NICMOS Data Handbook

5.4.1 Impact of Instrumental Effects on PSF Subtraction

Accurate PSF subtraction is a prime concern for an observer wishing to study faint features around bright objects. Typical situations are: a host galaxy harboring a bright quasar; circumstellar nebulosity around a bright star; faint companions of a bright star, etc. PSF subtraction for NICMOS data can be very effective, especially for images from cameras 1 and 2, thanks to a few important features of this instrument:

•	Cameras 1 and 2 provide images which are diffraction limited, and well sampled (see Figure 5.4 and Figure 5.5).

•	The MULTIACCUM mode automatically provides sub-exposures with different exposure times, and calnica processing will use only unsaturated data values when constructing the calibrated image.

•	For a saturated bright central point source, there is no blooming along the detector array columns as there is with a CCD.

One way to get a high quality PSF for subtraction is to measure an isolated bright unsaturated star in the same image or to construct a composite PSF using good stars in the image. This can be accomplished using the IRAF digiphot package as described in A User’s Guide to Stellar CCD photometry with IRAF¹. A PSF obtained from the same image ensures that effects of telescope focus and pointing jitter on the image quality are properly taken into account. In particular, this approach will take care of the breathing effect: variations of focus position due to thermally induced mechanical displacements in the HST optical path. One disadvantage of this method, however, is that it is difficult to account for positional variations in the PSF over the field of view of the NICMOS cameras (see discussion in Section 5.2.3, and also in Suchkov & Galas 1998, NICMOS ISR 98-005). Although these variations do not have a large affect on NICMOS aperture photometry, they can be large enough to cause problems for PSF matching and subtraction. If no suitable star can be found in the image, one can resort to a synthetic PSF computed using TinyTim.

Figure 5.4: NIC1 Image with PSF Radial Profile

Figure 5.5: NIC2 Image with Radial Profile

5.4.1

Impact of Instrumental Effects on PSF Subtraction

There are a number of factors which affect the NICMOS PSF and thus can impact the results of image analysis relying on PSF subtraction. The factors are: focus variations due to OTA breathing, cold mask irregular motion (“wiggling”) on an orbital time scale, PSF color dependence.

The limits to the accuracy of PSF subtraction imposed by these factors have been assessed using model PSFs generated by TinyTim software and synphot generated blackbody spectra. The results for Camera 2 are given in Table 5.5 below. The mean and standard deviation of the PSF subtraction residuals are given a percentage of the nominal PSF pixel values. The standard deviation can be interpreted as the average relative spatial noise expressed as a percent of the PSF pixel value at any given distance from the PSF center. The mean represents a systemic component in the PSF residuals.

Table 5.5: Effect of focus variation, cold mask wiggling, and PSF color dependence on PSF subtraction.


effect of:	mean sigma F187W	mean sigma F187N	mean sigma F160W	mean sigma F187N	mean sigma F110W	mean sigma F187W
focus variation	0.62% 6.81%	1.86% 17.29%
cold mask wiggling			0.10% 14.38%	3.15% 37.67%
color dependence (8000K-3000K)					23.50% 8.91%	0.58% 1.57%
color dependence (8000K-6000K)					-1.52% 1.32%
color dependence (6000K-3000K)					25.31% 7.32%

The effects of both cold mask wiggling and focus breathing introduce errors in a PSF-subtracted image, well above 20% of the PSF signal in a narrow band filter, with a spatial scale of a few pixels. The main effect of PSF color dependence is adding a systemic component to the PSF subtracted image. The effect is quite large if the color of the PSF used for subtraction is very different from the image PSF color. It is pronounced in blue, wide filters, like F110W, while in filters like F187W or redder/narrower it is essentially negligible. The best way to cope with this effect is to use PSFs with colors well matched to the sources.

PSF subtraction with NICMOS Camera 3 is extremely difficult due to the large pixel scale which severely undersamples the point spread function, and because of the strong intrapixel sensitivity variations (see Section 5.2.3) which affect the structure of point source images.

Figure 5.6: Effect of cold mask wiggling on PSF subtraction in broad band and narrow band filters.

Left panels: normalized counts (log scale) in the central column of the two TinyTim PSFs with 0.005 difference of the cold mask offset (thin lines). The thick line is the difference between the two. Right panels: absolute value of counts difference as a percent of the counts for the PSF at the “nominal” cold mask offset.

Figure 5.7: Illustration of PSF color dependence in different filters.

In F110W, the 8000 K blackbody spectrum has a very different slope from that of the 3000 K blackbody. Therefore the PSF resulting from images of point sources with these temperatures will be very different from one another. The F187W filter, however, samples the Rayleigh-Jeans regime for both temperatures, resulting in spectra with about the same slope, so the difference between the 8000 K PSF and 3000 K PSF will be very small.

Figure 5.8: Effect of PSF color dependence on PSF subtraction in different filters.

Left panels: normalized counts (log scale) in the central column of TinyTim 8000 K and 3000 K blackbody PSFs (thin lines). The thick line is the difference between the two. Right panels: absolute value of counts difference as a percent of counts for the 8000 K PSF.

This and other relevant IRAF documents can be obtained from the IRAF Web site at: http://iraf.noao.edu/.