STScI Logo

Hubble Space Telescope
What are ACS subarrays? How can I use them?

WFC subarrays

WFC observers can read out regions of either CCD that are smaller than the full 4K x 2K pixel array. Because the read time and data volume of subarray images scale with their dimensions, the large fractional overhead associated with short, full-array exposures can be significantly reduced by using subarrays. WFC subarrays are either pre-defined or user-defined, but calibration support for the different types of subarrays has varied from cycle to cycle. Since Cycle 18, STScI has supported nine pre-defined subarray apertures (five of which have distinct sizes and locations) and has discontinued support for user-defined subarrays. Unsupported subarrays are available for scientific use, but observers must provide a justified request for their use that includes a plan to obtain bias calibration images as part of their science program. Fully-supported, pre-defined subarrays are automatically assigned to WFC exposures using any polarizer or the F892N filter.

The locations, sizes, and output nodes of WFC subarrays are governed by several rules. The first and last rows of pixels in the light-sensitive region of each CCD (i.e., rows 1 and 2048) and the first and last pixels of the extended serial register (i.e., prescan columns 1 and 4144) in each CCD are unavailable to subarrays. Also, the first pixel read from a subarray must lie in an odd-numbered column of the full CCD array. Consequently, subarrays have at most 22 columns of physical prescan, 22 columns of physical overscan, and no rows of virtual overscan. All subarrays must have even numbers of rows and columns, and they can be read through only one output amplifier.

The sizes and locations of the subarrays are recorded as keywords in the headers of the raw science image files and the telemetry and engineering data files (i.e, the files with "raw" and "spt" suffixes, respectively) contained in each OPUS dataset. Note that the values of these keywords in the headers of some CALACS output files (e.g., files with "flt" and "drz" suffixes) are altered by CALACS to reflect operations that alter the image format. Consequently, the following keyword definitions and constraints apply only to the original values contained in the "raw" and "spt" file headers.

The "spt" file header contains the keywords NUMROWS, NUMCOLS, XCORNER, and YCORNER, which respectively list the row and column dimensions of the subarray and the row and column coordinates of the corner of the subarray closest to amplifier A in WFC1 (Figure 1). The "raw" file header contains the keywords SIZAXIS1, SIZAXIS2, CENTERA1,and CENTERA2, which respectively list the column and row dimensions of the subarray and the column and row coordinates of the center of the subarray with respect to a fiducial pixel in either WFC1 or WFC2, depending on where the subarray is located (Figure 2). Because SIZAXIS1 and SIZAXIS2 must be even numbers, CENTERA1 and CENTERA2 are defined to be on the high and low sides of the subarray bisectors, respectively.

The arithmetic relationships between (NUMROWS, NUMCOLS, XCORNER, YCORNER) and (SIZAXIS1, SIZAXIS2, CENTERA1, CENTERA2) are:

SIZAXIS1 = NUMCOLS
SIZAXIS2 = NUMROWS
CENTERA1 = 1 + YCORNER + NUMCOLS/2
CENTERA2 = 1 + 2048*[1 + (XCORNER > 2048)] - (XCORNER + NUMROWS/2),

where (XCORNER > 2048) equals 1 or 0 if the condition is true or false, respectively. Note that (XCORNER, YCORNER) coordinate system is a single 4144 x 4096 grid with origin (0,0) that spans both WFC1 and WFC2, while the (CENTERA1, CENTERA2) coordinate system comprises two 4144 x 2048 grids, each with origins (1,1), that span WFC1 and WFC2 separately.

Figure 1. WFC subarray coordinate system used in telemetry and engineering data file headers.


Figure 2. WFC subarray coordinate system used in raw science image file headers.


Although user-defined subarrays are currently unsupported by STScI, we describe how they are defined within an observing proposal for archival reference and potential future use.

User-defined subarrays are specified by the optional exposure parameters SIZEAXIS1, SIZEAXIS2, CENTERAXIS1, and CENTERAXIS2 in the Phase II proposal. The SIZEAXIS1 and SIZEAXIS2 parameters have the same meaning as the SIZAXIS1 and SIZEAXIS2 keywords in the "raw" image header. The SIZEAXIS parameters must be even numbers greater than or equal to 16. The (CENTERAXIS1, CENTERAXIS2) parameters and the (CENTERA1, CENTERA2) keywords describe the same central point within the subarray, but their coordinate systems are defined differently. The (CENTERAXIS1, CENTERAXIS2) coordinate system is a single 4144 x 4096 grid which has the same origin as the (CENTERA1, CENTERA2) coordinates for WFC2 subarrays (Figure 3). The following boundary conditions must be satisfied:

(a) CENTERAXIS1 - (SIZEAXIS1)/2 is an odd number

(b) (SIZEAXIS1)/2 + 3 <= CENTERAXIS1 <= 4143 - (SIZEAXIS1)/2

(c) (SIZEAXIS2)/2 + 2049 <= CENTERAXIS2 <= 4095 - (SIZEAXIS2)/2 for WFC1 aperture

(d) (SIZEAXIS2)/2 + 1 <= CENTERAXIS2 <= 2047 - (SIZEAXIS2)/2 for WFC2 aperture

The values of CENTERAXIS1 and CENTERAXIS2 need not be specified by the user. If unspecified, the Astronomers Proposal Tool (APT) will center the subarray at the reference pixel of the desired exposure aperture, provided the boundary conditions are satisfied. APT will make any necessary adjustments to the position of the subarray and report these adjustments as warnings to the user.


Figure 3. WFC subarray coordinate system used in Phase II proposal preparation with APT.


Before ACS's electronic failure in 2007, most WFC subarray images were calibrated from the same bias, dark, and flat-field reference frames that were used to calibrate the full-array WFC images. Special bias frames were required only for user-defined subarrays that crossed the boundary between adjacent quadrants. Except in these special cases, the OPUS pipeline extracted the regions of the full-array bias, dark, and flat frames that matched the footprint of the subarray on the full array. Unfortunately, the bias gradients produced by the new CCD electronics installed in SM4 vary with readout mode, so the full-array bias frames cannot be used to calibrate post-SM4 subarray images. STScI now obtains subarray bias frames for all supported subarray formats. To offset the increased calibration time needed to obtain these subarray bias frames, STScI supports only five distinct subarray types in the post-SM4 era.

CALACS version 2012.1 (released in Spring 2012) features new pixel-based corrections for post-SM4 electronic artifacts (amplifier crosstalk, signal-dependent bias shift, 1/f noise in bias offset) and degraded charge-transfer efficiency (CTE) from prolonged exposure to HST's radiation environment. These corrections have been derived from, and developed for, full-frame WFC images only. Because the electronic and CTE effects depend on CCD readout patterns and timing, the new CALACS corrections are not applied to WFC subarray images. Consequently, users should beware of possible photometric and astrometric inconsistencies between full-frame and subarray WFC images of the same science target after CALACS processing.

----------------------------------------------------------------------------------------------------------------

HRC subarrays

For the now defunct HRC, AXIS1 ranged from 1 to 1062, including 19 pixels of physical prescan and overscan. AXIS2 ranged from 1 to 1024. SIZEAXIS1 and SIZEAXIS2 were even numbers ranging from 16 to 1058 and 1022, respectively. The CENTERAXIS positions could not cause the array to overlap the edge pixels and CENTERAXIS1-SIZEAXIS1/2 had to be odd. The CENTERAXIS1 and CENTERAXIS2 coordinates were the reference position for the observation. The boundary conditions were:

(a) (SIZEAXIS1)/2 + 3 <= CENTERAXIS1 <= 1061 - (SIZEAXIS1)/2

(b) (SIZEAXIS2)/2 + 1 <= CENTERAXIS2 <= 1023 - (SIZEAXIS2)/2

HRC had two pre-defined subarrays. Subarray HRC-512 was a 512 x 512 subarray adjacent to amplifier C with 18 columns of prescan pixels. Subarray HRC-SUB1.8 was a 512 x 512 subarray centered on the 1.8 arcsecond coronagraphic spot. Because this subarray did not reach the detector edge, no overscan pixels were included.