Energy release in the solar atmosphere from a stream of infalling prominence debris
DD RAFT VERSION N OVEMBER
7, 2018
Preprint typeset using L A TEX style emulateapj v. 01/23/15
ENERGY RELEASE IN THE SOLAR ATMOSPHERE FROM A STREAM OF INFALLING PROMINENCE DEBRIS
A. R. I
NGLIS , H. R. G
ILBERT AND
L. O
FMAN
1. Solar Physics Laboratory, Heliophysics Science Division, NASA Goddard Space Flight Center, Greenbelt, MD, 20771 and2. Physics Department, The Catholic University of America, Washington, DC, 20064
Draft version November 7, 2018
ABSTRACTRecent high-resolution and high-cadence EUV imaging has revealed a new phenomenon, impacting promi-nence debris, where prominence material from failed or partial eruptions can impact the lower atmosphere, re-leasing energy. We report a clear example of energy release and EUV brightening due to infalling prominencedebris that occurred on 2011 September 7-8. The initial eruption of material was associated with an X1.8-classflare from AR11283, occurring at 22:30 UT on 2011 September 7. Subsequently, a semi-continuous streamof this material returned to the solar surface with a velocity v > 150 km/s, impacting a region remote fromthe original active region between 00:20 - 00:40 UT on 2011 September 8. Using SDO/AIA, the differentialemission measure of the plasma was estimated throughout this brightening event. We found that the radiatedenergy of the impacted plasma was L rad ∼ ergs, while the thermal energy peaked at ∼ ergs. Fromthis we were able to determine the mass content of the debris to be in the range 2 × < m < × g.Given typical promimence masses, the likely debris mass is towards the lower end of this range. This clearexample of a prominence debris event shows that significant energy release takes place during these events,and that such impacts may be used as a novel diagnostic tool for investigating prominence material properties. Keywords:
Sun: filaments, prominences — Sun: UV radiation — Sun: corona — Sun: activity INTRODUCTIONIt has been known for some time that solar prominences,or filaments, exhibit a wide range of eruptive behaviour, upto and including the full ejection of significant material fromthe solar corona into the heliosphere. More commonly, eithera partial or failed eruption is observed (Gilbert et al. 2007),where some or all of the eruptive prominence debris fails toescape the solar atmosphere and falls back towards the sur-face. These types of prominence eruptions are closely associ-ated with coronal mass ejections and are key towards improv-ing our understanding of CME initiation.Despite decades of research, several properties of promi-nences remain poorly constrained (see Labrosse et al. 2010;Mackay et al. 2010, for recent reviews). For example, a novelmethod of determining prominence mass was presented byGilbert et al. (2005), yet the uncertainties in column den-sity remained substantial. Additionally, the filling factor ofprominences, as with many other solar phenomena, remainspoorly known (Kucera et al. 1998). In recent years how-ever, increased availability of state-of-the-art solar instrumen-tation, including those on-board the Solar Dynamics Obser-vatory (SDO), STEREO and IRIS, has dramatically increasedthe potential for detailed studies of this phenomenon.In particular, recent observations have shown that descend-ing prominence debris from failed eruptions can cause sub-stantial energy release and plasma heating upon impact withthe solar atmosphere (e.g. Gilbert et al. 2013; Reale et al.2013, 2014). This energy release is directly observable bySDO at EUV wavelengths via the Atmospheric Imaging As-sembly (AIA), providing a new diagnostic opportunity for un-derstanding the properties of CME-associated material andprobing the response of the solar atmosphere. The best ex-ample of this phenomenon observed to date is the flare- andCME-associated 2011 June 7 eruption (Reale et al. 2013;Gilbert et al. 2013; Innes et al. 2012, 2016; Inglis & Gilbert2013; Carlyle et al. 2014; van Driel-Gesztelyi et al. 2014; Yardley et al. 2016; Li et al. 2012), where localized EUVbrightening was observed at multiple impact points due to de-scending prominence debris. Such brightening patches arespatially and temporally resolved by SDO/AIA at multiplewavelengths, indicating that the plasma is multithermal andheated to several MK (Gilbert et al. 2013; Reale et al. 2013).Reale et al. (2013) compared these observations with the pro-cess of stellar accretion observed at UV and X-ray wave-lengths. Recently, such EUV brightening was observed inanother flare by Li & Ding (2017).Other phenomena sometimes compared to prominence de-bris include sequential chromospheric brightenings (SCBs;Balasubramaniam et al. 2005; Kirk et al. 2012, 2017), andcoronal rain in active regions (Antolin et al. 2012, 2015;Vashalomidze et al. 2015). Both these phenomena are sub-stantially different; SCBs generally occur only in the par-ent active region of the eruption, while the small descendingblobs associated with coronal rain have much lower down-ward speeds than eruptive prominence debris, do not causeobservable emission due to impacting in the solar atmosphere,and are not associated with prominence or CME material.In this work, we present a new case study of prominencedebris impacting the lower atmosphere, from 2011 Septem-ber 7-8. To the best of our knowledge, this is only the secondexample of this phenomenon subject to detailed study, and themost energetic observed to date. This event took the form of asingle, near-continuous material stream leading to continuousbrightening of an atmospheric region at multiple wavelengths.By examining its energetic and kinematic properties we con-strain the properties of the incoming stream, including the to-tal mass of the deposited material. In Section 2 we describethe initial observations of this eruption, while in Section 3 wepresent the methodology for estimating the radiated and ther-mal energy of the plasma, and infer the prominence mass. Theimplications of these estimates are discussed in Section 4. OBSERVATIONS a r X i v : . [ a s t r o - ph . S R ] S e p Figure 1.
The initial eruption beginning on 2011-09-07 is shown in panel a), where the white box indicates the impact point of the prominence debris. Panelsb), c) and d) show the descending material stream over an hour later, at 00:15 on 2011-09-08. The color contrast in panels b), c) and d) has been stretched toenhance the cool material.
The initial prominence eruption occurred at ∼ ∼ ANALYSIS AND RESULTS3.1.
Impact evolution
Figure 2 shows the evolution of the bright region caused bythe impacting prominence stream as observed by SDO/AIA.Panels a)-d) show the appearance of the bright region at 4 dif-ferent times. Clearly, the bright source undergoes substantialevolution over time during the stream impact. Hence, it isnecessary to construct a scheme to estimate the area of thebrightening over time. To achieve this we first find the pointin space and time corresponding to the maximum brighteningvalue in the SDO/AIA 193Å channel. From this we establisha threshold of 5% of this maximum for a pixel to be includedin the bright region. Using this threshold, for each image frame we find the point of maximum intensity in the bright-ening region, and expand in all directions until the thresholdis reached. Thus, for each frame we estimate the area of theinstantaneous bright, heated plasma. The white contour inpanels a)-d) shows the area defined by this scheme during theselected times. Figures 2e-g show the full evolution of thisbrightening region over time. Figure 2e shows the integratedflux from a constant area that encompasses the entire bright-ening (shown as Box A in panel a). This illustrates that allsix EUV channels experience an increase in flux during im-pact. However, the 171Å flux peaks slightly later than mostother channels, reaching a maximum at ∼ ∼ of thebright region as it evolves. Together, these panels show thatbrightening begins at ∼ ∼ ∼ × cm . For comparison, the largest ofthe impacts observed on 2011 June 7 was ∼ × cm .The descending prominence debris is visible in several con-secutive AIA image frames prior to impact with the lowercorona, as shown in Figure 1. Thus, we can estimate the Figure 2.
Evolution of the bright impact area due to the prominence material stream. Panels a)-d) show the hot plasma observed by SDO/AIA at 193A at differenttimes during impact. The white contour indicates the estimated area of the brightening at each time. Box A denotes a constant-area region used to illustratethe overall change in flux. Panel e) shows the integrated, normalized flux from box A in the 6 optically thin EUV channels. Panel f) shows the normalized,pixel-averaged lightcurves for the area contained within the contour. Panel g) shows the estimate of the brightening area during the event, as described in Section3.1.
Figure 3.
Velocity estimate of the descending prominence debris stream. a)The location of a distinct piece of the material stream in successive AIA 193Åimage frames. b) Linear fit to the position estimates of the material, yieldinga plane-of-sky velocity v = 146 km/s. plane-of-sky velocity of the infalling material. Although thetrue observing angle of the material stream cannot be deter-mined due to a lack of triangulation, plane-of-sky measure-ments can place lower limits on the stream velocity. In Figure3a, we estimate the position of a distinct piece of the descend-ing stream in successive AIA 193Å images. These locationsare shown by the white diamonds, propagating from solarwest to east. In Figure 3b, we perform a linear fit to thesepositions, finding the best-fit plane-of-sky velocity v ∼ ◦ in the z-direction, the true velocity would be ∼
220 km/s.Nevertheless, these values are similar to estimates of mate-rial velocity found by Gilbert et al. (2013) for the 2011 June 7 event, who used triangulated measurements from AIA andSTEREO-A, finding v ∼
150 - 300 km/s. For the same event,Reale et al. (2013) estimated v ∼
300 - 450 km/s. For the2011 September 7-8 event, the material also does not appearto experience significant acceleration during this time period,suggesting it may have already reached critical velocity.3.2.
Differential emission measure, energetics, and massestimation
Given the enhancements in emission from multipleSDO/AIA channels, we can investigate the energy release dur-ing the impact process by estimating the differential emissionmeasure (DEM) of the bright plasma. To estimate the DEM,we use the forward fitting technique developed by Aschwan-den et al. (2013), which was used by Gilbert et al. (2013) toestimate the energy of prominence debris impacts in the 2011June 7 event. We choose a DEM distribution of the form,
DEM ( T ) = EM exp (cid:18) log T − log T c σ (cid:19) (1)i.e. a Gaussian emission measure distribution with peaktemperature T c and width σ , as utilized by Aschwanden et al.(2013, 2015).The temperature response functions of the AIA channelsare the source of significant uncertainty (e.g. Aschwandenet al. 2013), particularly the 94Å and 131Å channels at lowtemperatures. To account for this, we include a 25% uncer-tainty in the measured AIA flux due to instrument response, assuggested by Boerner et al. (2012); Guennou et al. (2012a,b). Figure 4.
Examples of the DEM fits to the AIA flux data, at three differenttimes; 00:21:31 UT (blue), 00:25:31 UT (black), and 00:36:07 UT (red). Thetop panel shows the ratio of the modelled AIA flux to the observed flux ineach channel. The bottom panel shows the best-fit Gaussian DEM functionsfor the three example times.
This is combined in quadrature with the statistical uncertaintyassociated with the AIA flux measurements.The best fit to the observed flux is achieved at each timeinterval via a search over the parameter space given by thevariables EM , T c and σ using the χ test. Figure 4 showsexamples of the best-fit DEM results at three different times,near the start, peak and end of the brightening. This showsthat the majority of the brightening comes from increased EM at moderate temperatures, with log T c < dL rad dt = (cid:90) T T DEM ( T ) (cid:48) × Λ ( T ) dT erg s − (2)where Λ ( T ) is the radiative loss function and DEM ( T ) (cid:48) = DEM ( T ) × A is the differential emission measure multipliedby the emitting area A , and hence is in units of cm − K − . Herelog T = 5.5 and log T = 7.0. To find the total energy radiated,we estimate Λ ( T ) using the CHIANTI database (Landi et al.2012; Del Zanna et al. 2015) assuming coronal abundances,and integrate Equation 2 over the duration of the impactingstream, hence, L rad = (cid:90) t t dL rad ( t ) dt dt , (3)where t = 00:20 UT and t = 00:40 UT.Using the DEM, it is also possible to calculate the peakthermal energy U th . For an isothermal plasma, this is givenby (e.g. Veronig et al. 2005; Hannah et al. 2008; Emslie et al.2012; Inglis & Christe 2014; Warmuth & Mann 2016), U th = 3 k B T (cid:112) EM tot fV (4) Figure 5.
Energetic properties of the impact region shown in Figure 2. a)The radiated energy rate dL / dt of the heated plasma as a function of timeduring the impacting stream. b) The radiated energy rate dL / dt normalizedto a per-pixel basis. c) The instantaneous thermal energy of the evolvingimpact plasma. where EM tot is the total emission measure in cm − of theplasma with single temperature T , V is the plasma volumeand f is the plasma filling factor. For a multi-thermal plasma,we must account for the energy at all T . Hence, Equation 4becomes (Inglis & Christe 2014; Aschwanden et al. 2015), U th = 3 k B V / (cid:82) T DEM ( T ) (cid:48) × T dTEM / tot (5)where DEM ( T ) (cid:48) is differential emission measure expressedin cm − K − as before, and a filling factor f = 1 is assumed.The temperature bounds are the same as in Equation 2. Forboth Eqn 4 and 5 the volume V must be estimated, whichis complicated by a lack of observational information of thesource in the z-direction. In this work, we use the simpleestimate V ∼ A / , hence the estimated volume varies overtime with the area (see Figure 2).Figure 5 shows the energetic properties of the debris im-pact region, the same region illustrated by the white contoursin Figure 2. Panel a) illustrates the estimated radiated energyrate dL / dt as a function of time. The radiated losses showan order of magnitude increase, beginning at 00:20 UT. Thiscoincides with a substantial increase in the estimated area ofthe impact region (see Figure 2); on a per-pixel basis, theincrease in emission is smaller, approximately a factor ∼ ∼ ergs.We can compare these energetic properties with the pre-vious observations of impacting prominence debris from the2011 June 7 event (Gilbert et al. 2013; Reale et al. 2013). InGilbert et al. (2013), the radiated energy was estimated for 5observable brightening regions affected by impacting debris.Combining all of these regions, the estimated total radiatedenergy L rad was ∼ × ergs. For the 2011 Septem-ber 7-8 observation, we find from integrating Figure 5a thatthe total energy radiated is ∼ ergs, at least a factor of 2higher. This is consistent with the observations for two rea-sons; firstly, the 2011 September 7-8 observation consists ofa relatively continuous material stream that impacts a largerarea than the 2011 June 7 impacts, and secondly the brighten-ing duration is substantially longer, with significant emissionlasting for ∼
20 minutes.However, the total estimated radiated energy loss is an or-der of magnitude lower than estimated peak thermal energyof the plasma U th . This could be the result of two major fac-tors. Firstly, when calculating the thermal energy, a value of f ∼ f is unknown, and may be substantially less thanunity (e.g. Cargill & Klimchuk 1997; Guo et al. 2012). This,combined with uncertainties in the plasma volume V , leads toa large uncertainty in the estimate of U th , possibly an overes-timate. The second factor is that conductive losses could playan important role in the energy evolution of the bright plasma.Despite these uncertainties, we can use these energy esti-mates to constrain the minimum and maximum kinetic energy,and thus the mass, of the impacting prominence material. Wecan assume that the estimate of L rad gives us a lower limit onthe kinetic energy requirement from the infalling prominencematerial. Hence KE ≥ ergs. Given our plane-of-sky ve-locity estimate of the falling material of v ∼
150 km/s, we canestimate the minimum value of mass m required to producethis kinetic energy. From this we estimate m low ∼ × g. Alternatively, we can assume that conductive losses playa significant role in the energy budget of this event, and thatthe estimate of U th provides a good approximation of the totalenergy deposited by the prominence material. In this case, wefind that KE ∼ ergs, which requires m high ∼ × g. CONCLUSIONSWe have analysed a new example of energy release andEUV brightening due to prominence debris, occurring on2011 September 8 at ∼ ∼ ∼ U th ∼ ergs. The estimated total radiated en-ergy was an order of magnitude smaller, at L rad ∼ ergs.The disparity between these values may be due either to theimportance of conductive losses in the plasma, or the uncer-tainty in the plasma volume V and the filling factor f , both ofwhich effect the estimate of U th . Comparing these values toestimates from the well-known 2011 June 7 event (e.g. Gilbertet al. 2013; Reale et al. 2013), we see that the 2011 Septem-ber 7-8 event was more energetic overall, lasting for longerand radiating more energy.Using SDO/AIA images we also estimated the plane-of-skyvelocity of the descending stream at v ∼