heat things and induce possible systematics. TEST, TEST, TEST Get lots of calibration pixels and/or starts Reaction wheels: they 'rumble' when they transition through 'zero-motion crossing.' If possible, schedule these crossings away from transits. M. Rieke: Thermal stability has been one of our main goals. So, we will keep the detectors 'active' and 'clocking' all the time. T. Greene: After re-pointing, thermal drifts will occur and PSF will change. J. Bean: Are temperature problems due to focus changes, or detectors? Christiansen: BOTH! -------------------------------------------------- Spitzer Photometry Sean Carey (and Ian Crossfield) 2014-03-11 12:35 -------------------------------------------------- IRAC photometry comes to within 10% of photon noise. But things get worse for timescales longer than ~100 s. But things get better if you co-add multiple transits/eclipses. Pre-flashing: use large, bright regions to fill all charge traps. IRAC4 charge-trapping: slow & linear for faint pixels, quickly-saturating then flat for brighter pixels. Intra-pixel variations: IRAC pointing is 0.03 arcsec amplitude (vs. 1.2 arcsec pixel), but systematic amplitude is still up to ~0.5%. Clampin: JWST's PSF will vary, but not *so* much and only on long timescales. Beichmann: probably 10x less variation than HST. Varying PSF will not significantly impact even *coronagraphic* observations. -------------------------------------------------- Precise Exoplanet CHaracterization with HST Laura Kreidberg 2014-03-11 13:44 -------------------------------------------------- GJ 1214b: high-precision transit spectroscopy WASP-43b: emission spectroscopy & phase curve. HD 97658b: two transits. HST Spatial Scanning: brightest objects suffer from flux loss all the way out to detector edges (though this is 30 pixels away from main spectral trace!) name H_mag spectral_precision GJ 1214 9.1 30 ppm HD 97658 5.8 20 ppm Earth 2.0 ??? 10 ppm (around 0.2 R_sun M dwarf) Brightest objects cause persistence! Can ruin & waste much time for extragalactic observations. -------------------------------------------------- Transit Observations with Spitzer/IRS Jeroen Bouwman 2014-03-11 14:05 -------------------------------------------------- IRS: Very similar to what JWST/MIRI will provide. Need to properly deal with pointing correction & slit losses. -------------------------------------------------- Mark Clampin & John Stansberry JWST Operations 2014-03-11 14:22 -------------------------------------------------- Image quality: --2 micron diffraction limit --Several short-wavelength detectors are undersampled JWST required to be thermally stable: 57nm RMS drift over 14 days The most drift-sensitive case will be multi-day phase curves Wavefront measured every 2 days (but could use phase curve target star for this) Image motion requirement: shall be typically <7 mas for ~3hr observations Spatial Scanning Options: Nominally, could scan up to 60 mas/sec (much slower than HST!). But watch out for losing your guidestars. Data rates are not a problem, except maybe for NIRCam (2 detectors full-frame --> 135 GB/day, vs. 57 GB/day limit for science data) Longest integrations: high-gain antenna nominally requires integrations <9000 seconds... but they have proposed to allow transit to 'observe through' antenna repointings. Momentum issues could occur as frequently as every ~5-25 days. But full-frame exposures could continue uninterrupted for >500 hours. High-gain antenna repointing: causes ~70 mas pointing disturbance (up to 1-2 pixels for shortest wavelengths!) lasting for <1 minute. Plan is to continue observing right through these events. Will allow 'event-driven' observing: e.g., can specify to observe a transit of a given planet without specifying a particular event. Probably no dedicated "exoplanet transit" Exposure Time Calculator. Timing: spacecraft clock accurate to <0.5 sec. (audience thinks that people will care about *millisec* timing!) Parallel modes: this is being evaluated on its science merits... currently undetermined. Flight segment is capable -- but probably no mechanism movements in parallel mode. For subarray observations, it may be possible to get every sample-up-the-ramp. For the brightest objects in every mode, this will *not* be possible! Valenti: it might be possible to add more 'resets' between each read to extend the total, uninterrupted possible duration of observations. Beichmann: can we specify that we always go to the ideal 'sweet pixel' (however it is chosen)? Dean Hines: NO, for MIRI. (unless you just do it via a pointing offset) -------------------------------------------------- NIRCam Detector Behavior Marcia Rieke 2014-03-11 16:20 -------------------------------------------------- Ten detectors! Flight detectors are in the ISIM, so they can only be tested there. Various ways to read out detectors more quickly: stripes, subarrays. Undersampled (a bit) below 3 microns. Various patterns & structure evident in flat-field illumination Brightness Limit: NIRCam spectroscopy: limit is L > 2.6 (Johnson mag) (32x2048 subarray) NIRCam photometry: limit is L > 7.1 (Johnson mag) (32x32 subarray) Can go brighter for photometry using narrowband filters If necessary, can 'increase the back bias' to allow observations of even brighter targets. But you incur stronger persistence, and have to wait for things to settle in detector. Latents are strongest in the shortest-wavelength detectors (1.7 micron-cutoff), weakest in the 5 micron-cutoff detectors. Has done initial 6-hour stability tests using relatively low count rate illumination (~70 DN/sec). Plans an upcoming test: will take multiple NIRCam sample-up-the-ramps representing transit observations... Sources in NIRCam observed simultaneously through dichroic: Could observe spectroscopy in one arm and (defocussed) photometry in the other arm. Must be 2.1 micron medium-band filter. (Because NIRCam has no spectroscopic mode for short-wavelenth arm). Other issues: --- Inter-pixel (adjacent) capacitance of ~0.5% --- "Popcorn" (random telegraph) noise: spontaneous DC-level jumps! (probably a change in gain of individual pixel) --- "Snowballs" -- occasionally, a spheroidal galaxy appears in your detector for a frame or two. --- Missing charge between pixels -------------------------------------------------- MIRI Detectors Mike Ressler 2014-03-11 17:07 -------------------------------------------------- Mainly duplicating IRAC3+4 detectors. Primary differences: -- Larger format (1024x1024) -- Smaller pixels (25 micron, not 30 micron) -- Added reference pixels (of questionable utility) Photometry at shorter wavelengths (~5 micron) will interact with array as a diffraction grating and make 'star of Bethlehem' PSFs MIRI detectors are cold (6K): so there are few charge carriers available, and funky things can happen. Did a 4-hour constant-illumination test. Sees a ramp, like in IRAC/MIPS If they implement a pause between sets of 5 exposures, they see a 'hook' just like in HST/WFC3. (But not clear why you would observe like this in practice.) They don't see strong latents. S. Carey: With IRAC, we could never reproduce on-sky latents in lab arrays even after we knew about this effect. T. Greene: In our arrays, we were able to make a latent with a 4hr time constant. Annealing: this will still be a problem, just like it was for MIPS. But with 'cold anneal' (@ 13 K) latent-free imaging might be restored within ~0.5 hr. Very bright sources (5e4 electrons/sec) cause all sorts of detector-wide problems. Changing readout mode (e.g., to/from subarray) leaves latents on the detector (!!). Maybe these decay in ~20 minutes. Consider: avoid subarrays if your science allows it! -------------------------------------------------- Jacob Bean RV & Transit Surveys and Bright Targets for JWST 2014-03-12 08:32 -------------------------------------------------- GJ 1214b vs. HD 97658b --- both give flat transmission spectra, but correlated noise is more evident in the spectrum of (the brighter) HD 97658b. Also, residuals are ~3x photon limit in HD97658b case. Also, absolute transit depth is less certain. Bean's opinions: --We should not merely target brightest stars. --The photon-limited S/N may be too optimistic for brightest stars --Instead: we want the brightest stars for which we expect the biggest atmospheric signals. We will probably be limited by instrumental systematics on the brightest stars. [P. McCullough: this could cause you to not observe more interesting objects, if you select for the 'biggest signals.'] Bean: We should be looking for planets around the smallest stars. Swain: You could extend your argument to gas giant regime, re. 'new' molecules we haven't yet studied in exoplanet atmospheres. Ground-based surveys: --MEarth --APACHE --SPECULOOS Beichmann: Also, keep K2 in mind! NIR RVs: state-of-the-art is 5 m/s with CRIRES & NH3 gas cell. Good, but not enough for habitable planets. 3 funded ongoing efforts for stabilized, 1 m/s RV precision: SPIRou, HPF, Carmenes. Lunine: will these RV efforts be ready in time for JWST? Or could they be delayed and we have to play 'catch-up' ? T. Greene: Don't hope to go for the most targets with JWST in Cycle 1! C. Beichmann: NASA is finally starting to recognize importance of ground-based RV efforts (e.g., for TESS startup). -------------------------------------------------- Steve Howell Targets for JWST: Kepler and K2 2014-03-12 08:53 -------------------------------------------------- Kepler produced many planet candidates. We have developed better understanding of main Kepler stars Kepler Field: contains very few bright M dwarfs. Even FGKs are KepMag>11, and M dwarfs are typically KepMag>13-15. Yesterday, "K2" started its first "Campaign 0" field and so far is performing well. Pointing is 8-10x worse than Kepler, but photometric precision is just 2-3x worse. Look at each field for ~75 days. K2: all fields are equally good for M dwarfs (b/c they are all local) Can reach 80 ppm in 6hr mean (precision will be worse for shorter transits) (with 3 transits). Roughly 4000 M dwarfs per field: should find ~50 rocky planets per year (down to 0.5 R_Earth for KepMag=9 and Teff=3400 K). Need spectra for all host stars! T. Greene: these numbers are for assumed distribution of planets? C. Beichmann: one challenge is the long (30 min) cadence for shorter transits. J. Valenti: are you more sensitive to longer-period planets? S. Howell: If you need 3 transits, then your period has to be P < 25 days. The "3-transit rule" doesn't quite get out to the Habitable Zone! Lunine: HZ definitions are fuzzy! FOr bigger/hotter stars, you can also expect ~50 bright-star planets per year. Real K2 engineering data: KepMag<13 reaches 70-100 ppm in 6 hr; reaches 500 ppm for KepMag=14-15. Probably some room for improvement. Lunine: Is K2 a good outreach activity? Howell: K2 is the first spacecraft we can know about that balances using solar pressure. Q: What is brightest source K2 can observe? Howell: With Kepler we looked at ~4th mag. We have done 6th mag with K2 -- it should be okay. -------------------------------------------------- TESS & JWST Targets George Ricker 2014-03-12 09:12 -------------------------------------------------- TESS: will target 200,000 stars, including M dwarfs out to 60 pc. TESS will find many Neptunes , quite a few super-Earths/sub-Neptunes , and *some* rocky planets around bright stars. TESS will provide full-frame images at 30min cadence. And 1-minute cadence for ~200K stars. TESS will provide *HOT* candidates. Observes most of sky for just 27 days each. A small fraction (including JWST continuous viewing zone) observed for ~350 days! Josh Winn et al. have run numerous yield simulations. 38 Earths (0.8-1.25 R_E), 335 Super-Earths (1.25-2 R_E), 1460 sub-Neptunes (2-4 R_E) ~17 "small, HZ planets" ~7 of these in JWST Continuous Viewing Zone Planet yield rises (weakly!) from ecliptic toward ecliptic pole. Earth/super-Earth yields (near ecliptic poles): 127 super-Earths, 19 Earths N Kmag 5 5-8 13 8=9 30 9-10 ~20 small planets with 51 billion stars with V<20 Launched Dec 2013 Data releases: ~Oct 2015: G mag, PM for brightest stars, some ecliptic pole data. ~Apr 2016: 5-param. astronomery, bright RV, more photometry Could conceivably detect transit for some objects (but mostly V>14 mag). Expect ~500,000 M dwarfs within 100 pc (with G<20). GAIA might find transiting giant planets by measuring astrometric orbits!! Measure inclination to within ~3-5 deg... need to follow these up! Ephemerides (from astrometry) typically determined to within ~20 *DAYS* for P=2yr. -------------------------------------------------- Stellar Variability & JWST Peter McCullough 2014-03-12 11:06 -------------------------------------------------- Effective stellar radius affected by surface spots. Bad for transits & phase curves! Not so bad for secondary eclipses. Both occulted & unocculted spots. Cool starspots can have complicated spectral features (even H2O!) HD 189733b: Teff = 5000 K, Tspot ~ 4000 K. Ground-based photometry was essential to interpret the planet's transmission spectrum! Says that if we mis-estimate stellar spot coverage, we could mistake spot effects for planetary atmospheres (e.g., Rayleigh scattering). D. Sing: I agree that we need long-term ground-based monitoring. -------------------------------------------------- Precursor data Needed for JWST transit/eclipse Observations David Ciardi 2014-03-12 11:32 -------------------------------------------------- Ephemerides: Should be OK for transits: if no strong TTVs, current ephemerides would propagate to ~1hr by 2018. Secondary eclipses could be quite uncertain! Especially, smaller planets (or longer orbits) may have larger orbital eccentricity. Need RV followup & good orbital elements for Kepler/TESS targets. Will be somewhat easier for TESS, since stars are brighter & periods are typically shorter. Stellar variability can corrupt phase curve measurements! This is especially bad for long-period planets or very massive planets. HD 189733b is more variable than typical stars (even for its Teff) Important bias: we typically find the smallest planets only around the quietest stars -- this makes things easier for JWST! For secondary eclipses, need good RV elements! Ricker: SDSS M dwarfs are active for first 11. Slit losses will be ~few percent, and chromatic. Aperture is 16 pixels. Pixels are 100 mas. JWST jitter is ~10 mas (~0.1 pixel). Antenna repointing will perturb by ~100 mas (happens every ~3hr) but fine-guidance should bring the pointing back to within <10 mas. Jitter may be okay, but what about longer-term drifts? NIRSpec noise sources: --Photon+readout (benchmark, noise floor) --Slit losses (b/c using aperture) --Intra-pixel sensitivity variations & undersampling --Flat-fielding? --Variable PSF? Intra-pixel effects could be as big as ~few x 10^-4 Pointing jitter & slitloss will be less, *at the best wavelengths* -------------------------------------------------- NIRISS Transit Spectroscopy Rene Doyon 2014-03-12 14:01 -------------------------------------------------- 0.7-2.5 micron. Weak lens to spread out light. Throughput not as good as NIRSpec, but higher spectral resolution. Two modes: Standard: J > 8.1 Subarray: J > 6.9 (maybe brighter with specialized subarrays) Inter-order flux contamination: ~5e-4, no matter what. Pupil "clocking wheel" uncertain rotation by ~0.15 deg -- induces movement of spectra of 3-5 pixels from one transit to the next. Transit simulations: achieve Kreidberg-like performance on GJ 1214b with 3 transits @ R~60, 0.7-2.5 micron. Ran simulations of TESS yields. Predicts very few GJ1214b-like planets, but many planets with depths <~ 0.4% and non-saturating. Finds that nearest HZ *deep transit* planet is ~20pc away. HZ Earth-like planet around M4.5 planet @ 13 pc takes <50hrs (<10 transits) asuming photon-limited precision. (unlikely!) Earth around M5.5 @ 3.2pc (Ross 238), J=6.9: 138 hr (50 tarnsits) to detect H2O, CH4, CO2, O3. (uses Kaltenegger+2009 model). If planet is at 13pc, requires 350hr (150 transits) for much lower S/N. Need to get down to ~5 ppm for these targets! Starspots could masquerade as planetary transit spectrum -- but expected differences are much greater at optical wavelengths (e.g., TiO, etc.) Evaluated test observation cases: --"TESS-01" 1.4 R_Earth transit simulation. Star too bright (Kmag=5.5). Even if could observe, S/N is too low!!! Earth-twin around M6 with J=7 *might* be feasible. --"Teegarden's Star Phase Curve" -- 1.4 R_Earth around ~M6 star. Need ~1200 days of data. Amplitude is <1 ppm. Expects 6-hr noise of >25 ppm for best modes & wavelengths (after binning by 10-40x). At R~700, 6-hr noise is ~100 ppm. Lunine: it would be useful to run some realistic simulations for Jupiters/Neptunes. Answer: These take <1 transit. Arguments for observations of standard stars. -------------------------------------------------- NIRCam Short-period exoplanet Observations Tom Greene (& John Stansberry) 2014-03-12 15:24 -------------------------------------------------- 0.6-5 micron.Photometry (all wavelengths) and R~1700 slitless spectroscopy (>2.5 micron in 2 frames) Includes defocusing-lenses for bright-object photometry. Many combinations of filters available. Good sampling of detector (vs. spectrum)! Transit modes: --faint-object photometry (K=11.0) --bright-object photometry (Weak Lens 8): K=5.5. Could observe TESS-001 Magnitude limits: K=3.0 for long-wavelength spectroscopy K=6.1 for photometry JWST pointing will be ~< 0.2 pixel RMS. Well develop "sweet spots" Wheel repeatability induces rotation <~ 0.1 pixels Can get simultaneous long-wavelength spectroscopy & narrowband short-wavelength spectroscopy. (But: caveats! Including different cadence for spectroscopy & imaging) Same detectors as HST/WFC3: expect noise floor <35 ppm. Model observations of transits & eclipses: --Bright hot Jupiters: expect terrific S/N at good resolution. --Maybe hope for GJ1214b, too. --Maybe higher hydrocarbons for GJ 436b? Can't take spectra of GJ436b: too bright! Could observe TESS-001 with 2.4-4.0 micron spectroscopy Not clear whether Weak-Lens photometry would use aperture photometry or some version of PSF-fitting. (Lots of background!) McCullough: HST does very little transit photometry. Is this an especially useful observation mode? Swain: Maybe for very high-cadence dayside mapping. Beichmann: This is a very versatile observatory -- a bad aspect for high-precision calibration! I suspect we will need to focus on just using a relative few well-calibrated modes (per instrument). Much of this should be done before launch. -------------------------------------------------- MIRI short-period exoplanet observations Tom Greene (again!) 2014-03-12 16:10 -------------------------------------------------- 5-28 micron Imaging & Spectroscopy: R=5, 70, 2000 R~70 long-slit spectroscopy (0.5 arcsec width) R~2000 slitless (slicer-based!!) spectroscopy 5-12 micron IFU 11-29 micron IFU Many wide-, few ~medium-width filters. PSF shows diffraction "cross" shape because of array substrate regularities. Nyquist sampled for >7 micron. Transit modes: --R~100 (LRS prism) transmission spectra of big, bright planets --R~100 (LRS prism) thermal emission spectra cool-->warm planets (GJ 436b, probably GJ1214b too) --Photometric phase curves (~500K plnets with 770W filter): GJ436b, HD80606 --Emission of warm planets down to 2 R_E & 400 K (Imaging --Silicate features of fainter planets (MRS: Band2 B+C, 10-12 micron. Undersampled!) IFU spectra are 'chopped up' -- interleaved at each of 3 settings! (Like Keck/NIRSPEC at longer wavelengths). So need three events for a continuous spectrum. IFU 'spaxels' -- how much light is lost between the slicers? MIRI super-Earth secondary eclipses (Deming+2009) for near-habitable 2 R_E around M5V at 10pc. (Would take ~50 hours) Various subarray modes, down to ~0.07 sec. Reference pixels do not seem so useful in current data... Bright limits: K=6 for 8 micron (F770W) imaging. K=3-4 for 15 micron(F1500W) imaging K=3-4.5 for spectroscopy. JWST pointing jitter --> <0.05 pixel Technically possible to use imager during MRS observations Systematics: --Expect systematic noise floor <100 ppm (cf. Spitzer) --MRS is undersampled --Photometric stability reaches 100 ppm over ~1 hour (not photon-noise limited?!) --spaxel/slicer losses! -- Maybe PSF changes, etc. --2.4hr limit with 64x64 subarray! --5.5 hr limit with 'slitlessprism' subarray Background noise becomes important for K >~ 7.5mag Retrieval simulation by Mike Line (based on Fortney input model). Retrieves T/P profile. Errors in planetary radius can systematically affect retrieval. "Homework": GJ 436b dayside mapping: works well at F770W photometry! S/N ~ 30/hr (??). Probably need photometry to get time resolution of ingress/egress. Silicate cloud phase curve: MRS 7-12 micron spectroscopy: might require >120hr. (Only get 10-11.8 in one setting!) Ressler: When is background important? Did you include that? Greene: I estimated it; becomes important for K > 7.5 mag and lambda>10 micron. Greene: we have not yet mapped out spaxel/slicer maps. Swain: Could we dither a star around the IFU to map losses? Answer: We will measure that. Bouwman: We measured this in ISIM tests but we lacked a good point source. Now we have done this for 1 spaxel. We will investigate this. We will also have fringing. Ferruit: IFUs are why we aren't focusing on this (IFU) mode for transits with NIRSPEC. Valenti: If we can downlink FGS centroids, we still lack info on roll angle of observatory. How does roll translate onto detectors of all instruments? This would tell us whether this info will be useful for various observations and how much we need simulatneous imaging, etc. Benneke: Question for all teams: Currently we're talking about bright-star limit and systematics. But still unclear is that, in the end, integration-time efficciency is so important! So far, most space-based observations come close to photon limit. So we need to be *sure* that we collect most photons. Thus 33% vs. 67% vs. 100% efficiency matters a lot. Would like to see: efficiency vs. stellar magnitude for all instruments. Greene: You only lose this efficiency for the very brightest objects. E.g., 10x from the bright limit is ~90% efficient. Discussions about readout methodology: reset-read-read, or reset-read? -------------------------------------------------- Optimum Combination of Instruments 2014-03-12 16:52 -------------------------------------------------- Can we come up with a small set of instrument modes for a small set of targets for a representative set of observations? Valenti: I like high-spectral resolution. I would use HD189733 in all NIRSpec modes -- I can't, so I'll look at GJ1214b at highest resolution. Kempton: what does medium-resolution gain you over low-resolution? Valenti: I like to see individual lines & bandheads to get better constraints on thermal structure or atmosphere. HD209458b: borderline too bright. Fortney: not clear that we need high-resolution! Thus far we have tried to stitch together many narrow-coverage snapshots. I would *MUCH RATHER* have first observations offer broad wavelength coverage. I would love high-resolution too... OBJECT 1: high-resolution case in many chunks of spectrum. OBJECT 2: low-resolution, broad-coverage spectrum. Benneke: we need a full spectrum as wide as possible. Then we can neglect many stellar effects, 'recon' the spectrum, and follow up on anything interesting. Lunine: we may never observe these objects again. Most useful thing we can do is get a high-resolution spectrum across entire wavelength range. This gives best chance of diagnosing all aspects of atmosphere. Beichmann: No, we will come back to some of these planets. -------------------------------------------------- Broad-wavelength "synoptic" spectrum: NIRISS: 0.6-2.5 micron (J>8.1, R~700) NIRCam: 2.4-5.0 micron (2x settings; J~5-6, R~1700) MIRI: 5-12 micron, LRS, R~100 (J~8 NIRSpec: 0.6-5.0 micron (4x settings? or maybe 2x settings? J~8-9, R~1000) (NIRSpec? or NIRISS + NIRCam? ) So we need ~4 transits for a complete spectrum. ~25 hours Go for secondary eclipses (T/P profile) or primary (abundances)?? Kempton: secondary eclipses can be quite degenerate (see Miler-Ricci+2009) -------------------------------------------------- Lunine: it would be great to get some Neptune-sized planet spectra. Jovians will be great for comparison w/Juno & Cassini data ("standard stars" or "ground truth"). -------------------------------------------------- Kempton: One-mode, one-spectrum each for several sub-Neptunes w/known masses and radii. Maybe across transition from Neptune-like to real super-Earth-like. E.g., NIRISS: 0.6-2.5 micron. Just need to get down to ~40 ppm at R~100. McCullough: potentially, you could do this with HST. These targets do not exist yet! TESS + CHEOPS. -------------------------------------------------- -------------------------------------------------- MIRI Filters Phase/eclipse mapping? Periapse observations of eccentric planets? NIRSPec 2.9-5 micron. Get Methane, CO, CO2, H2O. MIRI/LRS 5-12 micron. -------------------------------------------------- Rieke: we will use on-orbit checkout to test these instruments. Beichmann: I'm not sure we'll be allowed! Benneke: On-orbit checkout transits should be immediate-public-access. Jitter & drift most likely to be a problem for MOS unit (with narrow slitlets) -------------------------------------------------- Smallest Planet JWST will Characterize Drake Deming 2014-03-13 08:39 -------------------------------------------------- Deming rephrases: Coldest & densest planet JWST will characterize? Transmission spectra: signal ~ T / rho Emission spectra: signal ~ T R^2 Simulated a water-vapor atmosphere for an Earthlike planet. --MIRI 11.3 wide filter. --1 R_Earth, 13pc, --> 0.36 micro-Jy --Star like GJ1214 --> 90 minute eclipses --MIRI sensitivity: 1.7 micro=Jy, 10-sigma, in 10^4 sec ---> Maybe get 3-sigma detection in 4 eclipses. -- *CHALLENGE IS ACHIEVING DYNAMIC RANGE*: depth of 2e-4 Beichmann: this won't work! You've ignored noise from the stellar flux. [Deming offers to bet bottle of Wine w/McCullough that this will work] Deming: we should just scale from HST: S/N ~ d ~ 2.4x Simulations of Natasha Batalha are too pessimistic, b/c her calculations predict that we shouldn't be able to detect transmission features in GJ 1214b. Her systematic noise assumptions are much higher than we get for HST. Scaling from Kreidberg results for GJ1214b: should be able to get ~4-sigma measurement for 300 K, earth-sized, water-atmosphere planet. conclusion: we can do it, if we find such a planet around a cool M dwarf! Lunine: We need to bound the possibilities! Simulator may be more pessimistic, but Deming's presentation may be too optimistic. Swain: In the past, we have always been able to push the instruments far better than anyone initially thought. There is a danger in being too optimistic, but we need not to be too pessimistic. Greene: I do not think the simulator is too pessimistic. There is no simulator noise floor. Ferruit: Simulator has never been cross-checked with instrument teams. Also: we *do* need to sell JWST, each & every year. Doyon: I looked into the habitable-case too, and 10 ppm photon-limited performance is is feasible with 10 transits. Lunine: Point is not to debate the simulator. --> White paper should have an explicit discussion about concerns re. the simulator. Knutson: remember that we don't do nearly well on absolute transit/eclipse depths as on relative (spectroscopic) cases. -------------------------------------------------- Lab Tests: Stability of NIRCam Grism Mode Chas Beichmann 2014-03-13 08:59 -------------------------------------------------- Lab testbed to quantify stability of NIR detector arrays (ELEKTRA, FINESSE, JWST, etc.). Simulating 'monochromatic grism spectra.' Project 'star field' grid onto detector. Can measure sub-pixel motions, intrapixel sensitivity variations, etc. Greene: how much of the residual noise is weird PSF vs. intrapixel effects? Smith: It is intrapixel effects. But we can decorrelate most noise. Ricker: This detrending approach is similar to that advocated by Jon Jenkins for Kepler data pipeline. Beichmann: Yes. Kepler was so stable that it did not need a flat field. K2 will move around enough that flat-fielding becomes important again. Updated simulator includes "star" and "planet" in pseudo-spectral format. Injects fake 'planet spectrum' using a simple mask with various 'lines' with different shapes & amplitudes. Decorrelate data using simple PCA: take out variations in lamp, motions on detector, etc. Already see some insight that we need to take data *continuously* -- when we take 100-frame chunks, we see reset anomaly at beginning of every chunk. Median behavior: can bin down to ~75 ppm, maybe as low as 52ppm. Maybe could do better without 100-s readout-reset anomaly. Swain: How close is testbed to actual NIRSpec electronics? Beichmann: Soon we will get a flight-like detector. It will be a while before we move to 'sidecar' readout electronics: it is expensive! (many people): Your main noise source is related to your (non-flightlike) readout electronics. Very important to "fly what you test, test what you fly." Smith: I agree and disagree. We see the same sort of 'ramp' that many other people have shown. What is the physics behind cadence of readout waveform interaction with ... (everything else)? McCullough: note that this is all monochromatic, at present!! Doyon: Would be interesting to simulate a NIRISS-like situation. Greene: Can you change sampling of system, F-number, etc.? Could to NIRISS, NIRSpec, NIRCam... Smith: When we move spot across detector, we see flux modulation. McCullough: You're not flat-fielding! So you might not be seeing intra-pixel variations, but just flat-field variations. -------------------------------------------------- Data Simulations Stephan Birkmann / Jeff Valenti 2014-03-13 09:28 -------------------------------------------------- Various instrument teams need to agree on common guidelines for inputs (readout rules, instrumental effects, how to present results, etc.) Examples: --PPM vs. wavelength at instrumental and/or binned spectral resolution. --Saturation vs. magnitude Proposes a *small* working group of 1-2 people per instrument. Greene: Yes! We have to do this. Valenti: Yes. it is essential to present consistent results across all instrument teams. Ferruit: HOw do we use the momentum from this meeting to get things running? Once you have use cases you must agree on inputs assocciated with that. Then challenge instrument teams to do the best based on their instrument parameters, and answer 'what they can do' for a given use case. Then can return to proposers and see how they react to the instrument's offer. Greene, et al.: We should agree on *ONE* model and forge ahead with that. But this is probably too much work for the imminent white paper. Christiansen: Maybe everyone could look at HD189733b? -------------------------------------------------- Community Data Challenges Jeff Valenti 2014-03-13 09:39 -------------------------------------------------- Types of simulated data: --raw images (w/instrument artifacts) --Reduced spectroscopy/photometry (limited by model fidelity!) Types of analysis: --Blind analysis --Recover inputs w/knowledge of input Benefits of Data Challenge: --Forces you to write down all data quality issues & model parameters --Build familiaritaty with observing templates & standard data products --Formulate observing strategy, based on: - - Ability to remove instrumental effects (persistence, ghosts, jitter) - - Ability to recover input parameters (composition, thermal structure, etc.) --Inform TAC (which proposals are feasible/useful?) --Facilitate pipeline/algorithm development Advantages: --Learn to write better proposals --Become 'recognized expert' --Advertise instrument capabilities --Obtain funding (??) Disadvantages: --Could help the competition! --Takes a lot of work. We will *probably* release simulated data. Deroo: We need an end-to-end simulation challenge. Input model, simulated data, reduce data, retrieve model parameters. Swain: I strongly encourage you to release the simulator! This allows maximum investigation of instrumental systematics. This is *REALLY* important for error budgets. McCullough: Maybe try this simulation for HST data. Is it anywhere close to what we actually observe? Swain: I agree!! Validation of the simulator is key. Ferruit: We cannot reliably generate HST data. McCullough: If you can't simulate something you *KNOW*, how can you trust anything you simulate? You must validate simulations against some sort of ground (or space) truth. Ferruit: A simulator is tuned to its instrument. It would be major work to write a new, separate HST-data simulator. Swain: If you can't validate truth of simulated data, what's the point? Valenti: Is it worth simulating a likely-incomplete system? Ferruit: NIRSpec is interested in simulating data for a challenge. --------------------------------------------------- Pipeline Data & Processing Challenges Pieter Deroo 2014-03-13 09:59 --------------------------------------------------- Ideas to improve our ability to correct data. We should make observatory diagnostic data available for all instruments! And make it easy to access: e.g., query "All spectra of stars ~30,000 hours per 5 yr. Guest Observer (G.O.) Program: details TBD GTO: 4000hr allocated in first 2.5yr (~10% of five years) DDT Time: allocated as ~10% of available time. "First-look" or "Early-release Science" proposals: might be possible in Cycle 1, in order to prepare for Cycle 2 proposals. Data are non-proprietary. Size/scope TBD! Not yet certain who will recommend/decide these programs. Supported by DDT, so could conceivably be ~large. -------------------------------------------------- What do we need from Commissioning & Early Science for Planning? 2014-03-13 11:28 -------------------------------------------------- ERS will serve several purposes: -- PR -- science & validation Greene: we want a signal. But maybe measuring a flat spectrum (for validation) Benneke/Rieke: we want to do something where we know we will succeed! And show what JWST can do that no other facility can do. Benneke: we should look at a star with little/no activity. Beichmann: detector problems are important. So we need to push hard on systematics: one NIR, one mid-IR. Rieke: there are big differences between readout electronics. Experiences with one instrument is not exactly applicable to other instruments. Ferruit: Sampling varies considerably between instruments Fortney: Secondary eclipse spectrum, e.g. 3-5 micron to cover previously observe IRAC photometry (NIRSpec/NIRCam). Knutson: eclipses are less susceptible to stellar activity, limb darkening, etc. Doyon/Rieke: We need to get SWG support for an exoplanet working group to make these sorts of decisions. ERO: PR, *small* programs, embargoed for a while. ERS: bigger; based on DDT, KNutson: if we want to advocate for this time, who must we pester? Lee: It must be a recommendation from the "JSTAC" (chaired by Garth Illingworth). Sara Seager is also a member. Valenti: we should want short-period planets near the ecliptic poles. Dressing: all precursor missions will come too late for these early observations. Bean: Yes, but a few observations of a few hot Jupiters should easily be feasible. Rieke: Yes: include some 'dead-certain' sources. Beichmann: is the minimum idea to get one eclipse w/each instrument? We don't need the absolute sexiest object. Knutson: We should prioritize based on EARLIEST POSSIBLE observations, rather than the absolute most interesting object. Bean: Probably using the 'four key modes' we identified yesterday. NIRISS: 0.6-2.5 micron (J>8.1, R~700) NIRCam: 2.4-5.0 micron (2x settings; J~5-6, R~1700) MIRI: 5-12 micron, LRS, R~100 (J~8 NIRSpec: 0.6-5.0 micron (2x settings? J~8-9, R~1000) Maybe NIRSpec with one setting and NIRCam with the other, to cover full wavelength range and then get one observation from each instrument. Beichmann: commissioning is about testing the instrument. Brown dwarf variability will not do this. Lunine: It would be interesting to have another, small workshop to investigate what JWST could do for Brown Dwarf/substellar variability. Knutson: BDs are fainter than exoplanet host stars, so observing modes tested woul be different for this case. Valenti: What about a one-setting observation of GJ 1214b? Doyon: Better if we can choose something that will show a signal. Bean: the field will change dramatically over the next 3 years! Is there any hope for a large-scale Treasury program proposal call in ~2016? You might not exactly know your noise floor, but you at least know that you have great new spectroscopic capabilities. McCullough: There is risk for such an idea. Valenti: the community should engauge the GTO teams! They have first-right of target selection (and understand the instruments). Valenti: I would like GTO data to go public before Cycle 2 Ferruit: GTOs will not try to 'block' programs but rather do good science. McCullough: Could we make out-of-transit data immediately public, but make proprietary the data in-transit? Valenti: Good idea! -------------------------------------------------- Plans for the White Paper 2014-03-13 13:09 -------------------------------------------------- ---> Potential Science Programs across instruments (J. Lunine) ---> Transit "Best Practices" (J. Christiansen) ---> MIRI: Instrument modes for transits (only me!) Lots of Tables & Figures! Send all notes to Ellen for Wiki. --Coordinate with co-authors & presenters for my sections by mid/end of April. --Agree on "homework assignments" test cases. Send format in LaTeX or do it 'in the cloud' http://www.stsci.edu/jwst/doc-archive/white-papers Big Science questions: --are giant planets metal-enriched? --how does this change w/panet mass, stellar type, migration history --what is impact of snow line condensation on abundance deviations? --how far out of radiative equilibrium are atmospheres? --what modulates surface brightness distributions (metallicity, rotation, incident flux) --Cloud opacity: what are they made of? How effect thermal emission? --Deviations from equilibrium chemistry --Vertical/horizontal mixing & chemical homogeneziation --Exoplanet analogues: Y dwarfs --Is there a clear dividing line (or any discriminators) between H2-dominated and heavy molecule-dominated super-Earth atmospheres? --How do rocky-planet atmospheres behave in tidally locked case? E.g., for habitable systems. --Look for hydrocarbon products w/high-resolution spectroscopy (HCN, C2H6, etc.) --Two-band (in/out of water) photometry of many super-Earths, 1-2 transits per target, for a large sample of super-Earths. (Recon program) --What is eccentricity distribution (and ecc for individual planets) for possible secondary-eclipse targets? J. Valenti: I don't recommend requesting 300hr in Cycle 1. People to talk to: J. Fortney