|
|
|
| Parameter | Type | Values | Default | Description |
|
|
|
|
|--------------|-----------|-----------|------------|----------------|
|
|
|
|
| flag.region | list | | | This parameter can be used to define regions to be flagged in the input data cube prior to processing. The flagging region must contain a multiple of six comma-separated integer values of the following format: x_min, x_max, y_min, y_max, z_min, z_max, ... (all in units of pixels). Pixels within those regions will be set to blank in the input cube. If unset, no flagging will occur. |
|
|
|
|
| input.data | string | | | Name of the input data cube on which to run the source finder. The absolute path to the data file must be provided. If only the file name is specified, the pipeline will assume the file to be located in the current working directory. Currently, only the FITS format is supported. |
|
|
|
|
| input.mask | string | | | File name of an input mask cube. Any additional pixels detected by the source finder will be added to the input mask. This can be useful if the results from two different source finding runs should be combined into a single mask. The mask cube must have the same dimensions as the input data cube. The absolute path to the mask file must be provided. If only the file name is specified, the pipeline will assume the file to be located in the current working directory. Currently, only the FITS format is supported. |
|
|
|
|
| input.region | list | | | This parameter can be used to specify a region of the input data cube to be searched. Only the specified region will be loaded into memory and processed. A region must contain six comma-separated integer values of the following format: x_min, x_max, y_min, y_max, z_min, z_max (all in units of pixels). If no region is specified, then the entire data cube will be loaded. |
|
|
|
|
| input.weights | string | | | Name of a data cube containing weights. The input data cube will be divided by the weights cube prior to processing. The weights cube must have the same dimensions as the input data cube. The absolute path to the weights file must be provided. If only the file name is specified, the pipeline will assume the file to be located in the current working directory. Currently, only the FITS format is supported. |
|
|
|
|
| linker.maxSizeX | int | ≥ 0 | 0 | Maximum size of sources in the x-direction in pixels. Sources that exceed this limit will be discarded by the linker. If the value is set to 0, maximum size filtering will be disabled. |
|
|
|
|
| linker.maxSizeY | int | ≥ 0 | 0 | Maximum size of sources in the y-direction in pixels. Sources that exceed this limit will be discarded by the linker. If the value is set to 0, maximum size filtering will be disabled. |
|
|
|
|
| linker.maxSizeZ | int | ≥ 0 | 0 | Maximum size of sources in the z-direction in pixels. Sources that exceed this limit will be discarded by the linker. If the value is set to 0, maximum size filtering will be disabled. |
|
|
|
|
| linker.minSizeX | int | ≥ 0 | 5 | Minimum size of sources in the x-direction in pixels. Sources that fall below this limit will be discarded by the linker. |
|
|
|
|
| linker.minSizeY | int | ≥ 0 | 5 | Minimum size of sources in the y-direction in pixels. Sources that fall below this limit will be discarded by the linker. |
|
|
|
|
| linker.minSizeZ | int | ≥ 0 | 5 | Minimum size of sources in the z-direction in pixels. Sources that fall below this limit will be discarded by the linker. |
|
|
|
|
| linker.radiusX | int | ≥ 0 | 1 | Maximum merging length in x-direction. Pixels with a separation of up to this value will be merged into the same source. |
|
|
|
|
| linker.radiusY | int | ≥ 0 | 1 | Maximum merging length in y-direction. Pixels with a separation of up to this value will be merged into the same source. |
|
|
|
|
| linker.radiusZ | int | ≥ 0 | 1 | Maximum merging length in z-direction. Pixels with a separation of up to this value will be merged into the same source. |
|
|
|
|
| output.directory | string | | | Full path to the directory to which all output files will be written. If unset, the directory of the input data cube will be used by default. |
|
|
|
|
| output.filename | string | | | File name that will be used as the template for all output files. For example, if `output.filename = my_data`, then the output files will be named `my_data_cat.xml`, `my_data_mom0.fits`, etc. If unset, the name of the input data cube will be used as the file name template by default. |
|
|
|
|
| output.writeCatASCII | bool | true, false | true | If set to `true`, an output source catalogue will be produced in human-readable ASCII format. The catalogue file will have the suffix `_cat.txt`. |
|
|
|
|
| output.writeCatXML | bool | true, false | true | If set to `true`, an output source catalogue will be produced in VO-compatible XML format. The catalogue file will have the suffix `_cat.xml`. |
|
|
|
|
| output.writeCubelets | bool | true, false | false | If set to `true`, then individual source products for each detected source will be created, including sub-cubes, masks, moment maps and integrated spectra. The source products will be written to a sub-directory with the suffix `_cubelets`. Each source product will be labelled with the source ID number for identification. |
|
|
|
|
| output.writeFiltered | bool | true, false | false | If set to `true` and any input filtering algorithm is enabled, then a data cube containing the filtered data will be written in FITS format. The filtered cube will have the suffix `_filtered.fits`. |
|
|
|
|
| output.writeMask | bool | true, false | false | If set to `true`, then a data cube containing the source mask produced by the source finder will be written in FITS format. The mask cube will have the suffix `_mask.fits`. |
|
|
|
|
| output.writeMoments | bool | true, false | false | If set to `true`, then images of the spectral moments 0, 1 and 2 will be written in FITS format. The moment maps will have the suffix `_mom0.fits`, `_mom1.fits` and `_mom2.fits`. Note that moments 1 and 2 will not be produced if the input data cube is only two-dimensional. |
|
|
|
|
| output.writeNoise | bool | true, false | false | If set to `true` and local noise scaling is enabled, then a data cube containing the measured local noise values will be written in FITS format. The noise cube will have the suffix `_noise.fits`. |
|
|
|
|
| parameter.enable | bool | true, false | true | If set to true, the parametrisation module will be enabled to measure the basic parameters of each detected source. |
|
|
|
|
| pipeline.pedantic | bool | true, false | true | If set to `true`, the pipeline will terminate with an error message if an unknown parameter name is encountered in the input parameter file. If set to `false`, unknown parameters will instead be ignored. |
|
|
|
|
| pipeline.verbose | bool | true, false | false | Determines the level of output messages produced by the pipeline. Additional warning messages can be enabled by setting the value to `true`. |
|
|
|
|
| reliability.enable | bool | true, false | false | If set to `true`, reliability calculation and filtering will be enabled. This will determine the reliability of each detection with positive total flux by comparing the density of positive and negative detections in a three-dimensional parameter space. Sources below the specified reliability threshold will then be discarded. Note that this will require a sufficient number of negative detections, which can usually be achieved by setting the source finding threshold to somewhere around 3 to 4 times the noise level. |
|
|
|
|
| reliability.fmin | float | ≥ 0 | 20.0 | Threshold of the summed flux, F_sum, divided by the square root of the number of pixels, N, of a source. Sources that fall below this threshold will be deemed unreliable and assigned a reliability of 0. The value is a proxy for the integrated signal-to-noise ratio, SNR = F_sum / sqrt(N * Omega), of a source, where Omega is the solid angle (in pixels) of the point spread function of the data. |
|
|
|
|
| reliability.plot | bool | true, false | true | If set to `true`, a diagnostic plot (in EPS format) will be created to allow the quality of the reliability estimation to be assessed. It is advisable to generate and inspect this plot to ensure that the outcome of the reliability filtering procedure is satisfactory. |
|
|
|
|
| reliability.scaleKernel | float | | 0.5 | When estimating the density of positive and negative detections in parameter space, the size of the Gaussian kernel used in this process is determined from the covariance matrix of the distribution of negative detections in parameter space. This parameter setting can be used to scale that kernel by a constant factor. A value of 1 will use the original covariance matrix scale without additional scaling. A value of around 0.5 has proven to give acceptable results in many cases. |
|
|
|
|
| reliability.threshold | float | 0...1 | 0.9 | Reliability threshold in the range of 0 to 1. Sources with a reliability below this threshold will be discarded. |
|
|
|
|
| scaleNoise.enable | bool | true, false | false | If set to `true`, noise scaling will be enabled. The purpose of the noise scaling modules is to measure the noise level in the input cube and then divide the input cube by the noise. This can be used to correct for spatial or spectral noise variations across the input cube prior to running the source finder. |
|
|
|
|
| scaleNoise.fluxRange | string | positive, negative, full | negative | Flux range to be used in the noise measurement. If set to `negative` or `positive`, only pixels with negative or positive flux will be used, respectively. This can be useful to prevent real emission or artefacts from affecting the noise measurement. If set to `full`, all pixels will be used in the noise measurement irrespective of their flux. |
|
|
|
|
| scaleNoise.gridSpatial | int | ≥ 0 | 0 | Size of the spatial grid across which noise measurement window will be moved across the data cube. It must be an odd integer value. If set to 0 instead, the spatial grid size will default to half the spatial window size. |
|
|
|
|
| scaleNoise.gridSpectral | int | ≥ 0 | 0 | Size of the spectral grid across which noise measurement window will be moved across the data cube. It must be an odd integer value. If set to 0 instead, the spectral grid size will default to half the spectral window size. |
|
|
|
|
| scaleNoise.interpolate | bool | true, false | false | If set to `true`, linear interpolation will be used to interpolate the measured local noise values in between grid points. If set to `false`, the entire grid cell will instead be filled with the measured noise value. |
|
|
|
|
| scaleNoise.mode | string | spectral, local | spectral | This parameter sets the noise scaling mode. If set to `spectral`, the noise level will be determined for each spectral channel by measuring the noise within each image plane. This is useful for data cubes where the noise varies with frequency. If set to `local`, the noise level will be measured locally in window running across the entire cube in all three dimensions. This is useful for data cubes with more complex noise variations, such as interferometric images with primary-beam correction applied. |
|
|
|
|
| scaleNoise.statistic | string | std, mad, gauss | mad | This parameter sets the statistic to be used in the noise measurement process. Possible values are `std`, `mad` and `gauss` for standard deviation, median absolute deviation and Gaussian fitting to the flux histogram, respectively. Standard deviation is by far the fastest algorithm, but it is also the least robust one with respect to emission and artefacts in the data. Median absolute deviation and Gaussian fitting are far more robust in the presence of strong, extended emission or artefacts, but will usually take longer. |
|
|
|
|
| scaleNoise.windowSpatial | int | ≥ 0 | 25 | Spatial size of the window used in determining the local noise level. It must be an odd integer value. If set to 0, the pipeline will use the default value instead. |
|
|
|
|
| scaleNoise.windowSpectral | int | ≥ 0 | 15 | Spectral size of the window used in determining the local noise level. It must be an odd integer value. If set to 0, the pipeline will use the default value instead. |
|
|
|
|
| scfind.enable | bool | true, false | true | If set to `true`, the Smooth + Clip (S+C) finder will be enabled. The S+C finder operates by iteratively smoothing the data cube with a user-defined set of smoothing kernels, measuring the noise level on each smoothing scale, and adding all pixels with an absolute flux above a user-defined relative threshold to the source detection mask. |
|
|
|
|
| scfind.fluxRange | string | positive, negative, full | negative | Flux range to be used in the noise measurement. If set to `negative` or `positive`, only pixels with negative or positive flux will be used, respectively. This can be useful to prevent real emission or artefacts from affecting the noise measurement. If set to `full`, all pixels will be used in the noise measurement irrespective of their flux. |
|
|
|
|
| scfind.kernelsXY | list | ≥ 0 | 0, 3, 6 | Comma-separated list of spatial Gaussian kernel sizes to apply. The individual kernel sizes must be floating-point values and denote the standard deviation of the Gaussian used to smooth the data in the spatial domain. A value of 0 means that no spatial smoothing will be applied. |
|
|
|
|
| scfind.kernelsZ | list | ≥ 0 | 0, 3, 7, 15 | Comma-separated list of spectral Boxcar kernel sizes to apply. The individual kernel sizes must be integer values and denote the full width of the Boxcar filter used to smooth the data in the spectral domain. A value of 0 means that no spectral smoothing will be applied. |
|
|
|
|
| scfind.replacement | float | ≥ 0.0 | 2.0 | Before smoothing the data cube during an S+C iteration, every pixel in the data cube that was already detected in a previous iteration will be replaced by this value multiplied by the original noise level in the non-smoothed data cube. This is to ensure that the smoothing operation does not smear out the flux of the source over too large a volume, which would otherwise result in a significant overestimation of the size of the source mask. Values of about 2 have proven to be most effective. |
|
|
|
|
| scfind.statistic | string | std, mad, gauss | mad | This parameter sets the statistic to be used in the noise measurement process. Possible values are `std`, `mad` and `gauss` for standard deviation, median absolute deviation and Gaussian fitting to the flux histogram, respectively. Standard deviation is by far the fastest algorithm, but it is also the least robust one with respect to emission and artefacts in the data. Median absolute deviation and Gaussian fitting are far more robust in the presence of strong, extended emission or artefacts, but will usually take longer. |
|
|
|
|
| scfind.threshold | float | ≥ 0.0 | 5.0 | Flux threshold to be used by the S+C finder relative to the measured noise level in each smoothing iteration. In practice, values in the range of about 3 to 5 have proven to be useful in most situations, with lower values in that range requiring use of the reliability filter to reduce the number of false detections. |
|
|
|
|
| threshold.enable | bool | true, false | false | If set to true, the threshold finder will be enabled. The threshold finder is a very basic source finder that simply applies a fixed threshold (either absolute or relative to the noise) to the original data cube. It can be useful if a simple flux threshold is to be applied to a pre-processed or filtered data cube. |
|
|
|
|
| threshold.fluxRange | string | positive, negative, full | negative | Flux range to be used in the noise measurement. If set to `negative` or `positive`, only pixels with negative or positive flux will be used, respectively. This can be useful to prevent real emission or artefacts from affecting the noise measurement. If set to `full`, all pixels will be used in the noise measurement irrespective of their flux. |
|
|
|
|
| threshold.mode | string | absolute, relative | relative | If set to `absolute`, the flux threshold of the threshold finder will be interpreted as an absolute flux threshold in the native flux unit of the data cube. If set to `relative`, the threshold will be interpreted in units of the noise level across the data cube. |
|
|
|
|
| threshold.statistic | string | std, mad, gauss | mad | This parameter sets the statistic to be used in the noise measurement process if `threshold.mode` is set to `relative`. Possible values are `std`, `mad` and `gauss` for standard deviation, median absolute deviation and Gaussian fitting to the flux histogram, respectively. Standard deviation is by far the fastest algorithm, but it is also the least robust one with respect to emission and artefacts in the data. Median absolute deviation and Gaussian fitting are far more robust in the presence of strong, extended emission or artefacts, but will usually take longer. |
|
|
|
|
| threshold.threshold | float | ≥ 0.0 | 5.0 | Flux threshold to be applied by the threshold finder. Depending on the `threshold.mode` parameter, this can either be absolute (in native flux units of the data cube) or relative to the noise level of the cube. | |
