L. Colina1, S. Holfeltz & C. Ritchie Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, Email: firstname.lastname@example.org, email@example.com, firstname.lastname@example.org
(1)Affiliated with the Astrophysics Division, Space Science Department, ESA
HST, NICMOS, near-infrared, standards, photometry, calibration
The near-infrared wavelength range was opened to HST with the installation of the Near-Infrared Camera and Multi-Object Spectrometer (NICMOS) in February 1997, during the second HST servicing mission. Unaffected by atmospheric absorption and emission, NICMOS covers the entire 0.8 m to 2.5 m wavelength range. Similarly, NICMOS, being above the atmosphere, is not forced to adopt filter bandpasses like those used at ground-based observatories matching the near-infrared atmospheric windows. In practice NICMOS does not have a set of filters matching any of the standard ground-based photometric bands and this poses a challenge when trying to achieve precise absolute photometry. On the other hand, NICMOS absolute calibration requires a set of faint spectrophotometric standards covering the entire 0.8m - 2.5m wavelength range. Such a set of standards didn't exist before and the selection and generation of NICMOS standards represented additional challenges.
This paper reviews several aspects of NICMOS photometric performance. Details on NICMOS absolute spectrophotometric standards are given in section 2. Section 3 discusses NICMOS absolute photometry. The photometric stability of the cameras and the relative photometry across the detectors are mentioned in sections 4 and 5, respectively. Special issues like intra-pixel sensitivity, red leaks and transformations between HST and ground-based systems are treated in sections 6, 7 and 8, respectively. Finally, several sources of uncertainty when performing absolute calibration with NICMOS are summarized in section 9. This paper does not mention the performance of NICMOS grisms and polarizers as they will be the topic of separate contributions by W. Freudling and D. Hines, respectively.
The reader is recommended to visit the NICMOS Photometry WEB page to check for the latest news regarding the photometric performance of NICMOS.
The absolute calibration of the ultraviolet and optical instruments onboard HST is based on the existence of absolute flux calibrated spectra of a few pure hydrogen white dwarfs (WD) and hot stars, the so called HST set of absolute spectrophotometric standards (Colina & Bohlin 1994; Bohlin, Colina & Finley 1995; Bohlin 1996). The absolute calibration of this set of HST absolute standards in the UV and optical is based on a detailed model spectrum of a hot pure hydrogen white dwarf standard (G191-B2B), transformed into an absolute flux scale using synthetic photometry techniques and accurate Landolt visual photometry (Bohlin, Colina & Finley 1995). This model spectrum covers also the entire NICMOS wavelength range and therefore G191-B2B has been selected as the primary NICMOS white dwarf standard.
An alternative method for calibrating NICMOS uses solar analogs (Campins et al. 1985). Three faint solar analogs were selected by the NICMOS Instrument Definition Team (IDT), and observed on the ground at JHK (E. Green and E. Persson, private communication). Spectra of these three solar analogs were taken with the HST Faint Object Spectrograph (FOS) in the 0.2 m - 0.8 m wavelength range to study how accurately their spectral energy distribution matched that of the Sun (Colina & Bohlin 1997). The absolute flux distribution of these three solar analogs covering the ultraviolet to near-infrared range was obtained combining a scaled version of an absolute flux calibrated solar reference spectrum (Colina, Bohlin & Castelli 1996a) with the FOS spectra (Colina & Bohlin 1997). In the near-infrared, the solar reference spectrum was generated computing the energy output of a solar photospheric model using the most recent version of Kurucz ATLAS code (see Colina, Bohlin & Castelli 1996 for details). Of the three solar analogs, the star P330E shows an optical spectrum consistent, within the observational errors, with that of the Sun. P330E has therefore been selected as the primary NICMOS solar analog standard.
A preliminary on-orbit absolute photometric calibration of all three NICMOS cameras was obtained in July 1997 soon after the end of the Servicing Mission Observatory Verification (SMOV) phase. This preliminary update was based on on-orbit measurements of a couple of standard stars in 4/5 filters per camera, covering the entire NICMOS wavelength range. The accuracy of the absolute calibration obtained in this way was 10% - 15% for most filters (Colina & Rieke 1997).
An update to the NIC1 and NIC2 photometry, and corresponding photometric keywords that get populated during the calibration pipeline process, was delivered in March 1998. This update is based on the complete NIC1 and NIC2 on-orbit absolute photometry calibration program executed in August 1997 (proposal ID 7691).
Images of G191-B2B and P330E were taken through all available NIC1 and NIC2 narrow-, medium- and broad-band filters (32 in total). Independent measurements of each of the stars were taken at three separated positions around the center of the detectors.
Countrates (i.e. counts per second) were measured in the final calibrated images using a 0.5 arcsec radius aperture in both cameras. The total countrates for a nominal infinite aperture have been computed applying an aperture correction equivalent to multiplying the 0.5 arcsec aperture countrates by 1.15, irrespectively of filter.
The observed total countrates obtained in this way were compared with the pre-launch predicted total countrates obtained using synthetic photometry (SYNPHOT) on the spectra of the primary standards for all NIC1 and NIC2 filters (see figure 1 for NIC2 results).
Images of G191-B2B and P330E were also taken with NIC3 during the first NIC3 campaign in January 1998 (proposal ID 7816). To minimize pixelation effects, the countrates were measured in the final calibrated images using a 1.0 arcsec radius. The total countrates for a nominal infinite aperture have been computed applying an aperture correction equivalent to multiplying the 1.0 arcsec aperture countrates by 1.075, irrespectively of filter.
For each filter, the average of the observed/predicted countrates for each of the two stars was used to correct the pre-launch predicted countrates and match them with the observed countrates. These corrections were applied in practice by renormalizing the filter throughputs without changing their shape and without modifying the QE curve of the detectors. Once the corrections were in place (see figure 2 for NIC2 results), SYNPHOT was used to compute the updated photometric countrate to flux conversion factors for all NIC1 and NIC2 filters.
The main results of the absolute photometry measurements are valid for all three cameras and can be summarized in the following points:
The long term photometric stability of all three NICMOS cameras is being monitored during Cycle 7 by taking images of a standard star (the solar analog P330E) every four weeks. The standard star is observed through several filters spanning the entire wavelength range of NICMOS.
During the period already covered by the monitoring (May 1997 - April 1998),
no significant deviations (i.e. larger than 2%) have been measured and the
cameras have been photometrically stable to better than 2% (see figures 3 and
4 for an example of the stability of cameras NIC1 and NIC2 at 1.6 m).
The temperature of the detectors has been rising continuously since the beginning of Cycle 7 and will continue to do so during the next few months. As the temperature increases, the sensitivity of the detectors changes and this could eventually be reflected in changes in the countrates, and therefore in the photometry. Updates with the new results of the photometric monitoring program are posted on a monthly basis on the NICMOS Photometry WEB page.
Relative photometry across the detectors has been measured for a few filters in all three cameras. Images of the absolute standard P330E (solar analog) were taken at 144 different detector positions using filter F160W for NIC1 and filters F110W and F222M for NIC2 and NIC3 (proposal ID 7693). Data with camera 3 were obtained during the first NIC3 campaign while this camera was in focus. All images were calibrated using on-orbit darks and flats obtained with internal lamps.
The results show that relative photometry for cameras 1 and 2 is better than 2% in the filters for which measurements are available (see figure 5 & 6 for the NIC2 results). Since the images obtained as part of this calibration program have been calibrated using the on-orbit darks and flats reference files available to the observers, there is no reason to believe that a relative photometry at the 2% level, or better, can not be obtained for the rest of the filters in these two cameras, if on-orbit flats are used in the calibration.
The results for camera 3 have to be treated with more caution. For the F110W filter, the one sigma deviation from the mean (11%) is dominated by intra-pixel sensitivity variations (see next section) and not by residuals in the large scale structure of the flats. The results for the F222M filter show that a differential photometry of better than 2% can also be achieved in this camera.
As with many other array detectors, the sensitivity of NICMOS detectors is lower near the edges of the pixels than in their centers. It is as though there were small regions of reduced sensitivity along the intra-pixel boundaries. In practical terms this effect means that for a source whose flux changes rapidly on a size comparable with or smaller than the pixel size, the measured countrate, and therefore flux, will depend on where the center of the source lies with respect to the center of the pixel. Because this position is not known a priori, this effect introduces some uncertainty in the flux calibration for a point source.
For NICMOS detectors, this uncertainty is largest for camera 3 at short wavelengths, in which the PSF is largely undersampled. Calibration tests show that uncertainties are indeed largest (11%) in the blue filter (F110W; figure 7), decrease as the wavelength increases ( 6% for F160W) and are not measurable at long wavelengths (2 m and beyond). The effect of the intra-pixel sensitivity on the photometry is less important for observations of extended targets and outside the NIC3 campaigns when the camera is out of focus.
For high precision photometry and to compute the amount of photometric uncertainty in a given filter, in particular filters in the 1.0 - 1.8 m wavelength range, several measurements with subpixel dithering are recommended.
A calibration program designed to quantify the impact of the intra-pixel sensitivity in the NIC3 photometry will be executed in June 1998 during the second NIC3 campaign. The results of this program will be available on the NICMOS photometry WEB page soon after the analysis of the data is completed.
Many very red targets will be observed with NICMOS at short wavelengths (i.e. 1 m). For these sources the flux at 2.2m - 2.5m could be orders of magnitude larger than at 1.0m and therefore exceptionally good out-of-band blocking would be required. The filters for which red leaks might be a problem are F090M, F095N, F097N, F108N, F110M, F110W, and F113N. Images of a red star (Oph S1, J-K 3) have been obtained as part of the Cycle 7 calibration plan, the analysis is underway and the results will be posted on the NICMOS Web page. However, photometry of a somewhat less red star (BRI0021, J-K 1.3) show no indications of red leaks. Strategies involving observations in multiple filters to model the source spectral energy distribution would be required for very red stars, if red leaks were present.
The standard HST JHK system is formed by the F110W, F160W and F222M filters. NICMOS images are calibrated in units of Janskys and conversions to magnitudes are obtained defining the flux of a zero magnitude star in these filters. The spectrum of a zero magnitude star is defined by the model spectrum of Vega (see Colina, Bohlin & Vastelli, 1996b for details), correcting its flux to match the observed fluxes (i.e multiplying it by 1.05) and assuming Vega has a magnitude equal to 0.02 in all NICMOS bandpasses, as per calibration of Campins, Rieke & Lebofsky (1985).
The set of standard stars observed by HST (Table 1) contains blue stars (white dwarfs), intermediate color stars (solar analogs) and very red stars, and covers a large range in color (-0.2 J-K 3.0). This set of standards, together with ground-based measurements, will be used by the NICMOS Instrument Definition Team (M. Rieke and collaborators) to transform the HST magnitudes onto the CIT Arizona system.
There are various sources of uncertainties when performing photometry with NICMOS that could affect the final accuracy of the measurements. In the following, several systematic uncertainties and special cases are mentioned.
The white dwarf G191-B2B and the solar analog P330E are NICMOS primary spectrophotometric standards. To assess the accuracy of the G191-B2B model in the near-infrared, two atmosphere flux distributions for exactly the same physical parameters were computed independently by two different experts in this field. The largest differences in the continuum fluxes of the two independent models are 3.5% in the near-infrared at 2.5m (Bohlin 1996).
The spectral energy distribution of P330E in the 0.4m - 0.8m range is the same as the solar reference spectrum, within the uncertainties of the FOS measurements (Colina & Bohlin 1997). Also, the near-infrared spectrum of P330E, created by rescaling the reference spectrum of the Sun (see Colina & Bohlin 1997 for details), agrees to within 2% - 3% with ground-based near-infrared photometry.
In summary, the accuracy of the absolute spectral energy distribution of NICMOS primary standards introduces a systematic uncertainty of about 2% - 3% in the absolute calibration of the filters.
The photometric values provided in the headers of the images are obtained from measurements of standard stars positioned in the central regions of the detectors. The results of the relative photometry characterization of NICMOS cameras (see section 5 above) indicate that relative photometry to better than 2% can be achieved for all filters in cameras NIC1 and NIC2, and for long wavelength filters ( 2.0 m) in NIC3.
No evidence of intra-pixel sensitivity effects has been observed in cameras NIC1 and NIC2. However, as mentioned already (see section 6), the intra-pixel sensitivity affects NIC3 photometry in the 1.0 - 1.8 m wavelength range and errors as large as 10-20% can be present, if images are taken without a subpixel dithering strategy. Therefore, subpixel dithering is recommended for high precision photometry.
Changes in focus are observed on an orbital timescale due mainly to thermal breathing of the telescope. In addition to this short term PSF variation there is an additional long term NICMOS component, as the cryogen evaporates and the dewar relaxes. This last effect is critical for NIC3 images where the focus of the camera has changed significantly during Cycle 7. In addition, PSF changes as a function of position in the detector. All these effects are believed to affect at the few percent level photometry obtained with small apertures. However, no quantitative measurements are available and therefore Tiny Tim simulations are recommended to study these effects, if high precision is required.
The photometric conversion factors provided in the header of NICMOS images have been obtained by doing aperture photometry on standards using fixed radius apertures (see section 3 for details). It is often difficult to measure the total flux of a point source using large apertures where the flux contribution from the extended wings of the PSF, diffraction spikes, and scattered light is also included. This is in particular true in crowded fields where the extended wings of well resolved sources could overlap with each other. To take into account aperture correction effects it is advisable to use Tiny Tim PSFs to measure the encircled energy curve of growth as explained in the HST Data Handbook (Voit 1997).
The strong wavelength dependence of NICMOS flat-fields limits the photometric accuracy of sources with extreme colors observed in broad-band filters. Simulations with a very red source (J-K = 5 equivalent to a 700K black-body) indicates that these photometric errors are small, around 3% or less (MacKenty et al. 1997). Targets redder than J-K 5 could have photometric errors in excess of 3% for some of the filters.
The photometric conversion factors for all NICMOS filters are obtained from observations of continuum, emission line free, standards. The integrated flux inerg sec-1 cm-2 can be obtained as a function of the full width half maximum of the filter and the PHOTFLAM parameter as explained in the HST Data Handbook (Voit 1997). However, if the target has large velocity shifts the emission line does not coincide with the peak transmission of the filter, the line flux will be in error (few to several percent, depending on the filter and velocity shift) and a correction to account for the displacement is required. A method is indicated in the HST Data Handbook (Voit 1997).
One of us (L.C.) has benefited from several discussions with Dr. R. Bohlin and Dr. M. Rieke regarding various photometric issues.