Wide Field and Planetary Camera 2 Instrument Handbook for Cycle 14
8.12 Cycle 5 Calibration Plan
A summary of the Cycle 5 calibration plan follows as a general guide to the calibration and monitoring program in place for WFPC2. The full proposals are available through STScI's proposal status web page.
The data that the calibration and monitoring program produced have no proprietary period and are immediately available through the HST archive.
Calibration information obtained by the start of Cycle 5 consisted primarily of the System Level Thermal Vacuum (SLTV) tests, the initial on-orbit tests conducted in SMOV, and the Cycle 4 calibration. These tests have shown that the instrument is stable with some important exceptions and have provided an initial calibration sufficient for routine processing of most data.
The Cycle 5 calibration was designed to enable users to maximize the scientific usefulness of their data, while at the same time minimizing the use of spacecraft time. This was done by designing efficient proposals that:
- Improved the existing calibration - in particular towards the goal of 1% absolute photometric accuracy.
- Assessed the accuracy of the existing and new calibrations.
- Recalibrated important known time variable features of the instrument.
- Calibrated some important instrumental effects that are not well understood.
- Monitored the instrument and telescope to ensure that no new problems or variability in their performance are missed.
- Maintained the instrument in a healthy state and ensured that in the event of partial instrumental failures, the calibration can be maintained when possible.
The calibration of the instrument is seen in a larger context than simply preparing reference files for a pipeline reduction and assessing the errors in them. Several calibrations (such as geometric distortion, CTE correction, PSF calibration, chip-to-chip alignments, polarization calibration) are very important to some observers, yet are not included in the pipeline. Other corrections frequently need to be done to the data after it is ADC, bias, dark, and flat field corrected, with a photometric calibration included in the header. These other calibrations are made available to users through this Instrument Handbook, journal publications, instrument science reports, and postings linked to the Institute's WFPC2 WWW home page. The address is:
A list of the most important calibrations consists of the following items:
- Photometric zero-point: converting count rates to flux units.
- Photometric transformations: converting DN values to magnitudes in standard systems. Two separate photometric calibrations can be used for this, a direct approach and a synthetic approach.
- Photometric temporal variations: particularly important in the UV where significant variability is seen.
- Photometric spatial variation: flat fields and charge transfer efficiencies.
- Dark current: including its time variability and hot pixels.
- Analog-to-Digital converter errors.
- PSF: crucial for PSF fitting photometry, PSF subtraction, PSF modeling, and deconvolution efforts. Because PSF subtraction of very saturated sources is specialized to a few very diverse programs, PSF calibration in the image halo (beyond about 0.5 arcsecond) is not supported and must be requested with the program as a special calibration.
- Polarization and Linear Ramp Filter calibrations.
- Geometric calibration.
The Cycle 5 program consisted of 15 proposals which used a total of 63 orbits of spacecraft time (compared to a total of about 1550 orbits of approved Cycle 5 GO time). The proposal summaries and their associated Phase II files largely speak for themselves. Table 8.5 lists all of the proposal numbers, titles, the schedule for the calibration execution, an indication of whether the output forms part of the pipeline data reduction (CDBS) or provides other information, usually documented in Instrument Science Reports (ISRs), the approximate calibration accuracy expected (see the summary forms for the interpretation of these numbers), the primary areas from the above 10 calibration types they address and in what ways (A-F from the above list). Following the table there are more details on each proposal individually, including purpose, observing description, accuracy, and data products. For a report on the final results from these programs, please see ISR WFPC2 97-02 ("Cycle 5 Closure Report") to be found at:
ID Proposal Title Schedule Results Accuracy External Time (orbits) Notes1 6179 Photomet. Zero. Late 95 1ABE, 2AB 6182 Photomet. Trans. 9/95, 3/96 2ABE 6183 Decontamination 1x per 4 wks. F 6184 Photometric Mon. 2x per 4 weeks 3E 6186 UV Throughput Early in Cyc. 5 1AB, 3C 6187 Earth Flats Continuous 4ABE 6188 Darks Weekly 5ABC 6189 Visflat Monitor 2x per 4 weeks 4E 6190 Internal Flats Early Cyc. 5 4F,7E 6191 UV flats 2x in Cyc. 5 4ABE 6192 CTE Calibration Early Cyc. 5 4ABD 6193 PSF CTE+2m 8ABD 6194 Polarizers+Ramps TBD 9DE, 1AB 6195 Flat field Check Late 95 4B 6250 Internal Monitor 2x per week 5,6,10F TOTALS
1Letters and numbers are keyed to lists in text.
6179: Photometric Zero-point
- Purpose: Set synthetic zero points of all WFPC2 filters.
- Description: GRW+70D5824 is observed through all filters not in the photometric calibration monitors and longward of F336W (inclusive). It is observed in both PC1 and WF3. This data-set is directly comparable to the corresponding results for Cycle 4 (program 5572). Because of possible errors in the spectrophotometry for this target, and in order to check synthetic color transformations over a reasonably wide range of colors, the observations are repeated with 3 red standards, and 2 other blue standards in WF3 only. This time, most of the 18 broad and medium bandpass filters longward of F336W are included (restricted to 1 orbit/target). If CTE calibration fails, this proposal may need to be run with preflash (and take more time or do less targets). This proposal is needed by all GO proposals that want to do quantitative photometry at the few percent level.
- Accuracy: Overall discrepancies between the synthetic photometric products and the results of this test should be reduced to 1% rms. Part of the point of the test is to measure this accuracy, which will largely depend on the accuracy of the spectrophotometric calibration sources.
- Products: After pipeline processing, each image will be reduced by aperture photometry to measure the throughput of each filter. These numbers can be directly compared to SYNPHOT predictions. Systematic differences will be corrected in the throughput database by tweaking the filter normalizations (already done for the primary target in Cycle 4), overall system response (which is quite uncertain particularly longward of 8000Å), and finally bandpass shapes. Residual differences will give an idea of the intrinsic accuracy of the calibration.
6182: Photometric Transformation
- Purpose: Update photometric transformations to Johnson-Cousins system.
- Description: A photometric standard star field in Cen is observed twice, once at the September 1994 orientation and once rotated by 180° degrees (to correct to first order for residual CTE effects). All broad and medium bandpass filters are used. Based on Cycle 4 program 5663, this proposal also gives a check on the long term full field photometric stability of the instrument.
- Accuracy: Independent of the synthetic photometry results this test gives direct transformations to the Johnson-Cousins system for a wide range of source colors (but excluding very blue stars). These transformations should be accurate to 2%. The stability of these transformations will be measured to the sub-percent level (because then errors in the ground based photometry do not enter significantly).
- Products: A comparison with the corresponding Cycle 4 monitor (which ran monthly) will verify the photometric stability of the camera. Direct transformations to the Johnson-Cousins photometry system can be derived for all filters. The observations also provide a check of the astrometric and PSF stability of the instrument over its full field of view.
- Purpose: Remove UV blocking contaminants from CCD windows.
- Description: A sequence of observations is defined that is run twice - once before and once after a DECON when the CCDs are heated to +20°C for 6 hours. The sequence consists of 2 bias frames at each gain setting, 5 1800 second darks, 2 INTFLATS through F555W at each gain, and two K-spot images. The observations are arranged so that the first sequence occurs about 33 hours before the DECON, and the second follows it by 12 hours. The proposal is run every 4 weeks. This is based on Cycle 4 program 5568 with bias frames added.
- Accuracy: This proposal is mainly designed to test aliveness, and monitor the instrument to ensure that no untoward effects from the DECON have occurred. It will identify all hot pixels that are annealed by the DECON.
- Products: The data is examined and checked for anomalies. The dark frames are processed to yield plots showing the growth and annealing of hot pixels.
6184: Photometric Monitor
- Purpose: Monthly external check of instrumental stability.
- Description: GRW+70D5824 is observed through F170W in all four chips. It is then observed in one chip for filters F160BW, F185W, F218W, F255W, F300W, F336W, F439W, F555W, F675W, F814W to fill out the orbit. The chip chosen is changed each month, so each chip is used three times during the year. One extra F555W exposure is taken through the PC to allow for focus monitoring. The proposal is run once before and once after each monthly decontamination, with the same chip selected. Based on Cycle 4 programs 5629 + 6143 + (5563) + 5564.
- Accuracy: Overall discrepancies between the results of this test should be less than 1% rms. The point of the test is to measure this variation.
- Products: After pipeline processing, each image will be reduced by aperture photometry to measure the throughput of the filters. These numbers can be directly compared to results for previous months. This will allow the long term performance of the instrument to be checked for changes, and verify that the decontaminations are satisfactory. This proposal will be run every 4 weeks in association with the DECON. Results reported in Baggett et al., ISR WFPC2 96-02 and Whitmore, et al., ISR WFPC2 96-04.
6186: UV Throughput
- Purpose: Update SYNPHOT database for UV throughput.
- Description: GRW+70D5824 is observed shortly before and after a DECON through all of the UV filters in each chip and through F160BW crossed with F130LP, F185LP, and F165LP (where applicable) to determine the wavelength dependence of the throughput across the bandpass (hence color terms). Based on no particular Cycle 4 program, this program is designed to characterize better the spectral response curve in the UV, and the spectral shape introduced by the contamination.
- Accuracy: Overall discrepancies between the updated synthetic photometric products and the results of this test should be 1-2% rms. This does not mean that the UV throughput will be known to this accuracy, primarily because of uncertainties in the flux calibration of the standard used (5%), uncertainties in the UV flat fields (maybe 3% near the chip center), and time dependent contamination corrections (3% error), and uncertainties in the CTE correction (2%). The derived UV absolute photometric accuracy at the center of the chips should therefore be about 10%.
- Products: After pipeline processing, each image will be reduced by aperture photometry. The throughput curves and their normalizations can be updated by trial and error.
6187: Earth Flats
- Purpose: Generate flat fields.
- Description: Four sets of 200 Earth streak-flats are taken to construct high quality narrow-band flat fields with the filters F375N, F502N, F656N, and F953N. Of these 200 perhaps 50 will be at a suitable exposure level for de-streaking. The resulting Earth superflats map the OTA illumination pattern, and will be combined with SLTV data (and calibration channel data in case of variation) for the WFPC2 filter set, to generate a set of superflats capable of removing both the OTA illumination and pixel-to-pixel variations in the flat field. In addition, more limited observations are made in F160BW and the broad bandpass photometric filters.
- The Cycle 4 plan is being largely repeated except: (1) UV filters are dropped because the measurement is generally only of the red leak. (2) F160BW is retained in order to check for developing pinholes. (3) Crossed filters used as neutral densities are eliminated (illumination pattern is wrong). (4) An attempt will be made to schedule some broad bandpass measurements on the dark Earth. This is based on Cycle 4 programs 5570 + 5571 + 6142.
- Accuracy: Overall accuracy of the flats derived from this test and the corresponding Cycle 4 observations should be below 1% RMS. Discrepancies between the results of this test and those from Cycle 4 should be 1% RMS. Differences between the two datasets analyzed separately will measure the flat field variability to this level. This data, together with the flat field check proposal should enable similar accuracy in the broad bandpass flats.
- Products: This proposal provides medium and narrow bandpass streak flats which can be combined with the high frequency information in the TV flats to yield accurate flat fields. The ratio of the TV and derived flats provides a correction that can also be applied to the TV broad bandpass data. We may also get broad band flat fields directly for comparison from the sky or from this proposal's exposures on the Earth's shadow.
- Purpose: Provide dark frames for pipeline reduction, and hot pixel lists.
- Description: Five dark frames are taken every week to provide ongoing calibration of the CCD dark current rate and to monitor the characteristics and the evolution of hot pixels. Over an extended period, these data provide a monitor of radiation damage to the CCDs. The dark frames will be obtained with the CLOCKS=OFF. In addition, four darks are taken per month with CLOCKS=ON (although there is no effect such as amplifier glow expected from running the serial register for these CCDs).
- Changes from the Cycle 4 program 5562 are that each group of five darks is constrained to be taken within a 2-day period, in order to simplify the data analysis (because fewer new hot pixels are involved), and the CLOCKS=ON exposures have been added.
- Accuracy: The accuracy of the super-dark computed from these data depends on the number of frames combined. The present practice is to combine them in groups of 10 frames for pipeline super-darks. This gives a median signal-to-noise of 16 and higher signal-to-noise on hotter pixels than the median (with the somewhat shaky assumption that the dark noise is Poisson). This means that the residual systematic error after super-dark subtraction on an 1800-second exposure is about 3 electrons - much less than the read noise. In principle, this residual can be further reduced to 0.4 electrons if a super-dark is generated from all of the dark frames, with suitable masking based on hot pixel lists.
- Products: The data is grouped into sets of 10 frames every two weeks. These are combined into super-darks for use in the pipeline. In addition, hot pixel lists can be generated with a time resolution of one week.
6189: VISFLAT Monitor
- Purpose: Monitor internal flat fields of instrument.
- Description: All use of the VISFLAT channel is concentrated in this proposal. It is based on Cycle 4 programs 5568 and 5655. The program takes one complete set of exposures using the visible calibration channel lamp (VISFLATS) at the start of the cycle through all visible filters. Monthly, VISFLATS will be obtained with the photometric filter set (F336W, F439W, F555W, F675W, and F814W) both before and after the DECON. A monthly VISFLAT exposure with the Wood's filter, F160BW, allows its visible transmission to be monitored. Two monthly VISFLAT exposures obtained through the LRF (FR533N), one at each gain, provide a monitor of the ADC's performance. The VISFLAT exposures should be packed together to optimize use of each lamp-on cycle.
- Accuracy: The internal flats, when well exposed, are each accurate to 0.6% in terms of the pixel-to-pixel (high frequency) variations in the CCDs. Thus, high frequency flat field stability can be verified to 1%. When the results from several filters are combined it will be possible to check that the CCDs are indeed relatively stable to much better than 1%.
- Products: The complete filter-set sweep will be compared to the corresponding Cycle 4 data-set primarily to verify that none of the filters are developing problems. Ratios of these flats will primarily indicate the stability of the channel itself, unless there are strong variations from filter to filter. Unless time dependence in the filters is seen, it is likely that flat fields for pipeline calibration will continue to be made by combining Earth-flat, SLTV-flats, and eventually sky-flats.
- The bi-monthly photometric filter-set observations will be used to monitor WFPC2's flat field response and to build a high S/N flat field database (primarily useful in tracking any changes in the pixel-to-pixel response of the instrument and any possible long term contamination induced changes). Histograms generated from the ramp filter flats will be used to trace the ADC transfer curve. F160BW can be checked monthly for pinholes. Results reported in Stiavelli and Baggett, ISR WFPC2 96-01.
6190: Internal Flats
- Purpose: Provide backup database of INTFLATS, in case VISFLAT channel fails.
- Description: Based on Cycle 4 program 5568. A complete set of illuminated shutter blade flats (INTFLATS) is taken close to the complete set of VISFLATS. Each filter is exposed on each shutter blade (A or B) at each gain setting (7 or 15). Thus, there are four exposures per filter which should be uninterruptedly sequenced as (A7, A15, B15, B7). There is a possible concern on thermal control, where an out-of-limit condition was almost reached when 5568 was run. This will be avoided by spacing the exposures suitably.
- Accuracy: The signal-to-noise per pixel is similar to that obtained in the VISFLAT program (0.6%) but there are much larger spatial and wavelength variations in the illumination pattern. As a result, this data-set will not form any part of the pipeline calibration. This baseline is necessary in case the VISFLAT channel fails, and there are temporal variations in the camera flat fields at the 1% level. The test does give a good measurement of the gain ratios and their stability, which should be accurate to much better than 1%, when all of the data is analyzed.
- Products: INTFLAT/VISFLAT ratios can be generated if there is a failure in the calibration channel. Gain ratios and stability will be assessed. Results reported in Stiavelli and Baggett, ISR WFPC2 96-01.
6191: UV Flats
- Purpose: Use UV calibration channel to monitor long term internal UV stability.
- Description: UV flat fields will be obtained with the calibration channel's ultraviolet lamp (UVFLAT) using the limited FUV filter set (F122M, F170W, F160BW, F185W, and F336W) twice in the cycle immediately after a DECON. The UV lamp is degrading with use, so its use must be minimized. The UV flats will be used to monitor the FUV flat field stability and the stability of the Wood's filter, F160BW, by using F170W as a reference. The VISFLAT/UVFLAT ratio from the F336W filter will provide a diagnostic of the UV flat field stability, and tie the UVFLAT and VISFLAT flat field patterns together. This program represents the entire use of the UV lamp in Cycle 5. This proposal is based on the Cycle 4 program 5568, but with two extra filters (F122M and F185W).
- Accuracy: Should verify stability of the UV filters and flat field to 2%. The overall flat field response is not measured because the lamp output is not uniform, and temporal variations in throughput are not measured because the lamp output varies.
- Products: Ratio images with the corresponding data-set from Cycle 4 and future cycles will verify the UV flat field stability.
6192: CTE Calibration
- Purpose: Calibrate CTE effect for a range of star brightness and background.
- Description: The crowded Cen field is observed for 40 seconds through F555W with gain 7 and preflashes of 0, 5, 10, 20, 40, 80 and 160 electrons. As a gain check and calibration, it is observed at the same orientation with gain 15 twice at preflash levels of 0 and 160 electrons. The whole sequence is repeated with filter F814W. Then a whole orbit is filled with 1 second exposures, in order to investigate the effect of CTE on low signal level stars (but with high accumulated signal-to-noise). This last orbit is repeated with a preflash of 40 electrons. This is based on Cycle 4 programs 5645 + 5646 + 5659.
- Accuracy: As this test is a differential measurement of the CTE slope, it should be very accurate (much better than 1%). As a large number of stars are involved, and the photon noise on each measurement is of order 1%, the slope derived should be much more accurate. The largest remaining uncertainty will be the absolute level of the slope, not differential effects caused by varying background.
- Products: After pipeline processing, each image will be reduced by aperture photometry to measure the stellar brightness and how it depends on the preflash level. This is a differential measurement and gives no direct information about the slope of the CTE effect at high background levels. The absolute CTE can be estimated from 5646 (already run once with a raster in this field).
6193: PSF Characterization
- Purpose: Provide a sub-sampled PSF over the full field to allow PSF fitting photometry.
- Description: This proposal measures the PSF over the full field in photometric filters in order to update the TIM and TinyTIM models and to allow accurate empirical PSFs to be derived for PSF fitting photometry. With one orbit per photometric filter, a spatial scan is performed over a 4x4 grid on the CCD. The step size is 0.025 arcseconds. This gives a critically sampled PSF over most of the visible range. The crowded Cen field is used. 40 sec images are taken through each of the photometric filters (F336W, F439W, F555W, F675W, F814W). Data volume will be a problem, so special tape recorder management will be called for. This is based partially on Cycle 4 program 5575, which used the same field. The proposal also allows a check for sub-pixel phase effects on the integrated photometry.
- Accuracy: The chosen field will have hundreds of well exposed stars in each chip. Each star will be measured 16 times per filter at different pixel phases. In principle, the proposal therefore provides a high signal-to-noise critically sampled PSF. This would leave PSF fitting photometrists in a much better position than now where pixel undersampling clearly limits the results. The result will be largely limited by breathing variations in focus. It is hard to judge the PSF accuracy that will result. If breathing is less than 5 microns peak-to-peak, the resulting PSFs should be good to about 10% in each pixel. PSF fitting results using this calibration would, of course, be much more accurate. In addition, the test gives a direct measurement of sub-pixel phase effects on photometry, which should be measured to much better than 1%.
- Products: This program provides sub-sampled PSFs for photometry codes, data for comparison with PSF codes, and measurement of pixel phase effect on photometry (sub-pixel QE variations exist).
6194: Polarization and Ramps
- Purpose: Perform residual calibration and check for stability of polarizer and ramp filters.
- Description: This proposal does not duplicate the existing ramp wavelength calibration or polarization calibration (Cycle 4 programs 5574 and 6140). Instead, it provides a full polarization calibration for a filter that was not supported in Cycle 4 (F300W), a check for polarization stability, and a throughput calibration for the linear ramp filters, by scanning the spectrophotometric standard along the ramps.
- Accuracy: The proposal should support polarimetry at the 3% level and measure the ramp throughput at the 2% level.
6195: Flat field Check
- Purpose: Check quality of flat fields and estimate errors in them.
- Description: The crowded Cen field is positioned with a bright star at the center of each CCD in turn. 40 second images are taken through each of the photometric filters (F336W, F439W, F555W, F675W, F814W) and as many supplementary filters (from F450W, F606W, F702W and F547M) as can be fitted in. If data volume is a problem, single chip readout is acceptable, but should be avoided as much as possible. This is based on Cycle 4 programs 5659 and 5646.
- Accuracy: Overall discrepancies between the existing flat fields and the results of this test should be 1-2% rms. Part of the point of the test is to measure this accuracy.
- Products: After pipeline processing, each image will be reduced by aperture photometry to measure the RMS errors in the flat fields. The RMS error will be determined by the additional noise in the independent measurements over the expected variance of less than 1% from photon statistics. The single bright star at the center of each chip independently estimates the chip to chip normalization error.
6250: Internal Monitor
- Purpose: Check for short term stability of instrument.
- Description: The routine internal monitor, to be run twice weekly for WFPC2 during Cycle 5, obtains two bias frames at each gain, two INTFLATs with the F555W filter at each gain, and two Kelsall spot images with exposure times optimized for the WF and PC, respectively. It is identical to the Cycle 4 program 5560.
- Accuracy: This monitor is not involved in generating quantitative calibration information.
- Products: The test provides a biweekly monitor of the integrity of the CCD camera chain electronics both at gain 7 and 15, a test for quantum efficiency hysteresis in the CCDs, and an internal check on the alignment of the WFPC2 optical chain. Stiavelli and Baggett, ISR WFPC2 96-01.
Space Telescope Science Institute
Voice: (410) 338-1082