Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.naic.edu/~gibson/thesis/fig/fig_ch2.html
Дата изменения: Sun Jun 27 07:16:42 1999
Дата индексирования: Sat Dec 22 05:20:10 2007
Кодировка:
Thesis Figures
Chapter 2: Nebular Dust Imaging and Photometry
The WISP field position as originally planned, shown on a reproduction of
O filter (blue) Palomar Sky Survey prints of the Pleiades region. North is up
and east is to the left. The WISP field box is 5° in length. The
central cluster is obscured by saturated nebulosity.
Fluxes for four of the brightest Pleiades stars, taken from the TD-1UV flux catalog of Thompson et al. (1978) at 1565, 1965, 2365, &
2740 Å (filled boxes), and SIMBADB & V
(4400 & 5500 Å) (filled circles) magnitudes, with conversion
factors from Zombeck (1990). No corrections have been made for reddening or
extinction. WISP1, WISP2, and Bj 50% transmission bandwidths are shown at
bottom. The TD-1 error bars refer to the 20% absolute
photometric error, a measure of systematic scaling uncertainty for the data as
a whole. Relative errors for individual TD-1 measurements are <=
2%. The dotted blue line shows a Kurucz (1993) model stellar spectrum for an
unreddened star similar to those plotted here, having Z/ZSun
= 1, log(g)=3.5, and Teff = 13000 K. While Hydrogen
Balmer continuum opacity shortward of 3647 Å causes fluxes at 2740
Å to be lower than those at 4400 Å, the depression of stellar
UV fluxes relative to the model spectrum is due to a modest amount of
selective extinction toward the cluster (see Fig. \ref{Fig:reddening} and
related discussion in Chapter \ref{Chap:intro}).
A typical WISP point spread function, shown with logarithmic brightness
scaling. Note the lack of a single central maximum, and what appears to be a
pattern of two overlapping ringlike structures, possibly defocused images. The
``tails'' hanging off the PSF core appear split as here in Q frames, but
joined in U frames. A bicubic spline surface was logarithmically fit to this
and other PSFs, for the purpose of shifting and resampling them to subpixel
accuracy. The stars treated by the CLEANing process are for the most part
separated by noninteger numbers of pixels, which, given the undersampled nature
of the PSF, necessitated cautious interpolation. The PSF Full Width at Half
Maximum (FWHM) is approximately 2.5x5 pixels (75''x150'').
Four steps in the treatment of PSFs. All images are displayed on a linear
scale ranging from (black) -1x10-18 to (white)
30x10-18 erg cm-2 s-1 Å-1
arcsec-2. To obtain sky orientation, the images as shown must be
rotated clockwise 30°. Top: WISP2 Q- image prior
to PSF removal. No identifiable diffuse light present. Upper middle:
Sum of all PSFs removed. Lower middle: The desired residuals, including
some definite signs of nebulosity. Bottom: Same, with bright stellar
cores masked in preparation for smoothing. Note glows around Merope, Maia, and
Alcyone (23, 20, & Eta Tau).
The IWISP1 and IWISP2 combined images.
Display scales and orientations are as in Fig. \ref{Fig:wsteps}. Top:IWISP1, the combined average of Q-(WISP1),
Q+(WISP1), U-(WISP1), and
U+(WISP1). Upper middle:IWISP2,
similarly obtained. Lower middle:IWISP1 with bad
areas masked in preparation for smoothing.
Bottom:IWISP2 with bad areas masked.
The IWISP1 masked image with increasing amounts of smoothing.
Window sizes are, from top to bottom, 3, 7, 15, and 31 pixels, or 1.5',
3.5', 7.4', and 15.3'. Display scales and orientations
are as in Fig. \ref{Fig:wsteps}.
The IWISP2 masked image with increasing amounts of smoothing.
Window sizes are, from top to bottom, 3, 7, 15, and 31 pixels, or 1.5',
3.5', 7.4', and 15.3'. Display scales and orientations
are as in Fig. \ref{Fig:wsteps}.
The effects of culling WISP data with signal-to-noise information.
S/N maps are shown for 15x15-smoothed WISP1 (Top) and WISP2
(Upper Middle) data, with a display range of 5 <= S/N <= 25.
Regions with S/N < 5 are excised. Note the ``good'' remainder is more
poorly shaped in WISP1 (Lower Middle) than WISP2 (Bottom). Both
images suffer from flatfield problems which manifest in their eastern halves,
but these are more severe for the WISP1 data.
The WISP photometric S/N ratio as a function of image data counts (ADU),
also expressed in intensity units for the two different filter calibrations.
The bottom curve is the S/N without any binning or smoothing. Higher curves
show the enhanced sensitivities from binning/smoothing by 3x3,
7x7, 15x15, and 31x31. 15x15 smoothing (heavy
line) is required to achieve S/N~5 at the single-ADU level. The SNRs
plotted here are for average images I/2 = (Q+ + Q- + U+ +
U-)/4). The ADUs on the horizontal axis are thus the same as those for a
single image (e.g., Q+), and the intensity units are for the same scale.
This schematic identifies 11 locations in the Pleiades nebula for detailed
photometric investigation. Descriptions are given in
Table \ref{Tab:phot_regions}. Regions 1-5 lie in the Merope nebula, and are
shown connected to the star of that name. Regions 6-11 select more general
parts of the larger nebulosity and are shown connected to Alcyone, which is
presumed to illuminate them for the discussion in
\S\ref{Sec:schmidt_results}. The dotted lines represent WISP and
Schmidt field boundaries. Stars brighter than V=7.0 are shown.
WISP1/WISP2 color ratio maps of 15x15-smoothed data. Top: Raw
WISP1/WISP2 ratio image, with no S/N consideration. The display range
is 0 <= ratio <= 2, where these limits correspond to the extremes of the
color scale bar at bottom. The color table was chosen to represent high ratio
values (blue colors) as blue and low values (red colors) as red.
Middle: The ratio S/N, displayed on a range of 5 (yellow) to 20
(blue). Bottom: The ratio map, masked to exclude all points with
S/N < 5.
The 40 Schmidt Bj mosaic fields, shown on the sky. Each field is 67.6' in
diameter. Also shown is the WISP field, as observed (solid line)
and as originally planned (dotted line). Since the Schmidt survey was
carried out first, its fields cover the original WISP area. The stars
plotted have magnitudes in the range 3 - 12 and were taken from the Guide
Star Catalog of the Hubble Space Telescope (STScI, 1992).
Schmidt Pleiades field 11, showing the central cluster region. This image has
been corrected for bias, flatfield, and scattered moonlight effects. Many
distracting bloom lines can be seen. These spill excess charge along columns
on the CCD, which is mounted west-up south-left on the telescope.
Transposition to a proper north-up east-left orientation makes these
artifacts appear horizontal.
The
same field shown in Fig. \ref{Fig:schmidt_bloom}, with saturation bloom lines
removed. The next tasks are photometric calibration and PSF subtraction.
Least-squares linear regression fit to standard star photometry. The 39
selected observations are plotted, with the secant of the zenith angle on the
horizontal axis and the attenuated instrumental magnitude on the vertical.
This latter quantity is defined as the measured instrumental magnitude -2.5
log(Ncounts) minus the star's actual B
magnitude. Observations of the same star are connected with dotted lines to
show agreement between individual stars and the general fit. The slope of this
fit is the atmospheric extinction coefficient ABj, while the
intercept is the true instrumental magnitude (Bj)0.
The Schmidt photometric S/N ratio as a function of nebulosity data
counts (ADU), also expressed in intensity units and B magnitudes per
square arcsecond. The rightmost curve is without binning. Curves leftward of
this have 2x2, 4x4, 8x8, and 16x16 binning. 8x8 binning (heavy curve) brings
the Schmidt data very close to the WISP image scale; at this level,
features on the order of 30 ADU (muB ~ 26.7 ~ 2.5% of sky)
have S/N ~ 5. These are the faintest reliable structures discernable by
the eye in the final mosaic. However the Schmidt counterparts of the faintest
discernable WISP features are >~ 10 times brighter, with S/N near
its maximum.
Fit of a r-1.67 power law to the radial profile of the
B=6.8 magnitude PSF calibrator star HD 15333. The fit technique is more
conservative than a simple least-squares method; over ranges of r
determined to have no bad data, a power law is scaled in brightness and DC
offset to match the profile without exceeding it. The fluctuating residuals
show the fit is not perfect. However the profile measurement technique (5th
percentile), while fairly robust, is not immune to all image artifacts, nor is
the ``true'' PSF necessarily a single, pure r-beta function.
Thus a conservative method was adopted to avoid possible over-removal of the
PSF.
The same field shown in Figs. \ref{Fig:schmidt_bloom} &
\ref{Fig:schmidt_nobloom}, with r-1.67 stellar aureoles
scaled to the catalog magnitude of each bright star and subtracted.
Same as Fig. \ref{Fig:schmidt_fps}, except the PSF removal method is much
simpler and less successful. Rather than use a complicated fitting process,
the actual measured radial profiles have been subtracted directly. As a result
all nebular radial brightness has also been removed, leaving behind
only angular structure. The non-smoothness of parts of measured profiles also
creates a number of ringlike artifacts.
Mosaic of all 40 Schmidt fields, showing extended nebulosity. The brightness
scale is linear but discontinuous. Pixel values have been ``folded'' by
dividing each by 10 as many times as necessary to bring it under 0.8 ADU
sec-1 (muBj = 23.1 mag arcsec-2). In
this way the bright inner structures can be seen that would otherwise be washed
out in order to display the faint outer structures with proper dynamic
resolution. The result is three linear scales spaced at logarithmic intervals,
with boundaries at muBj = 23.1 and 20.6 (ADC saturation here
is 32.767 ADU sec-1, equivalent to muBj = 19.1).
Several artifacts from scattered light inside the telescope optics are present,
most notably the long parallel streaks.
Same as Fig. \ref{Fig:smos_str}, but scattered light streaks have been removed.
Same as Fig. \ref{Fig:smos_nstr}, but on a logarithmic brightness scale. This
allows a more coherent view of structures which range over several magnitudes,
at the price of degraded dynamic resolution of the faint outer nebulosity.
Plots of the measured photometry in the test regions and that predicted by
Andriesse et al. (1977); values listed in Table \ref{Tab:phot_vs_model}.
1-sigma error bars are shown; for the Schmidt data, these are smaller than
the symbol used to plot the points.
IRAS 60µm emission in the Pleiades vicinity. Image center and scale
are the same as the Schmidt mosaic maps. The brightness scaling is
logarithmic, ranging from 3 to 300 MJy sr-1. A DC level of
0.87 MJy sr-1 was added prior to log scaling to prevent negative values
from being used; a small zero-point error is present in IRAS data at this
wavelength.
IRAS 100µm emission in the Pleiades vicinity, with the same scaling as
Fig. \ref{Fig:iras060_log}. A DC level of 6.89 MJy sr-1 was subtracted
prior to log scaling to produce a color range similar to the 60µm data.
Whether this is genuine background, diffuse Pleiades emission, or a zero-point
error is unknown; a combination of these seems likely.
DIRBE maps of far-infrared emission in the same field. The 60µm
(upper left), 100µm (upper right), 140µm (lower left), and 240µm (lower right) bands are shown. The
minimum pixel value of each was subtracted to remove background emission in a
simplistic way. Zodiacal emission is significant at 60µm without this
correction. All of the images are displayed on the same logarithmic
brightness scale, ranging from 1-300 MJy sr-1.
The color temperature computed from IRAS 60 & 100µm data with a
beta=1 dust emissivity law. No DC level corrections were applied to the
data for this calculation, since the proper levels are not known. The display
scale is linear, ranging from 20 to 45 K.
DIRBEFIR color ratio maps, showing 60µm / 100µm
(upper right), 100/140 (upper left), and 140/240 (bottom).
The background correction made for Fig. \ref{Fig:dirbe_log} was also used here.