Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/nicmos/documents/isrs/isr_2009_005.pdf Дата изменения: Unknown Дата индексирования: Tue Feb 5 16:39:43 2013 Кодировка: Поисковые слова: п п п п п п п п

Instrument Science Report NICMOS 2009-005

NICMOS Temperature Dependent Flat-fields
Deepashri Thatte, Tomas Dahlen June 30, 2009

A

BSTRACT

The bias-derived detector temperature for all three NICMOS cameras showed a variation of about ~2K in both the era before the installation of the NICMOS Cooling System (NCS) 1997-1998, as well as in the post-NCS era, 2002-2008. This temperature variation affected the detector quantum efficiency (DQE) and hence the structure of the flat-fields. In this ISR we discuss the creation of new temperature dependent flat-field reference files that correct for the effect of temperature on the DQE. Temperature dependent flat-fields (*_tdf.fits) populate the keyword TDFFILE in the headers of post-NCS data retrieved after November 19, 2008 and pre-NCS data retrieved after January 23, 2009. The temperature dependent flat-fields are used by the FLATCORR step in calnica version 4.4 and later. For consistency check we also created epoch dependent flatfields for postNCS data and compared them with the temperature dependent flatfields. The result of this analysis is also included here.

1. Introduction
The detector quantum efficiency (DQE) of NICMOS detectors varies both on pixel-to-pixel scale and on large scale across the chip. This causes the science data to have large spatial variations even for a case with uniform illumination. Science data are corrected for this variation by multiplying by a flat-field reference image that takes into account the pixel-to-pixel DQE variation.

1


The DQE is furthermore a function of detector temperature (de Jong, 2009, NICMOS ISR, in preparation). A uniform change in the DQE across the chip due to a temperature change does not change the flat-fields since these are normalized to unity by construction. There are, however, higher order effects that will change the structure of the flat-fields. This is due to a relatively different response for individual pixels to a temperature change. E.g., hot pixels respond relatively more to a temperature increase than cold pixels. This will cause a change in the structure of the flat-field even after normalizing to unity. The instrument temperature varied throughout Cycle 7 and 7N (pre-NCS era, operating temperature ~62K) and in Cycle 11 and beyond (post-NCS era, operating temperature ~77K). The variation was approximately 2K in both eras. We therefore expect that the flat-field structure also changed due to this temperature variation. To correct for temperature dependent changes in the structure of flat-fields, we have created temperature dependent flats-fields for all filters in NIC1, NIC2 and NIC3. These temperature dependent reference files each consist of five different flat-field extensions, each valid at a different temperature range. Calnica, version 4.4 and later now uses these reference files to choose the flat-field extension appropriate for the temperature of the file that is being reduced. The effect of this temperature dependence on the photometry is at most ~1% and depends on the position on the detectors. Users who use data retrieved earlier than 2008 and for whom this effect is significant, should re-calibrate the data with calnica version 4.4 (or later) and the latest reference files or retrieve the data again from the archive using the OTFR. In this report we discuss the creation of the new temperature dependent flat-field reference files for all three NICMOS cameras, for both the pre-NCS and post-NCS eras. The detector temperature used here is the bias-derived temperature calculated by PyRAF task CalTempFromBias. This temperature populates the keyword TFBTEMP in primary headers. Temperature dependence of NICMOS bias levels is used by this task to calculate detector temperature for MULTIACCUM observations (Bergeron, 2009, NICMOS ISR, in preparation). The detector temperature measured by this method is believed to be more accurate than the temperature measured by mounting cup sensors. This bias-derived temperature showed variation in detector temperature for all NICMOS cameras during the years 1997-1998 as well as 2002-2008. As explained above, this is the main reason to create a new set of temperature dependent flat-field reference files. For more details about the CalTempFromBias code and the detector temperature measurements please refer to Dahlen et al. 2009, NICMOS ISR 2009-002, Bergeron, 2009 (NICMOS ISR, in preparation), Pirzkal et al. 2009, NICMOS ISR 2009-007 and Thatte, D. , Dahlen, T. et al. 2009, "NICMOS Data Handbook", version 8.0.

2


2. Data
Pre-NCS: The data obtained during the NICMOS flat-field monitoring from 1997-1998 was used to generate pre-NCS flat-fields. Data from program 8083 was used to calculate the temperature dependence of the DQE of the flat-fields. Post-NCS:-The data obtained during the NICMOS flat-field monitoring from 2002 to 2008 was used to regenerate post-NCS flat-fields. These programs represent the full span of post-NCS flat monitor programs. Table 1. Program information table for pre-NCS and post-NCS data used to make temperature dependent flat-fields
Program ID Program Title
7689 7690 7814 7956 7957 7961 8083 NIC2 Narrow Band Filter Pointed Super Flats NICMOS Pointed Super Flats NICMOS Camera 3 Pointed Super Flats NICMOS Camera 1 Narrow Band Filter Pointed Super Flats Lamp Flats II: NICMOS Pointed Flats FLATs: Warming Up FLATs: Warming Up continuation Post-NCS Data 8974 8985 9326 9327 9557 9640 9996 10379 10728 11016 NICMOS Flats and temperature dependence of the DQE NICMOS Internal Flats NICMOS Cycle 10 Early Calibration Monitor NICMOS Flats: narrow filters for NIC1 + NIC2, NIC3 in parallel NICMOS flats: Camera 3 narrow filters and grisms Flats Stability Flats Stability Flats Stability Flats Stability NICMOS Flats: narrow and broad filters for NIC1 + NIC2, NIC3 in parallel 11059 11321 Flats Stability NICMOS Cycle 16 Flats Stability N. Pirzkal T. Dahlen 1, 2, 3 1,2,3 11/7/2006 ­ 08/26/2007 05/02/2008 - 07/23/2008 A. Schultz A. Schultz A. Schultz A. Schultz N. Pirzkal 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3 9/10/2002 ­ 9/9/2003 10/15/03 ­ 9/4/2004 12/30/2004 ­ 8/2/2005 10/28/2005 ­ 8/6/2006 10/12/2006 ­ 09/22/2007 A. Schultz A. Schultz S. Arribas S. Arribas 1, 2, 3 1, 2, 3 1, 2, 3 3 5/13/2002 ­ 5/19/2002 5/30/2002 ­ 9/18/2002 7/26/2002 ­ 7/28/2002 6/2/2002 T. Boeker 1, 2, 3 4/20/2002 ­ 5/7/2002 D. Calzetti D. Calzetti 1,2,3 1,2,3 11/09/1998 ­ 12/13/1998 12/14/1998 ­ 01/10/1999 A. Schultz 1,2,3 08/03/1998 ­ 08/05/1998 A. Schultz 1 05/25/1998 ­ 05/26/1998 A. Schultz A. Schultz 1,2 3 07/08/1997 ­ 11/15/1998 12/22/1997 ­ 01/22/1998

Program PI Cameras
Pre-NCS Data A. Schultz 1,2

Date Range
02/26/1998 ­ 02/27/1998

3


11332

NICMOS Cycle 16 Time Dependent Flat Fields

T. Dahlen

1,2,3

06/14/2008 ­ 07/25/2008

3. Creating Temperature Dependent Flat-field images.
3.1. Pre-NCS temperature dependent flat-fields Temperature dependent flat-fields are created using three steps. 1) First we determine the temperature of the existing static pre-NCS flat-field images by calculating the mean temperature of the individual exposures that were used to produce the flat-fields. We use the bias-derived temperature given by the TEBTEMP keyword in the image header of each file. 2) We thereafter calculate the pixel-to-pixel DQE dependence of temperature and wavelength using flat-field images in F110W and F160W (and also including F222M for NIC3) taken during the end-of-life warm-up monitoring in January 1999. Adding to these a few images taken at lower temperatures earlier in the pre-NCS era, the available flat-fields span over the temperature range ~55-70K. The first fit of the DQE vs. temperature for each pixel is made at the particular wave-lengths of the available filters using an IDL script makefits written by Eddie Bergeron. Another script makewavefits is thereafter used to make wave-length fits to the output from above. The end products from these scripts are arrays with coefficients relating the DQE of each pixel in each camera with temperature and wave-length. An example of the DQE dependence with temperature for NIC3 at wave-lengths 1.1, 1.6, and 2.2 microns is shown in Figure 1. The symbols are measured points and the lines are fits to these points. Shorter wave-lengths have steeper temperature gradient. Curves are normalized to unity at 63K. 3) The coefficient arrays are used to wavelengths. The new flat-fields adjusting the latter by the relative in temperature. The central wavel field are used for this purpose. create new flat-fields at desired temperature and are created based on the existing flat-field by change in DQE calculated from the desired change ength and the temperature of the existing static flat-

For each filter and camera, flats are calculated at the following five different output temperatures: 61.2K, 61.6K, 62.0K, 62.4K, and 62.8K. This approximately divides the temperature range used in the pre-NCS era in five equally spaced bins. For consistency with previously created flat-fields, each flat-field is normalized to unity in [*,36:256] for NIC1 and NIC2 and in [*,56:256] for NIC3. The flat-fields are thereafter

4


inverted. Each flat-field, having an extension *_tdf.fits, consists of a stack of five flats for the above mentioned temperatures. When applying flat-field correction, the temp-from-bias, TFBTEMP, is used to choose the flat closest in temperature, which is thereafter used for the flat-field correction.

Figure 1. DQE response as a function of temperature, normalized at T=63K. Curves are shown for wave-lengths 1.1, 1.6 and 2.2 micron. Shorter wave-lengths have steeper slopes.

3.2. Post-NCS temperature dependent flat-fields. In Figure 2 we show the temperature variation of individual flat-field images taken during the NICMOS flat-field monitoring programs 2002-2008. From the figure it is clear that there is a strong evolution in temperature with time and that we therefore should expect that the shape of the post-NCS flat-fields will also change over this period. We therefore create temperature dependent flat-fields also for the post-NCS era, where each flat-field consists of a data-cub with five separate flat-fields extensions, each valid for a specific temperature range.

5


Figure 2. Temperature from bias for NIC1, NIC2 and NIC3 data

To do this we first create single exposure flat-field images by subtracting pairs of calibration images taken first with the NICMOS internal flat-field lamp on then off. This is done for all filters and all cameras using the available monitoring data taken 2002-

6


2008. In total there are about 800-900 single exposure images per camera divided among the different filters. We thereafter use the temperature (TFBTEMP) of these images to divide the flat-fields into five temperature bins so that approximately the same number of exposures falls in each bin for each camera. The limits for the temperature bins for each camera are given in Section 6. We finally combine the single exposure images in each temperature bin to create a median flat-field image. Each image is thereafter normalized to unity (using [*,36:256] for NIC1 and NIC2 and [*,56:256] for NIC3) and inverted. For filters monitored regularly, like F110W and F160W, there are enough single exposure flat-fields to create final images in all five temperature bins. For some less observed filters, mostly narrow band filters, images are not available in all five temperature bins. In those cases final images are generated by interpolating. We also generate a pedigree file that includes information on all individual images going into each final flat-field together with S/N for all the combined images. This pedigree file is available on the NICMOS webpage, www.stsci.edu/hst/nicmos.

4. Epoch Flat-fields
Figure 2 shows that the temperature changes close to linearly with time and an alternative approach to create flat-fields would therefore be to dive the flat-fields in temporal bins as discussed in Dahlen et al. 2007 (NICMOS ISR-2007-002). For consistency checks epoch dependent flat-fields were also created using a method similar to that of temperature dependent flat-fields. Instead of temperatures the flat-fields were divided into bins for different epochs.

5. Comparison of Epoch and Temperature flat-fields
To quantify the difference between the epoch and temperature flat-fields, the signal-to-noise ratio for flat-fields in all temperature bins was compared with signal-tonoise ratio for flat-fields in all epoch bins (not including bins for which interpolated flatfields were created). Table 2 shows the median S/N for temperature dependent and epoch dependent flat-fields for all three NICMOS cameras.

7


Table 2. Comparison of post-NCS epoch and temperature flat-fields

N N N N N N

I I I I I I

C1 C1 C2 C2 C3 C3

Number of Images temperature flat-fields 64 epoch flat-fields 52 temperature flat-fields 67 epoch flat-fields 57 temperature flat-fields 63 epoch flat-fields 55

Median S/N 733.510 722.220 728.600 713.710 823.690 772.180

The median of the S/N for temperature dependent flat-fields is higher than for the epoch dependent flat-fields. This can also be seen in Figures 3a, 3b and 3c. Note also that the total number of different temperature images created is larger than the number of epoch images. This means that on average fewer single flat-field images were used to produce the temperature dependent flat-fields. Despite that, the S/N for the temperature flat-fields are higher, assuring us that the temperature dependent flat-fields are a better choice to use.

Figure 3(a) Histogram of Signal-to-Noise ratio for flat-fields in epoch and temperature bins for NIC1

8


Figure 3(b) Histogram of Signal-to-Noise ratio for flat-fields in epoch and temperature bins for NIC2

Figure 3(c) Histogram of Signal-to-Noise ratio for flat-fields in epoch and temperature bins for NIC3

9


6. Structure of temperature dependent flat fields
Each flat-field image (*_tdf.fits) consists of five individual images (imsets) each having five extensions: SCI, ERR, DQ, SAMP, and TIME. The *_tdf.fits images therefore consist of 25 extensions with the SCI images, i.e., the temperature dependent flat-fields, in extensions 1, 6, 11, 16, and 21. Keywords TFBLOW and TFBHIGH show the temperature range for each extension. The full temperature range covered by all five bins is given by the TFBLOW and TFBHIGH keyword in the primary header (EXT=0). The extension for which TFBLOW < TFBTEMP < TFBHIGH, where TFBTEMP is the temperature from science data header, is used for the flat-fielding. The temperature ranges (in Kelvin) are: Pre-NCS extension extension extension extension extension extension NIC1, NIC2, NIC3 flat-fields 0: TFBLOW=52.0, TFBHIGH=70.0 1: TBFLOW=52.0, TFBHIGH=61.4 6: TBFLOW=61.4, TFBHIGH=61.8 11: TBFLOW=61.8, TFBHIGH=62.2 16: TBFLOW=62.2, TFBHIGH=62.6 21: TBFLOW=62.6, TFBHIGH=70.0 flat-fields 0: TFBLOW=64.95, TFBHIGH=86.85 1: TBFLOW=64.95, TFBHIGH=74.95 6: TBFLOW=74.95, TFBHIGH=75.55 11: TBFLOW=75.55, TFBHIGH=76.25 16: TBFLOW=76.25, TFBHIGH=76.85 21: TBFLOW=76.85, TFBHIGH=86.85

Post-NCS NIC1 extension extension extension extension extension extension NIC2 extens extens extens extens extens extens

ion ion ion ion ion ion

0: TFBLOW=64.65, TFBHIGH=86.80 1: TBFLOW=64.65, TFBHIGH=74.65 6: TBFLOW=74.65, TFBHIGH=75.30 11: TBFLOW=75.30, TFBHIGH=76.15 16: TBFLOW=76.15, TFBHIGH=76.80 21: TBFLOW=76.80, TFBHIGH=86.80

NIC3 extension 0: TFBLOW=65.60, TFBHIGH=87.05
10


ext ext ext ext ext

ens ens ens ens ens

ion ion ion ion ion

1: TBFLOW=65.60, 6: TBFLOW=75.60, 11: TBFLOW=76.08, 16: TBFLOW=76.56, 21: TBFLOW=77.05,

TFBHIGH=75.60 TFBHIGH=76.08 TFBHIGH=76.56 TFBHIGH=77.05 TFBHIGH=87.05

Figures 4 and 5 show the ratio images obtained by dividing each SCI extension in NIC3, F110W temperature dependent flat by the SCI extension in the static flat file, for pre and post-NCS data respectively. The temperature goes up as we go from panel 1 to panel 5. Panel 3 in figure 4 and panel 4 in figure 5 are relatively flat due to the fact that the temperature range of SCI[11] image is close to the temperature of the static flat file. From both the figures we can conclude that the temperature dependent flat-fields do have some structure as compared to the static flat-field and the structure of flat field does depend upon temperature.

Figure 4. Panels 1 to 5 show the ratio images obtained by dividing a flat-field corresponding to each temperature bin in the temperature dependent flat by the static flat for pre-NCS NIC3, F110W data.

11


Figure 5. Panels 1 to 5 show the ratio images obtained by dividing a flat-field corresponding to each temperature bin in the temperature dependent flat by the static flat for post-NCS NIC3, F110W data.

7. FLATCORR
Temperature dependent flat-fields populate the header keyword TDFFILE. This reference file contains the flat-field image for a given detector and filter (or polarizer) combination. As explained above this file contains five imsets, each with a flat-field image valid for a particular temperature range. In the FLATCORR step, calnica reads the detector temperature from the TFBTEMP keyword in the science data header and selects the appropriate imset from the TDFFILE. The imset used is written to the TDFGROUP keyword in the header of the calibrated (*_cal.fits) files. The temperature-dependent flat fields were delivered to the archive on November 19, 2008 (updated NIC3 flat-fields delivered on May 27, 2009) for Cycle 11 and later observations and on January 23, 2009 for data taken during Cycles 7 and 7N. There is no TDFFILE listed in the header for the data retrieved prior to these dates. In this case, the FLATFILE reference file which is a single static (non-temperature dependent) flat-field image is used instead. If both TDFFILE and FLATFILE reference files are given in the header, calnica chooses the TDFFILE. FLATFILE is used instead of TDFFILE in the cases where TFBTEMP keyword is missing or TFBTEMP lies outside the range given by TFBHIGH and TFBLOW keywords in the primary header. Error estimates and DQ flags

12


contained in the TDFFILE/FLATFILE are propagated into the processed images. The difference between the temperature dependent flat-fields and the static flat-fields is small but it can affect the photometry on a ~1% level.

8. Summary
Data from NICMOS calibration programs during 1997- 1998 and 2002-2008 is used to generate the pre-NCS and post NCS temperature dependent flat-fields. Temperature variation of T ~2K as shown by bias derived temperature (given by header keyword TFBTEMP), affects the DQE and the structure of the flat fields for all detectors. The new temperature dependent flat-fields (*_tdf.fits) correct for this effect in the FLATCORR step of calnica version 4.4 and later. The *_tdf.fits files populate the keyword TDFFILE in the headers of data retrieved after November 19, 2008 for postNCS flat-fields and January 23, 2009 for pre-NCS flat-fields.

9. Acknowledgements
We would like to thank L. Bergeron for providing the scripts to create pre-NCS temperature dependent flat-fields.

References
Bergeron, 2009, NICMOS ISR, in preparation de Jong, 2009, NICMOS ISR, in preparation Pirzkal, N., Grumm, D., Bergeron, E., ISR NICMOS 2009-007, STScI, CalTempFromBias:Implementation and Testing Dahlen, T., Barker, E., Bergeron, E., Smith, D. ISR NICMOS 2009-002, STScI, Temperature Dependent Dark Reference Files: Linear Dark and Amplifier Glow Components Dahlen, T., McLaughlin, H., & de Jong, R. 2007, NICMOS ISR-2007-002, STScI, NICMOS Time Dependent Flat-fields Thatte, D. and Dahlen, T. et al. 2009, "NICMOS Data Handbook", version 8.0, (Baltimore, STScI)

13


14