Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/nicmos/documents/isrs/isr_99_011.pdf Дата изменения: Tue Feb 12 14:43:37 2002 Дата индексирования: Sun Dec 23 01:25:13 2007 Кодировка: Поисковые слова: earth

Instrument Science Report NICMOS 99-011

Long-term trends in the NICMOS Camera 2 obscuration pattern and aberrations
John Krist November 15, 1999
ABSTRACT Phase retrieval measurements of well-exposed NICMOS Camera 2 star images show that long-term changes occurred in coma and the cold mask position over the 1997-1998 lifespan of the instrument. These caused time-dependent variations in the PSF diffraction pattern, notably the shape of the Airy rings and the positions of bands in the diffraction spikes. These effects are evident in both direct and coronagraphic images when two observed PSFs obtained at different times are subtracted from each other. The derived trends have been implemented in Version 5 of Tiny Tim. PSF monitoring programs for NICMOS after the cryocooler is in place should include similar exposures so that any further cold mask shifting can be tracked.

Introduction
The NICMOS dewar anomaly caused significant changes to the optical alignment of the instrument. The most obvious were variations in focus and plate scale as the dewar expanded and then contracted. These effects were detected soon after the instrument was activated on-orbit. Later, short-term changes in the alignments of the cold masks were identified from analysis of highly-exposed star images by Krist et al. (1998; hereafter K98). A subsequent study of the NICMOS focus monitoring data by Suchkov et al. (1999) showed that coma and astigmatism were varying almost linearly on long (multimonth) timescales. All of these variations lead to an unstable PSF, especially when evaluated over periods of months. This significantly complicated NICMOS PSF subtraction, which was often required for high-contrast imaging programs. It is common practice to subtract one observed star image from another in order to detect any faint objects that lie within the wings of the PSF. However, if there are mismatches between the two images caused by time-dependent variations in the optical system, then significant residuals may occur at levels that obscure these objects.

1


Such problems were the basis for the K98 report. Using NICMOS camera 2, a SNAPSHOT survey of low-mass companions to nearby stars was undertaken (GO program 7420 in cycle 7 and 7894 in cycle 7.5). Because any such companions would be significantly fainter than the primary stars, the images had to be well exposed to provide adequate signal for detection. This typically resulted in nearly overexposing the primary star to the point that diffraction structure could be seen throughout the entire 20" x 20" field. Subtraction of one star observed in the program from another revealed residuals in the diffraction structure indicative of obscuration pattern changes. These were traced to the apparently random motion of the cold mask between visits (termed wiggle). These errors produced unexpected asymmetries in the point spread functions (PSFs), including elliptical diffraction rings and asymmetric banding patterns in the diffraction spikes. K98 reported that camera 2 star images taken hours to months apart showed that the cold mask was varying randomly in position by up to 0.5% of the pupil radius on short (orbital) timescales. Presented here is an expanded analysis of the SNAPSHOT data, with specific emphasis on long-term changes to the NICMOS camera 2 PSF.

Long-Term PSF Variations
K98 did not detect any noticeable, long-term trend in the cold mask position, but they only looked at data taken early in the survey. The SNAPSHOT program would eventually have over 100 visits scattered between July 1997 and November 1998. Comparisons of star images taken early in the program with those from towards the end showed larger changes than those expected from mask wiggle. It was obvious that some larger, longer-term variations in cold mask position had occurred. Figure 1 shows the result of subtracting PSFs taken early and late in the program. The residuals in the diffraction rings and spikes are characteristic of cold mask shifting. Notably, the positions of the bands in the diffraction spikes do not match. These bands are dependent on the distance between the cold mask spiders and the telescope's, as projected into the pupil plane. Figure 2 shows the banding patterns from data taken over the NICMOS lifespan. The bands moved away from the PSF center with an approximately linear relation with time.

PSF Analysis
Phase retrieval analysis of star images is required to quantify the changes in the optical system, as described by Krist & Burrows (1995). A version of the Krist & Burrows software was used for NICMOS focus monitoring, as described by Suchkov et al. (1998).

2


Ob s e r v e d P S F

Pup il Pa t t e r n

19 97a - 1 997b

19 98 - 19 97

Figure 1. Comparison of good and bad NICMOS PSF subtractions. An F222M star image from the NICMOS snapshot program is shown in the upper left (full 20" x 20" NICMOS camera 2 field). Its corresponding pupil pattern is shown to the right. The thick spiders, rectangular tabs, and larger central obscuration are the NICMOS cold mask; the thinner spiders, circular pads, and smaller central obscuration are HST telescope obscurations. In the bottom row are PSF subtractions using two star images close together in time (1997a ­ 1997b) and two far apart (1998 ­ 1997). Note the increased residuals in the image on the right caused by mask shifting. The PSF images are logarithmically scaled.

3


Figure 2. Above are extracted and rotated images of one NICMOS camera 2 diagonal diffraction spike. These images span from July 1997 (left) to October 1998 (right). The PSF core is at the center of each spike. Above and below the core are bands caused by the HST and NICMOS cold mask spiders. Notice that the band positions with respect to the dotted horizontal reference lines change with time as the cold mask shifts.

The long-term variations in coma and astigmatism reported by Suchkov et al. (1999) were derived from the focus monitoring data. These consisted of images of an open star cluster taken at different focus positions. Unfortunately, the exposures had a limited range of defocus and rather low signal levels. While adequate for measuring low-order aberrations (focus, coma, and astigmatism), the data did not have the high-frequency spatial structure in the wings that was necessary to derive the obscuration pattern. Thus, the version of the phase retrieval software used on this data did not fit for obscuration changes. The camera 2 images from the companions survey had copious signal in the wings, and were the only data available that are suitable for deriving obscuration pattern changes over time. These images were processed using a version of the phase retrieval software modified to fit for cold mask position, rotation, and size (spider width, outer radius, and

4


inner radius were each fit separat of the NICMOS instrument were bandpass is fairly narrow and the longer wavelengths (though it is unsaturated images were chosen, extracted for analysis.

ely). A set of 27 star images that sampled the lifespan selected. F222M images were used, since that filter's PSF is more sensitive to obscuration pattern changes at less sensitive to aberration changes). Well-exposed, and the central 135 x 135 pixels (10.3" x 10.3") were

Each image was individually analyzed. In addition to cold mask changes, the following parameters were included : focus, coma, astigmatism, spherical aberration, clover, pixel size, uniform background level, and PSF centering. Clover (trefoil) exists in the HST optics and was expected to stay constant with time, as was spherical aberration. The blueprint-specified dimensions for the cold mask were used to define the initial mask pattern, assuming a beam diameter at the mask for an ideally aligned system.

0.015
RMS Error (µm)

0° Astigmatism
RMS Error (µm)

0.010 0.005 0.000 -0.005 -0.010 100 200 300 400 500 600 700 Days Since 1-Jan-1997

0.010 0.008 0.006 0.004 0.002

45° Astigmatism

0.000 -0.002 100 200 300 400 500 600 700 Days Since 1-Jan-1997

X Coma
0.030
RMS Error (µm)

Y Coma
RMS Error (µm)

0.025 0.020 0.015 0.010 100 200 300 400 500 600 700 Days Since 1-Jan-1997

-0.010 -0.015 -0.020 -0.025 -0.030 -0.035 -0.040 100 200 300 400 500 600 700 Days Since 1-Jan-1997

Figure 3. Astigmatism and coma measurements over time from phase retrieval analysis of NICMOS 2 images. Aberrations are in the detector coordinate system. The solid lines represent the trends used by Tiny Tim.

5


Results
The phase retrieval measurements (Figures 3-5) show significant trends with respect to time for coma and the cold mask position; there are weaker, compensating trends for the pixel scaling and cold mask size. The average position of the NICMOS camera 2 cold mask changed by 1.4% of the pupil radius over 500 days, almost exclusively in the vertical direction as viewed in the detector coordinate space. It had an average offset of 12% of the pupil radius. The level of coma in the vertical direction increased from about ­0.015 µm to beyond ­0.04 µm RMS, though the coma in the horizontal direction remained essentially constant at 0.022 µm, with some random scatter. Monitoring of the NICMOS coronagraph spot position showed that it also followed a generally vertical path.

Mask X
11.5 11.0 10.5 10.0 100 200 300 400 500 600 700 Days Since 1-Jan-1997 7.520·10-2 7.510·10-2 Arcsec 7.500·10-2 7.490·10-2 7.480·10-2 7.470·10-2 100 200 300 400 500 600 700 Days Since 1-Jan-1997 8 7 % Rpupil 6 5

Mask Y

% Rpupil

4 100 200 300 400 500 600 700 Days Since 1-Jan-1997

Pixel Scale
98.10 98.00 % Rpupil 97.90 97.80 97.70

Mask Radius

97.60 100 200 300 400 500 600 700 Days Since 1-Jan-1997

Figure 4. TOP : Camera 2 cold mask X and Y positions with respect to time as a percentage of the pupil radius. Coordinates are in the detector X-Y frame. BOTTOM : Pixel scale and cold mask outer radius as a function of time. These compensate each other, resulting in a constant Airy ring radius (neglecting cold mask shifting effects). The solid lines represent the trends used by Tiny Tim.

6


There is no definite trend in astigmatism with time, in contrast to the findings of Suchkov et al. (1999). As noted previously, Suchkov et al. did not fit for the position of the cold mask, and they also forced y-astigmatism to remain constant. Their astigmatism change may have thus been compensating for cold mask shifting. Clover and spherical aberration remained essentially constant over the timespan of the data. The average focus was off by the equivalent of 4 µm of breathing. The pixel scale and mask outer radius both changed by about 0.2%. The scale decreased while the outer radius increased. For a given pixel scale, an increase in the mask outer radius would cause the PSF to shrink. Conversely, for a fixed outer radius, a decrease in pixel scale would cause the PSF to expand (relative to the detector array size). When these two effects are combined, the net result is that the distance to a given Airy ring remained the same throughout the timespan of the data, neglecting changes caused by mask shifting. Why the measurements show a trend in scale and mask radius is unknown. Perhaps the software needed to compensate for some mask property that was not in the model (e.g. perhaps the mask is not coplanar to the pupil) and thus was forced to modify some other parameter instead.

X Clover
-0.0040 RMS Error (µm) -0.0050 -0.0060 -0.0070 100 200 300 400 500 600 700 Days Since 1-Jan-1997 RMS Error (µm)

Y Clover
0.0125 0.0120 0.0115 0.0110 0.0105 0.0100 0.0095 0.0090 100 200 300 400 500 600 700 Days Since 1-Jan-1997

Focus
0.02 RMS Error (µm) 0.00 -0.02 -0.04 -0.06 100 200 300 400 500 600 700 Days Since 1-Jan-1997 RMS Error (µm) -0.008 -0.010 -0.012

Spherical

-0.014 100 200 300 400 500 600 700 Days Since 1-Jan-1997

Figure 5. Measured clover, focus, and spherical aberration values. Note that the scale of focus is in microns of wavefront error, not HST secondary mirror (or breathing).

7


Measurements of the separations of stars in the focus monitoring data were used to derive plate scales for the NICMOS cameras, which were found to vary with time as the dewar expanded and contracted. The scale that defines the diameter of a PSF diffraction ring may be different from that which defines the separation on the detector of two objects at different field positions (ie. one scale is defined by diffractive optics while the other is geometrical). This may be a result of the complexity of the NICMOS system. The mask spider width, outer radius, and inner radius were all greater than the input estimates. This may be due to a different beam diameter at the mask than was assumed for an ideal system. The shifting cold subtraction of one mask positions. residuals resemble mask also affected coronagraphic observations. Figure 6 shows the simulated coronagraphic PSF from another, each with different cold Members of the NICMOS Science Team have indicated that the those sometimes seen in actual subtractions.

Figure 6. Subtraction of a simulated coronagraphic PSF from another one having a cold mask shifted by an additional 0.5% of the pupil radius.

8


Recommendations
A focus monitoring program will likely be undertaken after the installation of the NICMOS cryocooler. The program should retain the method of the previous one (imaging of stars at various PAM focus positions), but with some modifications. The through-focus sweep should use short wavelength filters, where the PSF is more sensitive to focus and other aberrations. Phase retrieval analysis of WFPC2 F555W and F814W PSF has shown that the shorter wavelength data provides better results, even if it might be undersampled. This may be due to the sensitivity of astigmatism to focus (the axis of ellipticity caused by astigmatism changes 90° through focus). An exception is camera 3, where the subpixel response function and undersampled core could significantly affect the measurements, so a longer-wavelength filter is more appropriate. The phase retrieval software generates monochromatic models, so it is important to use medium or narrow-band filters where polychromatic effects are limited. The signal levels in many of the images from the previous program left something to be desired. It is therefore suggested that another target be selected (a less dense, brighter cluster or isolated star). If necessary, fewer PAM positions could be used if longer exposures are needed. The images obtained in the survey program have proven to be a source of considerable insight to the NICMOS optical system. The focus monitoring data are not very sensitive to obscuration pattern changes, while the survey images are. Thus, a new focus monitoring program should include deep exposures of an isolated star, from which the cold mask position may be derived. Then, this new position may be used in the phase retrieval software to produce a better model for fitting the focus sweep data. A longwavelength, medium or narrow-band filter should be used. These data should be obtained in each camera.

References
Krist, J. and Burrows, C. (1995) Applied Optics, 34, 4951 Krist, J., Golimowski, D., Schroeder, D., and Henry, T., (1998) P.A.S.P., 110, 1046 Suchkov, A. & Galas, G. (1999) "NICMOS optical aberrations : coma and astigmatism." NICMOS ISR 99-003 (STScI).

9