TECHNICAL REPORT Title: NIRSpec Dithering Strategy Part 3: The Microshutter Array (MSA) Authors: J. Tumlinson 1.0 Phone: 410338-4553 Doc #: JWST-STScI-001769, SM-12 Date: June 3, 2009 Rev: - Release Date: 3 September 2009 Abstract This study concerns need for dithering of NIRSpec data, the strategies for obtaining dithered data, the effect of dithering on data calibration, and the effect of dithering on mechanism lifetimes. For simplicity the study has been broken into three parts for each of NIRSpec’s major observing modes (Fixed Slits, Part 1; IFU, Part 2; MSA, Part 3). The very high degree of configurability of the MSA allows the user to realize a slit mask for an arbitrary set of positions on the sky. Departures from an “ideal MSA” caused by failed open or closed shutters ensure that a given configuration cannot be shifted by a fixed offset in either dimension and always recover the same pattern of open apertures. For the MSA, we conclude that there is only a small set of restricted circumstances in which predefined dither patterns will be feasible. In particular, if the user has chosen to place a “slitlet” composed of multiple open shutters on each science target, then dithers between the open shutters are straightforward and can be standardized in advance for a specific slitlet shape. This strategy should be effective for a wide range of programs directed at faint point or marginally resolved sources. Predefined dither patterns will generally not work for programs that choose to depart from this simple strategy, or for programs that intend to observe large extended objects. In these cases the burden will fall on the user and on the MSA planning tool to define separate MSA configurations that accomplish the purpose of dithers by obtaining data on a given set of targets at different positions on the array. To capture this general case with a minimum of complexity, overheads, and mechanism movements, we recommend the implementation of a capability to observe multiple MSA configurations that refer to the same target acquisition and guide stars but which are separated by small angle motions (~< 20”). 2.0 Introduction NIRSpec data needs dithering for a number of important reasons. First, the PSF of the JWST OTE is undersampled at the FPA, where each 18 µm pixel spans 0.1˝ on the sky Operated by the Association of Universities for Research in Astronomy, Inc., for the National Aeronautics and Space Administration under Contract NAS5-03127 Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. JWST-STScI-001769 SM-12 compared with a PSF FWHM >~ 0.1 – 0.2˝ over 1-5 µm. Dithering by small spatial offsets (< 0.1˝) can improve the spatial sampling of the PSF. For many observations dithering with a larger offset will be needed to cover the gap in wavelength caused by the ~ 3 mm (17.8”) gap between SCA segments. Finally, dithering can reduce noise associated with pixel-to-pixel sensitivity variations by obtaining measurements of the same point on the sky at different places on the detector, averaging out the variations. This report assumes the small angle motion (SAM) accuracies stated in Section 3 of the fixed-slits and IFU chapters (JWST-STScI-001678 and JWST-STScI-001749 respectively). 3.0 MSA Properties: Slit Losses, the Sweet Spot, and Dithering The Microshutter Array (MSA) is the key technology that enables multi-object spectroscopy for NIRSpec. The MSA is a micro-electro-mechanical (MEMS) device with individually addressable shutters in 365 rows and 171 columns on each of 4 quadrants (for a total of 249,660 shutters). Each shutter has a rectangular open area that subtends 200 mas in the dispersion direction and 450 mas in the spatial direction. The “walls” that separate the open areas of the shutters subtend 60 mas in each direction, so the separation between the shutter centers (the “shutter pitch”) is 260 mas in the dispersion direction and 510 mas in the spatial dimension (this is within 4% of a 2:1 ratio for height:width). The “slit losses” caused by the MSA shutters are a key factor in planning MSA observations and in designing their dither patterns. A combination of geometric blocking and diffraction of light by the MSA causes loss of light that becomes worse near the edges of each shutter aperture. Slit throughput achieves a maximum value of ~60% near the center of each shutter and drops off to ~20% at the edges, and depends also on wavelength as shown in Figure 1. Figure 1: Slit losses for the MSA shutters as a function of distance from center for the x (dispersion) and y (spatial) dimensions. The slit losses with respect to the maximum are relatively flat in the center of the apertures. For spectrophotometric measurements loss of light at the slit must be corrected by dividing out the slit loss function. Because the exact position of the science target within the slit is only known to within some small positional uncertainty (owing to target acquisition errors and jitter), this correction carries some uncertainty. This uncertainty Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -2- JWST-STScI-001769 SM-12 becomes severe when the target is near the edge of the slit and a small positional error translates to a large uncertainty in the slit loss correction. To minimize the uncertainty in the slit-loss correction and to maximize the slit throughput, NIRSpec’s designers have introduced the concept of a “sweet spot” within each shutter where the uncertainty in the slit loss correction caused by errors in knowledge of the target position are constrained to < 10%. The sweet spot occupies the central 40-60% of the shutter area, depending on wavelength (see Figure 3). The concept of the sweet spot drives much of NIRSpec’s observing and dither strategies, since it plays a large role in the overall throughput and photometric accuracy of the data. 4.0 “Slitlet” Dithers for Maximum Throughput NIRSpec plans to routinely obtain spectra of astronomical sources near the faint limits of its sensitivity – objects with fluxes of 100 – 300 nJy, such as faint, high-redshift galaxies or main-sequence stars in the outer halos of galaxies. For these science cases maximum total instrument throughput is required, so it is desirable for every science target to be observed within the sweet spot of a microshutter. This requirement entails that the proposal planning system take into account the geometric filling factor of the sweet spot within each shutter, and break the input target list into as many target subsets as are needed to cover the field of the sky under study. Even if each science target is observed within a sweet spot, it is still desirable to place the same science target on Figure 3: “Sweet spot” for a different detector pixels for all the usual reasons to dither – single MSA shutter. From to correct for pixel-to-pixel sensitivity variations and dead NIRSpec OCD. pixels, and to improve the pixel sampling of the point- and line-spread functions. To accomplish these purposes and to maintain the high throughput offered by the sweet spots, the NIRSpec Instrument Science Team has defined a “slitlet” dithering strategy, illustrated in Figure 4. To implement this strategy, the MSA planning tool should place a 3x1 shutter “slitlet” for each science target in each subset. Three separate exposures will be obtained for each target, in the sequence illustrated in the right panel of Figure 4. For each exposure the two unoccupied shutters will obtain data on the local sky background. The small-angle motion (SAM) between each shutter will cover a full shutter pitch in the y-dimension (510 mas). This strategy places the target in the same point on the shutter in each exposure so that the slit loss is the same for each exposure (though small changes will result from pointing errors and jitter). Since the target is always in the sweet spot, the average slit throughput is maximized. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -3- JWST-STScI-001769 SM-12 Figure 4: At left, the MSA plane with five “slitlets”. At right, the observing sequence for the 3-shutter slitlet. Unoccupied shutters obtain data on the sky background. The interleaving of the MSA configurations, grating/filter wheel moves, and dithers are designed to minimize mechanism movements. We intend to conserve MSA reconfigurations as the highest priority, so all dithers and all gratings will be obtained before moving to the next MSA configuration for the next target subset. It is also important to minimize the number of GWA/FWA movements, so each dither position in the slitlet should be observed with a single GWA/FWA position before changing the grating (as for the fixed slits). If conceptualized as “nested loops”, the ordering would be: - MSA configurations (the “outer loop”, executed least frequently) - Changes of grating / filter - Small-angle motions to dither (the “inner loop”, executed most frequently) A variation of this strategy can be used to improve the sampling of the line-spread function of the NIRSpec gratings. In this variation each target is assigned a 3x2 shutter window and the slitlet pattern illustrated in Figure 4 is executed twice: once as shown in Figure 4 with the adjoining 3x1 slitlet closed, followed by an MSA reconfiguration, and then again in the right 3x1 side of the 3x2 window with the left 3x2 slitlet closed. This is equivalent to moving the complete 3x1 pattern over by exactly one shutter pitch and repeating the A-B-C pattern shown in Figure 4. The advantage of this repeated pattern is that it occurs 260 mas or 2.6 pixels away from the first pattern, and so effectively resamples the line-spread function of the grating with net 0.6 pixel separation. This variation is intrinsically less efficient overall since it requires 3x2 shutter clearing in the MSA planning tool, the number of targets that can be placed in a single configuration declines accordingly, and so the total number of target subsets must increase to cover the same field. The number of MSA configurations required to obtain a given target set is also effectively doubled by the shift of the pattern over by one shutter pitch. However, for some programs these inefficiencies will be worth the improved sampling1. 1 As noted in Part 1 (Fixed Slits), curvature of the iso-wavelength lines caused by optical distortion across the field mimics the effect of small spatial and/or spectral dithers. Thus two spectra taken with identical Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -4- JWST-STScI-001769 SM-12 Since the “slitlet” approach uses three shutters in the spatial direction, and moves each 3shutter slitlet by two shutters in each direction, five shutters worth of sky in the spatial direction will be observed. This may present problems for background subtraction, if a bright object or another target of interest falls within this 5-shutter region. Unintended light in the background shutters is not necessarily fatal to the background subtraction, but would present problems. To deal with this problem the planning software could either explicitly clear the 5x1 shutter region of the sky, or just mark it so that the user could check for problems. In most fields a couple of contaminated background shutters may not be prohibitive, but severe crowding of sources may force the user to define MSA configurations that depart from this simple slitlet scheme and use the more general schemes outlines in Section 7. 5.0 “Subshutter” Dithers for Efficient Multiplexing The “sweet spot” observing strategy and the simple dither pattern that goes along with it excel at obtaining data with maximum average throughput. However, since this strategy uses only a portion of each shutter’s area it is less efficient at covering a field of targets than a strategy that uses the full open area of the shutters. The cost is that a larger number of MSA configurations are needed to complete a given set of targets in a field. Observers who can tolerate a loss of average throughput or who desire greater photometric accuracy may instead opt to pursue a different strategy for configuration and dithering, which we will call “subshutter” dithering (Regan 2005). In this strategy, targets are placed by the MSA planning tool on any available shutter whether they fall into the “sweet spot” region or not. Then a set of exposures is obtained using dithers on a regular grid (4 x 4, 3 x 3, etc.) with total extent of one shutter pitch in each direction (see Figure 5). In principle this approach should lead to lower errors in the slit loss correction to the flux calibration, since the target will be observed on both sides of the slit loss curve. This strategy also in principle reduces the number of target subsets that are needed to cover a given field by using the full open extent of each shutter. The regular pattern of subshutter moves suggests that this observing strategy can be executed with a fixed and pre-programmed dither pattern. Simple modeling of this approach is adequate to show that this is not so. Figure 5 illustrates a simple field observed with this strategy. The initial location of each target within its shutter is at position 1. Each of the sixteen positions would be observed in turn using small angle dither motions between each exposure. Since some targets will move into adjoining shutters for some subset of the 16 positions, more than one MSA configuration is needed to realize this strategy. If only one open shutter is assigned to each target, then up to 16 different configurations may be needed. The number of unique configurations can be reduced if each target is assigned a 2x1 shutter open aperture – this allows positions in a single column (i.e. positions 1 – 4) to use the same configuration and leaves open a “sky shutter” adjoining the target shutter for each of the 16 positions. For the illustrated case, 4 conditions but at different y-positions will have slightly different pixel sampling of the PSF and LSF. However, the magnitude of these effects are not known for small offsets and may not be sufficient to accomplish the resampling that deliberate moves can achieve. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -5- JWST-STScI-001769 SM-12 unique configurations are needed to cover the 16 positions with 2x1 shutter aperture assigned to each target. The four configurations cover positions 1-4 (first configuration), 5-8 (second configuration), 9-12 (third configuration), and 13-16 (fourth configuration). Depending on the configuration of targets on the sky, the smallest, most efficient number of unique configurations can range from 1 to 16 just for the simple choice of 1x1 shutter apertures and a 4 x 4 pattern. The upshot of this exercise is that for the subshutter strategy there is no pre-programmed optimal set of positions and configurations that will work for every program. The exact minimum set of configurations and their correspondence to the offset positions can be worked out only when the exact set of targets is known – that is, at the time the observation is planned. The range of possibilities is driven by the user choice to adopt a specific pattern (3 x 3 or 4 x 4 moves to cover a shutter pitch) and by the actual positions of the targets on the sky. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -6- JWST-STScI-001769 SM-12 Figure 5 An example of subshutter dithering for a field of six targets observed with a 4 x 4 subshutter dither pattern. The sixteen dither positions for each of the six targets are numbered as shown in the inset. This set of targets and dither positions requires four configurations; more general fields could require up to 16. The subshutter dithering strategy therefore places the burden of constructing dither patterns onto the proposal planning system. Each program will want to observe a unique set of targets, which will require a unique set of MSA configurations and the corresponding sequence of small-angle motions that can only be worked out by the MSA planning tool. The tool may be able to speed the process significantly by, for example, automatically generating the set of configurations and dithers given the targets and user choices for the size of the subshutter grid. 6.0 The “Wavelength Gap” The ~19” gap between detector segments causes a gap in wavelength coverage for spectra obtained with the R = 2700 grating. A simple remedy for this gap is to shift the pattern of Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -7- JWST-STScI-001769 SM-12 targets on the MSA by 20” (approximately 100 shutters) to shift the missing wavelengths onto the detector. At first blush this dither appears so simple that if anything could be implemented as a pre-programmed dither, it would be this one. However, two features of the MSA prevent even this simple move from being defined uniformly for all cases. First, optical distortion of the field at the MSA plane ensures that while some targets would move 100 shutters over on the array for a move of 20”, others would move 99 or 101 or 100.54 shutters. Some target might move from the sweet spot of an open shutter to a failed closed shutter, or to a point behind an MSA wall. The cost is especially high if it is a high-priority target that is lost. Thus even for a simple move we cannot guarantee that all targets will be observed at both positions if the optimization is derived for only one of them, and that the wavelength gap will be filled for all. In this case the only recourse may be to perform two separate optimizations of shutter placement with different center positions and for the user, perhaps assisted by the software, to ensure that the missing wavelengths are obtained at one or more of the positions. Thus we reach the same conclusions as for the “subshutter” dithers: unique MSA configurations linked together with small angle motions in between is required to achieve the purpose of dithers in light of a non-ideal MSA. 7.0 Generic Dithers for Extended Objects and Difficult Cases The configurability of the MSA will enable new types of spectroscopic observations that have never been achievable in space. The diversity of astrophysical targets – from crowded fields of point sources to extended sources of arbitrary shape – means that we cannot plan in advance a canned set of dither patterns that will serve the need of every user in every case. For users with typical fields and standard requirements one of the two major strategies above will suffice, but even these will not admit one-size-fits-all dither patterns. For “hard cases”, we must enable another approach that still uses small-angle motions (as dithers) to combine multiple MSA configurations within a single visit. To implement this capability, we envision the MSA planning tool providing the option of linking together uniquely defined configurations into a custom pattern of configurations separated by small-angle motions, provided the whole pattern satisfies some constraint on the maximum angular separation. This approach is illustrated with a “hard case” in Figure 5, where we use the Cas A supernova remnant as an example of a complex extended source. In this schematic, three unique MSA configurations have been defined separately to cover bright knots of emission in the remnant. Because the center point of the MSA (the yellow stars) lie within a 20” circle (that is, their maximum separation < 20”), these three settings can be observed with the same guide stars and only small angle motions between them. This particular set of knots cannot be observed with a single configuration because it would require use of two shutters in the same column (along the dispersion direction), which is disallowed because the spectra would overlap. For example, observing the knots marked A, C, and E requires a different configuration from B, C, and D, since having, e.g. shutters for knots A and B open simultaneously would yield overlapping spectra if observed in the same configuration. In addition to this, small angle motions may be required to place the desired set of bright knots optimally in their Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -8- JWST-STScI-001769 SM-12 corresponding shutter, that is, a single position cannot capture their positions optimally, even with different configurations. In this general case, multiple configurations at multiple positions would be required. If for some reason the user desires additional exposures with small motions of order one shutter at each one of the marked center positions (for instance, those described as “subshutter” dithers above) these count as separate positions as well. Once the desire and/or need for custom sets of configurations and center points is established, it can easily accommodate nested patterns (analogous to “primary” and “secondary” motions from the FS and IFU chapters) within the generic scheme. The alternative is for these three configurations to be executed as separate visits, with the additional overhead of a new target acquisition and definite loss of efficiency. The baseline observing scheme already allows for multiple MSA configurations centered at the same point on the sky, within a single visit. This flexible dithering plan enables an arbitrary set of targets over a small area and adds only a small degree of complexity associated with the small-angle motions between the center points. Though these closely spaced MSA positions and their corresponding configurations arise from considering dithering strategies, and may prove to be very similar to dithers operationally, they differ in many respects from the typical usage of the term “dither”. For example, the data reduction pipeline will not always be asked to combine data from these different positions / configurations as is often done for dithered data. The different positions may request different exposure times, while traditional dithers use the same exposure at each position. For these reasons it is probably best to devise a different term, (“cluster target” is one possibility) for these patterns to be used in APT and in the user documentation, to avoid confusion. While this approach to extended sources by dithering does require flexibility in the planning of the observation, it need not add additional risk to the program. The exact pattern of moves and even their sequence can be derived uniquely and correctly by the planning software and passed downstream after checking for errors and constraints. There is no need for the user to specify the size, direction, and sequence of moves, if APT can derive these from the desired set of MSA configurations and center positions. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -9- JWST-STScI-001769 SM-12 Figure 6: A schematic view of how an observer might use the MSA to explore an extended source of arbitrary shape. There are three MSA configurations shown that have unique shutter open/closed patterns and different central positions. Provided the central positions of the MSA lie within a circle of diameter somewhat than the FGS field (say, less than 20” as a rule of thumb) the three positions can be treated as a custom dither pattern, derived by the software based on the users MSA setups. Note: the real Cas A is very large with respect to the microshutters, but the principle holds. 8.0 Conclusions and Recommendations Conceptual development of the various dithering strategies proposed for NIRSpec observations has revealed that even simple dither patterns are always difficult and often impossible to design in a predetermined fashion. It does not appear possible to pre-define dither patterns that will work for all programs, or even a representative subset of programs, while also ensuring that each target of interest will be observed at each and every dither position. Even if such canned patterns were possible, the desire of many users to observe extended objects or use long slits drives us toward a different conclusion: namely, that dither patterns for MSA observations should be constructed by the MSA planning tool in tandem with the configuration themselves. Furthermore, once this conclusion has been adopted it is probably most efficient and simpler to conceptualize and implement all of the strategies in this fashion. Put another way, there is no reason for the downstream ground system and flight software to know that it is being asked to execute the 3x1 slitlet strategy or the 4x4 subshutter strategy provided it receives a list of MSA configurations and their corresponding positional offsets and knows how to act on these in the proper sequence. This approach places the hard work of defining patterns into the proposal planning system. It is the effectively infinite complexity of the MSA as a device and the idiosyncrasy of each device as built that forces these conclusions. Our recommendations for NIRSpec MSA dithering strategies and implementation are: Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 10 - JWST-STScI-001769 SM-12 1) We recommend that NIRSpec MSA dither patterns be defined within the MSA planning tool as linked sets of MSA configurations and the corresponding offset positions. 2) We recommend that the linked sets of MSA configurations and offsets be passed downstream in a fixed formation that is common to all of the dithering strategies here, since they are all cases of a generic strategy. 3) We recommend that APT include the specific strategies and special-purpose dithers presented here as “shortcuts” or tools to assist the user in defining the necessary set of configurations. For many strategies the exact pattern can be derived deterministically for a given set of targets and so the software can relieve the user of the need to define completely unique and separate configurations once the basic strategy is chosen and the targets are known. The exact strategies and how they will be implemented require further study and quantitative modeling with the ongoing MSA optimization study. 4) We recommend that downstream systems and scripts adopt a common approach to receiving MSA configurations and offsets from the planning system in a generic way, so that a single algorithm can implement these various strategies based on the received configuration files and offsets without knowing the exact strategy (slitlet, subshutter, etc.) that they correspond to. 9.0 References Böker, T. and Valenti, J. 2008, “NIRSpec Operations Concept Document”, v5.0 Regan, M., 2005 “An Alternative Observing Strategy for NIRSpec and its Effect on NIRSpec Target Acquisition”, JWST-STScI-000674 Tumlinson, J. 2009 “NIRSpec Dithering Strategies, Part I: The Fixed Slits”, JWSTSTScI-001678 Tumlinson, J. 2009 “NIRSpec Dithering Strategies, Part II: The Integral Field Unit”, JWST-001749 Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 11 -