2.1 ACS File Structure
Data taken with ACS are processed and calibrated by the ACS calibration pipeline which assembles data received from HST into datasets, applies standard calibrations, and stores the uncalibrated datasets in the HST Data Archive. The structure of these data products is based on the STIS and NICMOS file format and consists of multi-extension FITS files which store science (SCI), data quality (DQ) and error (ERR) arrays, as shown in figure 2.1. The combination of SCI, ERR and DQ extensions for a single chip is called an IMSET.
ACS WFC data come from two CCD chips, resulting in two separate images for a single observation. The data from each chip are treated as a separate observation in the ACS FITS files, with a SCI, DQ and ERR array for each chip, and with both chips being stored in the same FITS file. This results in a FITS file with 7 extensions for a single WFC exposure. A single HRC or SBC exposure has only 4 extensions, similar to the NICMOS and STIS data format.
Figure 2.1: Data format for ACS Modes 
The ACS calibration software (CALACS) uses association tables to keep track of which image extensions should be combined to form sub-products. The inputs for pipeline processing are:
- raw exposure - FITS formatted, integer data
- association table (only for associated data)
- calibration reference files
- trailer file from Generic Conversion (optional for reprocessing)
Processing single exposures will result in the creation of a single fully calibrated exposure stored in the FITS format shown at left in figure 2.1, where the pixel array of the processed WFC data is twice the size of the raw WFC pixel arrays. The current version of the calibration pipeline will also recognize and correctly process cosmic-ray split (CR-SPLIT)or multiple, repeat (REPEAT-OBS) exposures by interpreting the association table and determining which exposures should be combined during calibration. Dithered observations, however, are not handled by CALACS and require further processing by a separate task called PyDrizzle.
In addition to dither-combining associated data, PyDrizzle performs a geometric correction on all ACS data (dithered or not) and combines multiple images into a single output image. All fully calibrated ACS observations will therefore have only 4 extensions. PyDrizzle converts the data to units of count rate when performing geometric correction and/or dither-combining ACS data.
