VEGAMAG is a widely used standard star system, defined as relative photometry to the actual star Vega; Vega magnitudes can be converted to absolute flux using the flux distribution of this HST secondary standard star1
in the CALSPEC database at:
The VEGAMAG system uses Vega (
α Lyr) as the standard star. The spectrum of Vega used to define this system is a composite spectrum of
empirical and synthetic spectra (Bohlin & Gilliland, 2004
). The “Vega magnitude” of a star with flux F
is the calibrated spectrum of Vega in synphot
. In the VEGAMAG system, by definition, Vega has zero magnitude at all wavelengths.
is expressed in erg
, and Fλ
. Another way to express these zeropoints is to say that an object with a constant Fν
will have magnitude AB
0 in every filter, and an object with Fλ
will have magnitude STMAG
0 in every filter.
to renormalize a spectrum to 1 count/second in the appropriate ACS passband and specify an output zeropoint value based on a selected magnitude system (Be sure to verify that the most updated throughput tables are being used). For instance, the calcphot task can be used with a renormalized 10,000
°K blackbody to calculate the photometric parameters for a passband, using the following syntax:
where, in this example, obsmode
is the passband, e.g. "acs,hrc,f555w"
is the output magnitude system: ABMAG
, or STMAG
is the inverse sensitivity (erg
); it represents the flux of a source with constant Fλ
which produces a count rate of 1
electron per second.
is the STMAG zeropoint, permanently set to -21.1.
The header keywords PHOTFLAM
, and PHOTPLAM
relate to the STMAG
zeropoints through these formulae (See also Bohlin, et al. 2011, AJ, 141, 173).
For point sources, ACS zeropoints (available in STMAG, ABMAG, and VEGAMAG) are applied to a measured magnitude that is aperture-corrected to a nominal “infinite” aperture of radius 5".5. For surface photometry, these zeropoints can be directly added to values in units of magnitude/arcsec2
. In order to facilitate the conversion for point sources, Sirianni et al., 2005, PASP
provides corrections derived from the encircled energy profiles of an intermediate aperture (0".5 radius) to an infinite aperture. The Sirianni paper has been superseded in a number of aspects by Bohlin 2012 (ACS ISR 12-01
) and the revised encircled energy fractions for both a 0".5 and 1".0 aperture from Table 1 of Bohlin (2012) are reproduced here as Table
Aperture corrections for near-IR filters present further complications because the ACS CCD detectors suffer from scattered light at long wavelengths. These thinned backside-illuminated devices are relatively transparent to near-IR photons; the transmitted long wavelength light illuminates and scatters in the CCD soda glass substrate, is reflected back from the header’s metallized rear surface, then re-illuminates the CCDs frontside photosensitive surface (Sirianni et al. 1998, proc SPIE vol. 3555, 608, ed. S. D’Odorico
). The fraction of the integrated light in the scattered light halo increases as a function of wavelength. As a consequence, the PSF becomes increasingly broad with increasing wavelengths. WFC CCDs incorporate a special anti-halation aluminum layer between the frontside of the CCD and its glass substrate. While this layer is effective at reducing the IR halo, it appears to give rise to a relatively strong scatter along one of the four diffraction spikes at wavelengths greater than 9000 Å (Hartig et al. 2003, proc SPIE vol. 4854, 532, ed. J. C. Blades, H. W. Siegmund
), see Section 5.1.4
and Bohlin (2012)
for more details.
The same mechanism responsible for the variation of the intensity and extension of the halo as a function of wavelength is also responsible for the variation of the shape of the PSF as a function of color of the source. As a consequence, in the same near-IR filter, the PSF for a red star is broader than the PSF of a blue star. Gilliland & Riess (2002 HST Calibration Workshop Proceedings, page 61)
. and Sirianni et al., 2005, PASP
provide assessments of the scientific impact of these PSF artifacts in the red. The presence of the halo has the obvious effect of reducing the signal-to-noise and the limiting magnitude of the camera in the red. It also impacts the photometry in very crowded fields. The effects of the long wavelength halo should also be taken into account when performing morphological studies and performing surface photometry of extended objects (see Sirianni et al., 2005, PASP
for more details).
The aperture correction for red objects should be determined using an isolated, same-color star in the field of view, or by using the effective wavelength versus aperture correction relation (Sirianni et al., 2005, PASP
). If the object’s spectral energy distribution (SED) is available, an estimate of the aperture correction is also possible with synphot
; the parameter aper
has been implemented to call the encircled energy tables in the obsmode synphot
field for ACS. A typical obsmode
for an aperture of 0".5 would be specified as "acs,wfc,f850lp,aper$0.5"
. A comparison with the infinite aperture magnitude using the standard obsmode "acs,wfc,f850lp"
would give an estimate of the aperture correction to apply. Please refer to the Synphot Users Guide
for more details.
In some cases, ACS photometric results must be compared with existing datasets in different photometric systems (e.g., WFPC2, SDSS, or Johnson-Cousins). Because the ACS filters do not have exact counterparts in any other standard filter sets, the accuracy of these transformations is limited. Moreover, if the transformations are applied to objects whose spectral type (e.g., color, metallicity, surface gravity) do not match the spectral type of the calibration observation, systematic effects could be introduced. The transformations can be determined by using synphot
, or by using the published transformation coefficients (Sirianni et al., 2005
). In any case, users should not expect to preserve the 1% - 2% accuracy of ACS photometry on the transformed data.
When ACS images are flat-fielded by the calacs
pipeline; the resultant flt.fits
files are “flat” if the original sky intensity was also “flat.” However, there is still very significant geometric distortion remaining in these images. The pixel area on the sky varies across
the field, and as a result, relative point source photometry measurements in the flt.fits
/flc.fits images are incorrect.
One option is to drizzle the data; this will
remove geometric distortion while keeping
flat. Therefore, both surface and point source relative photometry can be performed correctly on
the resulting drz.fits
files. The inverse sensitivity (in units of erg cm-2 sec-1 Å-1), given by the header keyword PHOTFLAM, can be used to compute the STMAG or ABMAG zeropoint, and to convert
flux in electrons/seconds to absolute flux units:
|PHOTFLAM3 is the
mean flux density (in erg cm-2 sec-1 Å-1)
that produces 1 count per second in the HST
observing mode (PHOTMODE
) used for the observation.
|PHOTZPT3 is the ST magnitude zeropoint (= 21.10).
Users who wish to perform photometry directly on the distorted flt.fits
files, rather than the drizzled (drz.fits
) data products, will require a field-dependent correction to match their photometry with that obtained from drizzled data. Only then can the PHOTFLAM
values in the flt.fits
images be used to obtain calibrated STMAG or ABMAG photometry. (Note: the corresponding drz.fits
image has identical PHOTFLAM
The correction to the flt.fits
images may be made by multiplying the measured flux in the flt.fits
image by the pixel area at the corresponding position using a pixel area map (PAM), and then dividing by the exposure time. The easiest way to do it is to simply multiply the flt.fits
/flc.fits image with its corresponding pixel area map.
/crc.fits and flt.fits
images have units of electrons/seconds, and have the same PHOTFLAM
To illustrate the concepts of extended source and point source photometry on flt.fits
images, consider a simple idealized example of a 3
3 pixel section of the detector, assuming that the bias and dark corrections are zero and that the quantum efficiency is unity everywhere.
For an extended object with a surface brightness of 2
/pixel in the undistorted case, an image without geometric distortion is:
The geometrical area of each pixel is imprinted in the flat field, along with its
photometric sensitivity. In this example, the quantum efficiency is unity everywhere, so the flat field is the equivalent of the PAM:
is run on a flt.fits
image, the output image is free of geometric distortion and is photometrically accurate.
When drizzling a single image, the user may want to user the Lanczos kernel which provides the best image fidelity for the single image case.
However, this kernel does not properly handle missing data and causes ringing around cosmic rays. Thus, the Lanczos kernel should not be used for combining multiple images where sections of the image lost to defects on one chip can be filled in by dithering.
For additional information about the inner workings of AstroDrizzle, please refer to The DrizzlePac Handbook
Point spread functions in the ACS cameras are more stable over the field of view than in any other HST
camera, especially when compared to WFPC2. Variations in the HRC are very small and probably negligible when using apertures greater than r
1.5 pixels or using PSF fitting. However, the WFC PSF varies enough in shape and width that significant photometric errors may be introduced when using small apertures or fixed-width PSF fitting. These effects are described in detail in ACS ISR 2003-06
The WFC PSF width variation is mostly due to changes in CCD charge diffusion. Charge diffusion, and thus the resulting image blur, is greater in thicker regions of the detector (the WFC CCD thickness ranges from 12.6 to 17.1
microns, see Figure 5.2
). At 500
nm, the PSF FWHM varies by 25% across the field. Because charge diffusion in backside-illuminated devices like the ACS CCDs decreases with increasing wavelength, the blurring and variations in PSF width will increase towards shorter wavelengths. At 500
nm, photometric errors as much as 15% may result when using small (r
pixel) apertures. At r
4 pixels, the errors are reduced to <
1%. Significant errors may also be introduced when using fixed-width PSF fitting (see ACS ISR 2003-06
Observers may use TinyTim
to predict the variations in the PSF over the field of view for their particular observation. TinyTim
accounts for wavelength and field-dependent charge diffusion and aberrations. The software is available at
Long wavelength (
nm) photons can pass entirely through a CCD without being detected and enter the substrate on which the detector is mounted. In the case of the ACS CCDs, the photons can be scattered to large distances (many arcseconds) within the soda glass substrate before reentering the CCD and being detected—it creates a large, diffuse halo of light surrounding an object, called the “red halo.” This problem was largely solved in the WFC by applying a metal coating between the CCD and the mounting substrate that reflects photons back into the detector. Except at wavelengths longer than 900
nm (where the metal layer becomes transparent), the WFC PSF is unaffected by the red halo. The HRC CCD, however, does not have this fix and is significantly impacted by the effect.
The red halo begins to appear in the HRC at around 700
nm. It exponentially decreases in intensity with increasing radius from the source. The halo is featureless but slightly asymmetrical, with more light scattered towards the lower half of the image. By 1000
nm, it accounts for nearly 30% of the light from the source and dominates the wings of the PSF, washing out the diffraction structure. Because of its wavelength dependence, the red halo can result in different PSF light distributions within the same filter for red and blue objects. The red halo complicates photometry in red filters. In broad-band filters like F814W and especially F850LP (in the WFC as well as the HRC), aperture corrections will depend on the color of the star, see Section 5.1.2
. for more discussion of this. Also, in high-contrast imaging where the PSF of one star is subtracted from another (including coronagraphic imaging), color differences between the objects may lead to a significant residual over- or under-subtracted halo.
In addition to the halo, two diffraction spike-like streaks can be seen in both HRC and WFC data beyond 1000
nm (including F850LP). In the WFC, one streak is aligned over the left diffraction spike while the other is seen above the right spike. For HRC, the streak is aligned over the right diffraction spike while the other is seen below the left spike. These seem to be due to scattering by the electrodes on the back sides of the detectors. They are about five times brighter than the diffraction spikes and result in a fractional decrease in encircled energy. They may also produce artifacts in sharp-edged extended sources.
WFC images of the standard star GD71 through filters F775W (9
sec., left), F850LP (24
sec., middle), and FR1016N at 996
sec., right). The CCD scatter, undetected below ~800
nm, grows rapidly with longer wavelength. In addition to the asymmetrical, horizontal feature, a weaker diagonal streak also becomes apparent near 1
Å, the low- and mid-spatial frequency aberrations in HST
result in highly asymmetric PSF cores surrounded by a considerable halo of scattered light extending 1
2 arcseconds from the star. The asymmetries may adversely affect PSF-fitting photometry if idealized PSF profiles are assumed. Also, charge scattering within the SBC MAMA detector creates a prominent halo of light extending about 1" from the star that contains roughly 20% of the light. This washes out most of the diffraction structure in the SBC PSF wings.
Anderson and Bedin (2010PASP..122.1035A
) have developed an empirical approach based on the profiles of warm pixels to characterize the effects of CTE losses for WFC. Such an algorithm first develops a model that reproduces the observed trails, and then inverts the model to convert the observed pixel values in an image into an estimate of the original pixel values. The pixel-based CTE correction, applicable only to WFC images, has been implemented in the ACS calibration pipeline (calacs
). Data products that are corrected for CTE losses have the suffix flc.fits
, which correspond to the (uncorrected) flt.fits
images. Additional information on the pixel-based CTE correction can be found in Section 4.6
The CTE-correcting formulae described in ACS ISR 2012-05
and ACS ISR 2009-01
can also estimate lost flux as a function of source brightness, sky brightness, x
position, and time. The pixel-based CTE correction on stellar fields is in general agreement with these photometric correction formulae. Statistically significant deviations are observed only at low stellar fluxes (~300
and lower) and for background levels close to 0
. In such a range of flux and sky levels, the correction formulae appear to be more accurate than the pixel-based correction.
images using calacs
, then use AstroDrizzle
to create drz.fits
data. Alternatively, drz.fits
files from the HST
Archive may be used if those images are acceptable.
|If the drz.fits
image was created from combining several flt.fits
images with the same exposure time, multiply the drz.fits
image by the exposure time of a single flt.fits
|If the drz.fits
file was made from input flt.fits
images with different exposure times, do not use that drizzle-combined image. Instead, run AstroDrizzle
to create single exposure drz.fits
files for each flt.fits
file. In these situations, the recommended method for photometry is to perform the measurements on each single exposure drz.fits
image. Those results can be corrected for CTE losses, then averaged to obtain a more accurate measurement of the stellar flux.
If the dithers have larger shifts, carefully check the original position of each star in the flt.fits
files, then derive the correction using the average value of Ytran
. While doing so, it is also recommended to verify that parameters used in AstroDrizzle
do not generate any biases when the cosmic ray rejection step is performed—the flux of a star located at different positions on the chip might significantly differ, and AstroDrizzle
might interpret such objects as cosmic rays.
is the stellar flux measured within the aperture radius (in electrons).
is the local background for each star (in electrons).
is the observation date in modified Julian days.
|Ytran is defined in step 5.
Any additional corrections (i.e., transforming the flux from e−
/sec and applying the zeropoints to transform the flux into AB or Vega magnitudes) should be performed after all of the steps above.
Δmag = 10A x SKYB x FLUXC x (Ytran/2000) x (MJD-52333) / 365
= -0.14 (0.04), B
=-0.25 (0.01), C
[p1 Log(SKY) Log(FLUX) t + p2 Log(SKY) Log(FLUX) + p’1 Log(SKY) t + q1 Log(Flux) t + p’2 Log(SKY) + q2 Log(FLUX) + q’1 t + q’2] * Ytran / 2000
Users should be reminded that the formula for WFC was calibrated using stellar fluxes between ~50 e−
and ~80,000 e−
(measured within the 3 pixel aperture radius), and for background levels between ~0.1 e−
and ~50 e−
. Therefore, to ensure the highest level of accuracy, the formula should be used to correct photometry of stars that are within the range specified above. However, note that for stellar fluxes and background levels higher than the above limits, the amount of the correction is < 2%, even for stars located at the edge of the chip, far from the amplifiers. For very bright stars, the CTE formula currently overestimates CTE losses. For the specific case of very bright stars, the use of flc.fits
files (i.e., those obtained with the pixel-based CTE correction included in the ACS pipeline) is a better option.
Δmag = 10A x SKYB x FLUXC x (Ytran/1000) x (MJD-52333) / 365
=-0.44 (0.05), B
=-0.15 (0.02), C
When designing a UV filter, a high suppression of off-band transmission, particularly in the red, had to be traded with overall in-band transmission. The very high blue quantum efficiency of the HRC, compared to WFPC2, made
it possible to obtain an overall red leak suppression comparable to that of the WFPC2 while using much higher transmission filters. The ratio of in-band versus total flux, determined using in-flight calibration observations, is given in Table 5.3
for the UV and blue HRC filters, where the cutoff point between in-band and out-of-band flux is defined as the filter’s 1% transmission points. This is described in ACS ISR 2007-03.
This ISR also reports on the percentage of in-band flux for seven stellar spectral types, elliptical galaxies spectrum (Ell. G), a reddened (0.61 < E(B-V) < 0.70) starburst galaxy (SB), and four different power-law spectral slopes
In an ongoing calibration effort, the star cluster NGC
6681 has been observed since the launch of ACS to monitor the UV performance of the HRC and SBC detectors.
Preliminary results for the HRC detector for the first year following launch were published in ACS ISR 2004-05
. For the three filters, F220W, F250W, and F330W, eight standard stars in the field were routinely measured, indicating a sensitivity loss of not more than ~1% to 2% per year.
Absolute UV sensitivity in the SBC
just after launch appeared to decline linearly for the first ~1.6 years in orbit and then leveled
off. Following the methodology for applying the STIS time-dependent sensitivity corrections (STIS ISR 2004-04
), this SBC data have been fit using line segments. Because the sensitivity losses moderate after ~1.6 years in all five SBC filters, this was chosen as the break point for the two line segments. The slope of the first line segment gives the sensitivity loss and is summarized in Table 5.5
(in percent per year).
A decline in UV sensitivity of ~2%
4% per year was
found for the SBC detector during this time. This result agrees with the sensitivity loss derived for the STIS FUV MAMA with the G140L filter for the first 5 years in orbit, where a 2% to 3% loss was observed. In the last two years before its failure4 prior to SM4,
the rate of the STIS MAMA sensitivity loss slowed significantly compared to the previous 5 years. Similarly, after 1.6 years the ACS SBC loss rate has been
consistent with zero.