The wide-field infrared transient explorer (WINTER)
Nathan P. Lourie, John W. Baker, Richard S. Burruss, Mark Egan, Gábor Fűrész, Danielle Frostig, Allan A. Garcia-Zych, Nicolae Ganciu, Kari Haworth, Erik Hinrichsen, Mansi M. Kasliwal, Viraj R. Karambelkar, Andrew Malonis, Robert A. Simcoe, Jeffry Zolkower
TThe wide-field infrared transient explorer (WINTER)
Nathan P. Lourie a , John W. Baker c , Richard S. Burruss c , Mark Egan a , G´abor F˝ur´esz a ,Danielle Frostig a,b , Allan A. Garcia-Zych a , Nicolae Ganciu c , Kari Haworth a , Erik Hinrichsen a ,Mansi M. Kasliwal c , Viraj R. Karambelkar c , Andrew Malonis a , Robert A. Simcoe a,b , andJeffry Zolkower da MIT Kavli Center for Astrophysics and Space Research, Massachusetts Institute ofTechnology, 77 Massachusetts Ave, Cambridge, MA 02139, USA b MIT Department of Physics, 77 Massachusetts Ave., Cambridge, MA 02139, USA c Division of Physics, Math, and Astronomy, California Institute of Technology, 1200 ECalifornia Blvd, Mail Code 249-17, Pasadena, CA 91125, USA d Caltech Optical Observatories, California Institute of Technology, 1200 E California Blvd.,Mail Code 11-17, Pasadena, CA 91125, USA
ABSTRACT
The Wide-Field Infrared Transient Explorer (WINTER) is a new infrared time-domain survey instrument whichwill be deployed on a dedicated 1 meter robotic telescope at the Palomar Observatory. WINTER will perform aseeing-limited time domain survey of the infrared (IR) sky, with a particular emphasis on identifying r -processmaterial in binary neutron star (BNS) merger remnants detected by LIGO. We describe the scientific goals andsurvey design of the WINTER instrument. With a dedicated trigger and the ability to map the full LIGO O4positional error contour in the IR to a distance of 190 Mpc within four hours, WINTER will be a powerful kilonovadiscovery engine and tool for multi-messenger astrophysics investigations. In addition to follow-up observationsof merging binaries, WINTER will facilitate a wide range of time-domain astronomical observations, all thewhile building up a deep coadded image of the static infrared sky suitable for survey science. WINTER’s customcamera features six commercial large-format Indium Gallium Arsenide (InGaAs) sensors and a tiled opticalsystem which covers a > Keywords:
WINTER, time-domain, infrared, LIGO, multi-messenger, wide-field, InGaAs detectors, robotictelescopes, kilonova
1. INTRODUCTION
WINTER (The Wide-field Infrared Transient Explorer) is a new robotic infrared time-domain survey instrumentwhich will operate behind a dedicated 1 m aperture telescope at Palomar Observatory. The instrument is cur-rently being assembled and integrated at the MIT Kavli Institute, and will be deployed to Palomar Observatoryin mid-2021. In this paper we describe the driving science goals of WINTER and how these goals drive the archi-tecture of the instrument. This paper is published alongside companion papers describing the derivation of theWINTER engineering requirements from scientific goals (Ref. 1), the design and testing of the optomechanicalstructures and lens mounts (Ref. 2), and the InGaAs sensors and readout electronics (Ref. 3).The design of the WINTER sensor and optical architectures are described in Secs. 2 and 3 respectively. Sec.4 describes the observatory robotic control system, and Sec. 5 gives an overview of the observation strategy andautomated scheduling. Finally, in Sec. 6, we describe the current status of the instrument and observatory as ofNovember 2020.
Further author information: (Send correspondence to N.P.L)N.P.L.: E-mail: [email protected], Telephone: +1 617 324 1194 a r X i v : . [ a s t r o - ph . I M ] F e b .1 Science Goals Time-domain infrared surveys offer a key window into a host of astrophysical phenomena, especially in the age ofmulti-messenger astrophysics. Infrared observations are one of the most powerful tools for detecting kilonovae,the thermal emission from rapid neutron capture ( r -process) nucleosythesis in the ejecta of binary neutron star(BNS) mergers. The LIGO-Virgo experiments’ detection of gravitational waves (GW) from the BNS mergerGW170817 and subsequent detection of electromagnetic (EM) counterparts
5, 6 across the spectrum from thex-ray to the radio suggest that these merger events may be the dominant mode of r -process element production inthe Universe. While GW170817 is the only confirmed observation of a kilonova associated with a BNS merger,improvements in sensitivity for the fourth observing run of LIGO (LIGO O4) are expected to yield 1-2 NS-NSmergers per month. Theoretical models predict that infrared kilonova are ubiquitous in these merger events,and unlike optical emission which is short-lived ( < >
9, 10
The primary science goal of WINTER is to undertake a systematic and unbiased search for infrared coun-terparts to detect lanthanide-rich kilonovae throughout the full LIGO O4 search volume. With three filtersoperating at the Y, J, and a shortened H (Hs) bands (centered at 1.02, 1.25, and 1.60 µ m respectively), WIN-TER will record weeks-long lightcurves of massive merger events detected by LIGO offering new insight intothe mass fraction of lanthanides in the ejecta of these events. This data will help answer fundamental questionssuch as (a) Are NS-NS mergers the only site of r -process nucleosynthesis? (b) Do BNS mergers produce thesame relative abundance ratios seen in the solar neighborhood? (c) Are the third-peak r -process elements syn-thesized? (d) What elements are made when a NS merges with a stellar black hole (BH)? WINTER’s observingplan (described in more detail in Section 5) will prioritize these LIGO detections, allowing follow-up of a largersample of GW events. With 100% of observing time available in both the bright and dark lunar cycle, it canfollow up all GW events, including those with lower LIGO detection significance. Additionally, WINTER’s large( ∼ ) field of view (FoV) is well matched to fast mapping of the typical 10-20 deg uncertainty contour froma LIGO-Virgo merger alert. WINTER’s sensitivity, wide FoV, dedicated telescope, and flexible scheduler makeit unmatched in ability to systematically survey kilonova emission in the infrared.Beyond kilonovae, WINTER’s synoptic time-domain infrared survey will provide a new perspective on intrin-sically red or obscured astrophysical events. Building on observational and analysis techniques developed for thepathfinder Palomar Gattini-IR (PGIR) survey, WINTER will be able to detect and classify transient eventssuch as stellar mergers, tidal disruption events, failed supernovae, and even exoplanet transits around low-massstars. WINTER is the first seeing-limited instrument dedicated to systematic infrared time-domain searchesand will survey the available Northern sky every two weeks.
Despite the large number of optical transient survey instruments commissioned (or soon to be) over the pastdecade (e.g. ZTF, PanSTARRS, ATLAS, DECam, HSC, Rubin Observatory/LSST ), there have beenfew comparable efforts in the infrared. This is largely due to the high cost ( ∼
30x the cost-per-pixel of CCDs)of state-of-the-art HgCdTe sensors and their associated cryogenic instruments. Fast optical systems can widenthe field of view, but are costly as well, and incur an associated reduction in sensitivity because of the high skybackground noise in the infrared.WINTER achieves background-limited low-cost wide-field imaging using a custom camera based on an arrayof newly-developed commercial indium-gallium-arsenide (InGaAs) sensors. While InGaAs sensors have been usedin defense applications for years, they have only recently realized the low dark current needed for background-limited astronomical applications, when cooled with a simple thermoelectric cooler (TEC). Using custom readoutelectronics based on commercially-available field programmable gate array (FPGA) modules, WINTER is ableto make multiple non-destructive sensor reads per second, helping to reduce read noise. WINTER’s 1.0 ◦ x 1.2 ◦ FoV matches that of VIRCAM on the VISTA telescope, currently the largest IR focal plane, but with a twofoldincrease in fill factor and at a ∼
10x reduction in cost.WINTER couples its split focal plane array to a commercial-off-the-shelf (COTS) 1 m aperture telescope.The WINTER camera’s unique optical system slices the telescope focal plane into six identical optical channels,achieving a near-100% fill factor despite not being able to directly abut the packaged InGaAs focal planes.2he camera optics reimage the telescope’s F/6 beam down to F/3 at the focal planes to optimally sample theexpected seeing at Palomar. The telescope is robotically controlled with custom software which autonomouslyparses weather data from the site, produces an optimized schedule for each night, executes observing routinesand continually logs housekeeping data. With the fast pointing response of the telescope and low instrumentaloverheads we expect an on-sky duty cycle of >
90% and correspondingly high survey efficiency. An overview ofthe WINTER instrument is shown in Fig. 1, with individual subsystems described in detail in the followingsections.Figure 1: CAD rendering showing an overview of the WINTER instrument mounted on its telescope. Insetshows a cutaway view of the instrument, with important subsystems and optical elements labeled.
2. DETECTOR ARCHITECTURE2.1 Detectors
The custom camera for WINTER features six commercial large high-definition (HD) format (1920 x 1080 pixels)AP1020 Indium Gallium Arsenide (InGaAs) sensors currently being developed for our group by FLIR Electro-Optical Systems. These hybridized CMOS focal plane arrays will be the largest and highest-performing InGaAsdetectors on the market.The system builds on lessons learned from the development of a prototype camera built by the MIT groupand tested on the 100-inch DuPont Telescope at Las Campanas Observatory in Chile. This prototype instrumentused a single 640 x 512 pixel FLIR AP1121 InGaAs sensor, which is a direct predecessor of the AP1020 sensorsbeing developed for WINTER, with similar component architecture at both the level of the individual CTIA pixelunit cell and the output amplifier. This “DuPont Prototype” instrument demonstrated sky-background-limitedperformance in J-band with InGaAs sensors for the first time. Each of WINTER’s sensors is an 8 × scaled-up version of the AP1121, with 1920 x 1080 pixels on a 15 µ m pitch, and eight output channels. Additionalmodifications to the sensor package were made to facilitate a closer packing of the sensors in the camera focalplane, including a redesign of the mounting footprint, and the location and orientation of the pinch tube (seeSection 2.3) from which the sensor package is pumped to vacuum during assembly. Each of the six sensors interfaces to its own readout system being developed at MIT, which features customamplifier electronics, an FPGA interface, and readout software to implement non-destructive (i.e. sample-up-the-ramp (SUTR) or Fowler ) sampling modes which reduce the effective read noise by using multiple readsto fit the real-time count rate. A Xilinx ∗ FPGA was selected in order to build on firmware developed for the ∗ Xilinx Inc., San Jose, CA USA
Left
Thermal model made with ThermalDesktop ∗ of a TEC in the WINTER sensor packaging, usedfor calibrating thermal contact resistances and sizing heat pipes. Right:
CAD rendering of a prototype heatsinking assembly for the WINTER sensors, showing the sintered copper/water heat pipe soldered to two highpurity copper heat sinks bolted to the sensor at one end, and a liquid-cooled heat exchanger on the other.DuPont Prototype. The particular FPGA interface, an OpalKelly ∗ ZEM7310-A200 featuring a Xilinx Artix-7FPGA, balances cost with memory (1 Gb of on-board RAM, and 13.4 Mb of FPGA RAM) in order to allowflexibility in selecting a sampling algorithm.During operation, multiple sensor reads per second are streamed from the FPGA to a readout PC, using theUSB 3.0 port on the interface board. Alternate approaches using the FPGA to analyze the sensor reads andsend processed data (ie SUTR slopes or Fowler averages) were also investigated, but were abandoned in favorof simpler firmware code after laboratory tests confirmed that the USB 3.0 bandwidth could handle the fulldata volume of streaming raw reads. Moving the data reduction to the FPGA may be appropriate for similarcamera systems with more stringent constraints on computing power or communication bandwidth such as onspace-based or suborbital platforms. For WINTER, the USB 3.0 output from each FPGA interface is convertedto an OM4 multimode fiber, using an Icron † USB 3.0 Spectra 3022 converter. The ∼
30 m fiber transmits thedata from the instrument to a pair of processing PCs (1 PC per 3 sensors) in an electronics shed separated fromthe telescope dome (see Section 4). These PCs process the raw sensor reads into FITS format images which aredownlinked from the observatory in near-real-time.The FPGA interfaces to the InGaAs sensors via a series of custom electronics boards designed by the WINTERteam (see Ref. 3 for further details). Each sensor has (a) a “sensor board” which interfaces with the FLIR sensorand serves as a first stage preamplifier and motherboard for the additional boards, (b) an individual “powerboard” to sequence the timing of applied voltages during power up/off, regulate the input voltages, and providetransient voltage suppression and overcurrent protection for the sensor in the event of static dissipation orlightning strikes to the observatory, and (c) an “analog front end (AFE)” board with a second stage amplifierand analog to digital converters for output to the FPGA. Each sensor has its own readout electronics boards,forming six individual focal plane modules (FPMs).
To reduce the dark current sufficiently to achieve sky-background-limited observations, the WINTER sensorsare cooled with a two-stage TEC ‡ mounted inside the vacuum-packaged sensor housing. Measurements with theDuPont Prototype showed that FLIR’s InGaAs dark current falls below the Y-band sky at -40 ◦ C, but halves for ∗ OpalKelly Inc., Portland, OR USA † Icron/Maxim Integrated, San Jose, CA USA ‡ Laird Thermal Systems, Durham, NC USA
20, 23
Additional cooling below this transition point yields direct increases insensitivity. A proportional-integral-derivative (PID) loop running in software on the sensor PC communicateswith the FPGA interface board to control the TEC current supplied by the power board on each FPM. Thesensor packages cooled below room temperature to ∼ ◦ C to improve the cooling capacity of the TEC, while stillkeeping the delicate sensor above the 90th percentile dew point ( ∼ ◦ C) at Palomar Observatory. ∗ We expectthe TEC to generate 10-12 W in operation, and a cold side temperature of ∼ ◦ C. The 10 ◦ C base temperatureis maintained with a Opti Temp † OTC 1.0A chiller running a 30% propylene glycol/distilled water cooling loop.To avoid coolant leaks within the instrument, a sintered copper/water heat pipe connects the sensor package toa liquid heat exchanger mounted outside the instrument enclosure (see Fig. 2).
3. CAMERA3.1 Camera Optics
The driving requirement of the WINTER camera design is to maintain a high fill-factor field of view ( > ∼ ◦ x 1 ◦ FoV) despite not being able to directly abut the sensor elements. The design approach makesuse of a novel “fly’s eye” approach built from a series of identical, replicated sensor channels. The layout isshown in Figure 3. In this replicated approach, the telescope focal plane is sliced into six individual channels,one per sensor, with a bonded array of plano-convex field lenses. Each individual lens has the same aspect ratioas the HD-format sensors, and the bonded array fills the full 1 ◦ FoV of the telescope. A right angle fold mirrorsplits the instrument into two separate arms, which improves access to the optical components during alignment,integration and testing (AIT), while also reducing the mechanical moment of the instrument by keeping theinstrument center of mass closer to the telescope interface (see telescope description in Section 4.1). The fieldlens array (FLA) feeds six identical, replicated lens channels, each with their own sensor. Each channel has 10lenses with spherical only surfaces (labeled L1 through L10 in the diagram), comprising a collimator section (L1through L5) which forms a pupil near L5, and a camera section (L4 through L10). The 2:1 focal ratio of thecollimator to camera sections reimage the F/6 telescope beam to F/3 at the sensor, providing a ∼ (cid:48)(cid:48) /pixel platescale. The degree of symmetry about the instrument pupil recalls opposing paired Petzval designs, with glassesselected to improve thermal stability.Figure 3: Ray trace from Zemax of the WINTER camera, showing the FLA near the telescope focal plane, thefold mirror which splits the instrument into two arms, and the individual lens arrays. A series of optimizations was carried out in Zemax ‡ based on the requirement flowdown from WINTER’s sciencerequirements. Key constraints in the design optimization were to achieve the 2:1 focal reduction, maintain an ∗ Based on 5 years of Palomar Observatory 60 inch telescope weather station data from http://bianca.palomar.caltech.edu/maintenance/weather/user_gen_file.tcl † Opti Temp Inc., Traverse City, MI USA ‡ Zemax LLC, Kirkland, WA USA ∼ A detailed study was carried out to establish tolerances on the optical glass selection, lens fabrication, optome-chanical support structures, and AIT. The tolerance study was carried out using Zemax’s Monte Carlo simulationcapability. The studies were scripted to simulate different alignment procedures to select the optimal approach.This approach enforces the sequential nature of laboratory alignment and includes random errors in the opticalcompensation to more accurately represent the range of expected results.Based on the tolerance studies, an approach was selected which prioritized reducing the complexity of opticalAIT at the cost of increased precision in the lens fabrication process. By maintaining strict fabrication toler-ances, and obtaining precise measurements of the glass indices over the full WINTER passband, the mechanicalalignment tolerances were loosened to the scale of typical machining tolerances ( ∼ µ m). This approachrequires a single compensating element which is precisely adjusted independently for each channel. Based onthe simulations, the horizontal and vertical (with respect to the optical axis) positioning of the L6/L7 bondeddoublet were selected as the compensating degrees of freedom, requiring ± µ m positioning accuracy. The mostsensitive aspect of the optical design is the index of refraction of the optical glasses. To reduce the effect ofuncertainty in the glass parameters on the performance of the camera, all glass blanks (for each type of glass)were purchased from the same melt batch from the glass manufacturer ∗ . Additionally, samples of each purchasedglass type were sent to a third-party measurement shop † and measured at select optical test wavelengths andat regular intervals over the complete WINTER passband. These measurements reduced the uncertainty in theindex of refraction from 0.5% to 0.05%. After recieving the measurements, a final round of optical optimizationwas carried out before finalizing the lens parameters. ∗ All glasses used in the WINTER design are from Ohara Corporation, Branchburg, New Jersey USA † M Measurement Solutions Inc., Escondido, CA USA .2 Optomechanics The close vertical packing of the individual channels within each arm requires that all lenses after the FLA inthe optical path have truncations cut into the upper and lower edges. Based on the tolerance analysis, all lensgroups except for the compensators, were cemented together into three-channel monoliths (the “tiled lenses”).This is shown in Figure 3, where within each of the two arms of the instrument, all L2/3 groupings are bondedtogether, as are L4/L5 and L8/L9/L10. This approach is enabled by precision fabrication and alignment ( < µ mplacement errors) capabilities of the lens manufacturer, and was developed in collaboration with the selectedvendor, Optimax Systems. ∗ Only the compensator lenses are independently mounted for each channel. Theoptomechanical implementation of the WINTER lens system is described in detail in Ref. 2.
Table 1: WINTER bandpass definitions. λ ( µ m) Center Cut-On Cut-Off Y 1.02 0.97 1.07J 1.25 1.17 1.33Hs 1.60 1.49 1.68The WINTER filters are based on the canonical Mauna Kea Ob-servatory (MKO) filter set, with a modified shortened H-band(Hs-band) with a long-wave cutoff tuned to the 1.7 µ m InGaAsbandgap cutoff. A single 3-position filter tray sits 50 mm in frontof the FLA in the converging, telecentric, F/6 beam of the tele-scope. While the ± ◦ angle of incidence (AOI) variation withthe F/6 beam somewhat reduces the slope of the band edges, plac-ing the filter in the converging beam ensures spectral uniformityacross the FoV, simplifying photometric calibration. The customWINTER bandpass filters were built by Asahi Spectra Co. † on10 mm thick fused silica, and achieve >
99% throughput in bandwith band-edge slopes ( λ
90% Trans − λ
10% Trans ) /λ
10% Trans ∼ < µ m to prevent blue leaks at shorter wavelengths where the InGaAsdetector is still sensitive. The bandpass specifications are listed in Table 1, and the measured spectra are shownin Fig. 5.
4. OBSERVATORY4.1 Telescope
WINTER uses a COTS 1 m corrected Dall-Kirkham (CDK) telescope with fast direct-drive altitude/azimuthpointing motors, and a instrument mount which compensates for sky rotation. The telescope has all fused silicaoptics, including two sets of three corrector lenses mounted in lens barrels within each of the two Nasmyth plat-forms. A flat, tertiary (M3) mirror can steer the beam towards either Nasmyth port. PlaneWave Instruments ‡ made several modifications to their base PW1000 1-Meter Observatory System for WINTER, including optimiz-ing the mirror antireflective (AR) coatings for infrared operation and developing a M2 focusing system. TheNasmyth port where the WINTER instrument is mounted has corrector lenses with IR-optimized AR coatings,while the opposite port is has optical-optimized coatings to allow for a future visible-light instrument to bemounted there. The main telescope components and instrument mount are shown in Fig. 1, and a photographof the telescope during laboratory testing is shown in Fig. 6.The telescope is controlled by a small embedded Windows PC § which runs a http command server fromPlaneWave. This server takes pointing and operating commands from the main WINTER observatory controlsoftware described in Section 4.3. ∗ Optimax Systems Inc., Ontario, NY USA † Asahi Spectra Co., Ltd. Tokyo, Japan ‡ PlaneWave Instruments, Adrian, MI USA § ARK-1550-9551, Advantech Co. Ltd - North America, Milpitas, CA USA .8 1.0 1.2 1.4 1.6 1.8 Wavelength [ m] T r a n s m i ss i o n [ % ] YJHs
Figure 5: Measured transmission for each of the three WINTER filters. Transmission data are measured witha parallel beam, and will deviate slightly ( < WINTER will operate at Palomar Observatory at an altitude of ∼ and PGIR. The dome communications and operating protocols derivedfrom WINTER’s observational goals are detailed in Ref. 1. A temperature-controlled shed a few meters fromthe dome houses all of the control electronics for the WINTER instrument, including the subsystem control PCs,instrument power supplies, and chiller for the sensor liquid coolant loop.
WINTER’s robotic operating system is controlled by a custom control program. This program, the WINTERSupervisory Program (WSP), is written in Python 3.7 and uses
PyQt5 ∗ , a set of Python bindings to the Qt C++application framework † , enabling multithreaded operation. A series of cron jobs on the observatory control PCrun schedule production code to generate a nightly schedule (see Section 5.1), and initiate the WSP. Elements of ∗ Riverbank Computing, † Qt Group,
Left:
The WINTER 1 m PlaneWave PW1000 telescope during laboratory testing at MIT.
Right:
Site layout of the WINTER site at Palomar Observatories indicating the location of the WINTER enclosure,and photo of the telescope pad and dome.the control software, including the weather decision-making and communications with the telescope are adaptedfrom the MINERVA project, telescopes.,
25, 26 which share heritage with the Robo-AO system which operates thePalomar 60 inch observatory. While the WSP is designed to make appropriate decisions about when to openthe dome and initiate observations, the robotic control can be overridden by the on-site operator at the nearby200 inch Hale Telescope who will close the WINTER dome in the event of inclement weather or other undetectedproblems.The WSP runs a series of threads to (1) load the current scheduled observation (2) listen for target of oppor-tunity (ToO) events (3) listen for commands from the terminal and externally over TCP/IP, (4) execute thesecommands from a priority queue and dispatch commands to the sensor and telescope PCs, (5) log housekeepingdata about the telescope and instrument state to a telemetry database, and (6) log all observations to a obser-vation database. The telemetry database is stored in dirfile binary format using GetData
PyGetData ∗ Pythonbindings. This binary database format can be used with a number of software tools developed at the Universityof Toronto for stratospheric balloon programs, including a realtime plotting tool, Kst † , and owl , ‡ a modulardashboard program. The observation log is a SQlite database which can easily be queried, or even completelycopied and transferred off the control computer for analysis. It is also used by the scheduler program to furtheroptimize observing plans for future nights, and recover from interrupted schedules. The structure of the databaseis defined in a separate configuration file to allow for easy modification of the database schema, without havingto modify the software responsible for recording observations.
5. OBSERVING STRATEGY
WINTER’s preliminary observing plan interweaves two systematic surveys: a
Shallow-Wide survey in WIN-TER’s deepest filter (J), and a
Deep-Fast survey in all three bands. The Shallow-Wide survey will cover 4050square degrees at a 9-day cadence, stepping fields every four epochs to cover the full northern sky in the firstyear of operations to J AB = 19.2 (single-visit) or 20.5 (coadded). Completing this survey in the first year is keyto building a sufficiently deep sky map to enable transient detections over WINTER’s full volume by the timeLIGO-Virgo O4 begins. The Deep-Fast survey in Y+J+Hs observes 450 deg at a 3-day cadence, rotating tonew fields after twenty epochs. ∗ The GetData Project, http://getdata.sourceforge.net/ † https://kst-plot.kde.org/ ‡ https://github.com/BlastTNG/flight/tree/master/owl .1 Scheduling To determine an optimized observing for multiple simultaneous observing programs, WINTER uses a customizedscheduler that builds on the extensive functionality of the ZTF scheduler. The ZTF scheduler balances manydifferent observing programs with unique cadences while also maximizing data quality by selecting fields throughvolumetric weighting. In this scheme, the observable sky is split into discrete fields matching the instrumentFoV. At any time, the most desirable field to observe is that which probes the greatest limiting volume for anygiven exposure. The limiting volume for an exposure is related to the distance at which a source of absolutemagnitude M will be detected. This is constrained by seeing, sky brightness, and instrument noise, which allfactor into the limiting magnitude (m lim ) for an exposure.The ZTF scheduler offers several modes for simulations and on-sky observing: (1) Queue observing ingests apredefined list of fields and steps through each field sequentially, (2) “Greedy” observing continuously selects thebest target (based on the volumetric weighting scheme) for a given time and recalculating before each target, (3)Optimized observing uses the Gurobi linear optimizer to solve a travelling salesperson problem to optimize eachnight of observing. All three modes are used in WINTER’s observing program: queue observing steps throughobservations for defined targets of opportunity (ToO), such as LIGO alerts, optimized observing balances sciencesurveys with varying cadences, and greedy observing fills in reference images of the sky between science surveysand ToOs.This scheduling code is used for simulating entire observing programs, as well as producing a nightly schedule.To develop and optimize observing program design, simulations of several years of observing are carried outusing historical weather data from Palomar Observatory to realistically estimate the typical observing conditionsthroughout the year. For actual observing, an optimized schedule is created each night which incorporates therecorded data in the observation database.For catching GW detection alerts and scheduling ToO observations, WINTER uses the GROWTH ToOMarshal.
31, 32
The GROWTH Marshal web portal ingests GW alerts distributed via the LIGO-Virgo GCNsystem ∗ , and uses the GW EM counterpart search package gwemopt as a backend to help produce optimizedobserving schedules for WINTER and other kilonova discovery engines such as ZTF, PGIR, and DECam. TheToO marshal is also integrated into WINTER’s analysis pipeline (see Section 5.2) to publish detections oftransient candidates. Because the GROWTH Marshal software is tuned towards visible wavelength followupobservations which peak in the first few days after a BNS merger, some modifications to the approach will benecessary to tune the observations for longer-lasting IR followup. Depending on the frequency of ToO events,and the performance of the upgraded LIGO O4 survey, it may be necessary to add additional features to theWINTER scheduling architecture, including a galaxy-weighted wide-field search mode which weighs individualfields based on published galaxy catalogs while still taking advantage of WINTER’s ability to cover the full LIGOpositional error contour, and the ability to interleave multiple ToO schedules to handle overlapping events. Thisongoing development effort will continue through WINTER’s commissioning and first year of observations.The ToO schedules are written in the same ZTF-style database format as the nightly schedules and are loadedthe same way into the WSP software. As soon as all possible observations in the ToO schedule are completed,the observatory will switch over automatically to the nightly schedule. The WINTER data processing approach draws on the heritage of the mature transient detection and analysispipelines developed at Caltech for the successful ZTF and PGIR ) programs. Similar to PGIR, each WINTERobservation will be a series of dithered exposures to facilitate longer exposure times on the bright sky background.Images acquired in each dithered sequence will be stacked using the Drizzle algorithm. The Drizzle algorithmwill enable reconstruction of the point spread function (PSF) in images where the PSF is undersampled. Astro-metric and photometric calibration will be carried out using the 2MASS point source catalog for the J and Hfilters and the Pan-STARRS catalog for the Y filter. For each field, data taken from at least five repeat visitsevenly spaced during the first year of observations will be stacked to build deep ( ≈
21 mag) reference images.All subsequent “science” images will be fed to the image differencing pipeline, where they will be subtracted ∗ LIGO Scientific Collaboration Public Alerts, https://emfollow.docs.ligo.org/userguide/index.html
Left:
WINTER’s available night sky split into ∼ ◦ x 1 ◦ fields, used forschedule optimization and image cataloguing. Right:
A sample hit map from a optimization run simulating athe first year of WINTER observations. The blue background represents the creation of a deep J-band seriesof reference images for the northern sky. The high-visit-density stripe represents a deep-fast observing programoptimized to select patches of sky that can be revisited every three nights. Historical weather data from PalomarObservatory is used to make reasonable estimates of favorable observing conditions.from the corresponding reference image using the ZOGY algorithm. The subtracted image will be then fed tothe Astromatic package SExtractor to detect candidate transient sources. A machine-learning based classifierwill be used to help automatically distinguish between real astrophysical transient sources and image subtrac-tion artifacts. The candidates that are flagged as real transients will be uploaded to a web-portal for humanvetting. Candidate vetting and follow-up by various telescopes will be coordinated by a web portal similar tothe GROWTH Marshal.
31, 32
A brand new Fritz system, the next generation of the GROWTH Marshal, iscurrently being developed as an open source project that integrates and extends two projects:
Kowalski andSkyPortal.
6. CURRENT STATUS
The WINTER instrument development effort is ongoing, with a planned commissioning date of summer 2021.The telescope and observatory control system is currently undergoing laboratory tests at MIT, and be installedat the site once infrastructure improvements at the Palomar Observatory site are completed in early 2021. Weare currently stress-testing the robotic system which takes in real-time weather and site data from the PalomarObservatory and the WINTER dome from dedicated telemetry servers. The WINTER camera lenses are currentlyin fabrication, and we expect the camera optics to be integrated during late spring 2021. The optomechanicalmounting approach for both the three-channel bonded lenses and the individually-mounted compensator lenseshas been demonstrated through detailed prototyping and characterization, as described in Ref. 2. The WINTERsensors are currently being fabricated, and the readout electronics and firmware have been prototyped and arecurrently being tested (see Ref. 3) ahead of upcoming systems-level tests of the focal plane modules withengineering-grade sensors.
7. CONCLUSION
WINTER will be a powerful new tool for researching the transient sky in the infrared, a relatively unexploredregime in which a host of astrophysical phenomena are most visible, including the thermal emission from r -process element synthesis in the ejecta of neutron star mergers. Observing in the Y, J, and Hs bands nearthe expected emission peak of BNS kilonovae, WINTER will robotically execute follow-up observations of GWevents detected by the LIGO-Virgo instruments. WINTER is the first seeing-limited instrument dedicated tosystematic infrared time-domain searches, and will perform multiple simultaneous observing programs to identifyIR transients. WINTER’s planned commissioning beginning in summer 2021 will allow sufficient time to builda full map of the available sky in J-band before the LIGO-Virgo O4 observing run, enabling WINTER’s fullkilonova search capability in its first year of science operations. WINTER is part of a new generation of IR11ime-domain instruments, including DREAMS at Siding Spring Observatory in New South Wales, Australia, PRIME on the South African Astronomical Observatory, and proposed Antarctic instruments like Cryoscope which will open new windows into the transient sky. ACKNOWLEDGMENTS
WINTER’s construction is made possible by the National Science Foundation under MRI grant number AST-1828470. We also acknowledge significant support from the California Institute of Technology, the CaltechOptical Observatories (COO), the Bruno Rossi Fund of the MIT Kavli Institute for Astrophysics and SpaceResearch, and the MIT Department of Physics and School of Science. The collaboration also acknowledges theongoing support and contributions to the observatory by the COO staff. Eric Bellm from the University ofWashington, Reed Riddle from COO, and Javier Romualdez from Princeton University contributed support andguidance on the WINTER controls software.
REFERENCES [1] Frostig, D. and The WINTER Collaboration, “Design requirements for the wide-field infrared transientexplorer (winter),” in [Proc. SPIE], (113) (2020).[2] Hinrichsen, E. and The WINTER Collaboration, “Opto-mechanical mounting design of the wide-field in-frared transient explorer (WINTER) “fly’s eye” camera lenses,” in [Proc. SPIE], (115) (2020).[3] Malonis, A. and The WINTER Collaboration, “Detector architecture of the wide-field infrared transientexplorer (WINTER) InGaAs camera,” in [Proc. SPIE], (105) (2020).[4] The LIGO Scientific Collaboration and The Virgo Collaboration, “GW170817: Observation of gravitationalwaves from a binary neutron star inspiral,”
Physical Review Letters (Oct 2017).[5] The LIGO Scientific Collaboration and The Virgo Collaboration, “Multi-messenger observations of a binaryneutron star merger,”
The Astrophysical Journal , L12 (Oct 2017).[6] Kasliwal, M. M., Nakar, E., Singer, L. P., Kaplan, D. L., Cook, D. O., Van Sistine, A., Lau, R. M.,Fremling, C., Gottlieb, O., Jencson, J. E., and et al., “Illuminating gravitational waves: A concordantpicture of photons from a neutron star merger,”
Science , 1559–1565 (Oct 2017).[7] Kasen, D., Metzger, B., Barnes, J., Quataert, E., and Ramirez-Ruiz, E., “Origin of the heavy elements inbinary neutron-star mergers from a gravitational-wave event,”
Nature , 80–84 (Oct 2017).[8] LIGO Scientific Collaboration and Virgo Collaboration, “Prospects for Observing and LocalizingGravitational-Wave Transients with Advanced LIGO and Advanced Virgo,”
Living Reviews in Relativity ,1 (Feb. 2016).[9] Kasen, D., Badnell, N. R., and Barnes, J., “Opacities and spectra of ther-process ejecta from neutron starmergers,” The Astrophysical Journal , 25 (Aug 2013).[10] Barnes, J., Kasen, D., Wu, M.-R., and Mart´ınez-Pinedo, G., “Radioactivity and thermalization in the ejectaof compact object mergers and their impact on kilonova light curves,”
The Astrophysical Journal , 110(Sep 2016).[11] De, K., Hankins, M. J., Kasliwal, M. M., Moore, A. M., Ofek, E. O., Adams, S. M., Ashley, M. C. B., Babul,A.-N., Bagdasaryan, A., Burdge, K. B., and et al., “Palomar gattini-ir: Survey overview, data processingsystem, on-sky performance and first results,”
Publications of the Astronomical Society of the Pacific ,025001 (Jan 2020).[12] Kasliwal, M. M., Adams, S., Andreoni, I., Ashley, M., Blagorodnova, N., De, K., Frostig, D., Furesz, G.,Jencson, J., Hankins, M., Helou, G., Lau, R., Moore, A., Ofek, E., Simcoe, R., Sokoloski, J., Soon, J.,Tinyanont, S., and Travouillon, T., “The dynamic infrared sky,” (2019).[13] Bellm, E. C., Kulkarni, S. R., Graham, M. J., Dekany, R., Smith, R. M., Riddle, R., Masci, F. J., Helou,G., Prince, T. A., Adams, S. M., and et al., “The zwicky transient facility: System overview, performance,and first results,”
Publications of the Astronomical Society of the Pacific , 018002 (Dec 2018).[14] Chambers, K. C. et al., “The pan-starrs1 surveys,” (2019).[15] Tonry, J. L., “An early warning system for asteroid impact,”
Publications of the Astronomical Society ofthe Pacific , 58–73 (Jan 2011). 1216] Flaugher, B., Diehl, H. T., Honscheid, K., Abbott, T. M. C., Alvarez, O., Angstadt, R., Annis, J. T.,Antonik, M., Ballester, O., Beaufore, L., and others., “The dark energy camera,”
The Astronomical Jour-nal , 150 (Oct 2015).[17] Miyazaki, S. et al., “Hyper Suprime-Cam,” in [
Ground-based and Airborne Instrumentation for AstronomyIV ], McLean, I. S., Ramsay, S. K., and Takami, H., eds., , 327 – 335, International Society for Opticsand Photonics, SPIE (2012).[18] LSST Science Collaboration, “LSST science book, version 2.0,” (2009).[19] Sutherland, W., Emerson, J., Dalton, G., Atad-Ettedgui, E., Beard, S., Bennett, R., Bezawada, N., Born,A., Caldwell, M., Clark, P., and et al., “The visible and infrared survey telescope for astronomy (vista):Design, technical overview, and performance,”
Astronomy & Astrophysics , A25 (Feb 2015).[20] Simcoe, R. A., F˝ur´esz, G., Sullivan, P. W., Hellickson, T., Malonis, A., Kasliwal, M., Shectman, S. A.,Kollmeier, J. A., and Moore, A., “Background-limited imaging in the near infrared with warm ingaassensors: Applications for time-domain astronomy,”
The Astronomical Journal , 46 (Jan 2019).[21] Glendinning, R., Beard, S. M., Mountain, C. M., Pettie, D. G., Pickup, D. A., and Wade, R., “Implemen-tation of a charge integration system in a low-background application,” in [
Instrumentation in AstronomyVII ], Crawford, D. L., ed., , 34 – 42, International Society for Optics and Photonics, SPIE (1990).[22] Fowler, A. M. and Gatley, I., “Demonstration of an Algorithm for Read-Noise Reduction in Infrared Arrays,”ApJ , L33 (Apr. 1990).[23] Sullivan, P. W., Croll, B., and Simcoe, R. A., “Precision of a low-cost ingaas detector for near infraredphotometry,”
Publications of the Astronomical Society of the Pacific , 1021–1030 (Sep 2013).[24] Tokunaga, A. T., “The Mauna Kea Observatories near-infrared filter set.” Institute for Astronomy Univer-sity of Hawaii [Online] (May 2005).(Accessed: 11 November 2020).[25] Swift, J. J., Bottom, M., Johnson, J. A., Wright, J. T., McCrady, N., Wittenmyer, R. A., Plavchan, P.,Riddle, R., Muirhead, P. S., Herzig, E., and et al., “Miniature exoplanet radial velocity array i: design,commissioning, and early photometric results,”
Journal of Astronomical Telescopes, Instruments, and Sys-tems , 027002 (Apr 2015).[26] Wilson, M. L., Eastman, J. D., Cornachione, M. A., Wang, S. X., Johnson, S. A., Sliski, D. H., Schap III,W. J., Morton, T. D., Johnson, J. A., McCrady, N., and et al., “First radial velocity results from the minia-ture exoplanet radial velocity array (minerva),” Publications of the Astronomical Society of the Pacific ,115001 (Sep 2019).[27] Riddle, R. L., Burse, M. P., Law, N. M., Tendulkar, S. P., Baranec, C., Rudy, A. R., Sitt, M., Arya, A.,Papadopoulos, A., Ramaprakash, A. N., and Dekany, R. G., “The Robo-AO software: fully autonomousoperation of a laser guide star adaptive optics and science system,” in [
Adaptive Optics Systems III ], Eller-broek, B. L., Marchetti, E., and V ˜A © ran, J.-P., eds., , 993 – 1001, International Society for Opticsand Photonics, SPIE (2012).[28] Romualdez, J., Design, Implementation, and Operational Methodologies for Sub-Arcsecond Attitude Deter-mination, Control, and Stabilization of the Super-pressure Balloon-borne Imaging Telescope (SuperBIT) ,PhD thesis, University of Toronto, Toronto, Canada (2018).[29] Bellm, E. C., Kulkarni, S. R., Barlow, T., Feindt, U., Graham, M. J., Goobar, A., Kupfer, T., Ngeow, C.-C.,Nugent, P., Ofek, E., and et al., “The zwicky transient facility: Surveys and scheduler,”
Publications of theAstronomical Society of the Pacific , 068003 (Apr 2019).[30] Gurobi Optimization, L., “Gurobi optimizer reference manual,” (2020).[31] Coughlin, M. W., Ahumada, T., Anand, S., De, K., Hankins, M. J., Kasliwal, M. M., Singer, L. P., Bellm,E. C., Andreoni, I., Cenko, S. B., and et al., “Growth on s190425z: Searching thousands of square degreesto identify an optical or infrared counterpart to a binary neutron star merger with the zwicky transientfacility and palomar gattini-ir,”
The Astrophysical Journal , L19 (Oct 2019).[32] Kasliwal, M. M., Cannella, C., Bagdasaryan, A., Hung, T., Feindt, U., Singer, L. P., Coughlin, M., Fremling,C., Walters, R., Duev, D., and et al., “The growth marshal: A dynamic science portal for time-domainastronomy,”
Publications of the Astronomical Society of the Pacific , 038003 (Feb 2019).1333] Coughlin, M. W., Tao, D., Chan, M. L., Chatterjee, D., Christensen, N., Ghosh, S., Greco, G., Hu, Y.,Kapadia, S., Rana, J., and et al., “Optimizing searches for electromagnetic counterparts of gravitationalwave triggers,”
Monthly Notices of the Royal Astronomical Society , 692–702 (Apr 2018).[34] Masci, F. J., Laher, R. R., Rusholme, B., Shupe, D. L., Groom, S., Surace, J., Jackson, E., Monkewitz, S.,Beck, R., Flynn, D., and et al., “The zwicky transient facility: Data processing, products, and archive,”
Publications of the Astronomical Society of the Pacific , 018003 (Dec 2018).[35] Fruchter, A. and Hook, R., “Drizzle: A method for the linear reconstruction of undersampled images,”
Publications of the Astronomical Society of the Pacific , 144–152 (Feb 2002).[36] Cutri, R. M., Skrutskie, M. F., van Dyk, S., Beichman, C. A., Carpenter, J. M., Chester, T., Cambresy, L.,Evans, T., Fowler, J., Gizis, J., Howard, E., Huchra, J., Jarrett, T., Kopan, E. L., Kirkpatrick, J. D., Light,R. M., Marsh, K. A., McCallon, H., Schneider, S., Stiening, R., Sykes, M., Weinberg, M., Wheaton, W. A.,Wheelock, S., and Zacarias, N., “VizieR Online Data Catalog: 2MASS All-Sky Catalog of Point Sources(Cutri+ 2003),”
VizieR Online Data Catalog II , 246 (June 2003).[37] Flewelling, H. A., Magnier, E. A., Chambers, K. C., Heasley, J. N., Holmberg, C., Huber, M. E., Sweeney,W., Waters, C. Z., Calamida, A., Casertano, S., Chen, X., Farrow, D., Hasinger, G., Henderson, R., Long,K. S., Metcalfe, N., Narayan, G., Nieto-Santisteban, M. A., Norberg, P., Rest, A., Saglia, R. P., Szalay,A., Thakar, A. R., Tonry, J. L., Valenti, J., Werner, S., White, R., Denneau, L., Draper, P. W., Hodapp,K. W., Jedicke, R., Kaiser, N., Kudritzki, R. P., Price, P. A., Wainscoat, R. J., Builders, P. S., Chastel, S.,McLean, B., Postman, M., and Shiao, B., “The pan-starrs1 database and data products,” (2019).[38] Zackay, B., Ofek, E. O., and Gal-Yam, A., “Proper Image Subtraction—Optimal Transient Detection,Photometry, and Hypothesis Testing,” ApJ , 27 (Oct. 2016).[39] Bertin, E. and Arnouts, S., “SExtractor: Software for source extraction.,” Astronomy and AstrophysicsSupplement , 393–404 (June 1996).[40] Duev, D. A., Mahabal, A., Masci, F. J., Graham, M. J., Rusholme, B., Walters, R., Karmarkar, I., Frederick,S., Kasliwal, M. M., Rebbapragada, U., and et al., “Real-bogus classification for the zwicky transient facilityusing deep learning,”
Monthly Notices of the Royal Astronomical Society , 3582–3590 (Aug 2019).[41] van der Walt, S. J., Crellin-Quick, A., and Bloom, J. S., “Skyportal: An astronomical data platform,”
Journal of Open Source Software (37), 1247 (2019).[42] Soon, J., Moore, A. M., Kasliwal, M. M., Lau, R. M., De, K., Travouillon, T. D., Jones, M. I., Ofek, E.,Smith, R., Terebizh, V., McKenna, D., Hale, D., Delacroix, A., Adams, S. M., Jencson, J. E., Ashley, M.,Burnham, J., Sokoloski, J. L., Bland-Hawthorn, J., Freeman, K. C., Marco, O. D., Cooke, J., Bland, P.,Ryder, S., Soria, R., Antoszewski, J., Heger, A., Spitler, L., and Simcoe, R., “Opening the dynamic infraredsky,” in [ Ground-based and Airborne Telescopes VII ], Marshall, H. K. and Spyromilio, J., eds., , 1406– 1417, International Society for Optics and Photonics, SPIE (2018).[43] Sumi, T. and The PRIME Collaboration, “PRIME prime-focus infrared microlensing experiment.” Na-tional Astronomical Observatory of Japan NIR Workshop [Online] (July 2019).[44] Smith, R. M., Kasliwal, M. M., Ashley, M. C. B., Beichman, C. A., De, K., Greffe, T., Hale, D., Hankins,M., Fucik, J. R., Mroz, P., Porter, M., Shapiro, C. A., and Travouillon, T. D., “Telescope design forunprecedented survey speed in the infrared,” in [Proc. SPIE],
Ground-based and Airborne Telescopes VII11445