New Flux Calibration for the WFC3 and ACS Detectors

A. Calamida (calamida[at], S. Baggett (sbaggett[at], J. Mack (mack[at], R. Avila (avila[at], R. Bohlin (bohlin[at], N. Grogin (nagrogin[at], and their teams

The revised flux distributions of the three primary HST standard white dwarfs and the increased reference flux of Vega (see related article by Bohlin in the current Newsletter) necessitated an updated flux calibration of all the HST instruments. In this article, we present the new calibrations of the WFC3 UVIS and IR detectors and the ACS detectors, available for all MAST calibrated data products starting October 15, 2020.

In the case of WFC3-UVIS, new inverse sensitivities (zero points) were calculated by using photometry collected from 2009 to 2019 for four CALSPEC1 white dwarfs, namely GRW+70D5824, GD153, GD71, G191B2B, and the G-type star P330E. The availability of this long baseline allowed us to measure low-level changes in sensitivity for all UVIS1 and UVIS2 filters (see Fig. 1, Calamida et al. 2020). These changes are of the order of ~0.10.2% per year according to the filter. The new inverse sensitivities were calculated at a reference time, May 2009, when WFC3 was installed in HST. After correcting for the sensitivity changes, values were also provided for each epoch of observations as inverse sensitivity keywords (PHOTFLAM, PHTFLAM1, PHTFLAM2) in the image headers. New inverse sensitivity ratios of UVIS2 over UVIS1, reported in the header keyword PHTRATIO, were also calculated and are in quite good agreement with values derived from dithered photometry of star clusters and white dwarfs.

New encircled energy (EE) corrections were derived for two filters, namely F275W and F814W, by correcting for time sensitivity changes of the UVIS detector before drizzling the Point Spread Functions to create a deep combined image for each filter. EE values derived for UVIS1 and UVIS2 are in very good agreement with each other and with the 2009 EE values. Given these favorable results, EE values for the other ultraviolet filters were adjusted by ~1% and EE for wavelengths larger than 7,500 Å were changed by ~0.5% to be in close agreement with the F275W and F814W EE values.

New inverse sensitivities were calculated for the WFC3-IR detector by using photometry collected from 2009 to 2019 for four CALSPEC white dwarfs and P330E (Bajaj et al. 2020). These data were calibrated by using new 'pixel-to-pixel' flats which correct for spatial sensitivity residuals up to 0.5% in the center of the detector and up to 2% at the edges (Mack et al. 2020). The flats were computed by stacking all available deep-IR exposures collected over ten years, providing high signal-to-noise images of the sky background. The change in the IR inverse sensitivities are primarily due to the new models, since the flats are normalized in the center of the detector where standard stars are observed.

Filter F814w and Modified Julian Date
Figure 1: Relative aperture photometry performed at 10 pixels on CTE-corrected (FLC) images collected in the F814W filter for the four CALSPEC white dwarfs, namely GRW+70D5824 (filled circle), GD153 (diamond), GD71 (square), G191B2B (triangle), and the G‑type star P330E (horizontal triangle) versus Modified Julian Date (MJD). Photometry performed on scanned images for GRW70 (blue) and GD153 (green) is also shown. Error bars are shown. The solid line is the fit to the photometry of all the stars, a sensitivity change of –0.11% per year.

Figure 2 shows the ratio of observed over synthetic count rates for the four CALSPEC white dwarfs and P330E for wide- and medium-band filters of both WFC3 detectors. The root mean square scatter around 1.0 is 0.5% for UVIS and 0.6% for the IR detector.

For both WFC3-UVIS and ‑IR detectors, all image header photometry keywords were updated with new values, i.e., the inverse sensitivity, PHOTFLAM, the inverse sensitivity for UVIS1 and 2, PHTFLAM1/2, and PHTRATIO.

It is important to note that the pixel values in the UVIS1 science arrays do not change; however, UVIS2 pixel values may change by a scale factor up to 1% due to the new chip sensitivity ratios. This ratio is used by the WFC3 reduction pipeline, calwf3, to flux correct the UVIS2 detector such that the count rates match those on the UVIS1 detector (for more details of the scaling process please see Section 3.2.12 of the Data Handbook).  WFC3 IR pixel values change slightly due to the flat-field updates.

Observed synthetic count (y) Piviot wavelength (x)
Figure 2: Ratio of observed to synthetic count rates for UVIS2 (Amp C) and IR wide- and medium-band filters for the four CALSPEC white dwarfs, namely GRW+70D5824 (black-filled circle), GD153 (red diamond), GD71 (cyan square), G191B2B (magenta triangle), and the G‑type star P330E (green triangle). Error bars are shown. The RMS scatter around 1.0 is 0.5% for the UVIS2 detector and 0.6% for the IR.

The new inverse sensitivities differ from the prior calibrations of 2017 (UVIS) and 2012 (IR) by ~0–1.5% according to the filter, with an average of ~1%. These differences account for all factors listed above, i.e., new models, time sensitivity corrections, new EEs and new flat fields. The new WFC3 inverse sensitivities are also available on the WFC3 Photometric Calibration webpage along with a Jupyter notebook illustrating the use of the new time-dependent UVIS solutions.

In the case of the ACS detectors, new inverse sensitivities were calculated for all the channels (WFC/HRC/SBC) using the updated models from Bohlin et al. (2020). Figure 3 shows the required change for the various ACS-WFC filters (from Bohlin, Ryon, & Anderson 2020; ISR ACS 2020‑08). The fitted smooth line in Figure 3 corresponds closely to the expected change in average flux of the primary standards.

Synthetic photometry (y) Wavelength (x)
Figure 3: Ratios of observed count rate C to synthetic photometry P vs. pivot wavelength for the ACS WFC. The black squares and their 3σ errors-in-the-mean represent the average of the three prime WDs, while the averages are shown for the G‑star P330E (green circle), K‑stars KF06T2 (purple star) and KF08T3 (inverted purple triangle), and GRW+70°5824 (red cross). The less precise results for the sparsely observed primary WD with the narrow‑ and medium-band filters are the open black squares.
Revised photometry (y) Wavelength (x)
Figure 4: As in Figure 3 for the revised ACS-WFC flux calibration.


After making the required changes to the ACS-WFC calibration, Figure 4 shows agreement to 0.1% of the observed stellar count rates with the synthetic photometry from the spectral energy distributions (SEDs) of the three primary standards (black squares).

The zero points for the SBC filters show a similar ~1–2% change.

The new ACS and WFC3 flux calibrations were incorporated in the calacs and calwf3 data reduction pipelines on October 15, 2020.

The authors would like to acknowledge their teams for the help and support provided for this work.


1 CALSPEC stars serve as the fundamental absolute flux standards for HST.