Discovery of the 2010 Eruption and the Pre-Eruption Light Curve for Recurrent Nova U Scorpii
Bradley E. Schaefer, Ashley Pagnotta, Limin Xiao, Matthew J. Darnley, Michael F. Bode, Barbara G. Harris, Shawn Dvorak, John Menke, Michael Linnolt, Matthew Templeton, Arne A. Henden, Grzegorz Pojmański, Bogumil Pilecki, Dorota M. Szczygiel, Yasunori Watanabe
aa r X i v : . [ a s t r o - ph . S R ] A p r Discovery of the 2010 Eruption and the Pre-Eruption Light Curvefor Recurrent Nova U Scorpii
Bradley E. Schaefer, Ashley Pagnotta, Limin Xiao
Physics and Astronomy, Louisiana State University, Baton Rouge, LA 70803
Matthew J. Darnley, Michael F. Bode
Astrophysics Research Institute, Liverpool John Moores University, Birkenhead, CH411LD, UK
Barbara G. Harris, Shawn Dvorak, John Menke, Michael Linnolt, Matthew Templeton,Arne A. Henden
American Association of Variable Star Observers, 49 Bay State Road, Cambridge MA02138
Grzegorz Pojma´nski, Bogumi l Pilecki
Warsaw University Observatory, Al. Ujazdowskie 4, 00-478 Warszawa, Poland
Dorota M. Szczygiel,
Department of Astronomy, The Ohio State University, 140 W. 18th Ave., Columbus OH43210
Yasunori Watanabe
Variable Star Observers League of Japan, Keiichi Saijo National Science Museum,Ueno-Park, Tokyo Japan
ABSTRACT
We report the discovery by B. G. Harris and S. Dvorak on JD 2455224.9385(2010 Jan 28.4385 UT) of the predicted eruption of the recurrent nova U Scorpii(U Sco). We also report on 815 magnitudes (and 16 useful limits) on the pre-eruption light curve in the UBVRI and Sloan r’ and i’ bands from 2000.4 up to9 hours before the peak of the January 2010 eruption. We found no significantlong-term variations, though we did find frequent fast variations (flickering) withamplitudes up to 0.4 mag. We show that U Sco did not have any rises or dipswith amplitude greater than 0.2 mag on timescales from one day to one yearbefore the eruption. We find that the peak of this eruption occurred at JD2455224.69 ± ± ± ∼
10 years, although at this time we can only predictthat the next eruption will be in the year 2020 ± Subject headings: novae, cataclysmic variables
1. Introduction
Recurrent novae (RNe) are ordinary novae (binary systems with mass accreting onto awhite dwarf until thermonuclear runaway is triggered) for which the recurrence time scale isbetween a decade and a century, such that more than one eruption has been observed (Payne-Gaposchkin 1964; Bode & Evans 2008; Evans et al. 2008). To have the fast recurrencetime scale, the novae must have the white dwarf near the Chandrasekhar mass and havea high accretion rate. These properties, at face value, imply that the white dwarf willsoon exceed the Chandrasekhar mass and become a Type Ia supernova, and thus RNe areone of the premier candidates for the progenitor class of these supernovae. RNe typicallyhave relatively fast eruptions, high ejection velocities, and small eruption amplitudes whencompared to ordinary novae. Only ten RNe are known with certainty in our Milky Way(Schaefer 2010).U Scorpii (U Sco) previously erupted in March 1999 with a peak at V=7.5 mag (Schaefer2010). In quiescence, it has V ≈ total eclipses taking it down to V=18.9mag (Schaefer 2010) with an orbital period of 1.23 days (Schaefer 1990; Schaefer & Ringwald1995). U Sco is the fastest of all known novae, fading by three magnitudes from peak in just2.6 days, while its rise from minimum to peak is 6-12 hours (Schaefer 2010). No light echowas detected to deep limits after the 1987 eruption (Schaefer 1988).U Sco has now had ten known eruptions, in the years 1863, 1906, 1917, 1936, 1945,1969, 1979, 1987, 1999 (Schaefer 2010), and now 2010 as we report in this paper. Withthe discovery of the 1917, 1945, and 1969 eruptions (Schaefer 2001; 2004), it has becomeapparent that U Sco has outbursts at intervals of 10 ± . ◦ from theSun every 28 November, so a significant fraction of its very fast eruptions must be missed.)With this, it became apparent that the next eruption of U Sco should occur in the year 3 –2009 ±
2. Schaefer (2005) made a better prediction, on the physical basis that the timebetween eruptions scales as the inverse of the average mass accretion rate between eruptions(as measured from the B-band flux), with the scaling determined by the inter-eruption lightcurves from prior eruptions. The predicted eruption date was 2009 . ± .
0. This is the firsttime that a specific star has been predicted to have an eruption on a specific date.With this advance notice, a large international collaboration was formed to providedetailed photometry and spectroscopy in the X-ray, ultraviolet, optical, and infrared bands.With U Sco going from quiescence to peak to one magnitude below peak in 24 hours, werealized that we must have frequent monitoring of U Sco to get a fast alert of an eruption.To this end, we mobilized daily and hourly photometry with the SMARTS 1.3-m telescopein Chile, the fully-robotic 2.0-m Liverpool telescope (Steele et al. 2004) in the CanaryIslands, and the four ROTSE 0.45-m telescopes in Australia, Texas, Namibia, and Turkey.In addition, we mobilized a large number of observers through the American Association ofVariable Star Observers (AAVSO). For the seven months each year centered on the oppositionof U Sco, we got hourly data. The headquarters of the AAVSO served as the internationalclearinghouse for discovery reports and delivery of alerts to the world. In addition, U Scowas heavily monitored from 2001 to 2009 with long time series photometry, where the maingoal was to precisely measure the timing of the eclipses. The result of all this activity from2000-2010 is the all-time best pre-eruption light curve for any nova. This paper presents allthe magnitudes and an analysis of this large data set.
2. The Observations
Since 1987, one of us (BES) has heavily monitored U Sco, with emphasis on the lightcurve around the time of the eclipses (Schaefer 1990; 2005; 2010; Schaefer & Ringwald 1995).These observations have been made with the McDonald 2.7-m, 2.1-m, and 0.8-m telescopesin Texas as well as with the 1.3-m, 1.0-m, and 0.9-m telescopes on Cerro Tololo in Chile.The typical integration times were 300 seconds in the B-band and I-band and 100 secondsin the V-band. Normal processing was carried out, and the photometry was done using theIRAF package PHOT, which performs aperture photometry on the stars in this uncrowdedfield. The magnitude of U Sco was determined relative to a selection of nearby comparisonstars, for which the primary comparison star, named ‘COMP’ (J2000 16:22:25.6 -17:51:34),has B=16.96, V=15.87, R=15.25, and I=14.59 (Schaefer 2010). The photon statistics, ascalculated by PHOT, are generally smaller than 0.01 mag, but the systematic uncertainties,as represented by the scatter in the measures of standard star magnitudes (Landolt 1992;2009), are typically 0.015 mag. The quoted uncertainty is the addition in quadrature of 4 –0.015 mag and the uncertainty from photon statistics. This data set consists of over 2100magnitudes, mostly as fast time series photometry centered on times of eclipses. The specificanalysis of the eclipse shapes has already been presented in Schaefer (2010), while a specificanalysis of the eclipse times is reserved for a separate paper. For the study in this paper, theeclipse effects would only hide the other variability, so we have not included any magnitudeswith orbital phase between -0.10 and +0.10. In all, we have 162 magnitudes in the U, B, V,R, and I filters from 2001 to 2006.Starting in early 2008, we (BES, AP, and LX) began frequent regular monitoring of USco with the 1.3-m SMARTS telescope on Cerro Tololo. This telescope is queue-scheduled,so an operator takes images of U Sco for us several times a week, thus allowing long-termfrequent monitoring without requiring us to be at the telescope year-round. Most of theobservations were 300 second exposures in the B-band, but we also made several sets ofnearly simultaneous BVRI images. The procedures and analysis were identical to thosedescribed in the previous paragraph. In all, we have 145 magnitudes from early 2008 untillate 2009.Beginning in early 2008, one of us (BES) started using the robotic ROTSE telescopesto monitor U Sco once every hour. The ROTSE telescopes (Akerlof et al. 2003) are fourautomated 0.45-m f/1.9 telescopes with 1.85 ◦ fields designed to provide very fast response tosatellite triggers on Gamma-Ray Bursts. The four telescopes are located at Coonabarabran,Australia; Mount Gamsberg, Namibia; Bakirlitepe, Turkey; and McDonald Observatory,Texas. This wide coverage in longitude gives the potential for complete time coverage. Nofilters were used, so the resultant magnitudes are similar to a very broad R-band. Theexposure time was 60 seconds in all cases. The requested cadence was one exposure everyhour from every ROTSE telescope, but problems such as clouds, dawn, daylight, a nearbyFull Moon, an altitude lower than 20 ◦ , higher priority alerts for Gamma-Ray Bursts, andthe usual equipment problems all make for a substantially lower cadence. In the monthsaround opposition, the ROTSE system achieved the ideal of nearly hourly coverage foraround a quarter of the days, while the average coverage was roughly 15 images in every 24hour interval. In the months approaching the conjunction of U Sco with the Sun, the dailycoverage decreased to one or two images per 24 hour interval. For example, in 2009, ROTSEfirst recorded that U Sco was not in eruption on 9 January (43 days after conjunction) andlast imaged U Sco on 18 October (40 days before conjunction). The limiting magnitudevaried widely (with clouds, altitude, focus, and the Moon), yet U Sco was visible at lowsignificance on about half the images. Even on the best images, U Sco did not get betterthan a 5-sigma detection, so in no case do we have accurate photometry from ROTSE. Inall, we have a set of roughly 7000 useable U Sco images from ROTSE. One of the goals of thehourly monitoring by ROTSE was so that we (BES, AP, and MT) could frequently check 5 –the images to try to discover the eruption as soon as possible. Another goal was to catchany pre-eruption rise (see Section 4) even if the amplitude was small and the duration wasshort. For the eruption in January 2010, there was no significant pre-eruption rise and our(BGH, SD, JM, ML) small-telescope monitoring produced a better light curve than ROTSE.A third reason for the ROTSE program was the hope that we would catch U Sco on therise. In all previous eruptions, U Sco has been recorded on the rise only three times, eachbeing close to the peak, with the rise from quiescence apparently lasting 6-12 hours (Schaefer2010). In the hours before the discovery of the 2010 eruption, the Namibia ROTSE telescopedid not look at U Sco due to a higher priority follow-up to a Gamma-Ray Burst, the TurkeyROTSE had clouds, and the Australia ROTSE was down with equipment problems. Withthe chance lack of any data on the rise and the poor photometric accuracy of the two yearsof monitoring, we are not presenting any of the ROTSE magnitudes in this paper.Beginning in early 2008, we (MT and AAH) organized a steady watch on U Sco by themany observers of the AAVSO. The primary goal was to catch U Sco’s eruption as quickly aspossible. The widespread distribution in longitude of the many AAVSO observers makes forfrequent monitoring, and this was the best chance of catching the eruption early. For the halfyear around opposition, U Sco was checked for outburst up to 6.7 times per day for monthlyaverages. A further requirement for getting fast reactions from the world’s telescopes was thatthe discovery had to be communicated from the discoverer to the rest of the world. For thisvital need, the AAVSO Headquarters served as an around-the-clock, every-day-of-the-yearcommunication center. Observers were instructed to report their discovery electronically,then automated services would alert key individuals who would test for validity and solicitfast confirmation. Once the eruption was discovered, we would immediately start notifyingthe world through IAU Circulars and long-prepared phone and email lists.As part of this effort, many AAVSO members made positive measures of the brightness ofU Sco during the pre-eruption phase. The AAVSO database contains 412 magnitudes (from29 observers) and 2853 limits (from 102 observers) between the end of the 1999 eruption andthe start of the 2010 eruption (JD 2451557.148 to 2455224.127). The limits were vital atthe time of the observation for knowing that U Sco had not erupted, but they are not nowhelpful for following the accretion rate. A further 77 magnitudes are not used here, primarilybecause the photometric system is not standard and the meaning of the magnitude wouldbe unclear. This leaves us with 335 positive detections in the pre-eruption time interval.Just over 90% of these magnitudes were made with unfiltered CCD imaging, where themagnitudes were calibrated differentially from nearby comparison stars using either the V-band or R-band magnitudes. These magnitudes (designated CV or CR) will not be exactlyon either the V or R magnitude systems, but the expected deviations (less than 0.1 mag)are always small compared to normal variations of U Sco. Our instrumentation is a 16-inch 6 –f/10 Schmidt-Cassegrain with a V filter located in New Smyrna Beach, Florida (BGH), an18-inch Newtonian telescope without filter located in Barnesville, Maryland (JM), and a 10-inch Schmidt-Cassegrain telescope with a V filter located in Clermont, Florida (SD). Duringthe critical month before the eruption, all positive detections in the AAVSO database areprovided by us (BGH and JM) from CCD images.Beginning in early 2009, we (MJD and MFB) started monitoring U Sco with the robotic2.0-m Liverpool Telescope (Steele et al. 2004) at the Observatorio del Roque de Los Mucha-chos on La Palma in the Canary Islands. The goals were to define the pre-eruption lightcurve in many bandpasses and perhaps to catch a pre-eruption rise or the eruption rise itself.The photometry was all differential with respect to the comparison stars given in Schaefer(2010). The images were usually taken through many filters in quick succession once eachnight. The filters we used were the B, V, Sloan r’, and Sloan i’. Each light curve pointconsisted of three 60 s exposures. The data were analysed using Starlink software. Thetypical photometric errors had an uncertainty of 0.01-0.02 mag. In all, we present here 173magnitudes from the Liverpool Telescope.An important practical question was whether U Sco erupted during its yearly conjunc-tion with the Sun every 28 November. The worst case scenario would be for U Sco to go up inearly November, fade back to its quiescent level before any detection was made, and for theeruption to be completely missed. If U Sco went up while behind the Sun, it would be vitalto know this so that our community would not be waiting anxiously with many resources,and also so that observations in the late tail could still be performed. For this, deep imageswould have to be made as far into twilight as possible. Professional telescopes do not go lowenough in the sky, so the push into twilight was made entirely by AAVSO observers. For theNovember 2008 solar conjunction, U Sco was lost on 2 November and deep images showedU Sco to be near quiescence (V > ∼
42 daysafter the peak. With this extremely short duration, the possibility for a missed eruption wasfor a peak from 3-23 November 2008 or 7-16 November 2009. As a chance to discover a USco eruption in the week around solar conjunction, one of us (SD) used the SOHO LASCOC3 instrument (which hides the Sun behind a white light coronagraph and produces imagesof stars, comets, and the corona out to 32 solar radii from the Sun) to demonstrate that thenova never came to peak (i.e., V > > ◦ on a side on a2048x2048 pixel CCD chip. The last known observation before discovery (with V > HJ D = 2451234 .
539 + N × .
3. Discovery of the 2010 Eruption
The 2010 Eruption was discovered by us (BGH and SD) as part of systematic nightlymonitoring aimed specifically at the discovery of the eruption. Harris imaged U Sco at2010 Jan 28.4385 UT (JD 2455224.9385), saw the bright star in the center of the field, andquickly realized that U Sco was in eruption. Her first act was to send the observation tothe AAVSO, and then she telephoned Schaefer. Schaefer could not get confirmation fromROTSE, so he took his 6-inch telescope out into the front yard and made direct visualconfirmation that U Sco was bright in eruption. Independently, Dvorak discovered theeruption, notified the AAVSO, and started a time series on U Sco to cover the short timeinterval until dawn got too bright to continue. These initial observations are included inTable 1. Circumstances, pictures, and anecdotes on the two independent discoveries aregiven in Simonsen & MacRobert (2010).In practice, our organization worked perfectly. The AAVSO automated alert systemwoke up MT and AP. Within an hour of the discovery, the eruption had been confirmedand worldwide notifications were started. The first was to the IAU Circulars (Schaefer et al.2010a). The sun had already risen in Chile, so we started with more western observatoriesas well as spacecraft. Within two hours, BES, AP, and MT had worked through all the long-prepared contact lists. The response to these contacts (both by members of our existingcollaboration as well as by independent observers) was excellent and fast.The discovery of the 2010 eruption was a fulfillment of the prediction in Schaefer (2005)that U Sco would next erupt in the year 2009.3 ±
4. Variations in the Light Curve
The folded light curve (see Figure 3) shows the primary eclipse at phases 0.0, 1.0, and2.0. (The magnitudes are double plotted so as to make the eclipse at phase 1.0 easily visible.)The out-of-eclipse brightness varies substantially, and this makes for a ragged eclipse lightcurve because each point is from a different epoch eclipse with a different amount of flickeringlight added. The scatter around the middle of the eclipse is much smaller than the out-of-eclipse scatter, which implies that the flickering region is small and centrally located.No secondary eclipse is visible in the B and V bands. However, in the I-band, thesecondary eclipse is visible with amplitude roughly 0.3 mag. This is readily understood as 9 –the companion star is much cooler than the accretion disk so eclipses of the companion canonly become noticeable at longer wavelength.All cataclysmic variables, including novae and recurrent novae, show fast flickering. USco is no exception, and this flickering causes the substantial scatter in Figures 1 and 2. Toquantify this, we have calculated the magnitude difference between pairs of magnitudes inthe same band, with the pairs being separated in time by some range of delays. When thedelays are shorter than one hour, the RMS scatter of the magnitude differences is 0.06 mag,which is consistent with the expected scatter as based only on the quoted error bars. Whenthe delays are longer than one day, the RMS scatter of the magnitude differences is 0.27,which corresponds to no correlation between the flickers. The RMS scatters are 0.09, 0.13,0.18, 0.19, and 0.21 mag for delays of 0.05-0.10, 0.10-0.15, 0.15-0.20, 0.20-0.50, and 0.50-1.00days, respectively. With this, the timescale for correlated variations is from one hour to oneday. The amplitude of these variations is given by the maximum values of the magnitudedifferences, which is roughly 0.4 mag for all delays longer than 0.05 days. Our observationsare not sensitive to fast, small-amplitude variations.Many of the recurrent novae have large secular variations in their quiescent light curves(Schaefer 2010). However, U Sco does not appear to have any long-term variations, as can beseen in Figure 1. To quantify this for the B and V light curves, Table 2 gives the averages forvarious time intervals. Again, no significant variations on time scales of one year or longer arefound, despite having small error bars due to the many magnitudes included in the averages.On the timescale of a tenth of a year, there is marginal evidence for variations, with theV-band average for 2009.7-2009.9 being 0 . ± .
05 fainter than average, but this variationis not reflected in other bands, so we think that this apparent change is not significant. Inall, we see no evidence for variations on timescales from 0.1 to 10 years.On longer times scales, U Sco has small, marginally-significant variations. During thelast four inter-eruption intervals, the average B magnitude was 18.44 ± ± ± ± ± ±
5. No Pre-eruption Rise or Dip
Robinson (1975) examined all known pre-eruption light curves of novae as based onreports in the literature. He found that five out of eleven novae have pre-eruption rises,lasting months to years in advance of the eruption, with amplitudes from 0.15 to 1.5 mag.Collazzi et al. (2009) have gone to the original archival photographic plates to measure manypre-eruption light curves, including the key novae with claimed pre-eruption rises. Four ofthe five claimed pre-eruption rises were found not to exist as based on our examination ofthe original plates, such that the claimed rises were caused by simple errors in the literature.Nevertheless, one of the rises (for V533 Her) was confirmed and extended, with the risebeing an exponential increase over ∼ classical nova eruption, and was second only to U Sco itself amongst all novae.) Inaddition, a complex pre-eruption dip was confirmed for the recurrent nova T CrB in the yearbefore the eruption, with the dip being 1-2 mag deep. In all, three out of 22 novae had eitherpre-eruption rises or pre-eruption dips.With other recurrent novae and the fastest classical nova having anticipatory rises anddips, we should investigate whether U Sco has any similar changes. (We were also hopefulthat any such phenomenon would allow us to anticipate the next eruption.) For this, we canuse our light curves to seek any rises or dips. A glance at Figure 1 quickly shows that thereis no significant rise or dip. Quantitatively, Table 2 shows that the B and V magnitudesfrom 2010.0-2010.1 are not significantly high or low. And looking at the bottom of Table 1,we see that the last positive detection of U Sco has V=18.2 (BGH) just 24 hours before thediscovery. With this, we can put strong limits on the presence of any pre-eruption rise ordip to be less than roughly 0.2 mag in amplitude on time scales from one day up to a year.
6. The Rise to Peak
U Sco rises from quiescence to peak in roughly 6-12 hours, although this is based on justthree pre-peak positive detections and one limit (Schaefer 2010). For the 1936 eruption, asingle Harvard plate shows B=10.75 at a time 0.25 days before peak. For the 1987 eruption,Dr. N. W. Taylor measured V=14.0 in the day before the peak. For the 1999 eruption, P.Schmeer measured V=9.5 at a time 0.31 days before the peak. The final rise to maximumis at a rate of around 19 magnitudes per day, which would imply a rise time of roughly 12hours if the rise is uniform throughout. For the 1999 eruption, B. Monard set a limit that
V > . ±
7. Predicting the Next Eruption
Schaefer (2005) presented a new method for predicting the date of the next eruptionof a recurrent nova based on the requirement that some constant amount of mass mustbe accumulated by the white dwarf between eruptions. Accretion rates vary substantially(Schaefer 2010; and see Figure 1), so the interval between eruptions ( T ) depends on theaverage accretion during that time. If the accretion rate is high then the interval will beshort, while if the accretion rate is low then T will be high. For U Sco, the blue light isdominated by the accretion disk, so the blue flux ( F B ) will be a measure of the accretionrate. In particular for U Sco, the accretion rate will be proportional to F . B (Schaefer 2005).By averaging F . B over each interval T , we can derive a quantity that is proportional to theaverage accretion rate. Then, h F . B i T should be proportional to the total mass accreted 12 –between eruptions, which should be a constant. Schaefer (2005) found that this quantity isindeed constant for four intervals for T Pyx and three intervals for U Sco (despite widelyvarying values of T for each system), with this providing a good test of nova trigger theory.These observed values for h F . B i T provide empirical measures of the mass required to triggerthe eruptions. Then, based on the observed B magnitudes up until 2005, Schaefer (2005)was able to predict that U Sco would next erupt in 2009 . ± .
0. Schaefer (2010) updatedthe situation to arrive at the same predicted date. As noted above, the actual eruption on2010.1 falls well within this prediction.Now, with the full pre-eruption light curve, we can better test the prediction and wecan refine the constant for use in predicting the next eruption. To this end, we have firstconverted the B-band magnitudes to flux units where B=18 is taken to be the unit flux( F B, ), which equals 10 (18 − B ) / . . The accretion rate will then be proportional to F . B, . Toget the time averaged value from 2000-2010.1, we should not simply average all the values,as this would produce a high weight to the behavior of U Sco during 2008 and 2009 (duringwhich the majority of the B-band magnitudes were taken). Instead, we have taken timeintervals and combined them with weights given by their duration. In Table 2, we list theaverage values of the measured F . B, for three intervals with roughly constant frequency ofobservations. The uncertainty in these averages is the RMS scatter divided by the squareroot of the number of observations. The durations of these intervals were then used asweights for averaging the intervals, and the uncertainty in the overall average comes fromthe usual propagation of errors. The resultant average over the entire interval between the1999.2 and 2010.1 eruption is listed in Table 2. In all, h F . B, i = 0 . ± . h F . B i for the three prior inter-eruption intervals.These need to be updated for four reasons. First, we need to standardize to the sameflux level (i.e., B=18) as the unit flux. Second, we should be consistent and not includeany magnitudes within phase 0.10 of the eclipse. Third, the individual observations shouldbe formed into the averages with equal weight (instead of weighted by the measurementuncertainty) because intrinsic fluctuations are substantially larger than the measurementerrors (so we would not want to give high weight to a bright point simply because it hasa small error bar). Fourth, the 1969-1979 interval has only two magnitudes, so instead ofdetermining the uncertainty based on the RMS scatter of just these two, we have equatedthe scatter to that during the 1987-1999 interval. The resultant h F . B, i values are given inTable 2.U Sco erupted in the years 1969, 1979, 1987, 1999, and 2010, with T values of 10.4,7.9, 11.8, and 10.9 years for the four inter-eruption intervals. The longest interval is a factorof 1.5 × the shortest interval. We see that the shortest interval has the highest average 13 –accretion rate, while the longest interval has the lowest average accretion rate, and the twomiddle intervals have the middle average accretion rates. The four intervals in time orderhave h F . B, i T values of 5 . ± .
2, 5 . ± .
4, 6 . ± .
4, and 6 . ± .
5. (The weighted average ofthese four values is 5.77 ± T varying by up to a factor of 1.5. Sowe have an improved confirmation of nova trigger theory.U Sco erupted in 1945 and 1969, with an inter-eruption interval of 23.7 years. The longinterval could be because one eruption was missed around 1957 (with intervals of around11.8 and 11.9 years) or because two eruptions were missed around 1953 and 1961 (withintervals or around 7.9, 7.9, and 7.9 years). These two possibilities can be distinguished duetheir greatly different prediction as to the quiescent B magnitude, roughly 18.52 versus 18.30respectively. The one measured magnitude (B=18.80 ± h F . B, i T will equal 5 . ± .
24. Such a predictionwill only be accurate to roughly 5 months out of ten years, which is fairly good. However,this method cannot be used yet, because we cannot predict the variations in the U Scoaccretion rate. For now, the best that we can do is to use the long record of U Sco where allof its inter-eruption intervals are 10 ± ± REFERENCES
Akerlof, C. W. et al. 2003, PASP, 115, 132Bode, M. F. & Evans, A. 2008, Classical Novae, 2nd edition (Cambridge, Cambridge Univ.Press) 14 –Collazzi, A.C., Schaefer, B.E., Xiao, L., Pagnotta, A., Kroll, P., L¨ochel, K. & Henden, A.A.2009, AJ, 138, 1846Evans, A., Bode, M. F., O’Brien, T. J. & Darnley, M. J. 2008, RS Ophiuci (2006) and theRecurrent Nova Phenomenon (ASP Conf. 401)Hachisu, I. & Kato, M. 2006, ApJSupp, 167, 59Landolt, A. U. 1992, AJ, 104, 340Landolt, A. U. 2009, AJ, 137, 4186Payne-Gaposhkin, C. 1964, The Galactic Novae (Dover, New York)Robinson, E. L. 1975, AJ, 80, 515Schaefer, B. E. 1988, ApJ, 327, 347Schaefer, B. E. 1990, ApJLett, 355, L39Schaefer, B. E. 2001, IAUCirc., 7749Schaefer, B. E. 2004, IAUCirc., 8279Schaefer, B. E. 2005, ApJLett, 621, L53Schaefer, B. E. 2010, ApJSupp, 187, 275Schaefer, B. E., Harris, B. G., Dvorak, S., & Linnolt, M. 2010a, IAUCirc., 9111Schaefer, B. E. & Ringwald, F. A. 1995, ApJLett, 447, L45Schaefer, B. E. et al. 2010b, ATel, 2507Simonsen, M. & MacRobert, A. 2010, Sky & Telescope, 119, No. 5, 18Steele, I. A. et al. 2004, Proc. SPIE, 5489, 679
This preprint was prepared with the AAS L A TEX macros v5.2.
15 –Table 1. U Sco Pre-eruption Light Curve
HJD Band Magnitude Source Phase Year2451702.1932 CV 18.1 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
16 –Table 1—Continued
HJD Band Magnitude Source Phase Year2452891.2341 CR 17.3 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
17 –Table 1—Continued
HJD Band Magnitude Source Phase Year2453191.6923 I 17.46 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
18 –Table 1—Continued
HJD Band Magnitude Source Phase Year2453196.5446 I 17.33 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
19 –Table 1—Continued
HJD Band Magnitude Source Phase Year2453282.1998 CR 17.6 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
20 –Table 1—Continued
HJD Band Magnitude Source Phase Year2454663.6044 B 18.36 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
21 –Table 1—Continued
HJD Band Magnitude Source Phase Year2454861.0752 CV 18.56 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
22 –Table 1—Continued
HJD Band Magnitude Source Phase Year2454919.5610 CR 17.2 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
23 –Table 1—Continued
HJD Band Magnitude Source Phase Year2454937.5482 V 18.20 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
24 –Table 1—Continued
HJD Band Magnitude Source Phase Year2454964.5124 Sloan i’ 17.74 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
25 –Table 1—Continued
HJD Band Magnitude Source Phase Year2454978.3238 CR 17.8 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
26 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455002.5382 V 18.48 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
27 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455020.7370 V 17.14 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
28 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455032.0822 CV 18.19 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
29 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455048.4355 Sloan i’ 17.52 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
30 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455060.5543 CV 18.19 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
31 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455085.3689 Sloan i’ 17.80 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± > ± > ±
32 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455158.9458 CV > ± > ± > ± > ± > ± > ± > ± > ± > ± > ± > ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
33 –Table 1—Continued
HJD Band Magnitude Source Phase Year2455224.1271 Visual > ± > ± > ± ± ± ± Table 2. Average Magnitudes, Colors, and Fluxes
Quantity Date Range Average RMS NumberU 2004.5 18.13 . . . 1B 2001.6-2010.1 18.45 0.26 145V 2000.4-2010.1 18.01 0.29 229R 2001.2-2009.8 17.67 0.26 127I 2001.5-2009.7 17.35 0.14 115Sloan r’ 2009.3-2010.1 17.86 0.18 26Sloan i’ 2009.1-2010.1 17.53 0.20 33U-B 2004.5 -0.27 . . . 1B-V 2005.7-2010.1 0.54 0.06 34V-R 2001.2-2009.4 0.34 0.05 10R-I 2004.4-2009.7 0.46 0.09 7r’-i’ 2009.3-2010.1 0.29 0.07 33B 2001.6-2006.2 18.45 0.29 11B 2008.1-2008.8 18.42 0.28 34B 2009.1-2009.8 18.46 0.26 98B 2010.0-2010.1 18.63 0.20 2V 2000.4-2001.7 18.07 0.33 12V 2003.4-2006.2 18.10 0.40 9V 2008.3-2008.8 17.91 0.20 18V 2009.1-2009.2 17.86 0.37 9V 2009.2-2009.3 17.92 0.26 25V 2009.3-2009.4 18.12 0.19 23V 2009.4-2009.5 17.93 0.46 17V 2009.5-2009.6 17.98 0.22 58V 2009.6-2009.7 18.06 0.22 27V 2009.7-2009.9 18.28 0.18 13V 2010.0-2010.1 17.97 0.33 17 F . B, F . B, F . B, F . B, a F . B, a F . B, a F . B, a F . B, a The quoted value is not the RMS scatter of the quantity,but rather is the one-sigma uncertainty in the average value.
34 –
U Sco (0.1 Year B ( m a g ) Fig. 1.— U Sco light curve in B, from 2000 to 2010. The B-band light comes almost entirelyfrom the disk and is a measure of the mass accretion rate. This light curve does not includeobservations within 0.1 phase of the eclipses. The light curve shows substantial short-termchanges but no significant long-term variations. 35 – U Sco (0.1 Year V ( m a g ) Fig. 2.— U Sco light curve in V for the last year before eruption. The observations within0.1 phase of eclipses are not included so as to concentrate on changes of the system brightnessalone. U Sco shows frequent short timescale variations, but long-term changes are apparentlynot significant. In particular, U Sco does not show any pre-eruption rise or dip on timescalesfrom one day to years before the eruption. 36 – U Sco (2000-2010) Phase B ( m a g ) U Sco (2000-2010) Phase V ( m a g ))