The columns and rows that include the various measurements for the various labels of the input in its HDUs were described separately in MakeCatalog output HDUs. But those raw numbers are not the only thing that MakeCatalog writes in its output! MakeCatalog will also write metadata (header keywords) in the 0th (first) HDU of the output FITS file which add a lot of value and help to interpret the raw numbers. The only keywords in the other HDUs are the column names, units and comments; generic metadata are written in the 0th HDU.
You can see the full list of keywords written by MakeCatalog in its output with Gnuastro’s Fits program, for example astfits out.fits -h0.
If you only want the value of certain keywords (in a script/pipeline for example), its --keyvalue option is pretty convenient see Keyword inspection and manipulation.
The keywords are grouped in the output based on context with a title above each group.
We’ll follow the same structure here, skipping the first three groups (that are generic to all Gnuastro’s programs, see Output FITS files).
Some of the values reported below will be repeated (are the same as the respective option keyword in the “Option values” group of keywords).
The “Option value” group contains the raw option names and values: the option (keyword) names are just written in full-caps according to the FITS standard.
This is done for the following reason:
Human readability is important for option names, so they tend to be long and force a HIERARCH string at the start of the line.
On the contrary, for automatic extraction and human readability, it helps a lot to have all keywords related to a certain metameasure start with the same characters and because the keyword description is written just after it, there is no problem if the name is cryptic.
INLAB and INLABHDU: main (object) label dataset.INCLU and INCLUHDU: clump label dataset.INVAL and INVALHDU: value dataset.INSTD and INSTDHDU: standard deviation dataset.INVAR and INVARHDU: variance dataset.INUPM and INUPMHDU: upper-limit mask dataset.The file name and HDU of all the possible inputs to MakeCatalog. Only the keywords that correspond to inputs which are actually used in a given run of MakeCatalog will be written in the output. This is based on the columns or metameasurements that you request, not if you gave them on the command-line (described in MakeCatalog inputs and basic settings), or the HDUs that exist in the main argument.
For the standard deviation or variance, in case a single number was given (instead of a dataset), that number will be written for INSTD or INVAR (without any INSTDHDU and INVALHDU).
Basic information about the pixel values and grid properties of the input:
PIXWIDTHThe width of one pixel on the sky (in units of arcseconds).
PIXAREAThe area of one pixel on the sky (in units of arcseconds squared) on the reference point. The difference in pixel area across the image will be negligible in most science images. In case you would like to check this for your input images, use the --pixelareaonwcs option of the Fits program (see Pixel information images).
ZEROPNTThe zero point of the values image (used to convert This is the same value you gave to the --zeropoint option of MakeCatalog inputs and basic settings.
STDUSEDPer-pixel standard deviation (used in noise-based metameasurements like Surface brightness limit of image and Noise based magnitude limit of image). This keyword will only be present when a standard deviation image has been loaded (done automatically for any column measurement that involves noise, for example, --sn, --magnitude-error or --sky-std). In case your catalog does not include any such columns and you want this keyword, you can use the --meta-measures option (see Metameasurements on full input).
If the MEDSTD keyword is present in the standard deviation dataset (see NoiseChisel output), it will be used.
Otherwise, the median of the standard deviation input is calculated, used for the metameasures and written in this keyword.
When any of the upper-limit measurements are requested, the input parameters for the upper-limit measurement are stored in the following keywords (see Upper limit measurements).
UPSIGMAThe multiple of sigma to measure the upper-limit. This is the same value given to the --upnsigma option of Upper-limit settings.
UPNUMBERThe number of random positions with a successful reading. This is the same value given to the --upnum option of Upper-limit settings.
UPRNGNAMName of the random number generator used for finding the random positions; see Generating random numbers.
UPRNGSEESeed used for the random number generator. This will be different on every run, unless --envseed is called. For more details, see Generating random numbers
UPSCMLTP\(\sigma\)-clipping parameter: multiple of sigma. Clipping is necessary to reject strong outliers that can affect the statistics. This is the first value given to the --upsigmaclip option of Upper-limit settings.
UPSCTOL\(\sigma\)-clipping parameter: tolerance level. Clipping is necessary to reject strong outliers that can affect the statistics. This is the second value given to the --upsigmaclip option of Upper-limit settings.
The following metameasurement are calculated purely based on the measured noise level. But they are not written by default, if you want them run with --meta-measure.
SBLMeasured surface brightness limit (in units of mag/arcsec\(^2\)); as described in Surface brightness limit of image.
NMLMeasured surface brightness limit (in units of mag/arcsec\(^2\)); as described in Noise based magnitude limit of image.
The confusion limit (CNL) is a measure of density of resolved sources in the input image. For a complete review on its goals and how to interpret the values to these keywords, see Confusion limit of image. In particular, if you would like to see the full distribution of nearest neighbors and their distances, you can use --cnl-check as described in MakeCatalog output HDUs. By default, the clumps catalog is used for the distribution, however if you would like to use the objects for any reason, you can use --cnl-with-objects as described below in this section (not recommended unless you understand the risks).
CNLThe confusion limit: difference between CNLP75 and CNLP25.
CNLP05CNLP25CNLP50CNLP75CNLP95Various percentiles of the distribution of distances to the nearest neighbor in units of pixels.
The P before each number is for percentile, so CNLP50 shows the median value of the distribution to nearest neighbors.
Note that a quantile is just the percentile after division by 100.
Percentiles are used in the keyword names because they are simple integers and do not need a floating point.
The following MakeCatalog options are specifically related to the various keywords above.
Value to multiply with the median standard deviation (from a MEDSTD keyword, if it is present in the Sky standard deviation image) for the measured surface brightness limit.
Note that the surface brightness limit is only reported when metameasurements are requested with --meta-measures.
See Metameasurements on full input for more on the basics of various metameasurements and Image surface brightness limit on a practical usage example.
Shape-agnostic area (in arc-seconds squared) to use for the measured surface brightness limit. Note that the surface brightness limit is only reported when meta-measurements are requested with --meta-measures. See Metameasurements on full input for more on the basics of various metameasurements and Image surface brightness limit on a practical usage example.
Use the object positions instead of clumps for measuring the distance to the nearest label. This is only useful if you have generated your labels image with something other than Gnuastro’s Segment, and that program doesn’t have the capacity to identify individual peaks and extended signal at the same time. Therefore, in case you have generated the input labels with Segment, we do not recommend using this option.
JavaScript license information
GNU Astronomy Utilities 0.24 manual, November 2025.