Space Telescope Science Institute
Table of Contents Previous Next Index Print

NICMOS Data Handbook > Chapter 4: Anomalies and Error Sources > 4.7 Pixel Defects and Bad Imaging Regions

4.7 Pixel Defects and Bad Imaging Regions
Each NICMOS array has a small number of bad pixels which may be either cold (i.e., very low or zero response) or hot (i.e., with very high or erratic dark current). Static bad pixel masks are used to set error flags (DQ=32) for these pixels in the DQ arrays during the MASKCORR step in calnica processing. The statistics on the cold/hot pixels present in each of the NICMOS camera are presented in Table 4.1. The cold pixels contain pixels that do not respond to incoming photons (“dead pixels”), and the hot pixels are those with increased dark current. Pixels with low responsivity or affected by "grot" (see below) are listed separately. These pixels get a flag value=16 in the MASKCORR step.
In addition to the bad pixels which were already known from ground-based testing, more pixels have shown low measured quantum efficiency in orbit. These are known as grot and they appear as small areas of reduced sensitivity, most likely due to flecks of anti-reflective paint which were scraped off the optical baffles between the dewars as they were forced against each other by the solid nitrogen cryogen expansion. The largest example, known as the battleship, is found in camera 1 and affects approximately 35 pixels. However, most grot seems to affect single pixels. Scrape tests conducted at Ball Aerospace shortly after NICMOS was launched reveal that the paint flecks can range in size from 25μm to greater than 100μm. Since NICMOS pixels are 40μm on a side, this means that there is potential for grot smaller than 1 pixel.
There was some concern that the warm-up and subsequent cool-down of NICMOS with the associated mechanical motions could produce an increased number of these particles. However, apart from one additional chunk of grot in the lower right corner of camera 1, no new contaminants were apparent in the data obtained during SMOV3b in 2002. The number of pixels with significantly degraded responsivity is roughly the same now as during Cycle 7.
Table 4.1: NICMOS Cosmetics (Cycle 11 and beyond, for masks produced in March 2009)
Dithering is the best way to guard against the effects of bad pixels and grot. If you have multiple dither positions, you can use the static masks to exclude bad pixels when combining the images (e.g., with calnicb, imcombine, or MultiDrizzle with the STSDAS dither.drizzle routines). Or, you may interpolate over masked pixels using, e.g., IRAF fixpix. For further description of the mask files, see (NICMOS ISR 99-008). For further updates please refer to the NICMOS Web page at:
Not all pixels affected by grot are unrecoverably ruined. For pixels which are only partially obscured, the effects of grot may flat field out. You may wish to compare your flat-fielded images with the grot flagged pixels to see which pixels have flattened well and which are potentially suspect.
Note that Cycle 7/7N data that were retrieved prior to April 2009 do not have the grot pixels flagged in the reference file given by the maskfile keyword. For these data, grot masks can be retrieved from the NICMOS Web page:
Alternatively, the latest mask reference files can be retrieved from the cdbs or NICMOS Web page, followed by an update of the maskfile keyword in the primary header of your data to point the these new reference files. Finally, you can retrieve your data again using the OTFR to ensure that all the latest update are included in your data.
In each NICMOS camera, there is one row or column which often deviates from others nearby. In NIC1 this is row 128; in NIC2 it is column 128, and in NIC3 it is column 129. Adjacent rows (or columns) are sometimes affected to a lesser degree. It is believed that this “photometrically challenged column” may be related to uncertainties in detector shading corrections. The affected column/row contains the first pixels read out in each quadrant. Since the shading function is very steep and highly nonlinear during this part of the readout, it is the part of the array most sensitive to changes in the detector environment. Thus it is thought that the pixels in these rows or columns are not any less sensitive, but that they have just had an incorrect bias subtracted from them. The result is a row or column of pixels that is either under or over corrected for the shading by the dark reference file.
For some data, it is possible to fit a function to the affected column and add it back in as a delta bias, but the data in the affected pixels tends to be rather noisy as well. Given dithered data, you may be better off just treating the affected pixels as “bad” and using spatial dithers to recover that information (just as you would do for other bad pixels). Adding the middle column/row to the DQ array flags of the calibrated image (*_cal.fits), or to the MASKFILE before calnica processing, will mark them for exclusion when dithered images are subsequently combined by calnicb or Multidrizzle.
NICMOS camera 2 includes a coronagraphic hole which can be used to partially obscure bright sources. In non-coronagraphic observations, however, the hole is still present, and will partially or completely blank out any objects which fall near it. For thermal IR images, the hole can appear as a positive “bump” due to excess background emission shining through the hole from warmer structures behind it. The position of the hole in detector coordinates has drifted with time, changing rapidly shortly after launch when the cryogen distortion of the dewar was varying quickly, then slowly stabilizing.
The hole should usually be treated as “bad pixels” when combining dithered NIC2 images, either by masking or interpolation. Masking a circular region with radius ~7 pixels should eliminate the effects of the hole for most data sets. For coronagraphic imaging science, see the discussion on reducing coronagraphic data in Chapter 5.
There are two effects which are sometimes described as vignetting in the NICMOS cameras.
First, the anomalous expansion of the NICMOS dewar moved the cameras in such a way that they image the edge of the Field Divider Assembly (FDA) at the bottom of all 3 cameras. The FDA is a fold mirror that sends the light from the foreoptics down each camera channel and through the filters and dewar window to the cameras. The bottom edge of the field of view for all 3 cameras looks at a black mask on the FDA. This means that there is less signal from the telescope along that edge, and thus the throughput is decreased. In NIC3 ~15 rows along the bottom edge are severely affected, and the throughput there ramps down steeply to about 30% of the mean for the rest of the array. For NIC2 it has about the same extent, but is only reduced to 95% of the mean of the rest of the array. The effect for NIC1 is even smaller. This hard edge moves around slightly with time, and thus does not flat field away perfectly, especially for camera 3 where the gradient is steep. Because the vignetting is small for cameras 1 and 2, the resulting flat fielding uncertainty is generally negligible. (The coronagraphic hole in camera 2 is also on the FDA, so its motions correlate with those of the vignetted region.)
In addition, the NICMOS cameras also image part of the forward bulkhead at the entrance aperture. This warm metal surface emits radiation in the longer wavelength bands. This emission extends over about 50 rows in NIC3 and about 20 rows in NIC2. Moving the Field Offset Mechanism (FOM) causes NICMOS to see a different part of the sky in the HST focal plane, and essentially to look farther away from the bulkhead causing the emission. This FOM repositioning was automatically implemented for all NIC3 observations from January 1998 and onward, so data taken after that time are unaffected. Data in the longer wavelength filters taken prior to that time may have an elevated gradient in the sky background. The FOM is moved forward of the FDA in the optical path, so the repositioning has no effect on the FDA vignetting described above.
Another consequence of the vignetting is its impact on image shape and geometric distortion for NIC3 images. This concerns all data taken out of focus which is the case for all NIC3 data except data taken during the two refocus campaigns in January and June 1998 (no refocus campaigns have been conducted since the installation of NCS in 2002). The vignetting introduces some additional image distortion when the detector is out of focus, especially over the first 50 rows of the detector (see e.g., Cox et al. Instrument Science Report OSG-CAL-97-07). Users should regard astrometry and PSF shape in that portion of NIC3 images with some caution.
For NIC3, it may be prudent to discard data from the bottom 10 to 15 rows. At the very least, photometry in that region should be treated with suspicion. Data taken before the FOM repositioning in January 1998 may have elevated background gradients for NIC3.

Table of Contents Previous Next Index Print