SuperCOSMOS Processing Software

A number of items of software have been developed by the ROE Software Group for the processing of SuperCOSMOS data. Although these are specific to the operation of the facility, and in general will only be used by the SuperCOSMOS operator, nonetheless it was considered useful to describe them here so that the reader may see the range of software available.

Also, although the SuperCOSMOS chain proceeds automatically to process the pixel data through to an object catalogue, any part of the chain can be started by the operator, or indeed the entire chain can be manually started. This has been found to be useful if the operator requires to reprocess the data for some reason (e.g. using a different threshold for object detection, etc.).

In order to be able to call any of the routines listed below, it is first necessary to "source" the SuperCOSMOS login file. This is done by calling:-

source /home/scosdev/v0xx/.login

where the numeric 'xx' refers to the version number of the SuperCOSMOS software currently regarded as the operational version. This is available from the SuperCOSMOS personnel.

List of the software


Conversion programme to convert an "Image Analysis Mode" data file (4 bytes per parameter) from DEC Alpha byte order into SUN UNIX byte order.


Conversion programme to convert a SuperCOSMOS Mapping Mode data file (2 bytes per pixel) from DEC Alpha byte order into SUN UNIX byte order. Needs to be run separately for each lane (or strip) of pixel data.


This is the command needed to initiate the COSMOS 'analyser'. The working of the analyser is described in: Beard et al. 1990, MNRAS, 247, 311. The output from the analyser is a set of 32 parameters for each object detected above the threshold for detection.


This is the main part of the digital stacking system. It combines 2 data sets from 2 different plates of the same part of the sky, and outputs a 'merged' (i.e. coadded data set). It is assumed that the pixel data are on the same rectilinear grid, and this is achieved using the programme 'regrid'.
Alternatively, the output from "bigmerge" together with another set of pixel data can be used as input in order to keep stacking more and more plates together. The running output is the arithmetic mean of all plates used for the stack.
Note that once a final coadded data set has been produced, one needs to run:- "sfudgelut" in order to create a Look-Up-Table of Transmission to Intensity values), "offsky" in order to determine the sky background for object detection, and "runbatch" in order to run the SuperCOSMOS processing chain (including analyser).


Given a set of pixel data, this programme can be used to extract any part of the data as a 2-D Starlink NDF image (with .sdf extension). The resulting image can be displayed directly using any of the standard Starlink packages (such as kappa).
Note that this programme assumes that the pixel data are not to be squashed down (i.e. 1:1 from the raw data); therefore care should be taken with the size of image extracted (e.g. a 4k x 4k image can be slow on some machines).


This is a more flexible version of bigimtondf. The same applies, viz. any area can be extracted, however in this case a compression ratio can be applied to "block down" the final output 2-D image. In the extreme, the whole of a Schmidt plate can be compressed in this way, a full plate compressed to 2k x 2k for example takes about 15 mins on a DEC Alpha 3000 model 600.


This programme is part of the SuperCOSMOS Quality Control suite of programmes and provides a postscript file with a 'dot' plotted for each object detected. This is useful, for example, in viewing the overall detection of objects over the plate, and highlights interesting objects such as star or galaxy clusters. It can be run standalone outside the QC suite.


This programme allows a pre-determined focus surface to be computed and used as the focus map for measurements. It is useful if one wants to, for example, constrain the focus surface to be a plane.


Allows the operator to 'poke' a value into the raw (NDF version) of the Housekeeping file. This programme is useful if one wants to change parameters for the processing (e.g. switching off 'deblending', adjusting parameters for transformation of xy to Ra,Dec, etc).


Produces a postscript file of histograms of various parameters from all of the objects in the IAM dataset (viz. frequency distribution of magnitude, area, ellipticity and orientation).


Produces the ASCII version of the Housekeeping file from the raw NDF version. (In practice, as data may be distributed to non-Starlink sites, although the SuperCOSMOS processing chain utilises the NDF version of the Housekeeping file it is recognised that this is not transportable generally and hence an ASCII version is produced for final data transportation).


Given the raw output from the SuperCOSMOS analyser, this programme sorts the objects into their respective 'lanes' in the data set and in ascending y order up each lane. As some objects may straddle lane boundaries, the lane in which they appear in the raw output is not necessarily the correct lane according to their centroids. If an object is a 'parent' object with 'child' objects, then the objects are placed into the lane corresponding to the position of the parent, and the corresponding child objects are positioned in the list immediately following their parent, and appearing in order of ascending y coordinate.
This programme may also be used if one just wants to output parent objects as well as non-deblended objects or just the child objects with non-deblended. This is acheived by using "hdspoke" to alter the parameter in the NDF version of the Housekeeping file.


Normally for a plate scan, the transputers determine the sky background (on a scale of 128 x 128 pixels) across the plate. However, if one has stacked several plates together, then one needs to redetermine the new sky background. This programme, therefore, performs the off-line equivalent of what gets done by the transputers.


Given two sets of IAM data from 2 different plates of the same field, then this programme takes the plate solution of the 'master' set to transform the 'slave' data set into the xy coordinates of the master. This is useful for some pairing methodologies, which pair up the different data sets in terms of x and y coordinates on the plate.


Given the celestial coordinates (referred to J2000 equinox) of a region of interest on the sky together with the scan of a plate of that region, then this program allows the user to extract the region of interest from the pixel map. The output is stored as an NDF file (with .sdf extension) and the data can be viewed and manipulated using the standard STARLINK utilities. The programme is similar to 'blocktondf' in that a compression factor can be provided to 'block' down the data; the main difference is that the coordinates are input in celestial rather than rectilinear units.


This programme takes the plate solutions of 2 different plate data sets of the same field and outputs the pixel data set for the 'slave' plate in the same xy reference frame as the pixel data for the 'master' plate data. This is a prerequisite for plate stacking, where the stacking methodology assumes that all plates are in the same xy system as the master.


This is the main program that submits the entire processing chain from the pixel data right up to the production of a final sorted output catalogue of objects (with their Right Ascension and Declinations provided). This is normally automatically performed by SuperCOSMOS, but this code allows for the chain to be instigated manually. It is useful, for example, if the operator wishes to reprocess the data with a different threshold for object detection.
Invocation:- runbatch {pathname}{directory name} 1.xx &
where the xx refers to the new % threshold required divided by 100.


This programme runs the suite of SuperCOSMOS Quality Control software (outputting dotplot, histogram plots, etc). The postscript files are automatically printed on the SuperCOSMOS laser printer.


This programme submits the automatic procedure for converting the object list from xy position on the plate into celestial coordinates, and creates a new output file with the values incorporated. This is done by cross-referencing the brighter stars in the SuperCOSMOS database with a bright star reference catalogue (currently the PPM is used). The output is in the FK5 system with J2000 equinox as standard.


Creates a Look-Up-Table for conversion of SuperCOSMOS transmission into Relative Intensity by means of a straight line in the Baker Density versus Log Intensity relationship. The operator specifies the input slope to be used (0.3 gives good results) together with the Transmission of chemical fog on the plate (24,000 typical) and the ambient 'black' (i.e. no light) transmission of the SuperCOSMOS system (typically 10).