gsaoiinfo -- General information of GSAOI, GSAOI data and GSAOIreduction tasks
The GSAOI package contains tasks for processing GSAOI imaging data. The specifics of the individual tasks can be found in their help files. This document describes the common features of the tasks and gives a description of GSAOI and GSAOI data format. GSAOIEXAMPLES provides a sample processing script.
The tasks are designed to provide a fairly complete and flexible reduction for the purpose of assessing data quality at the time of observation. Real-time reductions may not be optimal for a particular science application. The GSAOI package scripts can be optimized for a particular application using the hidden parameters to achieve the best possible results.
The tasks produce log files for the performed processing steps. The name of the logfile may be set in each individual task or at the package level by setting gsaoi.logfile.
The tasks add header keywords to the output images. These header keywords contain information about the performed processing steps and the values of the critical parameters that were used.
The GSAOI detector plane is composed of a 2x2 mosaic of Rockwell Hawaii-2RG arrays, 2048x2048 pixels each, with a ~2.8 arcsec pixel gap between the arrays. The pixel scale is 0.02 arcsec/pixel for a field-of-view of 85x85 arcsec^2. Four pixels around each array are not illuminated, so the final usable dimension is a 4080 x 4080 pixel detector (the mosaiced image itself is larger, given the gaps).
The gaps are not uniform, since the arrays are not mounted strictly parallel to each other. A minimum offset of 4 arcsec is required to cover the gaps.
All GSAOI data frames are in MEF (multi-extension file) format, where the primary header unit (PHU, ) contains all the information from telescope, instrument component positions, instrument detector configuration, Observing Database, weather tower, GeMS (guide stars, loop status, laser status). The extensions ( to ) contain the actual pixel information and also have smaller headers, which contain the information particular to each array (ROI, WCS, gain, readout noise). Note the distinction between detector and array: exposure time, coadds, LNRS are in the PHU, since they are the same for all the four arrays; the WCS is in each extension, since each array has an individually defined CD matrix.
The array layout is  to  clockwise, starting from the lower right corner. This means that, when a mosaic is constructed, pixel (1,1) in the mosaic corresponds to pixel (1,1) in array  (lower left corner).
The arrays can be read in three modes: bright, faint and very faint, corresponding to 1 Fowler sample (FS; 1 read at the start of each coadd, one at the end), 4 FS and 8 FS, respectively. The larger the number of FS, the lower the readout noise, but the larger the overheads. The number of FS also define the minimum exposure time possible: when reading the full-frame in bright mode, the shortest exposure is 5.3 sec, increasing to 43 sec with 8 FS. Note: the keyword LNRS in the PHU of a GSAOI image reflects the number of non-destructive reads and is described as such. Therefore, it can take values of 2, 8 and 16. This is NOT the same as the other near-IR instruments in Gemini, for which the LNRS keyword refers to the number of FS instead.
There are three different regions of interest (ROI) that can be used to read the GSAOI arrays: full-frame (2048x2048), Detector and Array. Detector is defined as a square region of side X pixels centred in the detector (i.e., where the gaps cross). The "X" value does not include the gaps and can be 64, 128, 256, 512, 1024 and 2048 pixels. Array mode is defined as a square region of side X pixels, centred on each array; X can take the same values as for Detector, with Array 2048 being the same as full-frame. Detector and Array ROIs have not been commissioned and may not be handled properly by the GSAOI reduction package.
In addition, there is the option to use On-Detector Guide Windows (ODGW) during an observation. These are small square regions in each array (up to 128x128 pixels) that can be read at a much faster rate while exposing, to act as natural guide stars for the AO system. Up to three can be used at any time, only one per array. The use of ODGWs will result in a square area of the array containing no information about that area on the sky. These areas will be flagged as bad pixels in data quality extensions if they are created (it is recommended to create DQ extensions for this reason). Note that the ODGW follows the star it will be guiding on and therefore will move around when dithering. However, for sky frames where the guiding is frozen, the ODGW does not move and will leave a "hole" in the sky frame.
The ODGWs can be read at different rates (different exposure times) but even if only one is being used to guide, the other three will be also be reading out. ODGWs that are not in use are parked in the external corner of the corresponding array. The information about the size and position of actively guiding ODGWs in use are contained in the PHU, under the "GWFS?SIZ", "GWFS?X" and "GWFS?Y" keywords, respectively. For which, the "?" is a value of 1-4 given by the "GWFS?CFG" keyword that contains a value of "ODGWa" where a designating the array number. "?" and "a" may not be the same but "a" always refers to the array for which the ODGW was guiding. The positions stored in the "GWFS?X" and "GWFS?Y" keywords refer to the center of the ODGW for that array.
There is a static BPM that is supplied with the package, which masks out hot pixels and "depedded" diodes that create perfectly round dark regions 10 or so pixel across.
Since the dark current in itself is very low, subtracting darks with the same exposure time as the data is not really necessary - "warm" pixels are handled well by the normal sky subtraction.
Abbreviation Explanations --------------------------------------------------------------- GSAOI Gemini South Adaptive Optics Imager MEF Multi-extension FITS BPM A MEF bad pixel mask PHU Primary header unit (extension  of a MEF file) CCDSEC Section of an array relative to the origin of that array. DETSEC Section of an array relative to the origin of the entire detector METACONF PHU keyword that stores reduction information ODGW On-detector guide window ROI Region of interest SCI Science data pixel extension VAR Variance pixel extension DQ Data quality extension ---------------------------------------------------------------
All raw GSAOI data must be processed with GAPREPARE before any further processing. The GSAOI tasks rely heavily on the METACONF keyword stored in the PHU of a prepared image. For all tasks that create an output file, the METACONF keyword will be updated in the output file. The updated METACONF value will be expected by tasks that use the outputs from the task that created that file.
- GAPREPARE - Prepare raw GSAOI data for reduction
All raw GSAOI data must be processed with GAPREPARE before any further processing. GAPREPARE updates certain keywords used in further reduction. In addition, GAPREPARE can correct non-linear pixels and add variance and data quality extensions. GAPREPARE can flag non-linear and saturated pixels in the DQ extensions, if requested by the user. If a BPM is supplied the bad pixels will also be flagged in the output DQ planes. Finally, GAPREPARE will trim the data too.
Note: GAPREPARE can be called from within other reduction tasks.
- GADARK - Process and combine GSAOI dark images
GADARK creates a master dark frame that can be used to dark subtract science and other calibration data. Currently it is not expected that GSAOI data need be dark subtracted due to it's low dark current, meaning warm pixels should be handled well during sky subtraction.
- GAFLAT - Derive flat field for GSAOI images
GAFLAT is used to construct flat field images using either GCAL, DOME or Twilight flats. For GCAL data, a set of exposures with the shutter closed is combined and subtracted from a set with the shutter open. Note: GCAL flats do not use the same optical path as the science data - GCAL flats do not pass through the AO system.
- GASKY - Derive sky image for GSAOI, includes masking of objects
GASKY creates a sky frame for sky subtraction of science frames. It will perform a quick sky background subtraction of the input frames using the output from GAFASTSKY and flat field these images (a flat frame can be supplied otherwise a modified version of the GAFASTSKY output will be used) to perform object detection. The (unprocessed) input frames are then combined using the recently created object masks to mask out any objects in them.
- GAFASTSKY - Derive sky image for GSAOI, median or min/max filtering
GAFASTSKY is used for constructing sky images by median or min/max filtering. There is made no attempt to specifically flag objects in the sky images. This task is used primarily to create a quick sky frame to subtract from sky frames when about to perform object detection when reducing sky frames with GASKY.
- GAREDUCE - Reduce images from GSAOI (trim, dark and flat correct)
GAREDUCE will reduce science data to an state at which it can be analyzed. This includes preparing the data with GAPREPARE (trimming too), dark subtraction, flat division, sky subtraction and conversion to electrons. The user has the option to run one or more of these steps at a time. GAREDUCE can be called on a file that was created by GAREDUCE, assuming that the second call requests different reduction steps to that of the first. It is possible to use the input science images as sky frames in the call to GAREDUCE. In this case, for a given science frame, the appropriate science frames will be selected (based on input parameters) to be used to create a sky frame (via GASKY) for that science frame.
- GAMOSAIC - Mosaic the 4 GSAOI arrays into one image
GAMOSAIC will create a MEF file with a single pixel extension that contains the data from the four arrays corrected for their position on the sky, using the WCS of the second extension as the reference. NOTE: the geometric transform (i.e., when not just pasting) in GAMOSAIC is not performed well - the output WCS is currently not very good and it is possible that the static distortion of arrays may not be handled well.
- GADISPLAY - Display GSAOI images
GADISPLAY displays GSAOI MEF images, images with four SCI extensions will be displayed as a single image. The user has the option to examine the image using imexam.
- GACALFIND - Create a table of calibration information
GACALFIND creates a table of information for calibration data within a given directory. This allows the automatic selection of processed calibrations from within that directory.
- GACLTRIM - Trim calibration images to match science images
GACLTRIM trims calibration images to match the dimensions of input science frames, if required, after comparing their dimensions using the CCDSEC of each science extension
- GASTAT - Calculate statistics for a GSAOI image.
GASTAT will calculate a requested pixel statistic on the input image. GASTAT can use DQ or input BPMs to mask pixels in the calculations. It is also possible to determine the requested statistic subsection of an individual array, each individual array, or the whole detector.
- GADIMSCHK - Check supplied statistic section for an image and or
- the dimensions of two GSAOI images against each other
GADIMSCHK has two functions. The first to check whether a supplied statistics region is valid and or to check the dimensions (and detector footprint) of two GSAOI images against one another. The latter allows calibrations taken with a different (larger) ROI be used to reduce science data. The checks can only be preformed on one extension at a time.
- GAIMCHK - Check the obstype etc., of input GSAOI images
GAIMCHK is the task that performs the majority of the smart sorting features available in some of the tasks. It also performs the general input and output checks.
The tasks in the GSAOI package are designed to operate on MEF images that have been processed using GAPREPARE. The task GAPREPARE will not run on data from instruments other than GSAOI and will not run on simple FITS files.
Testing of how the package handles ROI data is ongoing and as such it is possible that the task outputs for ROI data may be incorrect.
Finally, the registering and stacking of mosaiced images (e.g., by WCS) is not currently handled well by the IMCOADD task, which is partly due to the incorrect output from GAMOSAIC after a geometric transform. It is likely better, if the image has enough point sources to paste the images using GAMOSAIC then interactively run IMCOADD with alignment="user" and geofitgeom="rxyscale". This is not optimal and will likely give an incorrect WCS, at best. The user should inspect the output from this method thoroughly.