Determines the parameters of a model star profile by fitting star images in a
two-dimensional NDF PSF
The model has a Sérsic radial profile:
where is calculated from the true radial distance from the star centre allowing for image ellipticity, is the Gaussian precision constant or profile width. The application combines a number of star images you specify and determines a mean seeing-disc size, radial fall-off parameter (), axis ratio, and orientation of a model star image.
A table, giving details of the seeing and ellipticity of each star image used can be reported to an output text file. This table indicates if any star could not be used. Reasons for rejecting stars are too-many bad pixels present in the image, the star is too close to the edge of the data array, the ‘star’ is a poor fit to model or it could not be located.
An optional plot of the mean profile and the fitted function may be produced. The two-dimensional point-spread function may be stored in an NDF for later use, as may the one-dimensional fitted profile.
TRUE
if labelled and
annotated axes are to be drawn around the plot. The width of the margins left for the
annotation may be controlled using Parameter MARGIN. The appearance of the axes
(colours, founts, etc.) can be controlled using the Parameter STYLE. The dynamic
default is TRUE
if CLEAR is TRUE
, and FALSE
otherwise. []
TRUE
the current picture is cleared before the plot is drawn. If CLEAR is FALSE
not
only is the existing plot retained, but also an attempt is made to align the new
picture with the existing picture. Thus you can generate a composite plot within a
single set of axes, say using different colours or modes to distinguish data from
different datasets. [TRUE]
!
) value. Each line should contain the formatted axis values for a
single position, in the current Frame of the NDF. Columns can be separated
by spaces, tabs or commas. The file may contain comment lines with the first
character # or !
. Other columns may be included in the file, in which case the
columns holding the required co-ordinates should be specified using Parameter
POSCOLS. !
, the dimension of the square NDF will be the size of the
region used to calculate the radial profile, which usually is given by RANGE
width
in pixels
AXISR, unless truncated. If CUT has a value it is the threshold which must be included
in the PSF NDF, and it is given as the fraction of the peak amplitude of the PSF. For
example, if CUT=0.5
the NDF would contain the point-spread function to half maximum. CUT
must be greater than 0 and less than 1. The suggested default is 0.0001. [!]
!
) name indicates that no
plot is required. [
current graphics device]
TRUE
, the
coefficient
is fixed to be 2; in other words the best-fitting two-dimensional Gaussian is evaluated.
If FALSE
,
is a free parameter of the fit, and the derived value is returned in Parameter GAMMA.
[FALSE]
!
) value is
supplied Parameter COFILE will be used to get the star positions from a simple text
file. [15]
!
) name indicates that no
log file is required. [!]
!
) value is supplied, the value used is
0.15
(for all edges) if either annotated axes or a key are produced, and zero
otherwise. [
current value]
[
current value]
TRUE
the horizontal axis of the plot is annotated with
distance along the minor axis from the centre of the PSF. If MINOR is FALSE
,
the distance along the major axis is used. [TRUE]
TRUE
, the model PSF is normalized so that it has a peak value of unity.
Otherwise, its peak value is equal to the peak value of the fit to the first
usable star, in the data units of the input NDF. [TRUE]
!
, is entered no output NDF will be created. The dimensions of the
array are controlled by Parameter CUT. The pixel origin is chosen to align the
model PSF with the first fitted star in pixel co-ordinates, thus allowing the
NDF holding the model PSF to be compared directly with the input NDF. A WCS
component is stored in the output NDF holding a copy of the input WCS component. An
additional Frame with Domain name OFFSET is added, and is made the current Frame.
This Frame measures the distance from the PSF centre in the units in which
the FWHM is reported. These changes allows the NDF holding the model PSF to
be compared directly with the input NDF. [!]
!
) value. If a null (!
) value is supplied for POSCOLS, the
values [1,2] will be used. [!]
!
, is entered no
output NDF will be created. The DATA component of this NDF holds the fitted
PSF value at each radial bin. The VARIANCE component holds the square of the
residuals between the fitted values and the binned values derived from the input
NDF. An AXIS component is included in the NDF containing the radial distance
as displayed in the plot. [!]
[4.0]
A comma-separated list of strings should be given in which each string is either an
attribute setting, or the name of a text file preceded by an up-arrow character
""
.
Such text files should contain further comma-separated lists which will be read and
interpreted in the same manner. Attribute settings are applied in the order in which
they occur within the list, with later settings overriding any earlier settings given
for the same attribute.
Each individual attribute setting should be of the form:
name=value
where name
is the name of a plotting attribute, and
value
is the value to assign to the attribute. Default values will be used for any
unspecified attributes. All attributes will be defaulted if a null value (!
)–-the
initial default–-is supplied. To apply changes of style to only the current invocation,
begin these attributes with a plus sign. A mixture of persistent and temporary style
changes is achieved by listing all the persistent attributes followed by a plus sign
then the list of temporary attributes.
See Section E for a description of the available attributes. Any unrecognised
attributes are ignored (no error is reported). The appearance of the model curve is
controlled by the attributes Colour(Curves), Width(Curves), etc. (the synonym Line may
be used in place of Curves). The appearance of the markers representing the
real data is controlled by Colour(Markers), Width(Markers), etc. (the synonym
Symbols may be used in place of Markers). [
current value]
!
, is entered the NDF will not contain a title. ["KAPPA - PSF"]
Its integer index within the current Frame of the input NDF (in the range 1 to the number of axes in the current Frame).
Its Symbol string such as "RA"
or "VRAD"
.
A generic option where "SPEC"
requests the spectral axis, "TIME"
selects the time axis,
"SKYLON"
and "SKYLAT"
picks the sky longitude and latitude axes respectively. Only
those axis domains present are available as options.
A list of acceptable values is displayed if an illegal value is supplied. If a null (!
)
value is supplied, the axes with the same indices as the two significant NDF pixel axes
are used. [!]
!
) value is supplied,
a suitable default value will be found and used. The value supplied may be
greater than or less than the value supplied for XRIGHT. [!]
!
) value is supplied, a suitable default value will be
found and used. The value supplied may be greater than or less than the value
supplied for XLEFT. [!]
!
) value is supplied,
a suitable default value will be found and used. The value supplied may be
greater than or less than the value supplied for YTOP. [!]
!
) value is supplied, a suitable default value will be found and
used. The value supplied may be greater than or less than the value supplied
for YBOT. [!]
"X"
) towards the
second current Frame axis ("Y"
). FALSE
. When NORM=TRUE
, TOTAL is just
measured in the squared radial unit. Therefore, for direct comparison of total
flux, the same units must be used. starlist.FIT
. A
plot of the profile is drawn on the current graphics device. starlist.dat
) instead of a
positions list. fit.log
. The full-width half-maximum is written to the ICL
variable SEEING rather than the parameter file. "Point spread function"
. Values for the FWHM seeing are given in arcseconds if the Current co-ordinate Frame of the NDF is a SKY Frame.
The stars used to determine the mean image parameters should be chosen to represent those whose magnitudes are to be found using a stellar photometry application, and to be sufficiently bright, uncrowded, and noise-free to allow an accurate fit to be made.
It is assumed that the image scale does not vary significantly across the image.
The method to calculate the fit is as follows.
Marginal profiles of each star image are formed in four directions: at 0, 45, 90 and 135 degrees to the axis. The profiles are cleaned via an iterative modal filter that removes contamination such as neighbouring stars; moving from the centre of the star, the filter prevents each data point from exceeding the maximum of the two previous data values.
A Gaussian curve and background is fitted to each profile iteratively refining the parameters until parameters differ by less than 0.1 per cent from the previous iteration. If convergence is not met after fifteen iterations, each fit parameter is approximately the average of its last pair of values. The initial background is the lower quartile.
Using the resulting four Gaussian centres, a mean centre is found for each star. Iterations cease when the mean centroid position shifts by less 0.001 from the previous iteration, or after three iterations if the nominal tolerance is not achieved.
The four Gaussian widths of all the stars are combined modally, using an amplitude-weighted average with rejection of erroneous data (using a maximum-likelihood function for a statistical model in which any of the centres has a constant probability of being corrupt). From the average widths along the four profiles, the seeing-disc size, axis ratio and axis inclination are calculated.
The data surrounding each star is then binned into isophotal zones which are elliptical annuli centred on the star–-the ellipse parameters being those just calculated. The data in each zone is processed to remove erroneous points (using the aforementioned maximum-likelihood function) and to find an average value. A Gaussian profile is fitted to these average values and the derived amplitude is used to normalise the values to an amplitude of unity. The normalised values are put into bins together with the corresponding data from all other stars and these binned data represent a weighted average radial profile for the set of stars, with the image ellipticity removed. Finally a radial profile is fitted to these data, giving the radial profile parameter gamma and a final re-estimate of the seeing-disc size.
If a plot was requested the application stores two pictures in the graphics database in the following order: a FRAME of the specified size containing the title, annotated axes, and line plot; and a DATA picture, containing just the data plot. Note, the FRAME picture is only created if annotated axes have been drawn, or if non-zero margins were specified using Parameter MARGIN. The NDF associated with the plot is not stored by reference with the DATA picture. On exit the current database picture for the chosen device reverts to the input picture.
This routine correctly processes the AXIS, DATA, QUALITY, LABEL, WCS, and TITLE components of an NDF data structure.
Processing of bad pixels and automatic quality masking are supported.
All non-complex numeric data types can be handled. The output point-spread-function NDF has the same type as the input NDF.