AAVSO: American Association of Variable Star Observers

Proposal #110

Proposer (3360) Bradley Schaefer (schaefer@lsu.edu) obscode: SCK
Assigned To(3663) Dirk Terrell
Date SubmittedMay 7, 2019


My AAVSO Alert Notice #664 asked for continuous monitoring of the recurrent nova U Scorpii, because it is due to erupt anytime now. This has resulted in much activity by AAVSO observers and by professionals writing ToO and regular proposals for various ground-based telescopes and three spacecraft. As a result of the AAVSO observations coming from this Alert Notice, I now have two indirect-and-weak grounds for thinking that the eruption has already happened, perhaps around Thanksgiving 2016 or even around Thanksgiving 2017. This is the nightmare scenario where the entire 60-days of the eruption are invisible around the time of its close solar conjunction. There is nothing that can be done to help this possibility, and the nightmare scenario is that the eruption comes and goes completely-unobserved, and the world of observers keeps a long fruitless vigil, while many useless proposals are written. It is actually very hard to recognize such a nightmare scenario, except by a double length inter-eruption interval many years later. The best of the indirect methods is to see whether the O-C curve for the U Sco eclipses has undergone a sharp kink, with any such pointing to a missed eruption. But right now, I have no good eclipse timings since 2016.5. This proposal is to use the 0.61-m scopes of AAVSOnet to get four 3-4 hour time series centered on the times of mid-eclipse, so as to add good O-C points for 2019. This will provide a confident measure of the current O-C, and this can be compared to my very-well-measured parabolic O-C curve from 2010.4 to 2016.5. If I find good evidence that the O-C curve must have had a sharp kink sometime around 2017, then I will call off the AAVSO Alert, saving much effort. If I find no kink in the O-C curve, then the AAVSO Alert will keep going with redoubled importance. With the urgency of knowing whether the Alert should be continued, I am asking for these four eclipse times to be made before 18 May 2019 (i.e., 11 days from now).

Proposer: Brad Schaefer (member of AAVSO continuous since 1971)
E-mail: schaefer@lsu.edu
Target: U Sco (one of the most famous recurrent novae)
Coordinates: J2000 16:22:30.778 -17:52:43.16
Magnitudes: At mid-eclipse, usually V=19.0 (Schaefer 2010, ApJSupp, 187, 275)
AAVSOnet scopes: Either Tortugas Mountain Observatory 0.61-m scope
or Mount John University Observatory 0.61-m scope
(This is based on my very long experience at observing
U Sco eclipses with 0.8, 0.9, 1.0, 2.1, 2.7 meter scopes, as well as
with seeing data from many 0.4-m scopes in the AAVSO record.
That is, at least a 0.6-m scope is needed to get adequate SNR at
eclipse minimum. A 0.4-m scope just is too small. I have many
good eclipse times measured from Kitt Peak, so the New Mexico
observatory is not too far north. As U Sco is getting close to
opposition, it just means that the New Mexico observer needs to
pick an eclipse that happens around the time of meridian passage.)
Filter: I would prefer no filter, so as to get the best SNR at minimum. I have
already proven that the U Sco eclipse times are not dependent on the
optical filter used, or the lack of any filter (Schaefer et al. 2010, ApJ, 742,
113). If a filter must be used, use the V filter.
Exposure: 5-minutes (this can be pushed from a few minutes to 10 minutes)
Conditions: It need not be photometric, as I'll be using nearby comparison stars.
Thin clouds are marginally acceptable. For a 0.6-m scope, a gibbous
Moon nearby is a killer. This will have to be done with a mostly dark sky.
Cadence: Four time series, each lasting 3 to 4 hours, centered on mid-eclipse.
Eclipse Ephemeris: The exact time of the eclipses is the question at hand, and there
is now substantial uncertainty for 2019, which is the whole purpose of
this proposal. For purposes of scheduling observing times, use
HJD(mid-eclipse) = 2451234.5387 + N * 1.23054695.
Eclipses: U Sco has an orbital period of 1.23 days, so observing opportunities occur
roughly once every five nights. The total duration of eclipse is variable,
lasting 4-5 hours. The duration of the nearly-flat minimum is 1.2 hours.
To get long enough of a time series to be able to fit a parabola to derive
the time of mid-eclipse, my long experience shows that at least three hours
are needed for the time series.
Number of Eclipse Times: Eclipse times vary substantially from eclipse-to-eclipse,
due to the usual flickering on the CV, variously brightening the ingress
or the egress. This intrinsic scatter in the O-C curve is shown in Schaefer
et al. (2010, ApJ, 742, 113). This means that multiple measures of the
U Sco eclipse times are required to return a confident result. Four eclipse
times are a reasonable compromise between needing many eclipse times
and telescope resources.
When: The question of whether U Sco has undergone a missed eruption around
2017 affects all the work and proposals and monitoring of U Sco
associated with AAVSO Alert Notice #664. If we can provide a
convincing case that it went off already, then the AAVSO Alert must be
stopped. If U Sco still has yet to erupt, then we have to keep all these
activities going full blast. Thus, it would be good to get our answer
speedily. The next First Quarter Moon is in one week (14 May), and a
few days later will have the bright gibbous Moon shining when U Sco
is up. So I am requesting that the four time series be completed before
18 May, i.e., in the next 11 days.
Reduction: I expect to get the usual processed images (biased, dark-subtracted, and
flattened) in FITS format, presumably placed on some FTP site. My
analysis will be simple aperture photometry of the target with respect to
nearby stars of known magnitude. The magnitudes are from my measures
as reported in Schaefer (2010, ApJSupp, 187, 275). But really, especially
if no filter is used, the comparison star magnitudes are meaningless, as
relative photometry on some ad-hoc magnitude system is all that is needed
to see the eclipse and get a good minimum time. I do the photometry with
IRAF. I do the heliocentric conversion for the middle of each exposure to
get the HJD. The magnitudes of the ingress, minimum, and egress are
then fit (with a chi-square) to a parabola, returning an HJD eclipse time
along with the 1-sigma uncertainty. With the above ephemeris, I will
place the eclipse times onto an O-C curve. What I am looking for is
whether or not there is a significant deviation from my very-well
measured parabolic O-C curve from 2010.4 to 2016.5. If I see an O-C
curve after 2016.5 that requires a large break, then I will conclude that
U Sco must have had a missed eruption around Thanksgiving 2016 or
around Thanksgiving 2017.

U Sco is a famous recurrent nova, with eruptions in 1863, 1906, 1917, 1936, 1945, 1969, 1979, 1987, 1999, and 2010 (Schaefer 2010, ApJSupp, 187, 275). I was the discoverer of the 1917, 1945, and 1969 eruptions in archival data. Schaefer (1990, ApJLett, 355, L39) discovered that U Sco is a deep-eclipsing binary with a period of 1.23 days. This started a career-long program of measuring orbit period changes in recurrent novae and classical novae. Schaefer (2011, ApJ, 742, 112) gives U Sco eclipse times from 1989 to 2009, with measures of the sudden orbital period change across the 1999 eruption (with a kink in the O-C curve), plus the measured steady period change between eruptions (with a parabolic shape in the O-C curve). This is all big-time science, because it provides a direct refutation of the most popular single-degenerate model (i.e., recurrent novae) for the progenitors of Type Ia supernovae. The similar work with classical novae, combined with the recurrent nova work (dominated by period changes of U Sco) have provided sure refutations of the venerable Hibernation Model for CV evolution, and it has also provided two 'serious challenges' to the entire venerable model of Magnetic Braking, long the assumed primary guide to CV evolution. So the issues of the U Sco period changes have high importance.

U Sco goes off every ten years or so, with the shortest known inter-eruption time being 7.9 years. Schaefer (2005, ApJLett, 621, L53) came up with a way to predict upcoming eruptions of recurrent novae, where the B-band flux provides a measure of how much material has accreted onto the white dwarf, and prior eruption histories calibrates how much mass is required to reach the nova trigger mass. Applied to U Sco, I predicted the next eruption in 2009.3±1.0. In 2008, I started organizing a large international effort to prepare for the upcoming eruption. A key part of this was to get AAVSO observers to nearly-continually monitor U Sco on a hair-trigger waiting for it to go up, all set to notify the world fast. I had some grant money that I funneled to AAVSO HQ for help in serving as a central clearinghouse. B. Harris and S. Dvorak independently discovered U Sco rising in January 2010, and this triggered many spacecraft observations and many observers from the ground. For the next 65 days, we had photometry averaging once every 2.3 minutes, and had seven spacecraft observing at all wavelengths (Schaefer et al. 2010, AJ, 140, 925; Pagnotta et al. 2015, ApJ, 811, 32). This is even now the all-time best observed nova event of any type. We discovered two new classes of phenomena (the inexplicable short flares during the transition phase and the weird dippers late in the tail.).

U Sco last went off in 2010.1, so from its history, we'd expect an eruption around 2020, or at least after 2017 or so. But with the method of Schaefer (2005), we get a substantially earlier date. In particular, the AAVSO light curve shows that the B-band flux was particularly bright in 2011 and 2012, implying a lot of accreted matter coming onto the white dwarf, so the U Sco eruption should be ahead of schedule. Detailed calculations pointed to an eruption anytime now, or perhaps already in the past. It was on this basis that I put out AAVSO Alert Notice #664.

With this, I got two new data dumps, pointing to the eruption already having gone. First, additional data from 2011 and 2012, along with more detailed calculations, point best to an eruption date of 2016 or 2017. Second, MGW provided an ingress and egress for the U Sco eclipse, and this pointed to a sharp kink in the O-C curve. Both of these indirect arguments are weak. The theory model is just a model, with one great success (for U Sco in 2010) and one great failure (for T Pyx in 2011). The MGW eclipse time had huge error bars, so the existence of a kink in the O-C curve is not of high confidence. The AAVSO data shows two large seasonal gaps during which the U Sco eruption could easily be hidden. So this is the nightmare scenario, where an entire U Sco eruption can be lost as being completely inside the solar gap. This likely is what happened in 1927 and 1957. Suddenly, we have the horror story of not knowing whether to call off the AAVSO Alert or whether to redouble the efforts.

What to do? It is actually very hard to realize that an eruption is over and was missed. U Sco does not eject a visible shell, and it would take a long time to appear anyway. The photometry and spectroscopy just after an eruption is the same as just before. In principle, we could wait until 2030 and then declare that an eruption was missed sometime a decade previously. This basic nightmare dilemma was recognized and agonized over back in 2009 and earlier this year. Then and now, the best way to spot a missed eruption is to catch a kink in the O-C curve, but this is only possible a year or three after the eruption is long gone. In the meantime, all we can do is watch and wonder.

Well, maybe U Sco erupted a year or two ago, and we can now spot the kink in the O-C curve? I already have a wonderful O-C curve from 2010.4 to 2016.5, including 59 eclipse times from the Kepler spacecraft in 2014, and this forms a good parabola. So the kink can be spotted with eclipse times from 2017 or 2018 or 2019. The 2019 eclipse times would be most sensitive to any kink, as they would have the largest deviations from the 2010.4-2016.5 parabola. For various reasons (see next section), I have no good eclipse times from 2016.5 to now. So this proposal for AAVSOnet time is to get eclipse times to see whether the U Sco O-C curve has a sharp kink due to a missed eruption.

I have used about 150 nights of telescope time on U Sco eclipses, and this has produced a good O-C curve from 1989 to 2016.5 based on 141 eclipse times. After 2016.5, for various medical reasons plus my retirement and moving to Tucson, I was not able to keep up the eclipse timings. I am now retired and I have no access to any telescope time. This is the root reason why I need AAVSOnet time.

I have tried desperately to get archival light curves from 2017 and 2018 so as to find points near the deepest eclipse, so that I can get an approximate eclipse time for placing on the O-C curve. The AAVSO data base has no useable time series. (The best is the great HBB with a single random point very faint in February of this year, but a single point leaves much too large an uncertainty in the eclipse time to decide the question.) The best that I have is many faint magnitudes from the Catalina Sky Survey, but these are all on the egress, so I only have a limit on the eclipse time, and this goes in the wrong direction to decide anything. The ATLAS survey has many points going to 2017.4, but these have so much scatter as to be useless. Similarly, the CRTS has too much scatter to find any eclipse. The NOAO archives have nothing useable, while the DECam data has no light curve points in eclipse. Arne Henden emails that the APASS and AAVSOnet archival data is too shallow to be helpful. Fred Walter does not have SMARTS Nova Catalog data covering the time interval of interest. The LONEOS, NEAT, Palomar QUEST, LINEAR, and Spacewatch surveys are no longer active. I have examined the 15-days in 2016 and in 2017 where the SOHO LASCO C3 camera shows the position of U Sco passing through solar conjunction with a limiting magnitude of V=8.8. Pan-STARRS data are not publicly available. ZTF will sometime soon make public a batch of 2018 data, of unknown quality and coverage, and this is my only remaining hope. The point of this is to show that I have tried exhaustively to pull up archival data from 2017 or 2018 to get a good point or two on the O-C curve, but that this has largely failed.

But what about getting eclipse times in 2019 from other telescope resources? Well, my AAVSO Alert Notice #664 had an appeal for eclipse timings. But no one followed up, except MGW, who had too small a telescope. (His time series only show the ingress and egress, with nothing useable at the minimum, so the derived eclipse time has greatly too large an uncertainty to provide a confident enough of an answer to call off the worldwide U-Sco-eruption efforts.) At this point it seems as though no one with big enough of a scope is going for it. I tried appealing to Arto Oksanen, but his scope in the south has closed down for a move to Tasmania. I am not a member of any of the Las Cumbres institutions, so I cannot even put in for DD time. This is all too late for putting in any proposal (like at CTIO or KPNO) that could get done before middle 2020. SMARTS is closing down and I don't have any funding source to buy this time anyway. I could ask Dan Reichart for time on the PROMPT scopes at CTIO, but these are too small. All the other observers that I have used in the past for the big USCO2010 campaign (Schaefer et al. 2011, ApJ, 742, 113) have too small a telescope. (These were great for coverage during the eruption and its tail, but inadequate to cover the deep eclipse at minimum.) So I am out of options for how to get the current O-C for U Sco. That is why I am asking for AAVSOnet time.

Target RA (H.HH) Dec (D.DD) Magnitude Telescope Observation Frequency Expiration Date Proprietary Term
U Sco 16.375216 -17.87866 19.0–17.5 No


Comments on this proposal are closed.