D.2 IR Channel

The IR channel likewise has no bright-object constraints that are imposed due to instrument safety concerns. However, as discussed in Section 5.7.9 and Section 7.9.4, observers should bear in mind that there is a potential for image-persistence effects that could compromise observations made immediately following an exposure on a bright target. Such observations contain “afterglow” images at the location of the overexposed object, which gradually fades away over an interval of several hours.

Observers are able to be able to mitigate the effects of persistence from previous visits of other investigators. The Institute is minimizing the effects of persistence due to other observers as part of scheduling. We also provide estimates of the amount of persistence expected due to earlier exposures through MAST (as described in the WFC3 Data Handbook). These estimates allow the user to exclude regions of images that are affected by persistence and in some cases to subtract the persistence from an image.

However, observers are responsible for assuring that persistence within a visit (i.e. self-persistence) is not harmful to the science requirements of their own observations. Persistence is primarily a function of fluence, the total number of electrons accumulated by a pixel in an exposure. Very little persistence is observed when the fluence is less than about 35,000 electrons; thus, the best way to avoid self-persistence effects is to adopt an observing strategy that keeps fluence levels below that level. For example, an observing strategy involving a larger number of shorter exposures will result in less persistence than one with fewer longer exposures.

The WFC IR channel has a field of view seven times larger than NICMOS Camera 3 and also has higher sensitivity. This combination means that for WFC3/IR images it is often impossible to avoid saturating some portions of the image, especially since one is also constrained by the total volume of data that can be accumulated. In many cases this will not matter from a science perspective. For example, if your observing strategy involves small dithers and there are a few bright stars in the field, then persistence from earlier images will appear in the wings of the PSF of the bright star. Even if you have large dithers, you will probably not find persistence to be a major problem in the analysis, as long as you are willing to treat pixels with significant persistence as bad pixels.

Nevertheless, if you are planning a sequence of IR observations that may contain severely overexposed images, you may wish to estimate the degree of overexposure. Examples of types of observations for which this is commonly a problem are observations of regions in the Galactic plane or the LMC with large numbers of fairly bright stars, such as eta Car or 30 Dor. Significant amounts of persistence have also been observed in certain instances from globular clusters and nearby elliptical galaxies.

An IR observer might expect that the Two Micron All Sky Survey (2MASS) would be the appropriate catalog for examining the frequency of WFC3/IR saturation, but in fact the 2MASS catalog is generally not deep enough for this purpose. Although the depth of the survey varies across the sky, the faint limit is typically near 15th mag in J, H, and Ks (the formal Level 1 requirements on limiting magnitudes of the 2MASS catalog are J = 15.8, H = 15.1, and Ks = 14.3). Stars near this faint limit saturate the WFC3/IR detector in a relatively short time in both the medium and broad band filters. The STScI Guide Star Catalog (GSC), currently at version 2.3, generally goes much fainter (down to 22nd mag in V), but the extrapolation from the optical into the infrared depends on the accuracy of the spectral type implied by the optical colors and the assumed extinction (a large source of systematic errors along sightlines of high extinction).

As described in Chapter 9, the WFC3 Exposure Time Calculator (ETC) can be used to estimate the count rate in the central pixel for a given astronomical source. (Note that it is the rate for the central pixel and not the total count rate in an aperture that matters.) However, as a rough guideline, below we present tables of the count rates for two cases: a “hot star” with spectral type O3V, Teff = 45,000 K, [Fe/H] = 0, and log g = 4.0; and a “cool star” with spectral type M5V, Teff = 3500 K, [Fe/H] = 0, and log g = 5.0.

Tables D.1, D.2, D.3, and D.4 give the results for the cases where one normalizes to Johnson J, K, V, and Bessel H, as stated in the table captions. In each case a magnitude of 15 in the respective bandpass is assumed. The count rates are given in e^-/s for the central pixel of a star centered in a WFC3/IR pixel. These tables give the most reliable results when normalizing to a ground-based bandpass that overlaps with the WFC3 bandpass, regardless of the assumed spectral energy distribution. However, when normalizing to Johnson V, one must know the underlying spectral energy distribution to high accuracy in order to predict the count rate in the WFC3/IR bandpasses.

The Bright Object Tool (BOT) in the Astronomer’s Proposal Tool (APT) can provide a list of saturated objects for a potential WFC3/IR observation, given a Phase II proposal. Because the 2MASS survey is sufficiently deep for objects that would severely oversaturate the detector (by more than a factor of 100), the BOT uses 2MASS data where they are available, and the GSC2 where no 2MASS data are available. To use this feature, display a visit with the Aladin tool, loading the DSS image. Then, click on the BOT button in the main tool bar, which will bring up the Bright Object Tool. At this point you press the “Update 2MASS Display” button and stars likely to cause persistence problems are indicated in the Aladin Window, and can be looked at individually there, or shown as a list with the “Get Details...” button. The tool lists stars with different levels of saturation, computed using the time between pixel resets, which can be significantly longer than the exposure time for subarray exposures (WFC3 ISR 2011-09). One should probably be more concerned with the numbers of stars listed as saturated and their locations than with the crude categorization by saturation level, but keep in mind that extremely bright stars saturate not only the central pixel, but also pixels in the PSF wings.

One should be aware that neither the BOT nor other estimates based on star catalogs provide good information about persistence due to diffuse sources, e.g., a bright and extended galaxy nucleus. For that one can use the ETC, if one has an estimate of the surface brightness of the source.

The best strategy to reduce the effects of persistence is to reduce the exposure times. Two exposures of 350 seconds (SPARS25 with 15 full-frame readouts) will result in less persistence than one exposure of 700 seconds (SPARS50 with 15 full-frame readouts). If one cannot get short enough exposures (or enough readouts) with the full array and the science target does not fill all of the field, one should consider using a sub-array (see Section 7.7.4).

A second possibility if the program involves images that go to different depths is to place shallower exposures at the beginning of an orbit and deeper exposures at the end of an orbit. For example, if one is planning three 500 second exposures with a narrow band filter and one 500 second exposure through a broad band filter, one should almost always put the broad band filter last.

If neither of these strategies solves the problem, then one needs to very carefully examine dither patterns, and decide what is most important to the observing program. Sometimes, if persistence is isolated to one portion of the image, a line dither in the appropriate direction can limit the damage effects of persistence. Small dither patterns tend to keep bad pixels confined to small regions within the point spread function of the stars that cause persistence. On the other hand, these dither patterns do not cover enough area to step over IR blobs (Section 7.9.6). In some cases, one simply needs to decide which of these two problems it is more important to mitigate as part of planning an observation. An observer who cannot determine the best strategy for a program should consult the Contact Scientist of that program for advice.

Table D.1: Count Rates (e^–/s) for source with J=15 renormalized to Johnson/J.

¹ Cool star has spectral type M5V; hot star has spectral type O3V. See text for full definition.

Table D.2: Count Rates (e^–/s) for source with H=15 renormalized to Bessell/H.

¹ Cool star has spectral type M5V; hot star has spectral type O3V. See text for full definition.

Table D.3: Count Rates (e^–/s) for source with K=15 renormalized to Johnson/K.

¹Cool star has spectral type M5V; hot star has spectral type O3V. See text for full definition.

Table D.4: Count Rates (e^–/s) for source with V=15 renormalized to Johnson/V.

¹Cool star has spectral type M5V; hot star has spectral type O3V. See text for full definition.