The MultiDrizzle Handbook


2.5 More Detailed Considerations

2.5.1 Data Quality Issues involved in Dithering

The primary considerations in designing a dithered observational program are cosmic rays, hot pixels, spatial sampling and signal-to-noise. Finding the optimal strategy to deal with these issues is not always straightforward. Careful consideration must be given to the impact of breaking a long observation into multiple exposures, particularly in terms of increasing the overall read-out noise and reducing the amount of science exposure time due to observational overheads. The optimal strategy chosen will ultimately depend upon carefully weighing all these issues against one another, and also against the scientific questions involved such as: is the underlying structure totally unknown, is spatial resolution of paramount importance or is photometric accuracy the most crucial aspect?

2.5.2 How Many Dither Positions - 2, 3, 4 or more?

If integer-pixel dithers are all that is required, specifically to ameliorate the effect of hot pixels, then 2 or 3 different locations should be sufficient to guarantee that sources falling on hot pixels are not completely unrecoverable. The remainder of this discussion focuses on sub-pixel dithering, including the strategies and issues involved. The best choice for the number of sub-pixel dithers depends on the amount of time available and the goals of the project. Dithering requires a noticeable amount of spacecraft overhead, with each dither offset typically adding ~2-3 minutes of overhead to the total observing plan.

Figure 2.1: Sampling of the WFC3 IR Detector on the Sky

The sampling of the WFC3 IR detector on the sky.
Figure 2.2: Sampling for a 2-point Dither

The sampling produced by introducing a two-point dither using the WFC3 IR detector.
Figure 2.3: Sampling for a 4-point Dither

A four-point dither.
Figure 2.4: Sampling for a 3-point Dither

A three-point dither applied to the WFC3 IR detector.
Figure 2.5: A 6-point Dither

A six-point dither.

2.5.3 Data with Inaccurate Offsets in Position or Roll Angle

After the observations have executed, the pointing and orientation of the telescope can be determined directly by using cross-correlation techniques, as well as through examination of the jitter files that can be requested from the HST archive as part of the data products. The HST Data Handbook contains further details on the jitter file and other data products, and how to extract their information. The most recent version can be found at the following Web page:

For most programs, it is sufficient to determine the translational shift between images - the chance of a spurious roll angle offset is relatively small, and many long programs have now been performed without experiencing the roll offsets originally seen in the HDF. However, the drizzle software is capable of combining the shifts in rotation as well as position.

2.5.4 How many Images to Obtain at each Dither Location

It is generally possible to successfully remove cosmic rays using only a single image at each dither location, i.e, "singly-dithered" images, using the drizzle software that is incorporated with the dither package in STSDAS. If sub-pixel dithering is desired for small programs (less than about one orbit per target), or programs where reduction of read-noise is critical (e.g., narrow-band imaging of extremely faint sources), then the best approach is likely to involve obtaining only one image at each dither location. For larger programs, however, or when read-noise is not a serious issue, the user can choose between the slightly improved sampling of a larger number of independent dither pointings, or the relative simplicity and lower overhead of multiple exposures at a given dither pointing.

2.5.5 Specific Instrument Related Issues WFPC2

In addition to increasing information on the smallest spatial scales, dithering can be used to reduce the effect of flat-field errors in very deep images. Large dithers (tens of pixels) were used in the HDF for this purpose. Furthermore, dithers greater than one or two pixels can be used effectively to eliminate chip defects such as hot-pixels and bad columns.


The Effect of WFPC2 Geometric Distortion on Dither Offsets

The pixels near the edge of the CCD differ in size on the sky from those near the center. Thus a shift of (10,10) pixels at (400, 400) corresponds to a shift of about (10.2, 10.2) pixels at (700, 700). The default dither-line spacing produces a shift of (2.5, 2.5) WF pixels and (5.5, 5.5) PC pixels. Therefore, over nearly the entire field of view the difference in offset - even on the PC - is less than 0.1 pixels, and the shift will be essentially optimal across the whole field. However, the standard dither-box spacing offsets the telescope by as much as 0.75 arcseconds or 15.5 PC pixels. This means that at (700, 700) the shift differs from that at the center by ~0.3 pixels in and . While the drizzle software removes this geometric offset, it cannot change the fact that the sampling will not be optimal across the entire field of view.

The dither-box defaults were chosen to avoid repeating the placement of objects on the same columns (to reduce the effects of bad columns). However, if one is willing to live with the possibility that a given position of interest may fall twice on one of the several bad columns per chip, then one can use smaller offsets to produce a box that is more nearly perfect across the entire chip, for instance a square box with side of 2.5 WF pixels (equivalently 5.5 PC pixels).


The Exact Relationship Between POS TARGs and WFPC2 CCD Rows and Columns

For WFPC2 an additional complication is introduced by the fact that the four chips are not precisely aligned with one another, but possess small rotational offsets (<0.5 degree) from their nominal alignments. Thus, the POSTARG axes run exactly along the CCD rows and columns on whichever aperture is specified for the observations. For example, if aperture WF3 is specified, the POSTARG axes will run exactly along the rows and columns on WF3, and will run only approximately along the rows and columns of the other CCDs. Note that if WFALL is specified, then the rotation for WF3 is used since WF3 is the reference chip for the WFALL aperture.

The CCD rotation misalignments lead to errors when attempting to dither by certain pixel amounts. For small dithers (<0.3 arcseconds) the rotational offsets between the CCDs are unimportant, as they imply pixel registration errors less than 3 milliarcseconds, which is roughly the nominal pointing and guiding stability for HST. But such small dithers do not allow integral pixel stepping simultaneously on the PC as well as the WF chips. A dither of 0.5 arcseconds (5 WFC pixels or 11 PC pixels) gives near-integral stepping on the PC and the WF chips, though the CCD rotations will then introduce registration errors up to 5 mas. An offset of (1.993, 0.000) arcseconds in on WF3 would cause spurious motion in of 0.17 pixel on WF4, due to the rotation.

Two basic types of dither patterns are defined for WFPC2, and are implemented in the APT software that is used to process Phase II observing programs. These patterns can also be used with non-default spacings when necessitated by very specific types of observations, although in general we recommend that observers use the default spacings which are optimized for a wide variety of scientific programs. ACS

For ACS, an important issue to consider in designing a dither strategy is its relatively large distortion: up to 8% across the WFC camera. Moreover, the projections of the detector pixels on the sky correspond to parallelograms with interior angles that differ from 90o by up to 5 degrees, depending upon the location of the pixel on the detector. The differential distortion across the chip means that shifts of more than a few pixels produce noticeably different sub-pixel offsets across the entire chip. However, the two chips that compose the WFC are separated by a gap of order 2.5 arcseconds (~50 WFC pixels). As a result, many ACS dither strategies involve the use of offsets sufficiently large to allow the detectors to cover this gap. These will have differing sub-pixel effects across the detector. When taking several exposures of a field in a single filter, observers are generally encouraged to use dithers instead of CR-SPLIT exposures for a number of reasons: to change the placement of hot pixels on the field, to resample the point spread function and to reduce the impact of errors in the pixel-to-pixel flats.

Since dither offsets are achieved by specifying POSTARG shifts along the x and y axes of the detectors, this means that each POSTARG shift on the sky follows the edges of a parallelogram. The shifts have been defined so that displacements in rectilinear sky coordinates are aligned along the y-axes of the detectors (Mutchler and Cox 2001). Thus, for example, a displacement of one WFC pixel along the x- and y-axes of the detector is broken down as follows: the displacement of 1 pixel along the detector y-axis corresponds to 0.0497 arcseconds along the Y-POS direction; however, the displacement of 1 pixel along the detector x-axis corresponds to 0.0496 arcseconds along the X-POS direction, plus an additional 0.0038 arcseconds along the Y-POS direction, due to the non-orthogonality of the pixels on the sky.

For the WFC, HRC, and SBC, a number of pre-defined offset patterns have been created (Mutchler and Cox 2001). These are available in the APT Phase II software, and are aimed at covering a wide range of observing requirements:

The above pre-defined patterns should prove sufficient for the vast majority of scientific programs, however, other patterns can also be created simply by using a combination of POSTARG offsets. NICMOS

A wide variety of pre-defined patterns has been created for NICMOS, to allow an easy implementation of both integer-pixel and sub-pixel dithering. These are generalized extensions to the simple line and box dithers by including spiral and chopping dithers, which are necessary to allow successful removal of a number of NICMOS artifacts. The advantages offered by dithering with NICMOS are the following:

Dithering NICMOS observations may also have disadvantages that an observer should consider:

In general, the benefits of dithering greatly outweigh the disadvantages for NICMOS observations. Whenever possible without incurring excessive overhead, we recommend dithering as much as possible when taking NICMOS data. Note, however, that many NICMOS observations are significantly affected by read-out noise, especially for Cameras 1 and 2 and observations shorter than 1.8 micron. Therefore, the effects of read-out noise on multiply-dithered short exposures should always be carefully balanced against the benefits provided by extensive dithering. STIS

The concept of dithering as applied to STIS observations is multifaceted, since STIS can be used to obtain either images or spectra, and the best method for dithering depends upon the science goals for the observing program. The goal may either be to increase the spatial resolution or to ameliorate the effect of hot pixels or uncertainties in pixel-to-pixel sensitivity with respect to the reference flat-fields.


Imaging-Mode Dithering

Observers can reduce the effect of flat-field uncertainties (particularly for the MAMA detectors) by using a small step pattern with integral pixel shifts. This stepping, or dithering, effectively smooths the detector response over the number of steps, achieving a reduction of pixel-to-pixel non-uniformity by the square root of the number of steps, assuming the pixel-to-pixel deviations are uncorrelated on the scale of the steps. This approach requires sufficient signal-to-noise to allow image registration.

Alternatively, one may improve the spatial resolution somewhat with a dither pattern that includes sub-pixel shifts. Images obtained with the STIS/CCD (0.05 arcsec/pixel), have nearly the same spatial scale as those obtained with the WFPC2/PC camera (0.045 arcseconds/pixel), so that the improvement in spatial resolution would be similar. The spatial scale of MAMA images is half that of the CCD, so the gain in spatial resolution from dithering MAMA images will be more modest, and probably insignificant in the majority of programs. Although the PSF on the MAMA detectors should be narrower than on the CCD because of the shorter wavelengths at which the MAMAs operate, in practice this advantage is offset by additional complications introduced through the instrument optics. It is important to realize that the focus varies across the field of view for STIS imaging modes, with the optical performance degrading by ~30% at the edges of the field of view. Thus, the achievable spatial resolution is significantly compromised in those regions.

Whether or not the dither pattern includes sub-pixel shifts, the effects on CCDs of bad columns, hot pixels, etc., can be reduced or eliminated if the dither pattern is greater than a few pixels. Predefined dither patterns that are available for observers to use, these include:


Spectroscopic-Mode Dithering

Dither patterns can be used with STIS spectroscopic modes for the following purposes:

In first-order spectroscopic modes, improved S/N ratios can be achieved by stepping the target along the slit, taking separate exposures at each location. These separate exposures will subsequently be shifted and added in post-observation data processing. This dithering smooths the detector response over the number of steps, in a manner analogous to that for imaging. For echelle modes, stepping is only possible using the long echelle slit (6x0.2 arcseconds). Note that in the high dispersion echelle modes the Doppler shifting due to spacecraft motion will cause the counts from any output pixel to have been sampled at many independent detector pixels in the dispersion direction (for exposures comparable to an orbit visibility period and targets well away from the orbital pole of HST).

In slit-less or wide-slit mode, stepping along the dispersion would allow independent solutions for spectrum and flat-field, bearing in mind however the increased complexity due to the convolution of the spectrum with the spatial structure in the source. This technique is likely to be useful only if the constituent spectra have a good S/N ratio (perhaps 10 or better), so that the shifts between spectra can be accurately determined.

A variation on this technique involves using one of the contingent of fixed-pattern, or FP-SPLIT slits. These slits are designed to allow the wavelength projection of the spectrum on the detector to be shifted such that the fixed-pattern noise in the flat-field and the spectral flux distribution of the target can be computed simultaneously using techniques that have been successfully applied to data taken with GHRS. Note that this approach is likely to work best if the spectra have a good S/N ratio. More detailed information on the use of FP-SPLIT slits is provided in the STIS Instrument Handbook (Leitherer et al. 2001).

In many configurations the spectral line FWHM is less than two detector pixels. Possible solutions include stepping the target along the dispersion direction in a wide slit or slit-less aperture to subsample the LSF by displacing the spectrum. This technique can also be used to increase the S/N ratio. To employ this strategy, the observer will have to trade off the benefits of improved sampling with the negative impact of increased wings in the LSF when using a wide slit, particularly for MAMA observations. The use of high resolution (default) for MAMA observations may provide 15-30% better sampling, but flat-field variability may make it difficult to realize the benefits, particularly if high S/N ratio spectra are needed.

There are several pre-defined dither patterns that are available for observers to use, these include: WFC3



WFC3/UVIS images are in many ways similar to ACS/WFC images. The detector comprises two rectangular CCD chips separated by a gap approximately 35 pixels wide, so that a gap-stepping dither is needed to avoid having a gap across the center of the field of view. The projection of the pixels on the sky is in the shape of a rhombus, with an angle between the X and Y axes of 86 degrees. As with the ACS/WFC, a POSTARG in Y is along the Y axis of the aperture (along columns), and a POSTARG in X is perpendicular to the Y axis (not quite along rows). The plate scale is 0.04 arcseconds/pixel on each axis, and the FWHM of the point spread function is between 1.6 and 2.3 pixels, depending on wavelength. WFC3/UVIS images will thus benefit from half-pixel dithering, but not as much as WFC3/IR images. Non-linear distortion causes the projected area of the pixels to vary by +/-3% relative to that at the center of the detector, so POSTARGs and patterns will produce shifts in pixels that vary with location on the detector.

Five patterns have been installed in the phase 2 software to dither and mosaic WFC3/UVIS images:

Other patterns can be created by using POSTARG offsets or generic patterns, or by changing the spacings in the defined patterns. For programs requiring high precision small aperture photometry, observers should see WFC3 ISR 2008-10 for a discussion of features called "droplets", caused by contamination on the outer window of the UVIS detector. Dithers ~100 pixels are recommended to improve the photometry.

The WFC3 pipeline produces cosmic ray rejected (CRJ) images from input FLT images for CR-SPLIT exposures. When MultiDrizzle is run in the pipeline, it will use the FLT images as input, tagging cosmic rays in those images with a different value of DQI (4096) from the value used by CALWF3 (8192). Observers are generally encouraged to use dithers instead of CR-SPLIT exposures for a number of reasons: to change the placement of hot pixels on the field, to resample the point spread function and to reduce the impact of errors in the pixel-to-pixel flats.



The WFC3/IR pixels are projected as rectangles on the sky, with X and Y plate scales ~0.14 and 0.12 arcseconds per pixel. A POSTARG in Y is along the Y axis of the aperture (along columns), and a POSTARG in X is along the X axis (along rows). The FWHM of the point spread function is between 1.0 to 1.25 pixels, so sub-pixel dithering is needed to recover spatial resolution. Non-linear distortion causes the projected area of the pixels to vary by +/-4% relative to that at the center of the detector, so POSTARGs and patterns will produce shifts in pixels that vary with location on the detector.

Three patterns have been installed in the phase 2 software to dither and mosaic WFC3/IR images:

Other patterns can be created by using POSTARG offsets or generic patterns, or by changing the spacings in the defined patterns. Note that there is a ~45 pixel diameter dead spot near the lower edge of the WFC3/IR detector, centered at ~[358,54]. A dither larger than this diameter should be used if imaging in that area is required.

WFC3/IR exposures are made with predefined timing sequences of non-destructive reads. As in NICMOS, up-the-ramp fitting of the fluxes in the sequence is used to identify and remove cosmic ray flux from each pixel. The accuracy of the procedure depends on the timing sequence and the number of frames specified in the proposal, just as the accuracy of traditional cosmic ray rejection in CR-SPLIT exposures on a CCD detector depends on the number of exposures and the exposure time. The cosmic ray rejected FLT image is used as input to MultiDrizzle. As for WFC3/UVIS images, DQI values of 4096 and 8192 are used to tag pixels with cosmic rays identified by MultiDrizzle and by CALWF3, respectively.

Space Telescope Science Institute
Voice: (410) 338-1082