Recurrent Nova Science Tasks for Observers Worldwide

from Dr. Bradley Schaefer :

Recurrent novae (RNe) are those novae that erupt with a recurrence time scale faster than a century.  (This is an arbitrary, traditional, and useful cutoff.)  There are only ten known RNe, in our Milky Way Galaxy, yet many of the 392-known so-called classical novae in our galaxy really have recurrence time scales faster than a century, with only one of their 2-10 eruptions in the last century having been discovered.  RNe are one of the most important classes of variable stars, far past just that of being a cool and interesting type of system.  RNe are a critical phase in the lives of cataclysmic variables (CVs), with most-critical and broadest-impact coming from the RNe being progenitors of Type Ia Supernovae, with this being a highly controversial debate for the last 40 years amongst hundreds of workers.

Independent observers with `small' private scopes have provided the majority of knowledge about novae and RNe ever since the 1860s.  (Well, now-a-days, spectroscopy does tell a wonderful story on CVs, one which requires much too big of a telescope for almost all cases.  But see below for the upcoming eruption of T CrB.)  In the past century, observations collected and contributed to the AAVSO, and other such groups worldwide, form the majority of knowledge for most novae and RNe.  With the coming of CCDs ~2 decades ago, independent observers can go awesomely deep, and this opens up many targets, and allows for a high accuracy and good cadence that can make for new types of science return.  For the last two decades, the independent observers have been producing great science that has been defining the field of RNe.

Here is a table of the ten known RNe in our galaxy, with relevant properties:

RN  Vpeak Vquiescent Period (d)  #Eruptions (Last) Comments
T Pyx 6.4 15.5 0.076 7 (2011) Eclipsing
IM Nor 8.5 18.3 0.102 2 (2002) Eclipsing
CI Aql 9.0 16.7 0.62 3 (2000) Eclipsing
U Sco 7.5 17.6 1.23 12 (2022) Eclipsing
V2487 Oph 9.5 17.3 1.24 2 (1998) SUPERFLARES!
V394 CrA  7.2 18.4 1.52 2 (1987) Eclipsing
T CrB 2.5 9.8 228 2 (1946) Next eruption ~2023
RS Oph 4.8 11 457 9 (2021)  ---
V745 Sco 9.4 18.6 >500? 3 (2014)  ---
V3890 Sgr 8.1 15.5 747.6 3 (2019)  ---

Here, I will be pointing out the science tasks that any individual (with a CCD and a ten-inch scope or bigger) can readily perform that are most important for science return.  There is actually good science to be gotten from the default plan of getting isolated mags for a list of targets by cycling through them through the year.  But there are many types of detailed RN science questions that can be answered only with more focused programs.  Let me share outlines of a number of such programs for deep-science with RNe:


    #1.  ECLIPSE TIMINGS OF RN IN QUIESCENCE.  Simple, take a good time series of RNe that show eclipses, from which exact times of minima can be derived.  The times of minima can be placed onto an O-C diagram so as to measure with startlingly high accuracy the orbital periods.  The biggest topic at the forefront of CV astronomy is the mysterious evolution of the systems, for which the *period-changes* are the drivers and rulers for evolution.  These eclipse times are useful only on a long time scale, needing a decade or so to collect enough eclipse times over long-enough of a span to measure the fractional parts-per-million changes.  (I started this entire program back in 1989, with the idea coming ultimately from Joe Patterson, and it was only a few years ago that I finally accumulated enough eclipse times to reach confident conclusions, and these are earth-shaking for theorists.  So I am putting this program as #1, my favorite.)  Nowadays, the track record is in hand, so we are only awaiting the upcoming eclipse times.  These eclipse times give two unique types of measures;  First, the steady period change (Pdot) between eruptions comes from eclipse times spread out evenly, and this measures the actual evolution of the orbit.  Second, the sudden sharp period change across one nova eruption requires many pre-eruption eclipse-times combined with many post-eruption eclipse-times so as to see the period change (Delta P).  Both programs need >~6 eclipse times per year, well, for every year.

So what are the details?  For eclipse timing, the band pass is not important, so run with no filter so as to get better signal-to-noise.  The integration time must be better than roughly five minutes or so.  You'll need a big enough telescope to get ~0.10 mag error bars out of eclipse.  Your primary targets, those with the big payoff, are U Sco, T Pyx, IM Nor, and CI Aql.  (V394 CrA is infrequently observed, as it is hard and the eclipse depths are variable.)  Make sure that your times for each CCD image are accurate, and report the JD times of *mid-exposure*.  You must cover at least ~2-4 hours in your time series that is well centered on the eclipses.  (Catching only the minimum plus an egress [or ingress] does not result in a time of minimum light with enough accuracy to be useful any more.)  For T Pyx and IM Nor, your photometry will cover more than the entire orbit, so you can start your time series anytime.  For U Sco and CI Aql, approximate ephemerides for mid-eclipse are
        U Sco:    HJDmin = 2451234.55 + N * 1.23054695
        CI Aql:    HJDmin = 2451669.04 + N * 0.61836051
These ephemerides are good for 2022 (and the next decade) to twenty minutes or so, with the variations from these ephemerides being the whole science point.  Your light curve alone is adequate, and the analyst will measure the time of minimum with a chi-square fit to a parabola of near-minimum mags, and make the heliocentric correction.


    #2.  DISCOVER NEW RN ERUPTIONS.  RN eruptions are fast, which means that an inattentive world can catch the nova perhaps long after the most critical days around peak, or perhaps entirely miss the eruption completely.  The discovery rate for the fast RN events is horrifyingly low, and we can do nothing about catching eruptions entirely inside the yearly `solar gap' (like missing the 2016.78±0.10 eruption of U Sco).  The only way to catch eruptions is someone with a small scope checking frequently.  The fast rise is always in under one day, and most of the RNe have the already faded by 3 mags from peak within one week.  So daily checks by someone in the world are what is needed for this science task.  While this task will perhaps seem endless and fruitless, the importance makes up for it.

The high-priority targets to check are CI Aql, V2487 Oph, IM Nor, and V394 CrA.  These `should' be checked nightly, and any of these four can go up any night now.  Many workers have realized that the pre-eruption plateau means that T CrB will go off soon (like ~2023), and many tens of observers are checking it nightly, so your high-expertise is better spent with other targets.  U Sco, RS Oph, V745 Sco, and V3891 Sgr have all erupted recently, with no need to check for later eruptions, at least until ~2027 or so.  There is strong reason for knowing that T Pyx will not erupt again in our lifetimes, so do not use your time monitoring this target for eruption.  A luscious opportunity is posed by the several classical novae for which I am sure are really RNe with short eruption recurrence times that could go off any night now.  (I have calculated and compiled a list of novae that have very massive white dwarfs and very high accretion rates, i.e., those that must have a fast recurrence rate.)  These novae are V4643 Sgr, V1534 Sco, V1187 Sco, V5589 Sgr, and V2275 Cyg.  Nova discovery carries high prestige, yet discovery of a recurrent-nova system carries higher prestige, plus good and unique science.

Checking for the discovery of new eruptions is the one RN task that can be done by visual observers.  For catching RNe near peak, the visual observer is just as good as a CCD observer.  And you'll know the answer immediately, without waiting for processing.

Checking for the discovery of new eruptions with a CCD is easy.  Just take an image with any filter (or unfiltered), and immediately compare it to the finder charts, such as those from AAVSO VSP plotter.  If you see the star *significantly* above its normal quiescence level, then immediately shout to the world, loudly.  (With the demise of the IAU Circulars, it is unclear how best to do this.  Simply submitting a magnitude to the AAVSO database might not trip the alarm bells.  An ATel might not be appropriate unless you have independent confirmation.  Maybe you could send or call a pre-checked address for your national VS organization.)  A trap to fall into is to misidentify the RN counterpart and set off the alarms for some nearby mistaken star.  Another trap is to mistake some ordinary or rare upward fluctuation as being the start of a nova event.  (Having a track record for the RN with your equipment is valuable for this.  Or, you could wait an hour and see if the RN is significantly brighter.)

In practice, if your CCD images record the RN in quiescence, then your data is also useful for another science task, and that is to define the brightness level changes during quiescence.  For this to be useful, your images and comparison stars must be in some standard system, preferably V.


    #3.  MONITOR BRIGHTNESS IN QUIESCENCE.  The default task for CV observers is to measure magnitudes occasionally when the target is not in eruption.  This is pretty general, without needing much expertise, but there is still good science to be gotten from this.

CVs and RNe vary up and down by up to 2 mags on all time scales from minutes to decades.  The causes are many, including the ubiquitous flickering in accretion disks, eclipses, ellipsoidal variations, irradiation of the secondary star by the hot white dwarf, plus a variety of perplexing long-term changes in the accretion rate.  There is good astrophysics to be pulled out from these.  My favorite is special for RNe.  It is based on the time interval (DeltaT) between eruptions of a given RN being inversely proportional to the accretion rate (Mdot), with DeltaT x Mdot being a constant for each RN.  That is if the accretion rate is doubled, then the time for the white dwarf to accumulate the trigger mass required to erupt will be cut in half.  The accretion rate is what provides most of the RN brightness (but not for the four with long orbital periods), so a simple measure of the B or V magnitude in quiescence will provide a relative measure of the accretion rate in the system.  The total mass accreted onto the white dwarf will be an integral over time since the last eruption, and when this reaches the trigger threshold we will have a new eruption.  For RN only, we can calibrate this integral over Mdot by looking at prior inter-eruption intervals.  Then, we can *predict* the times of upcoming RN events.  For this, we need a fairly good continuous record of the B or V mag for many months over long years.

That is where you can come in.  Simply image the RN once a month (ongoing for years maybe), derive a B or V magnitude, and send in the value.  Identify the target with the AAVSO VSP finder charts, and use the AAVSO comparison star magnitudes.  It is important that you use B or V filters, as only these can be readily compared with each other over the years.  It is important that you use the AAVSO magnitudes (or equivalently, the APASS magnitudes) for your comparison stars, so as to avoid sytematic color corrections that are unavoidable otherwise.

For this RN science task, the high priority targets are U Sco and T Pyx.  For U Sco, the science is to understand the next eruption.  For T Pyx, we need to follow the weird secular decline as the accretion rate turns off after its ~1860 classical nova eruption.  The quiescent level of T CrB has big payoff for trying to understand its weird and mysterious pre-eruption plateau and predicting the date of the upcoming eruption, as well as for following the ellipsoidal variations so as to measure its orbital period, but this is already well covered, thankfully, by many good observers nightly.  I will not recommend CI Aql, V394 CrA, IM Nor, or V2487 Oph as targets, as my own personal vast collection of magnitudes in quiescence (plus that of the CBA) has not revealed any interesting science motives.  I will not recommend RS Oph, V745 Sco, or V3891 Sgr as targets, as my personal vast collection of magnitudes (plus that of the AAVSO) only shows chaotic `pulsations' that are always changing and I do not see how any astrophysics result can be gotten from these.


    #4.  TIME SERIES PHOTOMETRY THROUGHOUT EACH ERUPTION.  The above science tasks are all for the RNe in quiescence, that is most anytime.  But RNe are defined by their eruptions, and small-telescope photometry is what defines everything for the science.  So as soon as you hear about a new RN eruption, drop everything and start taking time series in some standard band (preferably V).  If you can, take a times series in two or three standard filters (no unfiltered or nonstandard mags please) where you keep alternating between filters.  Take as long a time series as you can, like ideally horizon-to-horizon or dusk-to-dawn.  Use something like five-minute integrations, and use the AAVSO comparison stars (equivalently, the APASS comparison stars) and their given magnitudes to calibrate your reported mags.  Keep this up all the way until the nova returns to quiescence.

The critical science return is to define the nova's light curve.  This includes getting the colors (say B-V) at peak so as to allow the extinction E(B-V) to be calculated.  It also includes getting the overall shape of the light curve, for example to test for differences between RN eruptions.  But the most important science return for your many time series might be in recognizing new and unique events.  That is, few novae and only 2 RNe in eruption have been closely studied with long time series photometry, so there are horizons yet to be guessed at.  The most forceful case for this comes from the 2010 eruption of U Sco.  Previously, no nova or RN had any useable amount of long time series photometry.  For U Sco in 2010, the world's amateurs banded together to have the entire 60 days of eruption (all the time from peak to quiescence) closely monitored with an average rate of once every 140 seconds, for all 60 days!  Wow, a large effort for the first of its kind of data on any nova.  This effort discovered *two* completely new and unexpected phenomena.  The first was flares of hours-duration and half-mag amplitude that occurred only during the transition time (when the light curve breaks from its initial fast decline to the slower steady decline of the late days).  The second was the weird `eclipses' at all orbital phases, only seen in the last few weeks of the eruption.  These two newly-discovered phenomena are still completely mysterious and have theorists scrambling.  To understand these two new phenomena, of course, we need more examples.  Two completely new phenomena were discovered the first time a nova was looked at with time series photometry, and this strongly suggests that there are yet other new phenomena waiting to be discovered.


    #5.  SUPERFLARES ON V2487 OPH.  In a whacko surprise, V2487 Oph was discovered to have superflares (up to 1.1 mag amplitude, typically one hour in duration) about once every day.  Where did this come from??  The discovery was made with the 67-day continuous photometry of the K2 mission of the Kepler satellite.  Holy cow, around once a day, these humongous 10^38 erg superflares blast away.

I think that the evidence is good that these superflares are caused by magnetic reconnection for field lines connecting the subgiant companion star to the accretion disk.  But the details form a horrifyingly hard problem for theorists.  (For full details, see Schaefer et al. 2022, MNRAS, 512, 1924, or .)   The way forward has to start with a good demographics study of the flares and how they change.  For this, a middle sized scope, say larger than 16-20-inches, can sit on V2487 Oph and get long time series photometry.  No ground based scope can hope to compete with the K2 light curve in hand.  But a long database from one observer can tell about how the flare frequency changes, and search for correlations (like how the flares and their frequency change with the quiescent brightness).  And you might be able to get a real handle on the orbital period from your photometry.  This science task will be long and hard.


    #6.  T CrB WILL ERUPT ~2023.  Everyone is in agreement (for various good and poor reasons) that T CrB will soon erupt.  Photometry will be likely well-handled by many people.  But *spectroscopy* is not being well handled by the professionals, as far as I know.  And only for the case of T CrB, independent spectroscopists can be perfect for recording a full record with high cadence while T CrB is still in quiescence.  This is the one chance for spectroscopists to do great work on any RN.

T CrB erupted in 1866 and 1946, 80 years apart.  The 1866 eruption was the first well observed real nova event, and it largely started and defined the field.  For decades, Leslie Peltier regularly monitored the star, hoping for a second eruption.  In a poignant tale in STARRY NIGHTS, he tells how he missed the 1946 eruption just barely, and that all love for that star was lost.  Part of the reason why he was expecting a recurrent event (at a time when the whole class of RN was basically unknown) was that he visually measured T CrB to be ~1.5 mag brighter than usual, starting a few years before 1946.  This is the unprecedented pre-eruption plateau that is now ongoing with T CrB.  Immediately before the eruption to peak at V=2.5 mag, T CrB dimmed out of its plateau state and returned to its usual brightness for about two weeks, *then* it erupted.  (So hopefully, we'll get a two-week notice for the upcoming eruption.)  The T CrB eruption is still in the top few in terms of peak brightness.  Its eruption was also one of the fastest.  Startlingly, after returning to its pre-eruption quiescent level for two months, it underwent an echo eruption getting up to V=8 for four months.  (I have pulled up and compiled records from 1866 that shows a similar event then.)  Then, for ~8 years, T CrB remains at its plateau level, until it dropped back to the long-time normal brightness level.  That is a lot of complex and completely un-explained phenomena, all for T CrB around the time of it eruptions.  I have collected a great light curve from historical records (like the Harvard plates), but the *spectroscopic* record is scant.  No one has any idea as to what is going on.  (Or better to say that 5 theorists have 5 widely divergent proposals.)  The trouble, I think, is that we do not have good spectroscopy throughout the years before and after the eruption.

Well, *now* is the time to get great spectroscopy of T CrB around the time of eruption.  This is not being done by professionals, as the target is `too bright' and they can get few nights of telescope time anyway.  So the task falls to observers with their own middle-sized scopes.  For this task, needed are both low resolution with very wide wavelength coverage, plus very high resolution for selected lines.  (Pick Halpha, Hbeta, the 5007Å oxygen line, the the 4686Å helium line.)  If you can, get regular spectra going down to below 3800Å so as to record the critical Neon III lines.  (An open and critical question from 1946 is whether T CrB is a neon nova.)  Get high SNR.  Observe maybe once a week for the next decade.  (Once a night during the fast action around its eruption.)  If you can, calibrate your spectra into physical units (like erg/cm^2/sec/Å.)  Make sure to upload your fully-processed and reduced spectra to some archive or data base where some professional-spectrum-analyst can be confronted with your measures.