Measured Object Parameters

File Formats

  1. Corrected Frames will be written to fits files containing images and masks, where the format of the corrected frames files will be:
  2. OBJCs will be split into three structures for output; the main motivation for this is to write fields in each of the ncolor OBJECT1s together (e.g. rowc[5] not 5 rowc columns). The structs are:
  3. Parameters describing photo's performance are also written out.

OBJC_IO: Photo's Main Tabular Output

First some outputs which describe the object as a whole:
id
An id for the object within the frame; this id is the same as the object's row number in the table, and is used to tie the OBJC_IO and ATLAS_IMAGE tables together.
parent
The id number of the object's parent, or -1 if it is a primary detection.
ncolor
The number of colours present in this table; should always be 5 during normal operations.
objc_type
An enum giving the object's overall classification. Current possibilities are:
UNK
An object of unknown type (the default).
CR
A cosmic ray
DEFECT
Some indeterminate chip defect.
GALAXY
A galaxy
GHOST
A ghost produced by the 2.5m optics.
KNOWNOBJ
A known object (from e.g. FIRST or ROSAT); its position and size are inputs to photo.
STAR
A Star
TRAIL
A satellite, aeroplane, meteorite, or asteroid trail.
SKY
A pseudo-object; a location where no object was detected in any of the survey bands.
These categories are repeated below for each band; the algorithm used to arrive at the overall classigfication is TBD.
catID
A catalog id number associated with KNOWNOBJs. We may want to revisit how this is done when some real known object catalogues are available.
objc_flags
The union of certain of the flag bits set in each individual band (see discussion of flags below). Specifically, if the flags BLENDED, CHILD, EDGE, INTERP, MANYPETRO, NOPETRO, NOTCHECKED, or SATUR are set in any band, they are also set in objc_flags.
objc_rowc, objc_colc, objc_rowcErr, objc_colcErr
The canonical position of the object (and 1-sigma errors), in the r' coordinate system. If an object is detected in r', this is the r' centre; otherwise it's a suitable average of the bands where the object was detected (Pixels).
The following fields are calculated for each band:
rowc, colc, rowcErr, colcErr
The position of the object (and 1-sigma errors) in each band. In the case that an object is not detected in a given band (say f'), the position is taken to be the r' position offset to the f' coordinate system (if detected in r'), and a suitable average of the other bands otherwise. (Pixels). We will describe the position assigned in the r' as the r' position, even if the object was not in fact detected in r'.
sky, skyErr
The sky level (and the 1-sigma), at the position of the object. (Counts/pixel^2).
psfCounts, psfCountsErr
The PSF-flux (and the 1-sigma error), at the position of the object (Counts). An aperture correction, to a fixed radius, calculated by the PSP, has been applied.
fiberCounts, fiberCountsErr
The 3''-counts (and the 1-sigma error), at the r' position of the object. These counts are supposed to be corrected to a canonical seeing, but this is not currently (Dec 1997) being done.
petroRad, petroRadErr
The Petrosian radius (and the 1-sigma error), measured using the r' position of the object (Pixels).
petroCounts, petroCountsErr
The Petrosian counts (and the 1-sigma error) within f_3 r_{P,r'} of the r' centre. Suitable measures must be taken if the object is not detected in r' (Counts).
petroR50, petroR50Err
The Petrosian 50%-light radius (and 1-sigma error) (Pixels).
petroR50, petroR90Err
The Petrosian 90%-light radius (and 1-sigma error) (Pixels).
Q, U, QErr, UErr
The values of <col^2/r^2 - row^2/r^2> and <col row/r^2> (and their 1-sigma errors), measured within r_{P,r'}. These are estimators of (a - b)/(a + b)cos(2 phi) and (a - b)/(a + b)sin(2 phi), and are named by analogy to the usual Stokes parameters. For more details, see The Estimation of Object's Ellipticities.
nprof
The number of points in the three succeeding measures of the radial profile, each of which refers to annuli with fixed outer radii of { 0.56, 1.69, 2.58, 4.41, 7.51, 11.58, 18.58, 28.55, 45.50, 70.51, 110.5, 172.5, 269.5, 420.5, 657.5 } pixels, that is { 0.23 0.68 1.03 1.76 3.00 4.63 7.43 11.42 18.20 28.20 44.21 69.00 107.81 168.20 263.00 } arcseconds. These radii correspond to circular apertures enclosing an integral number of pixels: { 1, 9, 21, 61, 177, 421, 1085, 2561, 6505, 15619, 38381, 93475, 228207, 555525, 1358149 }.
profMean
The mean surface brightness within the innermost nprof annuli, with fixed radii given above; these fluxes may be used to recover the annular counts exactly (counts/pixel).
profMed
The `median' surface brightness within the innermost nprof annuli, with fixed radii given above (counts/pixel).
profErr
An estimate of the uncertainty in the profiles (counts/pixel); note that this is not the photon noise (which can be recovered from profMean), but an estimate of the true uncertainty allowing for contamination by stars, HII regions, etc.
iso_rowc, iso_colc, iso_a, iso_b, iso_phi, iso_rowcErr, iso_colcErr, iso_aErr, iso_bErr, iso_phiErr, iso_rowcGrad, iso_colcGrad, iso_aGrad, iso_bGrad, iso_phi
The centre, major and minor axes, and position angle of a certain isophote (Pixels). These are determined from the 2-dimensional extracted profile. The Grad quantities are correction terms allowing us to correct for errors in the photometric calibration.
r_deV, I_deV, ab_deV, phi_deV, r_deVErr, I_deVErr, ab_deVErr, phi_deVErr
Parameters of the de Vaucouleurs profile that best fits the radial profile (as determined by the cell array), and errors. The r and I parameters are the effective radius and the surface brightness at that point.
r_exp, I_exp, ab_exp, phi_exp, r_expErr, I_expErr, ab_expErr, phi_expErr
Parameters of the exponential profile that best fits the radial profile (as determined by the cell array), and errors. The r and I parameters are the effective radius and the surface brightness at that point.
star_L, exp_L, deV_L
Likelihoods for the fits of the model by the PSF, an exponential disk, and a de Vaucouleurs profile. More specifically, the values quoted the probabilities of finding a value of chi^2 at least as large as that found for the model fits.
fracPSF
The fraction of the total light in the profile that can be assigned to a point source. Not currently calculated.
texture
A measure of the roughness of the object, based on the residuals after inverting the image and subtracting.
flags
Some more information about how the processing went. Note that these values are also used in objc_flags, where they are used to summarise information about the object as a whole, rather than a single band. Current possibilities are:
OBJECT1_NOTDETECTED
Object wasn't detected in this band
OBJECT1_BRIGHT
Object was found by findBrightObjects
OBJECT1_BINNED
Object contains pixels that were only marked as part of an object after the frame had been binned, and the object finder rerun
OBJECT1_EDGE
Object was too close to edge of frame to be measured
OBJECT1_PEAKCENTER
Given centre is position of peak pixel, rather than an MLE fit
OBJECT1_NOPROFILE
The object was too small for us to be able to measure a radial profile
OBJECT1_TOO_LARGE
The object is too large for us to measure its profile (i.e. it extends beyond a radius of approximately 263 arcseconds; see SDSS Profile Extraction.)
OBJECT1_BAD_RADIAL
The radial profile extends beyond where its S/N first drops to (??) 1.
OBJECT1_NOSTOKES
Object has no measured stokes parameters.
OBJECT1_BADSKY
The sky level is so bad that the highest pixel in the object is very negative; far more so than a mere non detection. No further analysis is attempted.
OBJECT1_ELLIPFAINT
The object's centre is fainter than the isophote whose shape is desired, so its properties are not measured.
Information about measuring Petrosian quantities:
OBJECT1_NOPETRO
The object has no Petrosian radius
OBJECT1_MANYPETRO
The object has more than one Petrosian radius; the largest found is adopted
OBJECT1_PETROFAINT;
At least one possible Petrosian radius was rejected as the surface brightness at r_P was too low. If NOPETRO isn't set, an (different) acceptable Petrosian radius was found.
OBJECT1_PETRO_SMALL
Object has no Petrosian radius; used f5 instead.
OBJECT1_PETRO_BIG
Not currently used
OBJECT1_MANYR50
An object has more than one 50% light radius
OBJECT1_MANYR90
An object has more than one 90% light radius
OBJECT1_INCOMPLETE_PROFILE
The circle r = r_{P,r'} intersects the edge of the frame
Information about pixels contained in the object:
OBJECT1_CR
Object contains at least one pixel flagged as belonging to a cosmic ray.
OBJECT1_INTERP
The object contains at least one pixel that has been interpolated
OBJECT1_NOTCHECKED
The object contains at least one pixel that is marked as having not been searched for objects
OBJECT1_SATUR
The object contains at least one saturated pixel
OBJECT1_SUBTRACTED
Bright wings were subtracted from this object (presumably a star).
Various flags associated with the deblender:
OBJECT1_BLENDED
Object was found to be blended, and has children
OBJECT1_CHILD
Object is a deblended child
OBJECT1_NODEBLEND
No deblending was attempted, although the BLENDED flag is set
OBJECT1_DEBLENDED_AS_PSF
The deblender treated the object as a PSF
OBJECT1_DEBLEND_PRUNED
The deblender deleted some of the peaks that we in the parent object
type
The type assigned to the object in this colour; the possibilities are described for the objc_type field.

Petrosian Quantities

(This discussion is stolen from and supersedes the corresponding parts of Michael Strauss' document Galaxy Selection Algorithm for SDSS).

Needs a rewrite! Let I(r) be (a spline fit to or other smooth representation of) the measured azimuthally averaged surface brightness profile of an object in r'. Define the Petrosian ratio R_P(r) as the ratio of the local surface brightness at radius r to the mean within r: 

	, 1.25r
        |
	|  I(r') 2 pi r' dr' / [pi (1.25^2 - 0.8^2) r^2]
        |
	' 0.8r
R_P == ---------------------------------------------------
                   , r
                   |
	           | I(r') 2 pi r' dr' / [pi r^2]
                   |
                   ' 0
Mark all the radii r_i, i = 1,..., N where R_P falls to a specified value f_1, and for which I(r) > f_2.

If there's at least one such radius (N > 0), the largest of the r_i will be taken as the Petrosian radius r_P; if N = 0, the adopted radius will be a Kent radius one given by the solution to <I(r)> = f_3, the point where the mean surface brightness falls to some value f_3. If more than one such Kent radius exists, the smallest is adopted; if no such radius exists, there are two possibilities: that the lowest surface brightness in the object is above f_3 (in which case we adopt r_P = r_max, the largest ``good'' radius in the profile), or that the highest surface brightness is below f_3 (in which case we take r_P = f_5).

We shall replace this use of a Kent radius in the near future, preferring to use photo's best fit model to extrapolate the Petrosian ratio until it reaches a value f_1.

The Petrosian flux F_P is defined as the total flux as measured within a certain number of Petrosian radii: 

                   , f_4 r_P
                   |
	     F_P = | I(r') 2 pi r' dr'
                   |
                   ' 0
The Petrosian half-light r_{50} is defined by the implicit equation: 
                   , r_50
                   |
	           | I(r') 2 pi r' dr' = 0.5 F_P
                   |
                   ' 0
The Petrosian 90% radius is defined by the implicit equation: 
                   , r_90
                   |
	           | I(r') 2 pi r' dr' = 0.9 F_P
                   |
                   ' 0
How should we set these surface brightness values? Michael Strauss suggests the following: For a Freeman disk (central surface brightness in r' of 20.85, using the B-r' colors of disks from Frei and Gunn), the Petrosian ratio falls to 1/4 at 3.21 scale lengths, and to 1/8 at 4.43 (with corresponding surface brightnesses of 24.33 and 25.66). The Kent surface brightness at these radii are 22.83 and 23.40. If we decide that f_2 should be a magnitude fainter than the value for a Freeman disk, and take f_5 to be twice the fibre radius, we arrive at the two possible sets of strawman values given in the following table: 
f_1	Petrosian Ratio				1/5	1/8
f_2	Minimum Surface Brightness at r_P	25.3	26.7
f_4	Multiple of r_P for Petrosian flux	3	2
f_3	Fallback Radius				3''	3''
These numbers have changed