Speckle Imaging at WIYN Using NESSI

The following is a guide to the capabilities of speckle imaging at WIYN and to preparing a speckle observing program using NESSI. Speckle observations using NESSI at WIYN are currently carried out in queue mode, which operates ~2 times per semester and is applied for through the NN-Explore program at NFS's OIR Lab or university partners. Speckle queue PI's receive a set of reduced data products in the month or so following the observations. For questions not answered here (or if this document is unclear), contact the support scientist,

Speckle imaging using the NN-Explore Exoplanet Stellar Speckle Imager (NESSI) at WIYN provides observers with diffraction-limited resolutions within small (few arcsec wide) fields. The two electron multiplying CCDs (EMCCDs) of NESSI operate simultaneously through a pair of filters (out of a current choice of 4 specialized speckle filters). A common use of speckle imaging has been to detect and provide relative photometry and astrometry for binary and higher-order multiple stars. The relative astrometry, brightnesses and colors from NESSI data help characterize the components and may be used to measure visual binary orbits or relative proper motions. At WIYN, speckle imaging should reach sensitivities to detect companions at least 3-5 magnitudes fainter than a target star depending on brightness, exposure time and conditions. Typically, the time spent integrating on a speckle target is 1-20 minutes.

A standard speckle observing method and data reduction pipeline has been developed over the years at WIYN and other telescopes using NESSI's predecessor instrument, DSSI. This procedure was employed during the 2016B speckle queue observations using NESSI. Once a target has been acquired, data are taken by reading a subsection of the full array (for NESSI the window is 256x256 pixels or 4.6x4.6 arcsec, although the speckle imaging works best at separations inside of ~1.5 arcsec). Normally the window is centered on the target of interest. Images are taken in sets of 1000 frames apiece where each frame is a 40 millisecond exposure (and each image set requires a little less than 1 minute to complete). To achieve deeper imaging or for faint targets and poor conditions, multiple image sets are taken.

In addition to science targets, at least one bright point source star in close proximity on the sky is observed using the same techniques. This point source standard accounts for the shifting PSF as a function of mount coordinates, airmass and focus.

Data reduction is done using a custom software pipeline run by the speckle team. Successful proposers to the speckle queue will receive 4 basic pipeline products for each target in each filter. The products were developed as part of projects to detect previously-unresolved secondary stars or fore- or background sources near stars of interest and to acquire temporal astrometric measurements for measuring binary star orbits. These data products are probably applicable to a wider variety of programs:

1. A reconstructed image at or near the telescope diffraction limit, centered on the target star. Sources detected within the field will show up as bright objects against a noise background (sky, effectively).
2. An evaluation based on visual inspection of each image (and in Fourier domain its power spectrum) to determine if there are multiple sources detected. When multiple objects are present, their relative astrometry and photometry are measured.
3. An estimate of the background sensitivity or the minimum magnitude difference relative to the target star any undetected sources in the field would have as a function of separation from the target star. This may be called a contrast curve. It is based on noise statistics in the reconstructed image as measured in a set of concentric annuli of various radii centered on the target star. The product consists of a table listing background limiting magnitude vs. angular separation.
4. A plot (PDF), showing the background limiting magnitude vs. angular separation.

See example data products here.

Users can ask for the raw data, although this will often be quite large (~100GB/night) and hasn't been requested by most users.

The effectiveness of speckle imaging is strongly dependent on observing conditions, especially seeing. As a rough guide, NESSI should be able to operate effectively on stars of V or R magnitude as faint as 14 under reasonably good conditions. In excellent conditions, stars of V=14.5 are observable.

Note that the magnitudes above refer to a "central" target star, around which fainter companions may be detected. Faint companions may be detected down to a background limit, which depends on the seeing and typically rises in its magnitude difference relative to the central star with increasing angular separation. A contrast curve reaching 5 magnitudes fainter than the central star is achievable for stars brighter than V=12. For fainter stars, longer integrations are used. Even so, the resulting images are typically less deep.

The field-of-view for speckle imaging is limited by the angular size of the atmosphere's isoplanatic patch. Speckle patterns of stars separated by more than ~1 arcsec become uncorrelated, lowering the sensitivity to detect and measure their relative properties. For this reason, DSSI was operated in a 128x128 pixel window (2.8x2.8 arcsec). For NESSI, which has a somewhat smaller pixel scale, our current plan is to cover this same field-of-view by observing with a 256x256 pixel window and provide data products for the inner 2.8x2.8 arcsec box. As our software gets refined for the new instrument, these details may change.

Speckle interferometry delivers imaging at or near the telescope diffraction limit. The diffraction limit in each speckle filter is listed in the table below. Note, however, that these numbers are not necessarily indicative of the filter choices that will yield the best-resolved images! Our ability to construct speckle images with small inner working angles, push for relatively deep background limits inside of a few tenths of arcseconds, and resolve (for example) close binaries, depends on closely matching the PSF of science targets to point source standards. Generally, the worse seeing and greater differential atmospheric diffraction at shorter wavelengths renders speckle in the bluer filters more difficult and the image quality may be adversely affected. For these reasons, the red filters often yield effectively superior angular resolution. For similar reasons, targets that can be observed at low airmass at WIYN will work best.

The performance of speckle imaging is quite sensitive to conditions like seeing, so there are no strict rules to follow for determining the ideal number of image sets to acquire on a target of given brightness.

Observers targeting stars fainter than V=13 should plan on acquiring multiple image sets and those observing brighter stars may also benefit from taking multiple sets. This will depend on how they balance better contrast depth/image quality vs. number of targets visited. Multiple image sets per star can also help under less than optimal observing conditions and, given the several minutes needed to set up observing of each new target, many users may want to devote comparable time to exposures.

Each image set requires 1 minute of telescope time. Acquiring a target with a short slew requires 3 minutes and with a long slew, 5 minutes. Since a science target requires a point source observation, additional time is needed for that (about 4 minutes). Refer to the guide on estimating observing time for more information.

Note that we have found the signal-to-noise ratio for detecting secondary sources in speckle images does not grow as rapidly with exposure time as it would in traditional CCD imaging (ie. with the square root of time). Proposers may not expect to achieve the same contrast limits on faint stars as bright ones (5 magnitudes may be achievable on 12th magnitude stars and 3 magnitudes on 14th magnitude stars.) The table below only suggests numbers of image sets to take for various magnitude stars:

To prepare an observing proposal, you should estimate the time needed to carry out the observations. Use the guide to estimating observing time (PDF) for help with that.

If you're unsure how to estimate your observing time or the guide doesn't seem to apply to your project, contact the support scientist for help.

NESSI uses a dichroic beamsplitter to separate the incoming light (at 686nm) into blue and red channels before focusing on the two identical cameras, which operate simultaneously. The speckle filter choice will be one of 467nm or 562nm paired with one of 716nm or 832nm. NESSI's SDSS filters are also listed below (although not used for speckle imaging). Data are in nanometers and fractional efficiencies as quoted by the manufacturer.

Other important efficiency curves:
QE of NESSI's Andor Ixon 888 "BV" EMCCDs
Andor's BB-VS-NR (wedged) dewar window transmission
Semrock Di03-R660 dichroic transmission/reflection (used for all NESSI data starting 2017 May 10)
older dichroic transmission/reflection (used for all NESSI data before 2017 May 10)

In addition to the usual information included in observing proposals, proposers to the speckle queue must submit a coordinate list sufficient to carry out the program. The coordinate list can be altered and resubmitted close to the observing date if the nature of the program warrants it and new objects are in keeping with the proposal's science. In some cases, where target lists have been submitted with your proposal, we may take those coordinates as a starting point and check with PIs to make sure they are finalized and whether there are special instructions. The speckle coordinate lists are not submitted to the Kitt Peak Coordinate Cache system (which the speckle program does not use), but rather are plain text format files which can be emailed to the support scientist, Mark Everett.

In addition to science targets, point source standards will need to be observed. It is likely easiest for the speckle queue team to pick good point sources for you, which we will do if you don't provide them. In any case, an estimate for the time required to carry out your observing program should include the time needed to observe the point source standards. The best point source standards are HR stars of 5-6th V magnitude within 5 degrees of each science target, and which are thought to be single. For efficient slewing, the science target and its associated point source should both lie either north or south of 31.96 degrees declination, the zenith at WIYN.

During a typical night, nearby science targets from different queue proposals will be observed sequentially and may share a single point source standard observation for their calibration.

Your coordinate list should be provided at least a few weeks prior to the observing run. You may be contacted by the support scientist to ensure that we have your list and additional instructions to carry out your program.

A proper coordinate list should include a target name, RA & DEC (equinox and epoch 2000), a V or R magnitude measurement or estimate, and any significant proper motions in RA and DEC. Note that with the small speckle field-of-view, it is important for us to correctly identify your targets! To that end, it can be helpful if the target name is identifiable to the observer using SIMBAD or other online sources of information such as Exofop, that in certain crowded fields finding charts could be helpful and that proper motions are needed only if they are rather large (e.g., target moved more than a few arcsec since 2000). If you have a priority order or ranking for your targets, or special instructions for different targets, it would be helpful if you indicate these things in your coordinates list.

A example coordinate list of 3 objects listing RA, DEC, Vmag, pmRA and pmDEC in milliarcsec/year respectively (if you use a comparable format, e.g., hexidecimal coordinates, it will be helpful):

HIP 1	     00:00:00.22 +01:05:20.4 9.08 -4.55 -1.19
HIP 40617    08 17 31.65 +08 51 58.7 6.3
BD+71 889A   18:20:45.43 +71:20:16.1 4.22 -5.03 37.86

A description of our data products may be found in Howell et al. (2011) and Scott et al. (2018). Scott et al. (2018) describes NESSI.

Howell et al. 2011, AJ, 142, 19
Scott et al. 2018, PASP, 130, 4502

A request for acknowledgement is made of authors using NESSI data in publications.