Bright Earth Persistence in NICMOS Adam Riess Eddie Bergeron Acknowledgements: Megan Sosey, Dave Grumm, Helene McLaughlin, Deepashri Thatte, George Chapman, Anton Koekemoer for helpful discussions (Another) NICMOS Anomaly NICMOS Persistence well-known (e.g., CRs post-SAA, planets, etc) + Limb of the bright Earth (NICMOS has no shutter) = Bright Earth Persistence (BEP) What’s new about BEP? Whole field pattern. Characterized in SHOES (GO 10802, P.I. Riess) • SHOES used NICMOS Camera 2 and ACS in parallel Visit b2, December 2006: large-scale pattern in sky, unlike Dark or Flat 1st dither Last dither Diagnosis: Bright Earth Persistence • SMS: B1 preceded B2, ACS dump to SSR delays NIC dump, NIC transition from “observe” to “operate” and placement of blank. NIC saw bright Earth for 45-70 seconds, array saturated • “Map” of trap density decays ~20% per 400 sec (consistent with Bergeron & Dickinson 2003) Good News: BEP can be modeled and removed Step 1: Make BEP Frame Using 10 clear examples of BEP in SHOES, register subtract matching frame w/o BEP & w/ sources, produce mean Target Field BEP Visit Matching visit w/o BEP NGC3370_green B2 B1 NGC3021_cyan 6F 6E NGC1309_cyan2 5D 5J NGC1309_green 59 5A NGC4639_cyan 1E 1D NGC4536_blue 24 25 NGC3982_blue 35 3G NGC3021_red 62 6H NGC3021_green 6B 6A NGC4536_green 27 28 BEP in Camera 2 •Level is arbitrary, mean in middle of impacted orbit, 0.047 e/sec, similar to true sky level • Ring median this to remove grot (from dithering, GS reacq) Persistence from Various Sources of Excitation BEP CR’s post-SAA Post Lamp •CR source different along edges •Signal-to-noise much better in BEP due to saturation, coverage BEP Removal • Why? Mottled, irregular sky impacts photometry, source shapes. And its highly predictable, ~noiseless removal yi s Pxi ni Image sky persistence BEP noise y1 1 x1 n1 y2 1 x2 s n2 P yN 1 xN nN Least-squares estimator from matrix inversion (X X) X Y T 1 T =[sky,persistence] •Different than SAAclean, not local--RMS around CR’s, global-all pixels •Leaves sky, easy to mask out sources pixels (rmedian, threshold, corners) •Run calnica and pedsub first Subset of SHOES Persistence Level Sky level Fraction of Pixels used Overall ~5% of SHOES orbits impacted, ~10 of 200 Before and After BEP corrected VS. Unaffected Example 2: GO 10258 (P.I. Kretchmer, GOODS NIC+ACS) Persistence Level Sky level Fraction of Pixels used 2 of 20 visits, 09 and 12 Result 99.4% Variance removed Some residual--fit errors,T dependence, angle illum, order of ops Other Cameras Camera 1 Camera 3 Post-SAA CR’s post-Lamp In Progress • Looking for examples of BEP in Camera 1 and 3 • D. Grumm coded algorithm in pyraf for STSDAS, D. Thatte experimenting with parameters • First Do No Harm: apply no correction if small or < 50% of pixels useful • Identifying conditions to predict BEP (G. Chapman) • Commanding: GO’s can insert blank, procedural change? • WFC3 has no shutter and similar detectors…