diff options
Diffstat (limited to 'noao/digiphot/daophot/doc/daopars.hlp')
| -rw-r--r-- | noao/digiphot/daophot/doc/daopars.hlp | 331 |
1 files changed, 331 insertions, 0 deletions
diff --git a/noao/digiphot/daophot/doc/daopars.hlp b/noao/digiphot/daophot/doc/daopars.hlp new file mode 100644 index 00000000..80f340b6 --- /dev/null +++ b/noao/digiphot/daophot/doc/daopars.hlp @@ -0,0 +1,331 @@ +.help daopars May00 noao.digiphot.daophot +.ih +NAME +daopars -- edit the daophot fitting parameters +.ih +USAGE +daopars +.ih +PARAMETERS +.ls function = "gauss" +The functional form of the analytic component of the PSF model computed by the +DAOPHOT PSF task. 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. + +.ls gauss +An elliptical Gaussian function aligned along the x and y axes of the +input image. +.le +.ls moffat15 +An elliptical Moffat function with a beta parameter of 1.5. +.le +.ls moffat25 +An elliptical Moffat function with a beta parameter of 2.5. +.le +.ls lorentz +An elliptical Lorentzian function with beta parameter of 1.0. +.le +.ls penny1 +A Gaussian core with Lorentzian wings function, 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. +.le +.ls penny2 +A Gaussian core with Lorentzian wings function, where the Gaussian core and +Lorentzian wings may be tilted in different directions. The Lorentzian wings +have a beta parameter of 1.0. +.le +.ls auto +The PSF task computes the analytic PSF model for each of the six analytic model +PSFs in turn and selects the one that produces the smallest standard deviation +for the model fit. +.le +.ls func1,func2,... +The PSF task computes the analytic PSF model for each of a subset of the six +defined functions in turn, and selects the one that produces the smallest +standard deviation for the model fit. +.le + +In general "gauss" is the best and most efficient choice for a well-sampled +ground-based image, "lorentz" is best for old ST images, and "moffat15" or +"moffat25" are best for under-sampled ground-based images. +.le +.ls varorder = 0 +The order of variability of the PSF model computed by the DAOPHOT PSF task. +Varorder sets the number of look-up tables containing the deviations of the +true PSF from the analytic model PSF that are computed by the model. +.ls "-1" +Only the analytic function specified by \fIfunction\fR is used to compute +the PSF model. The PSF model is constant over the image. +.le +.ls "0" +The analytic function and one look-up table are used to compute the +PSF model. The PSF model is constant over the image. +.le +.ls "1" +The analytic function and three look-up tables are used to compute the PSF +model. The PSF model is linearly variable over the image, with terms +proportional to 1, x and y. +.le +.ls "2" +The analytic function and six look-up tables are used to compute the +PSF model. The PSF model is quadratically variable over the image, with terms +proportional to 1, x, y, x**2, xy, y**2. +.le +.le +.ls nclean = 0 +The number of additional iterations the PSF task performs to compute the PSF +look-up tables. If \fInclean\fR is > 0, stars which contribute deviant +residuals to the PSF look-up tables in the first iteration, will be +down-weighted in succeeding iterations. +.le +.ls saturated = no +Use saturated stars to improve the signal-to-noise in the wings of the PSF +model 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. +.le +.ls matchrad = 3.0 (scale units) +The tolerance in scale units for matching the stellar x and y centroids in the +input photometry file with the image cursor position. Matchrad is currently +used by the PSTSELECT and PSF tasks to match stars shown on the image display +with stars in the photometry list. +.le +.ls psfrad = 11.0 (scale units) +The radius of the circle in scale units within which the PSF model 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 +set larger than the size of the PSF model but may set smaller in tasks +like GROUP, ALLSTAR, SUBSTAR, and ADDSTAR. +.le +.ls 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 a somewhat +smaller value may be used 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. +.le +.ls recenter = yes (peak, nstar, and allstar) +Compute new positions as well as magnitudes for all the stars in the input +photometry list? +.le +.ls 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 averaging 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. If fitsky = yes the ALLSTAR task computes new sky values for each +star every third iteration by subtracting off the best current fit for the star +and and estimating the median of the pixels in the annulus defined by +\fIsannulus\fR and \fIwsannulus\fR. The new group sky value is the average of +the new individual values. +.le +.ls 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 \fIgroupsky\fR 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. +.le +.ls sannulus = 0.0 (scale units, allstar) +The inner radius of the sky annulus used by ALLSTAR to recompute the sky +values. +.le +.ls wsannulus = 11 (scale units, allstar) +The width of the sky annulus used by ALLSTAR to recompute the sky values. +.le +.ls flaterr=0.75 (percent, peak, nstar, allstar) +The image flat-fielding error in percent used to compute the predicted +errors of the fit. +.le +.ls proferr = 5.0 (percent, peak, nstar, allstar) +The profile or interpolation fitting error in percent used to compute +the predicted errors of the fit. +.le +.ls maxiter = 50 (peak, nstar, allstar) +The maximum number of times that the PSF fitting tasks PEAK, NSTAR, and ALLSTAR +will iterate on the PSF fit before giving up. +.le +.ls 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. For values of clipexp greater than 1 a residual small compared to +cliprange standard deviations does not have its weight significantly altered, +one with exactly \fIcliprange\fR standard deviations is assigned half its +normal weight, and large residuals are assigned weights which fall off as the +standard deviation to the minus clipexp power. For normal applications users +should leave these parameter at their default value. +.le +.ls critsnratio = 1.0 (group) +The ratio of the model intensity of the brighter star computed 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 +signal-to-noise ratio parameter is used to group stars. In general if a small +value such as 0.1 divides all the stars in an image into groups less than +\fImaxgroup\fR, then the expected random errors will determine the accuracy +of the photometry. On the other hand if a value of critical 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. +.le +.ls mergerad = INDEF (scale units, nstar, allstar) +The critical separation in scale units between two objects for an object merger +to be considered. Objects with separations > mergerad will not be merged; faint +objects with separations <= mergerad will be considered for merging. The +default value of mergerad is sqrt (2 *(PAR1**2 + PAR2**2)), where PAR1 and PAR2 +are the half-width at half-maximum along the major and minor axes of the psf +model. Merging can be turned off altogether by setting mergerad to 0.0. +.le +.ls maxnstar = 10000 (pstselect, psf, group, allstar, substar) +The initial star list buffer size. If there are more than maxnstar stars in the +input photometry file buffer, DAOPHOT will resize the buffers as needed. +The only limitation is the memory and configuration of the host computer. +.le +.ls maxgroup = 60 (nstar, allstar) +The maximum numbers of stars that the multiple star fitting tasks NSTAR and +ALLSTAR will fit simultaneously. NSTAR will not to fit groups large than +maxgroup. ALLSTAR dynamically regroups the stars in large groups until the +group is either maxgroup or smaller in size or becomes too dense to group, +after which the faintest stars are rejected until the group is less than +maxgroup ins size. +.le + +.ih +DESCRIPTION + +DAOPARS is a parameter set task which stores the DAOPHOT parameters +required by all those DAOPHOT tasks which compute the PSF model, fit stars +to the PSF model, or evaluate the PSF model. + +Typing DAOPARS on the terminal invokes the EPAR parameter editing task. The +DAOPARS parameters may also be edited from within an EPAR command on task, +for example PSF, which references them. The DAOPARS parameters may also +be changed on the command line in the usual manner when any task which +references them is executed. + +Any given set of DAOPARS parameters may stored in a text file along with +the data being reduced by typing the :w command from within the EPAR task. If +the user then sets the value of the \fIdaopars\fR parameter to the name of +the file containing the stored parameter set, the stored parameters will be +used instead of the default set in the uparm directory. + +.ih +ALGORITHMS + +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 model generation process. + +.nf + 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 + moffat25 = 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)) +.fi + + +The predicted errors in 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. + +.nf + 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 +.fi + +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. + +.nf + wtr = 5.0 / (5.0 + rsq / (1.0 - rsq)) + rsq = (dx ** 2 + dy ** 2) / fitrad ** 2 +.fi + +The weight assigned each pixel in the fit then becomes the following. + +.nf + wtp = wtr / error ** 2 +.fi + +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. + +.nf + wt = wtp / (1.0 + (residual / error / chiold / + cliprange) ** clipexp) +.fi + +Pixels having a residual of cliprange sigma will have their weight reduced +by half. + +.ih +EXAMPLES + +1. Print the DAOPARS task parameters. + +.nf + da> lpar daopars +.fi + +2. Edit the DAOPARS parameters. + +.nf + da> daopars +.fi + +3. Edit the DAOPARS parameters from with the PSF task. + +.nf + da> epar psf + + ... edit a few psf parameters + + ... move to the daopars parameter and type :e + + ... edit the daopars parameters and type :wq + + ... finish editing the psf parameters and type :wq +.fi + +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. + +.nf + da> epar daopars + + ... type ":w daonite1.par" from within epar +.fi + +.ih +TIME REQUIREMENTS +.ih +BUGS +.ih +SEE ALSO +pstselect,psf,peak,group,nstar,allstar,substar,addstar,setimpars +.endhelp |