TIPS-JIM Meeting 27 January 2005, 10am, Auditorium 1. Cycle 14 Statistics Brett Blacker 2. NICMOS Grism Flux Calibration and Non-linearity Ralph Bohlin 3. The HST Pointing Drift Anomaly Stefano Casertano 4. MultiDrizzle Development Status and Plans Anton Koekemoer Next TIPS Meeting will be held on 17 February 2005. TIPS-JIM Meeting 27 January 2005, 10am, Auditorium 1. Cycle 14 Statistics Brett Blacker 2. NICMOS Grism Flux Calibration and Non-linearity Ralph Bohlin 3. The HST Pointing Drift Anomaly Stefano Casertano 4. MultiDrizzle Development Status and Plans Anton Koekemoer Next TIPS Meeting will be held on 17 February 2005. The following are checks on the NICMOS grism non-linearity. • Curvature in the non-flat fielded background subtraction over the region between the upper and lower background areas causes <1% error. • The non-linearity is still present for the brighter stars, even for no background subtraction. • The same results are obtained using the independent NICMOSLOOK spectral extraction package developed at the ECF by W. Freudling. • The differential count rates are constant between the NICMOS readouts and agree with the cumulative count rates to <1%. • The spectra overlap between dither positions and are susceptible to persistence; but the first dither positions show no systematic deviation. • Comparisons of NICMOS photometry of galaxies vs. ground based photometry is inconclusive (Mobasher). • Comparisons of NICMOS photometry of stars vs. ground based photometry on NIC2 has about 5 times less slope than the NIC3 grism non-linearity (Riess). • A proposal to compare photometry for the brightest and faintest stars has been submitted by the NICMOS group. • Grism spectra of 3-4 faint stars in the G191B2B field have NIC3 photometry for a direct comparison with the spectrophotometry. TIPS-JIM Meeting 27 January 2005, 10am, Auditorium 1. Cycle 14 Statistics Brett Blacker 2. NICMOS Grism Flux Calibration and Non-linearity Ralph Bohlin 3. The HST Pointing Drift Anomaly Stefano Casertano 4. MultiDrizzle Development Status and Plans Anton Koekemoer Next TIPS Meeting will be held on 17 February 2005. The HST “Observer Bias Convergence” Pointing Anomaly TIPS, 27 January 2005 Stefano Casertano With input from: Russ Makidon Ed Nelan Ron Gilliland Bill Zmek (Goodrich) The Observer Anomaly Working Group (STScI, GSFC, LMCO, Goodrich, L3) 21 ! ! ! ! ! What is the Observer Bias Convergence Anomaly? Do we understand how it arises? Does it affect science observations? Ongoing work To find out more: " http://edocs1.hst.nasa.gov/ota/Observer/Observer.html 22 What is the anomaly? ! ! ! ! The gyro drift rate appears to vary rapidly, especially during the transition from night to day The variation is seen in the V2 axis, and affects primarily Gyro 4 The variation was detected starting around June 2004, and is seen consistently since then The amplitude of the variation appears to be growing with time 23 An example of the anomaly nigh t da y nigh t da y V2 V3 Nighttime V2/V3 data (mean-removed) are strongly correlated; roughly 2-to-1 ratio (spikes are anti-correlated) V2 is poorly correlated to V3 during day Night-Day pattern repeats 24 What is it? (cont.) ! ! ! ! Total rate variation (interpreted as gyro bias drift) up to 40 as/hr in extreme cases Note that there is no change in pointing - the telescope’s optical axis is held steady by the FGS No obvious preference for one FGS over another Some apparent periodicity pattern vs. time - possibly due to observing program (still under investigation) 25 Gyro bias measured 20 min after fine lock 26 Are there impacts on HST operations? ! ! ! ! No immediate impact; there could be occasional losses of lock (none significant to date), but image quality is unchanged [possibly with some exceptions] Concern about HST health (possible thermal effects due to insulation degradation etc) If the drift continues to grow, eventually will affect reacquisitions (at about 70 as/hr) NOTE: impact may be reduced in 2-gyro mode 27 What is really going on? ! ! ! We know that the telescope optical axis axis is steady on these scales (no significant increase in PSF size) but Gyro 4 reports that the telescope is moving Three basic possibilities: " " " The optical axis is moving, and the inertial axis is not; the gyro reading is correct, and the FGS compensate for the true motion of the spacecraft The gyro is malfunctioning, and registers a motion where there is none The gyro mount (the equipment shelf) is moving with respect to the telescope; the gyro reading is correct, but it registers a local motion in the telescope that does not affect its functioning 28 Can we detect an optical tilt? ! If the optical axis is moving with respect to the telescope, the most likely reason is a tilt of the secondary " " " Tilting the secondary produces coma Coma can be detected by phase retrieval (ACS/HRC focus data) and in the FGS s-curves The expected effect is ~ 12-25 nm rms 29 ZEMAX Analysis: SM Drift Due to Pure Tilt Detachment of bipod* removes constraint on tilt, allowing drift; no decenter S M SM tilt necessary to reach 3 arc-seconds in object space was calculated SM tilt ! = 0.00377 degree Aberration: all coma, 0.012 um RMS Strehl loss ~ 2.3% @ 0.5um wavelength 4907.1 mm S = 6407.13 mm ! P M Focal Plane " " ~ 2 ! S = (3 arc-seconds) / (plate scale, arc-sec/mm) * Speculative 30 Zemax analysis - SM Drift Due to Metering Truss Strain Metering truss assumed to strain in ‘banana’ form* (thermal cause?) Quadratic bend assumed SM optical axis assumed to remain tangent to centerline of truss (see sketch) SM deflections necessary to reach 3 arc-seconds in object space were calculated Decenter result: $ ~ 60 um Tilt result: # ~ 0.0014 degree Aberration: all coma, ~0.025 um RMS Strehl loss ~ 10% @ 0.5um wavelength $ S M # s = 4907.1 mm ! S = 6407.13 mm $ P M Focal Plane # ~ 2 $/ s ~ 2 {# + $/ROC ! " SM } ROCSM = 1358.065 mm " ~ ! S – $ = (3 arc-seconds) / (plate scale, arc-sec/mm) * Speculative 31 Coma traced from focus data •Coma has varied systematically over time •Typical variation within an orbit much less than predicted •Y coma correlates with focus within a single observation – suggests deformation associated with breathing 32 Predicted change in S-curve with coma 33 34 Comparison of day and night fringes with FGS1R - no change Xaxis Yaxis 35 However... ! ! ! ! It is possible for tilt and decenter to compensate and produce a smaller coma variation Resulting optical effect undetectable at field center Astigmatism produced away from field center (at HRC, FGS) Some astigmatism change seen in focus data 36 Focus observations - anomaly (June 2004) 37 Phase retrieval (June 2004) •Focus changes smoothly through observation •No jump at day-night transition •Change consistent with breathing model •Y coma changes at transition, much less than predicted •X,Y astigmatism also change at transition 38 Further analysis needed ! ! ! ! Astigmatism measurement has large measuring error Astigmatism changes occurred before the anomaly In principle, can determine exact 3D position of secondary from focus data Optical modeling by Makidon and Zmek continuing 39 Focus adjustment •Concerns about secondary have delayed focus adjustment •Move executed December 22, 2004 (+4.16 mm) •Current focus nominal December 18, 2004 (-4 µm) January 14, 2005 (~ 0 µm) 40 How about gyros? ! The bias drift around V2 is almost entirely on Gyro 4 " " ! Analysis of the output from Gyro 4 shows no anomalies " " " ! ! Expected effect on 1, 2 is small Could be due to Gyro 4 itself Digital output remains normal during supposed bias variation; behavior similar to 2002 data No changes in data available through telemetry (temperature, input voltages) Other forcing functions (e.g., magnetic) do not have correct time scales Manufacturer concludes that gyro anomaly is unlikely More detailed analysis could separate V2 and Gyro 4 errors (some sensitivity to V2 in the other gyros) 41 Tilt in equipment shelf? ! ! ! Gyros and FHST are mounted on equipment shelf Shifts in shelf are believed to be the cause of V2 disturbances (impulsive tilts that affect telescope pointing) However: " " " " Shelf is located on -V3 side of Aft Shroud Not expected to respond to day-night temperature cycles Time behavior of anomaly inconsistent with predicted temperature profile No long-term drifts in temperature data 42 Location of equipment shelf FGS 2 STIS COSTAR FGS 3 FGS 1 NICMOS ACS RSU 2 FHST 3 RSU 3 FHST 1 FHST 2 43 Location of shelf (cont.) 44 Equipment shelf temperature trends 20 18 16 14 12 10 8 6 +V2 ESHLF1T E339 Mean +V2 ESHLF7T E520 Mean -V2 ESHLF3T E324 Mean -V2 ESHLF6T E519 Mean -V1/+V3 ESHLF2T E323 Mean -V1/+V3 ESHLF4T E425 Mean -V1/+V3 ESHLF5T E426 Mean 4 2 0 37257 37288 37316 37347 37377 37408 37438 37469 37500 37530 37561 37591 37622 37653 37681 37712 37742 37773 37803 37834 37865 37895 37926 37956 37987 38018 38047 38078 38108 38139 38169 38200 38231 38261 45 Other temperature trends ! ! Thermal analysis of sensors at light shield, forward shell, and metering truss assembly shows no obvious long-term changes More detailed analysis in correlation with anomaly needed 46 Also of interest… ! ! ! ! ! A small variation in PSF width has been detected in GO data (Gilliland, GO 10441) Variation appears to correlate with day/night cycle Possible jump in detected jitter when variation occurs Telemetry and pointing data are being analyzed Relation with Observer Bias Correction Anomaly unclear 47 Fault Tree Analysis ! The anomaly could be due to many causes: # # # # # # # # ! ! OTA outer shell or metering truss Secondary Mirror Primary Mirror SA3 Gyros Gyro Equipment Shelf FGS Flight Software At this point, Primary Mirror, SA3, FGS, Flight Software appear unlikely Further analysis required for OTA, Secondary, Gyros, Equipment Shelf 48 Summary ! ! ! ! No credible cause established for the Anomaly Effect is growing over time; may become a concern in ~ 6 months No current science impact Possible investigations: " " " " Optics: measure position of Secondary over time Thermal: correlate temperature data with variations Equipment shelf: study V2 anomalies, FHST data Gyros: study response of Gyros 1 and 2 49 50 Serendipitous monitoring from GO data •Ellipticity pattern of PSF depends on focus position •Pattern measured from GO observation of stellar field •Before focus adjustment: pattern consistent with model at -5 mm •After focus adjustment: pattern consistent with nominal focus •Note: deviant points are not significant (cosmic rays or other artifacts) 51 Support of Observer Bias Anomaly analysis ! Observer Anomaly: increased values of gyro drift rate are seen at night-day transition while on FGS lock " " " " ! Drift rate corrections (V2) up to 40 as/hr observed since June 2004 RMS correction up to 12 as/hr (<5 a year ago); some temporal structure over time scale of 1-2 months, possibly due to frequency of day-night transitions Worst-case disturbances correspond to an apparent integrated displacement of gyro to optical axis of about 3” in a few minutes No significant impact yet on science, efficiency, image quality Possible sources include: " " " Motion in the OTA metering truss and/or secondary mirror Gyro anomaly Motion in support structure (e.g., equipment shelf) 52 Observer’s Anomaly (cont.) ! Image data useful to assess possibility of optical displacement " A 3” displacement of the optical axis due to secondary tilt induces coma (~ 12 nm rms) and astigmatism # # # # Phase retrieval on focus data and FGS S-curves place upper limits of ~ 4 nm rms on coma variations during the anomaly A concurrent decenter of the secondary, due to a “banana” distortion of the metering truss, can reduce the coma impact Astigmatism measurements noisy; analysis not yet final Eventually, coma and astigmatism could help trace the position and tilt of the secondary vs. time 53 TIPS-JIM Meeting 27 January 2005, 10am, Auditorium 1. Cycle 14 Statistics Brett Blacker 2. NICMOS Grism Flux Calibration and Non-linearity Ralph Bohlin 3. The HST Pointing Drift Anomaly Stefano Casertano 4. MultiDrizzle Development Status and Plans Anton Koekemoer Next TIPS Meeting will be held on 17 February 2005. STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 55 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) Current Status $ STSDAS / Pipeline Near-Term Priorities $ $ $ Continue MultiDrizzle maintenance & testing Develop/test Tweakshifts (for off-line & pipeline use) Expand definition of ACS associations Longer-Term Strategy $ $ Science Working Group – prioritization / planning MultiDrizzle enhancements: – Refine CR rejection algorithms (CTE, single-image, etc) – STIS spectroscopy – NICMOS iterative background correction $ Archive/VO-related improvements: – Further expand associations to include multiple visits – Allow MultiDrizzle parameter changes in archive interface STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 56 Current Status STSDAS MultiDrizzle - current v2.4.2 (can be used off-line): $ $ $ $ ACS: all observation modes WFC, HRC, SBC, fully tested WFPC2: all observation modes, partially tested STIS: all imaging modes (CCD, MAMAs), fully tested NICMOS: all observation modes, testing in progress Pipeline MultiDrizzle - current v2.3.6 (runs in OTFR): $ $ $ $ Same code base as STSDAS version, frozen every ~3-6 months Runs on all ACS associations in pipeline Very few problems since OPUS15.4 installation in September 2004, all have been addressed or are currently being resolved MultiDrizzle pipeline products include: – CR information in FLT file DQ arrays – Improved astrometric header keywords in FLT files – Header values in drizzled image containing history of drizzle parameters STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 57 Near-Term Priorities - MultiDrizzle MultiDrizzle Maintenance: $ $ Improvements in response to user problems (often installation-related) Direct and efficient interaction between OPUS and SSB groups to resolve pipeline-related problems: – Occur rarely – Typically due to problems with data, eg science exposures with EXPTIME=0 MultiDrizzle Testing: $ Functional / regression testing (SSB: Hack/Hanley/Jedrzejweski): – Run a subset of datasets nightly, the full set is run once/week – Ensure that the code still successfully executes after modifications – Augment with new datasets that have revealed problems in the code $ Scientific testing (INS): – ACS (Gonzaga), WFPC2 (Platais), STIS (Dressel/Davies), NIC (Bergeron) – Explore optimum parameters to ensure good default/pipeline behaviour for the widest possible range of datasets – Quantify accuracy of astrometry, photometry, CR rejection STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 58 Near-Term Priorities - Tweakshifts Shift measurement method: $ $ $ catalog-based (currently DAOfind, expect later to also use SExtractor) cross-correlation wavelet transforms Two basic uses for Tweakshifts: $ $ Refine relative shifts between exposures in an association Refine absolute astrometry, by comparing with external catalog (eg GSC2) Initial pipeline implementation – improve relative shifts: $ $ Initially for data with guidestar problems, where header shifts can be bad If robust after extensive experience, may extend to all ACS data Absolute astrometric improvement: $ $ Most ACS images have several GSC2 objects % can improve to <0.3-0.5” absolute astrometry by using Tweakshifts in catalog-based mode Currently exploring pipeline implementation (Koekemoer/McLean/Jenkner) STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 59 Near-Term Priorities- ACS Associations Current definition of ACS associations: $ $ $ Observations obtained using dither pattern or CR-SPLIT Exposures are not associated if obtained using POS TARGs Often, observers use POS TARGS either because: – they are confused by the dither patterns – they have a specific need not covered by the dither patterns – BUT they still expect the data to be associated and combined Proposal to expand ACS associations: $ Build association from all exposures that satisfy the following: – same filter – within the same visit $ Benefits of initially restricting this to exposures in the same visit: – same guidestars, therefore header shifts can be used – provides limitation on the size of the largest offset (less than ~1’) $ $ TRANS implementation for future observations (tbd with Tony Krueger) Archive implementation for previous obs’s (talks with ASB, CADC, ECF) STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 60 Longer-Term Strategy Science Working Group – prioritization / planning $ $ $ $ Cross-section between Dither Group, SAR initiatives, and discussions with HST MO - current discussions on-going with SD mgmt Develop “wishlist” of ideas Propagate to ESS for resource estimates Carry out subsequent prioritization, interacting with HST MO MultiDrizzle enhancements: $ Refine CR rejection algorithms: – CTE – single-image (useful for datasets with only 2-3 exposures) $ $ STIS spectroscopy NICMOS iterative background correction Archive/VO-related improvements: $ $ Further expand associations to multiple visits (incl more Tweakshift work) Allow MultiDrizzle parameter changes in archive interface STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 61 MultiDrizzle Enhancements – CR rejection Current CR rejection scheme: $ Uses driz_cr, which has the following features: – essentially based on sigma-clipping, combined with a slight “softening” from flux gradients in the image to avoid clipping bright concentrated source – works best for large numbers of images, although with sufficient testing & exploration of parameter space, also gives acceptable results for 3-4 images $ Main limitations: – in 2-3 image datasets, doesn’t always succeed when there is just 1 good pixel – doesn’t reject CTE tails on CRs – to work well, need to create sub-sampled image % resource-intensive Two possible improvements: $ CTE tails can be rejected in a second more stringent pass, examining only pixels along the read-out direction from CRs identified in the first pass: – relatively easy to implement – some preliminary experiments have been done by Koekemoer/Busko $ Consider rejecting CRs on single images, eg by Laplacian edge detection, prior to running MultiDrizzle STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 62 MultiDrizzle Enhancements - STIS Current situation: $ $ MultiDrizzle works for all STIS imaging modes All 3 STIS cameras: CCD, NUV-MAMA, FUV-MAMA Eventual goal – spectroscopy: $ $ $ Many observers dither along the slit to mitigate hot pixels; CTE from hot pixels is along slit direction, thus dithering is important in obtaining good spectra Some observers also dither along wavelength direction to better sample the spectral line spread function Some studies currently underway within STIS group (Dressel/Davies): – examine changes required (e.g., new keywords) as well as the extent of code modifications – examine scientific usefulness of incorporating this capability: would the resulting products be usable as-is for the majority of science data? $ The STIS studies will determine whether or not to proceed with this STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 63 MultiDrizzle Enhancements - NICMOS Current capabilities: $ $ MultiDrizzle can now read all types of NICMOS data, and can perform the basic tasks of CR rejection and image combination Main limitation is that background correction is treated the same way as for optical CCDs, which is incorrect for NISMOS Improved background correction: $ $ $ $ $ $ Create a first-pass image, assuming static background correction Run object detection / identification on the resulting combined image Use the objects to create a “mask” excluding any regions with flux For each exposure, fit the remaining unmasked background and subtract Re-do the drizzle combination Iterate the above if necessary Status: $ Tests underway to investigate required software changes (Bergeron/Hanley) STScI TIPS/JIM 27 January 2005 MultiDrizzle Development Status & Plans Anton Koekemoer (INS) 64 Archive/VO-Related Enhancements Absolute Astrometry improvements: $ Experiments to date (Koekemoer/McMaster/McClean) find improvement in absolute astrometry accuracy to ~0.3”, ie about 10x improvement Further expand associations – include adjacent visits $ Benefits: – Move HST data products from large mosaic programs into the “VO era” $ Two potential concerns: – different guidestars, and the need to very accurately align to < 0.1 pixel; can be solved in principle by running Tweakshifts in a specialized fashion – Processing resource issues (much larger images than currently); can be solved by letting the user specify a central RA,Dec and a limiting radius around this Parameter control in archive interface: $ $ Can further increase scientific value of products (observers can select exactly the parameters they want, eg North=up, pixel scale, etc) Concerns about processing load – can be addressed by limiting the options TIPS-JIM Meeting 27 January 2005, 10am, Auditorium 1. Cycle 14 Statistics Brett Blacker 2. NICMOS Grism Flux Calibration and Non-linearity Ralph Bohlin 3. The HST Pointing Drift Anomaly Stefano Casertano 4. MultiDrizzle Development Status and Plans Anton Koekemoer Next TIPS Meeting will be held on 17 February 2005.