daopars
function = gauss
The functional form of the analytic component of the PSF model computed
by the DAOPHOT task PSF. The better this function
matches the true PSF, especially in the cores of the stars,
the smaller the interpolation errors will be. The choices are the following.
gauss
An elliptical Gaussian function aligned along the x and y axes of the
input image.
moffat15
An elliptical Moffat function with a beta parameter of 1.5.
moffat25
An elliptical Moffat function with a beta parameter of 2.5.
lorentz
An elliptical Lorentzian function with beta parameter of 1.0.
penny1
A Gaussian core with Lorentzian wings model, where the Gaussian core may be
tilted, but the Lorentzian wings are elongated along the x or y axes.
The Lorentzian wings have a beta parameter of 1.0
penny2
A Gaussian core with Lorentzian wings model, where the Gaussian core and
Lorentizian wings may be tilted in different directions.
The Lorentzian wings have a beta parameter of 1.0
auto
The PSF task computes the analytic PSF model using each of the previous
six analytic model PSFs in turn and selects the one that gives the smallest
standard deviation for the fit.
func1,func2,...
The PSF task computes the analytic PSF model using any subset
of the six previously defined functions in turn, and selects the one
that gives the smallest standard deviations for the fit, e.g.
"gauss,moffat25,penny1".
In general "gauss" is the best and most efficient choice for a well-sampled ground-based image, "lorentz" is best for ST images, and "moffat15" or "moffat25" is best for under-sampled ground-based images.
varorder = 0
The order of variability of the PSF model computed by the DAOPHOT PSF task.
Varorder
determines the number of look-up tables containing the deviations of the
true PSF from the analytic model PSF defined by function.
-1
Only the analytic function specified by function is used to compute
the model PSF. The model PSF is constant over the image.
0
The analytic function and one look-up table are used to compute the
model PSF. The model PSF is constant over the image.
1
The analytic function and three look-up tables are used to compute the
model PSF. The model PSF is linearly variable over the image,
with terms proportional to 1, x and y.
2
The analytic function and six look-up tables are used to compute the
model PSF. The model PSF is quadratically variable over the image,
with terms proportional to 1, x, y, x**2, xy, y**2.
nclean = 0
The number of additional iterations used in the PSF task to compute the PSF
look-up tables.
If nclean is > 0, stars which contribute deviant residuals to
the PSF look-up tables in the first iteration, will have these residuals
down-weighted in succeeding iterations.
saturated = no
Use saturated stars to improve the signal-to-noise in the wings of the
model PSF computed by the PSF task. This parameter should only be set to
"yes" where there are too few high signal-to-noise unsaturated stars
in the image to compute a reasonable model for the stellar profile wings.
matchrad = 3.0 (scale units)
The tolerance in scale units for matching the stellar x and y centroids in the
photometry file with the image cursor position. Matchrad is currently
used by the PSF task to match stars shown on the image
display with stars in the photometry list.
psfrad = 11.0 (scale units)
The radius of the circle in scale units within which the PSF is defined.
Psfrad should be a pixel or two larger than the radius at which the
intensity of the brightest star of interest fades into the noise.
Psfrad can never be larger than the value with which the
PSF was computed but may be smaller, in which case the value stored in
the PSF image header is overridden.
fitrad = 3.0 (scale units)
The fitting radius in scale units. Only pixels within the fitting radius of
the center of a star will
contribute to the fits computed by the PEAK, NSTAR and ALLSTAR tasks.
For most images the fitting radius
should be approximately equal to the FWHM of the PSF. Under severely
crowded conditions or if the sky is being fit in the least-squares sense
a somewhat smaller value may be chosen in order to improve
the fit. If the PSF is variable, the FWHM is very small, or sky fitting
is enabled in PEAK and NSTAR on the other hand,
it may be necessary to increase the fitting radius to achieve a good fit.
recenter = yes (peak, nstar, and allstar)
Compute new positions as well as magnitudes for all the stars in the
input photometry list?
fitsky = no (peak, nstar, and allstar)
Compute new sky values for the stars in the input list (peak, nstar, allstar).
If fitsky = "no", the PEAK, NSTAR,
and ALLSTAR tasks compute a group sky value by combining
the sky values of the stars in the group.
If fitsky = "yes", PEAK and NSTAR fit the group sky simultaneously
with the positions and magnitudes.
In the ALLSTAR task new sky values for each star
are computed every third iteration by subtracting off the best current fit
for the star and recomputing the sky using pixels in the region
defined by sannulus and wsannulus. The new group sky value is
the average of these new individual values.
groupsky = yes (nstar and allstar)
If groupsky is "yes", then the sky value for every pixel which
contributes to the fit is identical and equal to the mean of the sky
values of all the stars in the group.
If groupsky is "no", then the sky value for every pixel which
contributes to the fit is equal to the mean of the sky
values of all the stars in the group for which that pixel is within
one fitting radius.
sannulus = 0.0 (scale units, allstar)
The inner radius of the sky annulus used by ALLSTAR to recompute the sky
value.
wsannulus = 11 (scale units, allstar)
The width of the sky annulus used by ALLSTAR to recompute the sky
value.
flaterr=0.75 (percent, peak, nstar, allstar)
The image flat-fielding error in percent used to compute the predicted
errors of the fit.
proferr = 5.0 (percent, peak, nstar, allstar)
The profile or interpolation fitting error in percent used to compute
the predicted errors of the fit.
maxiter = 50 (peak, nstar, allstar)
The maximum number of times that the PSF fitting tasks PEAK, NSTAR
and ALLSTAR will iterate on the fit before giving up.
cliprange = 2.5, clipexp = 6.0 (peak, nstar, allstar)
The parameters of the down-weighting scheme in the fitting code
used to resist bad data by dynamically
reweighting bad pixels. For clipexp greater than 1 a
residual small compared to cliprange standard deviations
does not have its weight materially altered, one with exactly cliprange
standard deviations is given precisely half its normal weight, and
for large residuals the weight falls off as the standard deviation
to the minus clipexp power. For normal applications it is
recommended that the user leave these numbers alone.
critoverlap = 1.0 (group)
The ratio of the model intensity of a brighter star computed from
the scaled PSF at a distance of one fitting radius from the center
of the fainter star to the expected random error computed from the
readout noise, gain and value of the PSF. The critical overlap
parameter is used to group stars. In general if a small value such
as .1 divides all the stars in an image into groups less than
maxgroup, then the expected random errors will determine
the accuracy of the photometry. On the other hand if a value of
crtitcal overlap much greater than one is required to divide up the
stars crowding errors will dominate random errors. If a value of 1
is sufficient then crowding and random errors are roughly equivalent.
maxnstar = 10000 (psf, allstar, substar)
The maximum number of stars which PSF, PEAK, ALLSTAR and SUBSTAR
will process.
This should be set to a value somewhat larger than the
number of stars in the input photometry file.
maxgroup = 60 (nstar, allstar)
The maximum numbers of stars that the multiple star fitting tasks
NSTAR and ALLSTAR will fit simultaneously. NSTAR will simply refuse
to fit any groups large than maxgroup and set all the
output magnitudes of these stars to INDEF. ALLSTAR dynamically regroups
the stars
until the group is either maxgroup or smaller in size or becomes too
dense to
fit, in which case the faintest stars are rejected until the group is small
enough to fit.
DAOPARS is a parameter set task which stores the DAOPHOT parameters required by all the DAOPHOT tasks which compute the PSF or fit stars to the PSF. Typing DAOPARS on the terminal invokes the EPAR parameter editing task. The DAOPARS parameters may also be edited from within an EPAR command to any task such as PSF which references them. These parameters may also be changed on the command line in the usual manner during invocation of any task which uses them.
Any given set of DAOPARS parameters may stored in a text file along with the data being reduced by simply invoking the :w command from within the EPAR task. If the user then sets the value of the daopars parameter in say the PSF task to the name of the file containing the stored parameter set, the stored parameters will be used instead of the default set in the user's uparm directory.
The functional forms of the analytic PSF functions are as follows. The A is simply an amplitude or normalization constant The Pn are parameters which are fit during the PSF generation process.
z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 gauss = A * exp (-0.5 * z) z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p3 moffat15 = A / (1 + z) ** 1.5 moffat15 = A / (1 + z) ** 2.5 z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p3 lorentz = A / (1.0 + z) z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 e = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p4 penny1 = A * ((1 - p3) / (1.0 + z) + p3 * exp (-0.693*e)) z = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + p5 * x * y e = x ** 2 / p1 ** 2 + y ** 2 / p2 ** 2 + x * y * p4 penny2 = A * ((1 - p3) / (1.0 + z) + p3 * exp (-0.693*e))
The predicted errors in the the DAOPHOT photometry are computed per pixel as follows, where terms 1, 2, 3, and 4 represent the readout noise, the poisson noise, the flat-fielding error, and the interpolation error respectively. The quantities readnoise, epadu, I, M, p1, and p2 are the readout noise in electrons, the gain in electrons per ADU, the pixel intensity in ADU, the PSF model intensity in ADU, the FWHM in x, and the FWHM in y both in pixels.
error = sqrt (term1 + term2 + term3 + term4) (ADU) term1 = (readnoise / epadu) ** 2 term2 = I / epadu term3 = (.01 * flaterr * I) ** 2 term4 = (.01 * proferr * M / p1 / p2) ** 2
The radial weighting function employed by all the PSF fitting tasks is the following, where dx and dy are the distance of the pixel from the centroid of the star being fit.
wtr = 5.0 / (5.0 + rsq / (1.0 - rsq)) rsq = (dx ** 2 + dy ** 2) / fitrad ** 2
The weight assigned each pixel in the fit then becomes the following.
wtp = wtr / error ** 2
After a few iterations and if clipexp > 0, a clipping scheme to reject bad data is enabled. The weights of the pixels are recomputed as follows. Pixels having a residual of cliprange sigma will have their weight reduced by half.
wt = wtp / (1.0 + (residual / error / chiold / cliprange) ** clipexp)
1. Print the DAOPARS task parameters.
da> lpar daopars
2. Edit the DAOPARS parameters.
da> epar daopars
3. Edit the DAOPARS parameters from with the PSF task.
da> epar psf
... edit a few psf parameters
... move to the daopars parameter and type :e
... edit the daopars parameters and type :q
... finish editing the psf parametera and type ^Z
4. Save the current DAOPARS parameter set in a text file daonite1.par. This can also be done from inside a higher level task as in the above example.
da> epar daopars
... type ":w daonite1.par" from within epar