StrayCats: A catalog of NuSTAR Stray Light Observations
Brian W. Grefenstette, Renee M. Ludlam, Ellen T. Thompson, Javier A. Garcia, Jeremy Hare, Amruta D. Jaodand, Roman A. Krivonos, Kristin K. Madsen, Guglioelmo Mastoserio, Catherine M. Slaughter, John A. Tomsick, Daniel Wik, Andreas Zoglauer
DDraft version February 3, 2021
Typeset using L A TEX twocolumn style in AASTeX63
StrayCats : A catalog of NuSTAR Stray Light Observations
Brian W. Grefenstette , Renee M. Ludlam , ∗ Ellen T. Thompson , Javier A. Garc´ıa ,
1, 3
Jeremy Hare , † Amruta D. Jaodand , Roman A. Krivonos , Kristin K. Madsen ,
6, 7
Guglielmo Mastroserio , Catherine M. Slaughter , John A. Tomsick , Daniel Wik , andAndreas Zoglauer Cahill Center for Astronomy and Astrophysics, California Institute of Technology, Pasadena, CA 91125, USA Space Sciences Laboratory, 7 Gauss Way, University of California, Berkeley, CA 94720-7450, USA Dr. Karl Remeis-Observatory and Erlangen Centre for Astroparticle Physics, Sternwartstr. 7, 96049 Bamberg, Germany NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA Space Research Institute, Russian Academy of Sciences, Profsoyuznaya 84/32, 117997 Moscow, Russia Space Radiation Laboratory, Caltech, 1200 E California Blvd, Pasadena, CA 91125 Astrophysics Science Division, NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA Caltech Summer Undergraduate Research Fellowship Department of Physics and Astronomy, University of Utah, 201 James Fletcher Building, Salt Lake City, UT 84112, USA (Accepted Jan 27, 2021)
Submitted to ApJABSTRACTWe present
StrayCats : a catalog of
NuSTAR stray light observations of X-ray sources. Stray lightobservations arise for sources 1–4 ◦ away from the telescope pointing direction. At this off-axis angle,X-rays pass through a gap between optics and aperture stop and so do not interact with the X-rayoptics but, instead, directly illuminate the NuSTAR focal plane. We have systematically identified andexamined over 1400 potential observations resulting in a catalog of 436 telescope fields and 78 straylight sources that have been identified. The sources identified include historically known persistentlybright X-ray sources, X-ray binaries in outburst, pulsars, and Type I X-ray bursters. In this paperwe present an overview of the catalog and how we identified the
StrayCats sources and the analysistechniques required to produce high level science products. Finally, we present a few brief examples ofthe science quality of these unique data.
Keywords: surveys INTRODUCTIONCompact objects in our galaxy provide an excellentlaboratory in which to study matter in extreme con-ditions. Of most interest are neutron stars (NS) andblack holes (BH) in binary systems, where the compactobject accretes material from its companion star eitherthrough Roche lobe overflow of through a stellar windfrom the companion. The inflowing material forms anaccretion disk around the compact object with tempera-
Corresponding author: Brian [email protected] ∗ NASA Einstein Fellow † NASA Postdoctoral Program Fellow tures hot enough to produce copious amounts of thermalX-rays and giving rise to a corona of non-thermal elec-trons emitting in the hard X-ray band.The hard X-ray (E ≥ ≥
20 keV) spectrumof the X-ray binaries in the Galactic plane have beensurveyed with low spectral resolution instruments onthe
INTErnational Gamma-Ray Astrophysics Labora-tory ( INTEGRAL , Winkler et al. 2003) and the NeilGehels
Swift
Observatory (Gehrels et al. 2004).Targeted observations with
NuSTAR (The
NuclearSpectroscopic Telescope ARray
Harrison et al. 2013)have demonstrated the diagnostic power of a sensitive a r X i v : . [ a s t r o - ph . H E ] F e b Grefenstette et al.
Figure 1.
Schematic of the path of stray light photons. (
Left ): CAD rendering of the focal plane and the aperture stopassembly. (
Right ): Red traces show the stray light paths that survive to the focal plane after passing around the aperture stop(AP1, AP2, and AP3) rings and the “can” housing the detectors. The height offset from the focal plane to AP1 is shown onthe right. Figures adapted from Madsen et al. (2017a). instrument over the 3–80 keV bandpass. However, whenthese sources go into an X-ray bright state they resultin extremely high count rates and correspondingly hightelemetry loads. Because of this, many observations ofbright sources are short in duration ( ≈ Swift , NuSTAR is not a rapidly slewing instrument, sorepeated short monitoring observations of to the sametarget are not generally possible due to scheduling con-straints and require “Target of Opportunity” programsthat can take days or a week to get on target once anobservation is trigger.Fortunately,
NuSTAR can also serendipitously ob-serve bright X-ray binaries through “stray light.” While
NuSTAR is well-known as the first focusing hard X-ray satellite in orbit, the open geometry of the mastthat connects the optics to the detectors allows for thepossibility of stray light (light that has not been fo-cused by the optics) illuminating detectors. This is typi-cally referred to as “aperture flux” since the light passesthrough the open area of the aperture stops (see Figure 1) and occurs for sources that are roughly 1–4 ◦ fromthe center of the NuSTAR field-of-view (Madsen et al.2017a).For most
NuSTAR observations, the dominant sourceof aperture X-ray emission is the cosmic X-ray back-ground (hereafter “aperture” CXB, or aCXB). This isthe superposition of X-ray light from a uniform back-ground of (unresolved) AGN in the 1–4 ◦ annulus. Thiscontribution to the NuSTAR background has been welldocumented (e.g., Wik et al. 2014) and is generally de-scribed by a spatial gradient in the
NuSTAR backgroundacross the field of view.When stray light comes from a single off-axis sourcethe emission geometry is much simpler. Instead of a“gradient” in the background, we instead observe aneasily-identified shadow of the aperture stop ring sharplycutting off the source (Figure 2). Because the X-rays donot interact with the
NuSTAR optics, the response ofthe instrument is somewhat more straight forward aswell. This comes at the reduced effective area for straylight observations compared with pointed observations. trayCats intentionally placing a targetso that it is observed via stray light have been under-taken for a number of bright X-ray binaries. This wasdone to provide contiguous observations while reducingthe count rate (and thus the telemetry load) and to po-tentially extend the spectral range covered by
NuSTAR beyond the 78.4 keV cutoff in the optics response. Oneexample is the observation of the Crab nebula seen viastray light which allows for a simple, unique measure-ment of the spectral shape and flux of the Crab (Madsenet al. 2017b).In this paper we describe the
NuSTAR
StrayCats :a catalog of NuSTAR stray light observations (bothserendipitous and intentional) throughout the mission.In § § StrayCats spectroscopic analysis as well as the toolsthat we have developed for streamlining the extractionof
StrayCats high level science products, such as spec-tra and lightcurves. In § § StrayCats data sets to give a demonstration of the typeand quality of data. However, we generally will reservea more detailed follow-up analysis of individual sourcesto future work. DATA PROCESSING AND STRAY LIGHTIDENTIFICATIONIdentifying observations contaminated by stray light isnon-trivial, due to the variability in the
NuSTAR back-ground contributions, the presence of multiple sourcesin the field of view (FoV), and the different amounts ofdetector area illuminated by the stray light sources atdifferent off-axis angles. We utilized two complementarymethods: an a priori approach based on the locationof known bright X-ray sources detected by
Swift -BATand
INTEGRAL ; and a “bottom up” approach usinga statistical approach to identify potential stray lightcandidate observations.2.1.
An a priori approach
We use the
Swift -BAT 105-month all-sky catalog (Ohet al. 2018) of sources along with the
INTEGRAL | b | < ◦ ) catalog (Krivonos et al.2012). These catalogs are both used by the NuSTAR
Science Operations Center (SOC) to identify and miti-gate sources of stray light contamination for science ob-servations. To estimate the amount of stray light in a https://nustarstraycats.github.io/ given observation, we utilize the nustar stray light IDL code . This contains a model of the size, shape,and relative positions of the focal plane structures (seenin Fig 1) and the bench that holds the NuSTAR optics.For a given
NuSTAR pointing orientation and a givenstray light target, the “shadow” from the aperture stopand the optics bench are projected onto the focal planefor each detector to estimate the stray light contribu-tion.Estimating the strength of the stray light is done byextrapolating the measured spectrum in the
Swift -BAT/
INTEGRAL bands down into the
NuSTAR straylightbandpass (3–20 keV); a process which frequently resultsin overestimating the
NuSTAR flux for sources that havecurvature in the hard X-ray bandpass or have a predomi-nantly thermal spectrum. Nonetheless, there is usually areasonable match between the brightest catalog sourcesand the stray light in
NuSTAR .As a first step, we produce an estimate catalog ofall
NuSTAR observations within 4 ◦ of a “bright” X-ray source in one of our reference catalogs where wetypically define the minimum flux level for a persistent,bright source to be > INTEGRAL and
Swift . This re-sults in several hundred
NuSTAR stray light candidateobservations. For each observation we produce the esti-mated stray light map, and visually compare the resultsto the observed data. As many of these sources are vari-able and the internal model of the structures may notbe entirely accurate, this does require a human-in-the-loop for positive identification of a stray light candidate.While this process is able to positively identify dozensof stray light observations, it is both inefficient and doesnot catch any stray light observations of new or inter-mittently transient sources.2.2.
A more statistical approach
Rather than requiring any prior knowledge of a nearbybright target, we instead use the observed data to iden-tify stray light candidates. Since the area of the skyaccessible to each
NuSTAR telescope for stray light aredifferent, we treat the two separately.We first remove contributions from the primary tar-get by first excising all counts from within 3 (cid:48) of the es-timated target location. This large exclusion region at-tempts to account for any astrometric errors between theestimated J2000 coordinates for the target and wherethe target is actually observed to reduce the “PSF bleed”from bright primary targets. For bright primary tar- https://github.com/NuSTAR/nustar-gen-utils Grefenstette et al.
Figure 2.
NuSTAR quick look images in “sky” coordinates from the
HEASARC showing the stray light fromGRS1915+105 along with the X-rays from the targeted source for two epochs (the intended source target name is given inthe figure titles). Unlike the point source which is contained on one detector, the stray light spans multiple detectors on the
NuSTAR focal plane. gets (those with focused count rates rates >
100 cps) wefind that the primary source dominates over the entireFoV, so we exclude these observations from considera-tion. Once this is complete, we compute the 3–20 keVcount rate for all four detectors on each FPM and com-bine them to account for the fact that the stray lightpatterns tend to illuminate one side (or all) of the FoV.For the remaining sources we flag observations wherethe count rate measured by a particular detector com-bination deviates from the mean. Unfortunately, dueto extended sources, fields with multiple point sources,and intrinsic variation in the
NuSTAR background, allof the candidate
StrayCats observations had to be fur-ther checked by eye. We do this by constructing
DET1 images in the 3–20 keV bandpass and look for the sig-natures of stray light. Figure 3 shows a selection of
StrayCats observations where the SL can clearly beseen.We continue the iterative process to identify candi-dates described above until all of the candidates ap-pear to be simply variations in the
NuSTAR backgroundand not clearly associated with stray light. Overall,more than 1400 candidate stray light observations werechecked by hand for the presence of stray light.We feel confident that we have thus identified all ofthe stray light sources that could (a) produce a strongenough signal to impact science analysis of the primary target and (b) be useful for scientific analysis in theirown right. These fully vetted
StrayCats sources formthe basis for the full catalog. In addition to stray light,we have also identified a number of observations wheretargets just outside of the
NuSTAR
FoV result in “ghostrays”, where photons perform a single-bounce photonsoff of the
NuSTAR optics rather than the double-bouncefor focused emission (Madsen et al. 2017a). These areincluded in
StrayCats for completeness.We do note that this human-in-the-loop approach doesresult in a bias where faint stray light sources are moreeasily seen during long exposures. Similarly, sourceswith transient flaring behavior on timescales of a few100- s will be difficult to identify unless the quiescentflux level is greater than that of the standard NuSTAR background. We anticipate that a further investiga-tion for transients could produce a number of additional
StrayCats candidates, though this is beyond the scopeof this first work. THE
StrayCats
CATALOGThe
StrayCats
Catalog is intended to be used by ob-servers looking for serendipitous observations of brightgalactic (including the LMC and SMC) sources beyondwhat is available through traditional monitoring obser-vations. The catalog is available via a simple web inter- trayCats Figure 3.
A rogue’s gallery of 3–20 keV
NuSTAR images in
DET1 pixel coordiantes (1 pixel = 2.54 (cid:48)(cid:48) = 120.96 µ m) forthree StrayCats observations showing some of the variety of the stray light patterns in FPMA ( left column ) and FPMB ( rightcolumns ). The primary source has been masked out and the linear colorscale shows the fluence (counts per second per cm )across the field of view for each detector. ( Top ) One of the cases where stray light (here from the LMXB 4U 1624-490) isseen in both FPMs. (
Middle ) A more complex geometry where multiple overlapping or partially blocked stray light sources(the strongest being 4U 1708-2 in FPMA and 4U 1700-377 in FPMB) overlay the extended primary source (RX J1713.7-3946).(
Bottom ) Strong and overlapping stray light from GX 5-1 (lower SL) and GX 3+1 (upper right in FPMB).
Grefenstette et al. face or simply through a FITS file that identifies which NuSTAR sequence IDs contain
StrayCats sources. Forobservations that contain multiple
StrayCats sourcesthe web interface also contains diagnostic informationthat can be used to determine which stray light patternis associated with a particular source (i.e., the imagesshown in Fig 3). An excerpt of the table is given in theAppendix in Table 4.The first version of
StrayCats includes the followingcolumns: • StrayID: The
StrayCats catalog identifier, whichis StrayCatsI XX where XX is the row numberafter the catalog is sorted the RA and Dec for the
NuSTAR sequence ID. • Classification1. SL: The source has been positively identifiedas a
StrayCats target2. Complex: Stray light is present, but there aremultiple overlapping stray light regions thatmake the sources difficult to identify3. Faint: Stray light is present, but is too faintto be positively identified.4. GR: The observation contains ghost-raysfrom sources just outside of the FoV5. Unkn: A stray light pattern is present, butthe source of the stray light remains un-known. • SEQID: The
NuSTAR sequence ID • Module: The
NuSTAR
FPM that contains thestray light (A or B) • Exposure: The exposure time for this observationin seconds • Multi: Whether the sequence ID contains multiplestray light patterns (Y or N) • Primary: The name of the primary target for thepointed science observation • TIME / END TIME: The MJD start/end of theobservation • RA/DEC Primary: RA/Dec of the primary target • SL Source: The name of the source of SL if wehave identified it https://nustarstraycats.github.io/ • SL TypeFor sources with a positive identification, we havemade an effort to sample the literature and pro-vide a source classification. Many of these arerelatively famous sources identified by
GINGA or Uhuru with a large literature background, so wedo not provide prime references for the classifica-tions in
StrayCats . For sources with Classifica-tion other than SL, this defaults to “??”. Classifi-cation types are:1. AGN: Active Galaxy2. LMXB (low-mass X-ray binary) with -NS or-BH if the compact object type is known3. HMXB (high-mass X-ray binary) with -NS or-BH if the compact object type is known4. Pulsar / PWNe (Pulsar Wind Nebula) / NS5. BHC (Black Hole Candidate)6. SNR (Supernova Remnant)7. Cluster (Galaxy cluster)8. Radio Galaxy • SIMBAD ID: The identifier that can be used viaSIMBAD to identify the source. This can often bedifferent than the source name in the all-sky cat-alogs used to identify the source (if known, other-wise defaults to NA) • RA/DEC SL: RA/Dec of the source of the straylight (if known, otherwise defaults to -999).
StrayCats contains 436 telescope fields (with A and Bcounted separately) containing stray light from 78 con-firmed
StrayCats sources. During the visual inspectionof the stray light candidates, we compare the observedstray light patterns with those predicted for that obser-vation using the same code used in § not present in either catalog. Thiswas either because the source was a new transient (e.g.,a number of MAXI-identified transients that went intooutburst over the last few years), the source is only oc-casionally detected by the all-sky hard X-ray detectors(e.g., sources contained in the “ Swift -BAT historicallydetected” list), or the source is typically too soft to bedetected by
Swift -BAT or
INTEGRAL . We have not yetidentified any previously unknown
StrayCats sources.We can esimate the source location using the projectedshape of the aperture stop on the focal plane. Fig 4 givesan example of this for a simple case. Here, the curva-ture of the aperture stop shadow is clearly seen on the trayCats (cid:48) , which isgenerally good enough to identify the source. For caseswhere multiple overlapping stray light patterns are seenand we cannot unambiguously identify the source weassign the “Complex” classification pending a detailedanalysis.
Figure 4.
An example of the “stray light” (SL, green)region and the “aperture stop” region (red, dashed) that canbe used to identify the source location on the sky. See textfor details.
The catalog contains seven AGN and one galaxy clus-ter, several pulsar wind nebulae and supernova rem-nants, roughly 17 accreting black holes (including blackhole candidates), as well as over forty accreting neu-tron stars including several pulsars and a number ofknown Type I X-ray bursters. Figure 5 shows the galac-tic distribution of these sources, where the density ofsources near the galactic plane and the LMC and SMCcan clearly be seen. StrayCats
DATA ANALYSIS TOOLS ANDRESPONSE FILES
StrayCats require subtly different analysis methodsthan those typically used for focused
NuSTAR obser- vations. Rather than working in “SKY” coordinateslike focused observations, for stray light observations weinstead work in “detector” coordinates (
DET1 coordi-nates in
NuSTAR vernacular). This coordinate systemis fixed with respect to the
NuSTAR
CdZnTe detectorsand, in these coordinates, the pattern of stray light onthe focal plane is predominantly sensitive to the obser-vatory orientation and is extremely weakly coupled toany motion of the
NuSTAR mast. For pointed obser-vations, the ∼ mm-scale motion of the NuSTAR mastaffects the throughput of the optics by changing the dis-tance of the source from the optical axis (“vignetting”Harrison et al. 2013). In non-focused observations themast motion only minimally changes the shadow patternas observed by the detectors and can be neglected.Producing high-level science products for a
StrayCats observation is relatively straightforward. These mostlydeal with properly tracking the production of “source”regions files and applying spatial filtering on the
NuS-TAR data in
DET1 coordinates. Our goal is to makethe resulting products as similar to standard
NuSTAR products as possible for the ease of use.To date, we have contributed a number of high-level“wrappers” to the
NuSTAR community-contributedGitHub page . These are largely written in pythonand significantly leverage the existing astropy frame-work (Robitaille et al. 2013; Collaboration et al. 2018),as well as the multi-mission FTOOLs distributed by theHEASARC, such as XSELECT . Final high-level prod-ucts are mostly generated using nuproducts from the
NuSTARDAS software with a number of non-standardconfiguration settings. This allows a user to easily pro-duce standard spectrum (PHA) and lightcurve files aswell as response matrix functions (RMFs) which can di-rectly be loaded into downstream analysis software suchas
Xspec (Arnaud 1996) or
ISIS (Houck & Denicola2000) for spectral analysis or
Stingray (Huppenkothenet al. 2019) for timing analysis.4.1.
Response Files
The one unique requirement for the analysis of
StrayCats observations is the production of the re-sponse files. For a focused observation, each count isfirst “projected” onto the sky and the optics response(i.e. the ancillary response file, or ARF) is produced sothat it accounts for the time-dependent drift in the lo-cation of the optical axis due to the thermal motion ofthe
NuSTAR mast. The ARF is generated starting withan on-axis optics response, which is then convolved with https://github.com/NuSTAR/nustar-gen-utils Grefenstette et al.
150 120 90 60 30 0 330 300 270 240 210-75°-60°-45°-30°-15°0°15°30° 45° 60° 75°
Figure 5.
Distribution of the
StrayCats in galactic coordinates showing the clustering of these sources near the Galacticplane, the contribution from bright sources in the LMC and SMC, and a few AGN located out of the plane of the Galaxy. Thecoordinates shown here are the for the primary (focused target). energy-dependent vignetting function based on the off-axis angles sampled by the source. Finally, the ARF alsoincludes the attenuation along the photon path due tothe optics thermal covers, the Be window protecting thedetectors, and the absorption features in the CdZnTedetectors themselves .Since for StrayCats observations we are working in
DET1 coordinates, we no longer need to account for thetime-dependent variations in the ARF, nor (obviously)the response of the optics themselves. The
StrayCats
ARF, instead, only needs to account for the amountof illuminated area on the focal plane (for overall nor-malization, given in cm ) and any energy-dependent ab-sorption due to the Be window and losses in the CdZnTedetectors. All of these contributions are currently storedin the NuSTAR
CALDB files (with the exception of theBe window attenuation, which is subsumed into the on-axis ARF in the CALDB). The ARF generation tool for
StrayCats analysis properly reads these files from the
NuSTAR
CALDB and weights the response based on theilluminated area on each focal plane detector. The re-sulting file can be directly imported into
XSPEC alongwith the other spectral files above for analysis. This see the NuSTAR software user’s guide:https://heasarc.gsfc.nasa.gov/docs/nustar/analysis/nustar swguide.pdf approach has been validatd against observations of theCrab (Madsen et al. 2017b).Absorbed stray light (stray light that partially pen-etrates through the aperture stops, see Madsen et al.2017a) is not accounted for here. These response filesonly account for the unabsorbed stray light that reachesthe focal plane. In addition, two of the sources in StrayCats are extended sources (Cas A and the ComaCluster). Analyzing data from extended sources is morecomplex and beyond the scope of this analysis. Analyz-ing these sources in detail will likely require bespoke ray-trace simulations to properly interpret the stray lightspectrum. 4.2.
Region Files
While all of the
StrayCats clearly show the effects ofstray light, the scientific usefulness of the observationswill depend on how much of the FoV is covered by straylight. In the case of the intentional stray light observa-tions mentioned above, the
NuSTAR observations wasdesigned to maximize the amount of detector area illu-minated by stray light, which results in roughly halfof the 16 cm detector area being illuminated (com-pared with the on-axis effective area of ≈
400 cm foreach NuSTAR telescope). For standard observations,the
NuSTAR
SOC attempts to minimize this coveragewhen possible, so the illuminated detector area for theserendipitous
StrayCats observations varies dramati- trayCats Figure 6.
Example of a semi-automatically-generated re-gion for a
StrayCats observation of the Crab. cally. Because the stray light pattern depends on theshadowing of the detectors by the optical bench,
NuS-TAR is also rarely in an orientation where stray light ispresent on both NuSTAR
FPMs.Due to the large number of
StrayCats , and the geo-metrically complex region shapes, we developed a semi-automated approach to reduce the amount of manualeffort involved in generating the optimal extraction re-gion. The “wrapper” for this approach is available in theaforementioned
NuSTAR
GitHub page. For
StrayCats containing a bright point source the first step of thisprocess is point source removal. This is done first by de-termining the position of the targeted source in
DET1 coordinates (using the nuskytodet
FTOOL). This loca-tion depends on the motion of the
NuSTAR mast andany changes in the
NuSTAR pointing, so we determinethe radial distance from each observation count fromthis position. We screen events within r -arcminutes ofthe source (if necessary, and where the choice of r is cho-sen on a case-by-case basis) and generate an image in anadjustable energy band (the 3–10 keV band is default).We use Canny edge detection from scikit-image to generate the polygons used to estimate the sourceregion where the width of the Gaussian filter used by the https://scikit-image.org Canny edge detection ( σ ) is an adjustable parameter.Again, this is chosen on a case-by-case basis such thatthe filter accurately identifies the edges of the stray lightregion. Polygon region corners in image coordinates aredetermined from the detected edge pixels and used towrite a region file in SAOImageDS9 standard formatusing the regions astropy-affiliated module.This approach is particularly useful for stray lightregions with an angular cutaway resulting from theshadow of the optical bench (i.e., Fig 6). This processis most efficient for intentional stray light observationsand serendipitous observations containing a single straylight pattern only from the “SL Target” source (i.e., en-tries in the StrayCats catalog with the Classification“SL” and Multi value “N”). Currently, this approach ismost limited by the σ parameter, which approximatelyranges between 3 and 12 for optimal stray light observa-tions but can vary greatly for weak stray light regions.Discontinuities in the edges identified by the Canny filteroccasionally result in the created polygon region omit-ting (sometimes negligibly thin) slices of the stray lightregion; these anomalies can often be corrected by fine-tuning σ . However, there are no optimal σ values for theCanny filter to properly identify the stray light region forobservations in which the fluxes of the background andthe stray light are comparable. Future improvements tothis process that eliminate the manual determination ofthe point source removal limit and Canny edge detectorsigma would allow for fully-automated region extraction.4.3. Background
Dealing with background for
StrayCats sources isnot trivial. For standard
NuSTAR pointed oberva-tions, standard techniques such as using a neighboringsource-free region to estimate the background and/orestimating the
NuSTAR background through tools suchas nuskybgd (Wik et al. 2014) can be used “out of thebox”. However, as we are using
NuSTAR as a collimatorrather than a focusing telescope, the background mustbe treated with more care.The
StrayCats source regions cover a large region ofthe FoV (and there may be multiple
StrayCats sourcesas well as the primary source in the FoV), so select-ing a background region may be difficult. In addition,for bright
StrayCats sources, some stray light may alsobe transmitted through the aperture stop at higher en-ergies, making it impractical to select a neighboring“source free” region of the FoV to use to estimate thebackgrounud (see Madsen et al. 2017a,b, for further dis-cussion).Modeling the background contributions also must behandled with care. Because many of the
StrayCats Grefenstette et al. M A X I R a t e Figure 7.
The long-term 2-10 keV lightcurve of GRS 1915+105 as measured by MAXI (blue histogram) along with the timingof focused
NuSTAR observations (red lines) and the
StrayCats observations (dashed black lines). The final three epochs areclustered in the 16 days just before MJD 58600. sources are near the Galactic plane, the standard modelsof the spatial variation of the
NuSTAR background usedby nuskybgd to model the contributions from the Galac-tic ridge X-ray emission (Krivonos et al. 2007, GRXE)are largely untested and may need to be adapted for thenon-isotropic shape of the GRXE.The exact method used to handle the presence ofbackground will necessarily vary depending on the sci-ence goals for the individual analysis. For bright, hardsources, even without the aid of the
NuSTAR optics,the backgrounds in
NuSTAR are so low that the back-ground may be neglected up to high energies. For faintersources (or soft sources) the energy at which the back-ground starts to significantly contribute (and thereforethe background component which matters the most forspectral analysis) will depend on the details of the sourceflux. We do not expect there to be a universal solutionor recommendation for how to handle the backgrounds.In the selected preliminary results below, spectralanalysis is typically halted when the source flux fallsso that the background is estimated to be ∼
10% of thesource flux, but we stress that a thorough treatment ofthe background must be considered. SELECTED PRELIMINARY
StrayCats
RESULTS5.1.
GRS 1915+105
GRS 1915+105 is a LMXB system which has beenin outburst since its discovery in 1992 (Castro-Tirado
Figure 8.
The 3-20 keV
DET1 image for sequence ID30201013002. The faint primary source is shown, as is thestray light pattern for GRS 1915+105 along with the regionshowing the shadow of the aperture stop. et al. 1992) and shows a wide range of source spectraland timing states (e.g., Belloni et al. 2000). The sys-tem is known to host a near-maximally spinning blackhole (McClintock et al. 2006) and observations of the trayCats Table 1.
GRS 1915+105
StrayCats
ObservationsObs )1A 80001014002 2013-11-08T18:11:07 56604.8 A 45.15 3.91B - - - B 45.49 4.02 30101050002 2015-07-01T15:31:08 57204.6 A 41.34 **3A* 30201013002 2016-10-20T16:56:08 57681.7 A 122.3 2.23B - - - B 122.6 3.64 40301001002 2018-03-17T01:46:09 58194.1 A 125.6 5.95 30401018002 2018-08-01T12:41:09 58331.5 B 78.3 4.96 90501317002 2019-04-10T01:26:09 58583.1 A 40.8 5.77 30402026002 2019-04-22T00:11:09 58595.0 A 18.83 **8 30402026004 2019-04-26T13:41:09 58599.6 A 23.31 ***:Used for the analysis in this work; **:Small stray light area ModelFe LineDiskContinuum ν F ν R a ti o R a ti o Figure 9.
The integrated spectrum from Obs3A from aportion of the stray light region. A similar sized region wasused to estimate the background. The base model spectrumhere consists of a hot accretion disk component and a softnon-thermal power-law, though this leaves strong residualsnear the Fe line ( middle ). We find that after the addition ofa broad Fe line and absorption features associated with diskwinds in this system that we obtain a reasonable fit to thedata. absorption features also reveal the presence of a com-plex outflowing disk wind (Miller et al. 2016). How- - k e V c p s - k e V R a t e Figure 10.
The 3-20 keV lightcurve for the first 450-s ofthe first orbit, binned at 2-s resolution shows the presence oftransient slow (mHz) QPO signals. (
Bottom ) The 3-20 keVof two later orbits binned at 10-s resolution showing that thesource has transitioned to its θ -state. ever, the source began a decay to either a quiescent ora highly absorbed state between 2018 and 2020 (Milleret al. 2020; Neilsen et al. 2020). Since 2012, NuSTAR has observed the source a number of times at vary-ing flux levels (Fig 7). However, the high count ratesfrom this source present two key problems that affectthe scientific return from these data: (1)
NuSTAR hasa fixed 2.5-ms deadtime-per-event, resulting in a max-imum throughput of 400 cts s − . In high rate sources2 Grefenstette et al. this deadtime also results in the effective exposure be-ing much lower than the time spent observing the target;(2) As mentioned above, the high count rates result inhigh telemetry loads that require short duration obser-vations to avoid data loss on board. GRS 1915+105 alsoappears in 6
StrayCats epochs, covering a wide rangeof flux states (Figure 10) as measured by the Monitorof All-sky X-ray Image (MAXI) instrument on the
In-ternational Space Station (Matsuoka et al. 2009). Theduration of the
StrayCats observations vary, with sev-eral snapshots roughly 20-ks effective exposure to severaldeep observations with over 120-ks of exposure. A sum-mary of the
StrayCats for GRS 1915+105 is given inTable 1.As an example, we show preliminary results from oneepoch (Obs 3A, 30201013002, Figure 8), which had aneffective exposure of 122 ks spanning over roughly 240ks (over two and a half days) of clock time. The epoch-averaged source spectrum (Figure 9) shows that thesource is clearly detected up to at least 40 keV beforebackground becomes a significant contribution to thespectrum. At low energies we clearly see evidence for aFe-line features and absorption features typically asso-ciated with disk winds in this system (e.g, Miller et al.2016; Neilsen et al. 2018).However, the spectrum for this source is known to behighly variable with the source hardness varying withthe apparent emission states and throughout this ex-tended observation the source showed a variety of emis-sion states. For example, during the first orbit we clearlyobserve QPOs in the form of 10 to 20- s recurrent “pul-sations” of emission, while in later orbits during thesame observation the source has transitioned to its θ -state, showing emission building up over the span of afew hundred seconds before sharply dropping away (Fig10). A detailed analysis of the spectral changes through-out this system is beyond the scope of this work (e.g.,Zoghbi et al. 2016), but shows the utility of only one ofthe several observations of GRS 1915+105.5.2. GX 3+1
GX 3+1 is a persistently accreting ‘atoll’ source. Atollsources trace out regions on hardness-intensity dia-grams that resemble ‘islands’ (for which they are named:Hasinger & van der Klis 1989) or ‘banana’ shapes.GX 3+1 exclusively occupies the banana branch (Sei-fina & Titarchuk 2012) and was serendipitously observedvia straylight in
NuSTAR nineteen times between 2012July and 2020 May. Table 2 shows the sequence ID,observation date, FPM that the straylight occurred on,exposure time, and area on the FPM for observationswith an area greater than 1 cm of straylight from the ) H R ( - k e V / . - k e V ) obs1obs2obs3obs4obs5obs6obs7obs8obs9obs10obs11Aobs11Bobs12obs13obs14 Figure 11.
Hardness-Intensity diagram of the straylightobservations of GX 3+1. Observation numbers refer to thesequence IDs in Table 2. Data are binned to 300 s. The‘banana’ branch is traced out by the data. C oun t s k e V − ( da t a − m ode l ) / e rr o r Energy (keV)
Figure 12.
The 3–20 keV straylight spectrum of GX 3+1obs10 and residuals divided by the error. The orange dashedline indicates the power-law component, the blue dot-dashedline is the single-temperature blackbody, the dotted line isthe multi-temperature blackbody. A prominent Fe line fea-ture is present between 6 − source. Lightcurves were generated in three different en-ergy bands (3 −
20 keV, 6 . −
10 keV, and 10 −
16 keV)with a binsize of 300 s. Figure 11 shows the hardness-intensity diagram for GX 3+1. The hardness ratio (HR)is defined as the 10 −
16 keV band divided by the6 . −
10 keV band (Coughenour et al. 2018). The sourcetraces out the ‘banana’ branch.To demonstrate the spectral utility of straylight ob-servations for studying NS LMXBs, we extract a spec-trum from the longest observation, obs10. The dataare fit with the three component model of Lin et al. trayCats Table 2.
GX 3+1
StrayCats
ObservationsObs )1 30002003003 2013-06-19T09:31:07 B ∼
29 3.512 80002017002 2014-02-15T05:36:07 A ∼
39 4.643 90101012002 2015-08-11T22:51:08 B ∼
49 1.464 90101022002 2016-02-18T22:26:08 A ∼ . ∼
52 1.356 80102101002 2016-09-29T21:21:08 B ∼ . ∼
28 7.158 80102101005 2016-10-31T20:11:08 B ∼
29 6.669 80202027002 2017-02-18T14:31:09 A ∼
31 4.6910* 40112002002 2017-04-03T18:31:09 A ∼ . ∼
61 3.40- - B ∼
61 3.4312 90501329001 2019-06-22T07:51:09 B ∼
40 3.3513 90501343002 2019-10-01T22:36:09 B ∼
37 1.6514 90601317002 2020-05-07T07:06:09 A ∼
49 4.12*:Used for the analysis in this work (2007) that was used in Ludlam et al. (2019) for thepointed observation of GX 3+1. This is comprised of amulti-temperature blackbody for thermal emission fromthe accretion disk, single-temperature blackbody for aboundary layer or emission from the NS surface, andpower-law for weak Comptonized emission. For directcomparison to the intentional
NuSTAR observation, wemodel the continuum emission by fixing the absorptioncolumn along the line of sight, blackbody temperatures,and photon index to the values reported in Table 2 ofLudlam et al. (2019) while allowing for the normaliza-tions of each spectral component to vary. The spectrumand continuum components are shown in Figure 12. Thecolor scheme and line types correspond to those in Lud-lam et al. (2019). Indeed, a prominent Fe line emissionfeature can be seen in the straylight observations akinto the one observed from the pointed observations (seeFig 1 of Ludlam et al. 2019). Further details of thevariations in this source over time will be addressed infuture work. 5.3.
GS 1826-24
GS 1826-24 is a LMXB which showed remarkableconsistent Type I X-ray bursts since its discovery by
GINGA (e.g. Ubertini et al. 1999). The Type I X-raybursts were so regular as to earn this source the “ClockedBurster” moniker. A sudden dip in the
Swift -BAT 15-50keV lightcurve resulted in a
NuSTAR
ToO observationof this source in 2014 (Chenevez et al. 2016). Afterbriefly returning to a hard state, the source appears tohave transitioned into a “soft” state in 2016 with the M A X I R a t e B A T R a t e Figure 13.
The long-term 2–20 keV MAXI lightcurve(blue) and the
Swift -BAT transient monitor 15-50 keVlightcurve for GS 1826-24 (grey). The timing of the focused
NuSTAR observations are shown in solid red lines while thetiming of the
StrayCats observations are shown in dashedblack lines.
MAXI lightcurve increasing to a plateau in 2018 and the
Swift -BAT lightcurve in an apparently quiescent state(Fig 13). While there have not been any subsequenttargeted observations with either
NuSTAR or XMM-Newton , NICER has monitored the source and foundevidence for mHz QPOs (Strohmayer et al. 2018).The
StrayCats observations (Table 3) span both thepre-dip observations and include several long observa-4
Grefenstette et al. R a t e R a t e
40 60 80 100 120Seconds0102030 R a t e
60 80 100 120 140Seconds0510152025303540 R a t e Figure 14.
All panels show the 3–20 keV lightcurve of Obs7 and show: (
Top Left ) The full observation using 1-s bins clearlyshows the two Type I X-ray bursts; (
Top Right ) The same data, but using 1-ks bins; (
Bottom panels ) The zoomed in view ofthe first ( left ) and second ( right ) Type I X-ray burst. tions during the BAT X-ray minimum. We highlightone of these (Obs7), which had a substantial amount ofstray light covering over half of FPMB and a long expo-sure of over 150-ks, resulting in nearly 300-ks of elapsedclock time. During this observation
NuSTAR clearly de-tected two Type I X-ray bursts lasting ∼
10s of seconds(Fig 14). Simultaneously, the X-ray flux in the 3–20keV lightcurve dipped leading up to the burst itself. Weonly find two Type I X-ray bursts, while we would haveexpected over a dozen had the source been regularlybursting with a recurrence time of ∼ NuSTAR observations that the “clocked”nature of the source has disappeared in the soft state(Chenevez et al. 2016). A more complete survey of thebursting state over all 7 epochs and correlations withthe spectral changes in the source will be the topic of afuture paper. SUMMARY AND FUTURE WORK In this paper we have presented a summary of aunique, untapped set of
NuSTAR observations. The
StrayCats observations found thus far are predomi-nantly associated with known bright sources and tran-sient X-ray binaries as they go into outburst.
StrayCats is based on a systematic approach to min-ing the database of
NuSTAR observations. While previ-ously these observations were considered a nuisance, wehave now produced a set of publicly available tools foranalyzing these data and producing high-level scienceproducts. In addition, we provided access to scriptsthat help in the generation of region files. which of-ten requires some fine tuning based on the projected“shadow” of the optics bench.The
StrayCats catalog that we present here we con-sider to be version 1.0. We intend to extend the cur-rent version of
StrayCats to include additional sum-mary data products (such as count rates, hardness ra-tios, and source and background extraction regions) forall
StrayCats observations where the source is brightenough and enough of the focal plane is covered by stray trayCats Table 3.
GS 1826-24
StrayCats
ObservationsObs )1 80002012002 2014-02-14T00:36:07.184 56702.0 A 24.05 2.22 80002012004 2014-04-17T22:46:07.184 56764.9 A 26.42 2.33 30101053002 2015-06-17T16:06:07.184 57190.7 A 131.32 2.754 30101053004 2015-06-21T07:11:07.184 57194.3 A 51.52 2.55 60160692002 2016-04-14T18:26:08.184 57492.8 B 21.88 1.76 10202005002 2017-04-18T13:06:09.184 57861.5 A 156.51 2.527* 10202005004 2017-09-23T08:36:09.184 58019.4 B 156.54 8.88 80460628002 2019-03-08T20:21:09.184 58550.8 B 41.39 1.6*:Used for the analysis in this work light. This work is on-going and will be provided in afuture release.Finally, our brief survey of the science potential from StrayCats observations shows the power of these ob-servations. Through these highlights of a few selectedobservations we have shown that these data can be usedto track sources over long periods of time and providea unique window into their behavior by providing im-proved sensitivity and finer spectral resolution comparedto other all-sky monitors such as MAXI and
Swift -BAT.ACKNOWLEDGEMENTSThis work was supported by the National Aeronauticsand Space Administration (NASA) under grant number80NSSC19K1023 issued through the NNH18ZDA001NAstrophysics Data Analysis Program (ADAP). R.M.L.acknowledges the support of NASA through Hubble Fel-lowship Program grant HST-HF2-51440.001. R.A.K.acknowledges support from the Russian Science Foun-dation (grant 19-12-00396). JH acknowledges supportfrom an appointment to the NASA Postdoctoral Pro- gram at the Goddard Space Flight Center, administeredby the USRA through a contract with NASA.Additionally, this work made use of data from the
NuSTAR mission, a project led by the California In-stitute of Technology, managed by the Jet PropulsionLaboratory, and funded by the National Aeronauticsand Space Administration. We thank the
NuSTAR
Op-erations, Software and Calibration teams for supportwith the execution and analysis of these observations.This research has made use of the
NuSTAR
Data Anal-ysis Software (NuSTARDAS) jointly developed by theASI Science Data Center (ASDC, Italy) and the Cali-fornia Institute of Technology (USA). This research hasmade use of data and/or software provided by the HighEnergy Astrophysics Science Archive Research Center(HEASARC), which is a service of the Astrophysics Sci-ence Division at NASA/GSFC.
Facilities:
NuSTAR, Swift, MAXI, HEASARC
Software: astropy (Robitaille et al. 2013; Collabo-ration et al. 2018), astroquery (Ginsburg et al. 2019),HEASoft/FTOOLS, IDL , matplotlib(Hunter 2007),numpy , pandas(Pandas Development Team 2020),plotly , scikit-image , Veusz APPENDIX https://numpy.org https://plotly.com/python/ https://scikit-image.org https://veusz.github.io Grefenstette et al. T a b l e . StrayCats E x c e r p t S T R AY I D C l a ss i f . S E Q I D M o d . P r i m a r y T S T A R T E x pS L S o u r ce S L T y p e R A S L D E C S L R A P r i D E C P r i S tr a y C a t s I F a i n t B I C X . . NA ?? - - . . S tr a y C a t s I U n k n A I G R J p . . NA ?? - - . . S tr a y C a t s I S L A S W I F T J d m . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L A S M C X . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L A S X P d . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L A S X P d . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I F a i n t B S X P d . . NA ?? - - . - . S tr a y C a t s I F a i n t A S M C D ee p M O S . . NA ?? - - . - . S tr a y C a t s I S L A I R A S m . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L A S M C X . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L A S M C X . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L B S M C D ee p M O S . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L B S M C X . . S M C X - H M X B - N S . - . . - . S tr a y C a t s I S L A S M C X . . R X J . - H M X B - N S . - . . - . S tr a y C a t s I S L B S M C X . . R X J . - H M X B - N S . - . . - . S tr a y C a t s I S L B S M C X . . R X J . - H M X B - N S . - . . - . S tr a y C a t s I S L A G K P e r . . N G C A G N . . . . S tr a y C a t s I S L A G K P e r . . N G C A G N . . . . S tr a y C a t s I S L B G K P e r . . N G C A G N . . . . S tr a y C a t s I S L AN D . . L M C X - L M X B - N S . - . . - . S tr a y C a t s I S L B N D . . L M C X - N S . - . . - . S tr a y C a t s I S L B N D . . L M C X - L M X B - N S . - . . - . S tr a y C a t s I S L A R X S J d p . . C r a b P W N e . . . . S tr a y C a t s I S L B R X S J d p . . C r a b P W N e . . . . S tr a y C a t s I S L A S G R m . . L M C X - L M X B - B H . - . . - . S tr a y C a t s I S L B S G R m . . L M C X - L M X B - B H . - . . - . trayCats17REFERENCES