1 CLEAN CLEAN This task contains implementations of several CLEAN-like deconvolution algorithm. The algorithm is selected through the METHOD$ keyword, which can be : SIMPLE : The straightforward direct method. CLARK : The Barry Clark major-minor cycle method. MRC : a Multi-Resolution Clean. THRESHOLD : Components are not restricted to be point sources. The input map and beam should have the same sizes, except for the SIMPLE method where different beam sizes are allowed, and must have the same grid spacing. Sources are searched only in the inner quarter of the map, but their sidelobes are removed in the full map. A smaller search box can be specified if desired. The program can process all planes of a data cube at the same time, but a separate list of component will be produced for each plane. The cleaning cannot be restarted. 2 METHOD$ TASK\CHARACTER "Method for Clean" METHOD$ Specify the algorithm to use among : SIMPLE : The straightforward direct method. Safe, but slow. CLARK : The Barry Clark major-minor cycle method. Safe and fast. MRC : a Multi-Resolution Clean. Recommended for extended sources or large images. THRESHOLD : Components are not restricted to be point sources. NOT RECOMMENDED, and NOT DEBUGGED. 2 DIRTY$ TASK\FILE "Input dirty map" DIRTY$ This is the name of the input "dirty map". If the map has more than two dimensions, all planes will be cleaned successively. 2 BEAM$ TASK\FILE "Input dirty beam" BEAM$ This is the name of the input "dirty beam" (Point Spread Function in optical jargon). The beam can be a cube. 2 RESIDUAL$ TASK\FILE "Output residual map" RESIDUAL$ This is the name of the output residual map. The residual map is not deleted at the end of the run. 2 CLEAN$ TASK\FILE "Output clean map" CLEAN$ This is the name of the output "clean map". 2 COMPONENT$ TASK\FILE "Output clean component table" COMPONENT$ The name of the table to received Clean Component coordinates and values. Default extension is .CCT 2 GAIN$ TASK\REAL "Loop gain" GAIN$ This is the gain of the subtraction loop. It should typically be chosen in the range 0.05 and 0.3. Higher values give faster convergence, while lower values give a better restitution of the extended structure. 2 NITER$ TASK\INTEGER "Maximum number of clean components" NITER$ This is the maximum number of components the program will accept to subtract. Once it has been reached, the program starts the restoration phase. 2 FRES$ TASK\REAL "Maximum value of residual (Fraction of peak)" FRES$ This is the minimal fraction of the peak flux in the dirty map that the program will consider as significant. Alternatively, an absolute threshold can be specified using ARES$. Once this level has been reached the program stops subtracting, and starts the restoration phase. This parameter is normalised to 1 (neither in % nor in db). It should usually be of the order of magnitude of the inverse of the expected dynamic range. 2 ARES$ TASK\REAL "Maximum value of residual (Absolute)" ARES$ This is the minimal flux in the dirty map that the program will consider as significant. Alternatively, the threshold can be specified as a fraction of the peak flux using FRES$. Once this level has been reached the program stops subtracting, and starts the restoration phase. The unit for this parameter is the map unit. The parameter should usually be of the order of magnitude of the expected noise in the clean map. 2 BLC$ TASK\INTEGER "Bottom left corner of cleaning box" BLC$[4] These are the (pixel) coordinates of the Bottom Left Corner of the cleaning box. Only the first two coordinates are actually used. The actual cleaning window will be the intersection of the specified window with the inner quarter of the map. 2 TRC$ TASK\INTEGER "Top right corner of cleaning box" TRC$[4] These are the (pixel) coordinates of the Top Right Corner of the cleaning box. Only the first two coordinates are actually used. The actual cleaning window will be the intersection of the specified window with the inner quarter of the map. 2 POSITIVE$ TASK\INTEGER "Minimum number of positive components" POSITIVE$ The minimum number of positive components before negative ones are selected. 2 KEEP$ TASK\LOGICAL "Keep cleaning after convergence" KEEP$ This is a logical flag to keep cleaning after an approximate convergence has been reached. It should usually be set to .TRUE., except may be for the SIMPLE method. 2 MAJOR$ TASK\REAL "Clean beam major axis" MAJOR$ This is the major axis (FWHP) in user coordinates of the gaussian restoring beam. If 0, the program will fit a gaussian to the dirty beam. 2 MINOR$ TASK\REAL "Clean beam minor axis" MINOR$ This is the minor axis (FWHP) in user coordinates of the gaussian restoring beam. 2 PA$ TASK\REAL "Position angle of clean beam" PA$ This is the position angle (from North towards East, i.e. anticlockwise) of the major axis of the gaussian restoring beam (in degrees). 2 BEAM_PATCH$ TASK\INTEGER "Size of cleaning beam in pixels" BEAMPATCH$[2] The dirty beam patch to be used for the minor cycles in CLARK and MRC method. It should be large enough to avoid doing too many major cycles, but has practically no influence on the result. This size should be specified in pixel units. Reasonable values are between N/8 and N/4, where N is the number of map pixels in the same dimension. If set to N, the CLARK algorithm becomes identical to the SIMPLE algorithm. 2 CLARK A Major-Minor cycles method in which clean components are selected using a limited beam patch, and deconvolved through Fourier transform at each major cycle. Usually faster than the Simple Hogbom method. 2 MRC A Multi Resolution Clean. Two maps are cleaned instead of only one: a smooth dirty map, and the difference between this smooth map and the original map. The same process is applied to the dirty and clean beams. The final clean map is obtained adding the two clean and two residual maps with proper weighting. Since the difference map contains no flux, a limited Clean can be performed on it. The smooth map contains only larger structure, and is compressed before cleaning. Deeper clean can be performed on extended structures, because they look more point-like and have better signal to noise in the smoothed map. Hence the MRC algorithm is able to recover more extended structure than a standard Clean. 1 END