IPP Software Navigation Tools IPP Links Communication Pan-STARRS Links

Changeset 40305


Ignore:
Timestamp:
Dec 27, 2017, 7:40:35 AM (9 years ago)
Author:
eugene
Message:

updates to remove superfluous figures and adjust text as needed

File:
1 edited

Legend:

Unmodified
Added
Removed
  • trunk/doc/release.2015/systematics.20140411/diffusion.tex

    r40304 r40305  
    1919%\newcommand\newtext[1]{\textbf{\color{blue}#1}}
    2020
    21 \newcommand\oldtext[1]{\textbf{\color{lightgray}#1}
    22 \newcommand\newtext[1]{\textbf{\color{blue}#1}
    23 \newcommand\fixtext[1]{\textbf{\color{red}#1}
     21\definecolor{light-gray}{gray}{0.50}
     22\newcommand\oldtext[1]{\textbf{\color{light-gray}#1}}
     23\newcommand\newtext[1]{\textbf{\color{blue}#1}}
     24\newcommand\fixtext[1]{\textbf{\color{red}#1}}
    2425
    2526\usepackage[T1]{fontenc}% (2) specify encoding
     
    514515of the PSF magnitude tree-rings would certainly have been obvious.
    515516
     517\newtext{Figure~\ref{fig:all.effects.rband} shows the complete set of
     518  measured effects for the \rps\ filter.  In addition to the PSF and
     519  aperture photometry, this figure shows the astrometric residuals,
     520  the high-frequency flat-field structures, along with two
     521  measurements derived from the second moments: the ``smear'' and the
     522  ``shear'', discussed below.}
     523
    516524\subsection{Astrometric Residuals}
    517525
     
    533541%% \end{figure*}
    534542
    535 Figure~\ref{fig:astrom.by.filter} shows a similar type of measurement
     543\oldtext{Figure~3} \newtext{Figure~\ref{fig:all.effects.rband} (middle-left)}
     544shows a similar type of measurement
    536545for astrometric residuals.  To generate this plot, we use the same
    537546selection of measurements for astrometry as for photometry.  In this
     
    550559: tangential component).
    551560
    552 Figure~\ref{fig:astrom.by.filter} shows the 2D patterns of $\delta R$
     561\oldtext{Figure~\ref{fig:astrom.by.filter} shows the 2D patterns of $\delta R$
    553562for each filter (\grizy).  The dynamic range of the color scale is
    554 from -20 to +20 milliarcseconds for all 5 plots.  A tree-ring
     563from -20 to +20 milliarcseconds for all 5 plots.}  A tree-ring
    555564pattern is visible for all five filters, with systematic structures
    556565following a circular pattern centered on the chip corner; the finging
    557 pattern is not apparent in the \yps\ astrometry.  The per-pixel
     566pattern is not apparent in the \yps\ astrometry.  \oldtext{The per-pixel
    558567standard deviations of these plots are listed in
    559 Table~\ref{table:sigmas.by.filter}.  The signal-to-noise of these
     568Table~\ref{table:sigmas.by.filter}.}  The signal-to-noise of these
    560569structures is again somewhat weak, but the pattern is clearly visible
    561570in these figures.
     
    579588
    580589% 2012ApJ...750...99T = Tonry et al PS1 phot system
    581 Figure~\ref{fig:flats.by.filter} shows the high-spatial-frequency
     590\oldtext{Figure~4} \newtext{Figure~\ref{fig:all.effects.rband} (middle-right)}
     591shows the high-spatial-frequency
    582592structures in the flat-field images.  For this measurement, we have
    583593used a set of monochromatic flat-field images obtained with a tunable
     
    597607pixels associated with each superpixel. 
    598608
    599 Figure~\ref{fig:flats.by.filter} shows the superpixel images for the
    600 flat-fields in the five filters.  These flat-field images are
     609\fixtext{Figure~\ref{fig:flats.by.filter} shows the superpixel images for the
     610flat-fields in the five filters.}  These flat-field images are
    601611displayed as fractional deviations relative to the median flat-field
    602612image and can thus be compared to the magnitude residuals.  When
     
    699709PSF ellipticity from the $e_1$ term.
    700710
    701 Figure~\ref{fig:smear.by.filter} shows the spatial trend of the smear,
     711\oldtext{Figure~5} \newtext{Figure~\ref{fig:all.effects.rband} (lower-left)}
     712shows the spatial trend of the smear,
    702713$e_0$.  The dynamic range of these images is -0.3 to +0.3 pixel$^2$. A
    703714tree-ring pattern is visible for all 5 filters, though \yps\ is
     
    705716spatial frequencies can also be seen.
    706717
    707 Figure~\ref{fig:shear.by.filter} shows the spatial trend of the shear,
     718\oldtext{Figure~6} \newtext{Figure~\ref{fig:all.effects.rband} (lower-right)}
     719shows the spatial trend of the shear,
    708720$e_2$.  This value is positive definite and is plotted with a color
    709721scale ranging from -0.02 to 0.22 pixel$^2$.  Overlayed on
    710 Figure~\ref{fig:shear.by.filter} is a set of vectors representing the
     722\oldtext{Figure~6} \newtext{Figure~\ref{fig:all.effects.rband} (lower-right)}
     723is a set of vectors representing the
    711724ellipse orientation as a function of postion.  The length of the
    712725vectors corresponds to the value of $e_2$.  The tree-ring structure is
     
    729742  much higher frequencies than the previous two effects.  Aperture
    730743  magnitude (upper-right) and shear residuals (lower-right) do not
    731   show a strong signal consistent with either of the two patterns.} \label{fig:all.effects.rband}
     744  show a strong signal consistent with either of the two patterns.}
     745\label{fig:all.effects.rband}
    732746\end{center}
    733747\end{figure*}
     
    778792For a given type of measurement, the systematic effect is strongly
    779793correlated between filters.  The strongest correlation is the smear
    780 term: Figure~\ref{fig:smear.trends} shows the correlation of the smear
    781 pattern between \gps\ and the other four filters. Even \yps\ is
     794term\oldtext{: Figure~8 shows the correlation of the smear
     795pattern between \gps\ and the other four filters}. Even \yps\ is
    782796strongly correlated with \gps\ despite the presence of the fringe
    783797pattern.  PSF photometric residuals are also correlated between
    784 filters, as shown in Figure~\ref{fig:psfmag.trends}.  Here, the
     798filters\oldtext{, as shown in Figure~9}.  Here, the
    785799\yps\ correlation with \gps\ is quite weak: the fringing pattern
    786800dominates the tree rings for PSF photometry.  The radial component of
     
    792806pattern.
    793807
    794 For all four types of measurements, the slope of the fitted lines are
    795 listed in Table~\ref{table:correlation.by.filter}.  There is a
    796 consistency in the trend from \gps, with the strongest systematic
    797 tree-ring effects to \yps, with the weakest effects.  Note that the
    798 second moment smear and astrometry terms have different relative
    799 strength in \yps\ compared with \gps.
     808For all four types of measurements, the \oldtext{slope of the fitted
     809  lines} \newtext{amplitudes relative to \gps} are listed in
     810Table~\ref{table:correlation.by.filter}.  There is a consistency in
     811the trend from \gps, with the strongest systematic tree-ring effects
     812to \yps, with the weakest effects.  Note that the second moment smear
     813and astrometry terms have different relative strength in
     814\yps\ compared with \gps.
    800815
    801816% smear trends by filter
     
    849864pattern between the different types of measurements.  Different models
    850865for the tree-ring structures make different predictions about the
    851 correlations between different effects.  Note the very different
     866correlations between different effects.
     867%
     868\newtext{Figure~\ref{fig:effects.vs.radius} shows the radial run of the
     869  four effects which show clear tree rings (in \rps).  Since the tree
     870  rings are relatively narrow, this figure shows only the radial range
     871  of 150 - 300 pixels to allow the reader to see the relationship
     872  between structures in the different effects. }
     873%
     874Note the very different
    852875spatial structure between the different measurements in a given
    853876filter: the radial variations do not all follow the same patterns.
     
    863886  signal for \gps\ (upper-left), \rps\ (upper-right), \ips\ (lower-left),
    864887  \zps\ (lower-right).
    865 } \label{fig:smear.vs.psfmag}
     888} \label{fig:effects.vs.radius}
    866889\end{center}
    867890\end{figure*}
     
    876899  signal for \gps\ (upper-left), \rps\ (upper-right), \ips\ (lower-left),
    877900  \zps\ (lower-right).
    878 } \label{fig:smear.vs.psfmag}
     901} \label{fig:dsmear.and.astrom}
    879902\end{center}
    880903\end{figure*}
     
    889912  signal for \gps\ (upper-left), \rps\ (upper-right), \ips\ (lower-left),
    890913  \zps\ (lower-right).
    891 } \label{fig:smear.vs.psfmag}
     914} \label{fig:dastrom.and.flat}
    892915\end{center}
    893916\end{figure*}
     
    897920mean tend to have smaller measured PSF fluxes than the mean (note that
    898921$\delta m_{psf}$ is defined so that positive values correspond to
    899 larger fluxes).  These trends are shown in
    900 Figure~\ref{fig:smear.vs.psfmag}. 
     922larger fluxes).  \oldtext{These trends are shown in Figure 12.}
    901923
    902924Second, the radial derivative of the smear is anti-correlated with the
    903 radial component of the astrometric residuals: $\frac{\partial
    904   (\sigma^2_{major} + \sigma^2_{minor})}{\partial radius} \sim \delta
    905 R$ (see Figure~\ref{fig:dsmear.vs.astrom}).
     925radial component of the astrometric residuals
     926\newtext{Figure~\ref{fig:dsmear.and.astrom} shows the radial run of
     927  $\frac{\partial (\sigma^2_{major} + \sigma^2_{minor})}{\partial radius}$
     928  and $\delta R$ together to illustrate this relationship.}
     929\oldtext{: $\frac{\partial(\sigma^2_{major} + \sigma^2_{minor})}{\partial radius} \sim \delta R$. (see Figure~13).}
    906930
    907931Finally, the radial derivative of the radial component of the
    908 astrometric residual is anti-correlated with the flat-field residual
    909 errors: $\frac{\partial \delta R}{\partial radius} \sim \delta flat$
    910 (see Figure~\ref{fig:dastrom.vs.flat}).  This last relationship is
    911 somewhat weakly measured.  Because of the periodic nature of the tree
    912 rings, it is also difficult to be completely certain that the
    913 flat-field is proportional to the derivative of the astrometry
    914 residual, rather than the astrometry residual being proportional to
    915 the derivative of the flat-field.  The correlation is somewhat weaker
    916 for derivative of the flat-field vs astrometry residual.  The
    917 correlation is very weak between the flat-field and the astrometry
    918 residual values without a derivative.  We are convinced that we have
    919 the sense of the derivative correct by examination of specific
    920 features in each image.
     932astrometric residual is correlated with the flat-field residual
     933errors.
     934\newtext{Figure~\ref{fig:dastrom.and.flat} shows the radial run of
     935  $\frac{\partial \delta R}{\partial radius}$ and $\delta flat$ together
     936  to illustrate this relationship.}
     937\oldtext{: $\frac{\partial \delta R}{\partial radius} \sim \delta flat$ (see Figure~14).}
     938
     939This last relationship is somewhat weakly measured.  Because of the
     940periodic nature of the tree rings, it is also difficult to be
     941completely certain that the flat-field is proportional to the
     942derivative of the astrometry residual, rather than the astrometry
     943residual being proportional to the derivative of the flat-field.
     944\newtext{Careful examination of Figures~\ref{fig:effects.vs.radius}
     945  and \ref{fig:dastrom.and.flat} convince us that we have the sense of
     946  the derivative correct.}
     947%
     948\oldtext{The correlation is somewhat weaker for derivative of the
     949  flat-field vs astrometry residual.  The correlation is very weak
     950  between the flat-field and the astrometry residual values without a
     951  derivative.  We are convinced that we have the sense of the
     952  derivative correct by examination of specific features in each
     953  image.}
    921954
    922955\begin{table}
     
    9871020
    9881021First, if we consider the smear pattern
    989 (Figure~\ref{fig:smear.by.filter}), the measurement shows that the
    990 intrinsic sizes of the stellar images are varying in a radial sense
    991 between the different tree-ring regions.  Although images experience
    992 an average image quality (due to seeing and focus) across the chip
    993 which may vary substantially from exposure to exposure, stars landing
    994 in the different tree-ring regions are consistently somewhat
     1022\oldtext{(Figure~5)}\newtext{(Figure~\ref{fig:all.effects.rband}, lower-left)},
     1023the measurement shows that the intrinsic sizes of the stellar images
     1024are varying in a radial sense between the different tree-ring regions.
     1025Although images experience an average image quality \oldtext{(due to
     1026  seeing and focus)} across the chip which may vary substantially from
     1027exposure to exposure \newtext{(due to seeing and focus)}, stars
     1028landing in the different tree-ring regions are consistently somewhat
    9951029larger or somewhat smaller than that average.
    9961030
    9971031Next, we can explain the correlation between the PSF photometry
    998 residuals and the observed smear (Figure~\ref{fig:smear.vs.psfmag}).
     1032residuals and the observed smear
     1033\newtext{(Figure~\ref{fig:effects.vs.radius})}\oldtext{(Figure~12)}.
    9991034In the photometry analysis, we model the PSF allowing for some spatial
    10001035variation in the shape.  However, we have a limited number of stars to
     
    10041039interpolation between the $3 \times 3$ grid points.  Thus, the spatial
    10051040scale on which we model PSF variations is much larger than the spatial
    1006 scale on which PSF variations are actually occuring, as illustrated
    1007 by the changes in the smear plot (Figure~\ref{fig:smear.by.filter}).
     1041scale on which PSF variations are actually occuring, as illustrated by
     1042the changes in the smear plot
     1043\oldtext{(Figure~5)}\newtext{(Figure~\ref{fig:all.effects.rband}, lower-left)}.
    10081044When the true PSF is larger than the model PSF, our model fits
    10091045systematically underestimate the amount of flux in a given object.
    1010 Conversely, when the true PSF is smaller, we overestimate the flux -- this
    1011 type of offset is a typical effect when mis-estimating the PSF size.
    1012 The slope of the trend depends on the mean typical seeing for the
    1013 given filter.  For example, the \gps\ seeing is typically 1.3\arcsec,
    1014 corresponding to a Gaussian $\sigma$ of 2.15 pixels.  A smearing of
    1015 $\sigma^2_{major} + \sigma^2_{minor} = 0.1$ pixels$^2$ would increase
    1016 the size by about 0.02 pixels, or 1\%, roughly consistent with the
    1017 observed photometric deviation of about 5 to 10 millimags for this
    1018 amount of smearing.
     1046Conversely, when the true PSF is smaller, we overestimate the flux --
     1047this type of offset is a typical effect when mis-estimating the PSF
     1048size.  The slope of the trend depends on the mean typical seeing for
     1049the given filter.  For example, the \gps\ seeing is typically
     10501.3\arcsec, corresponding to a Gaussian $\sigma$ of 2.15 pixels.  A
     1051smearing of $\sigma^2_{major} + \sigma^2_{minor} = 0.1$ pixels$^2$
     1052would increase the size by about 0.02 pixels, or 1\%, roughly
     1053consistent with the observed photometric deviation of about 5 to 10
     1054millimags for this amount of smearing.
    10191055
    10201056The correlation between the flat-field structures and the radial
    10211057derivative of the astrometric residual displacements in the radial
    1022 direction (Figure~\ref{fig:dastrom.vs.flat}) is consistent with radial
    1023 variations in the plate-scale.  The tree-rings observed by DES are
    1024 completely attributed to effective plate scale changes.  Effective
    1025 plate scale changes result in flat-field deviations because the
    1026 flat-field illumination is a source of constant surface brightness.
    1027 Pixels see a varying amount of flux depending on their effective area.
    1028 This changing plate scale also affects the astrometry since these
    1029 variations occur on spatial scales much smaller than the astrometric
    1030 model.  In this description of the tree rings, the flat-field
    1031 deviations are $-1 \times \frac{\partial \delta R}{\partial r}$.  The
    1032 best-fit slopes of our correlations are \approx 0.5, but the
    1033 signal-to-noise is rather low.  A slope of -1 appears to be consistent
    1034 with our measurements.
     1058direction
     1059\oldtext{(Figure~14)}\newtext{(Figure~\ref{fig:dastrom.and.flat})}
     1060is consistent with radial variations in the plate-scale.  The
     1061tree-rings observed by DES are completely attributed to effective
     1062plate scale changes.  Effective plate scale changes result in
     1063flat-field deviations because the flat-field illumination is a source
     1064of constant surface brightness.  Pixels see a varying amount of flux
     1065depending on their effective area.  This changing plate scale also
     1066affects the astrometry since these variations occur on spatial scales
     1067much smaller than the astrometric model.  In this description of the
     1068tree rings, the flat-field deviations are $-1 \times \frac{\partial
     1069  \delta R}{\partial r}$.  The best-fit slopes of our correlations are
     1070\approx 0.5, but the signal-to-noise is rather low.  A slope of -1
     1071appears to be consistent with our measurements.
    10351072
    10361073The fact that the PSF ellipticity changes are {\em not} correlated
    1037 with the tree-ring structure (Figure~\ref{fig:shear.by.filter}) tells us
    1038 that, unlike the case for DES, the effective plate-scale changes seen
    1039 in the flat-field and astrometry signals are not the dominant cause of
    1040 the PSF photometry errors.  Also, the fact that we do not measure
    1041 significant aperture photometry errors correlated with the tree rings
    1042 confirms this point.  The amplitude of the flat-field errors are 1-2
    1043 millimagnitudes, much smaller than the PSF photometry errors, and far
    1044 below the pixel-to-pixel noise in the aperture magnitude residuals.
    1045 It is likely in our opinion that the plate-scale changes causing the
    1046 flat-field and astrometry effects are affecting both the ellipticity
    1047 and the aperture magnitudes, but the level of the effect is too small
    1048 to see given the other systematic structures (in the shear plot) and
    1049 the noise level (in the aperture magnitudes).
     1074with the tree-ring structure
     1075\oldtext{(Figure~6)}\newtext{(Figure~\ref{fig:all.effects.rband})}
     1076tells us that, unlike the case for DES, the effective plate-scale
     1077changes seen in the flat-field and astrometry signals are not the
     1078dominant cause of the PSF photometry errors.  Also, the fact that we
     1079do not measure significant aperture photometry errors correlated with
     1080the tree rings confirms this point.  The amplitude of the flat-field
     1081errors are 1-2 millimagnitudes, much smaller than the PSF photometry
     1082errors, and far below the pixel-to-pixel noise in the aperture
     1083magnitude residuals.  It is likely in our opinion that the plate-scale
     1084changes causing the flat-field and astrometry effects are affecting
     1085both the ellipticity and the aperture magnitudes, but the level of the
     1086effect is too small to see given the other systematic structures (in
     1087the shear plot) and the noise level (in the aperture magnitudes).
    10501088
    10511089Finally, the correlation between the smear structures and the
    10521090astrometry residuals shows that these two effects are connected.
    1053 Although the correlation is weak in Figure~\ref{fig:dsmear.vs.astrom},
     1091Although the correlation is weak in
     1092\oldtext{Figure~13} \newtext{Figure~\ref{fig:effects.vs.radius}},
    10541093careful inspection of the location of these two tree ring patterns
    10551094shows that the locations of the rings in the radial astrometric
Note: See TracChangeset for help on using the changeset viewer.