Free Access
Issue
A&A
Volume 587, March 2016
Article Number A123
Number of page(s) 5
Section The Sun
DOI https://doi.org/10.1051/0004-6361/201526682
Published online 29 February 2016

© ESO, 2016

1. Introduction

The Lyman-alpha (Lyα; 1216 Å) transition (2p–1s) of hydrogen results in the strongest emission line in the solar spectrum. It is an optically thick line formed between the mid and upper chromosphere and recent studies have suggested that it is responsible for radiating a significant fraction of the non-thermal energy deposited in the chromosphere during solar flares (~10%; Nusinov & Kazachevskaya 2006; Rubio da Costa et al. 2009; Milligan et al. 2014). Lyα is also known to be a driver of changes in terrestrial ionospheric density in the D-layer (80–100 km; Tobiska et al. 2000) in conjunction with soft X-rays (SXR), although the Lyα contribution is smaller for flares that occur closer to the solar limb owing to absorption along the line of sight (Woods et al. 2006; Qian et al. 2010). Understanding variations in Lyα are therefore a major priority for both flare physics and space weather research. However, despite the importance of Lyα as a solar diagnostic, there are relatively few papers in the literature that discuss changes in Lyα emission during solar flares. (For a recent review of Lyα and other chromospheric EUV flare observations, see Milligan 2015.) This paper aims to highlight the inconsistencies in currently available Lyα datasets.

Early irradiance measurements in Lyα were routinely carried out by instruments such as the Solar Stellar Irradiance Comparison Experiment (SOLSTICE) on board the Solar Radiation and Climate Experiment (SORCE), and the Solar EUV Experiment (SEE) on board the Thermosphere Ionosphere Mesosphere Energetics and Dynamics (TIMED) satellite. These, and other, instruments often only had a duty cycle of a few percent so flares were either missed or averaged over. One notable exception was during the Halloween flares of 2003 when spectrally resolved Lyα line profiles were obtained at high cadence using SORCE/SOLSTICE (Woods et al. 2004; see Sect. 3). In 2010, NASA launched the Solar Dynamics Observatory (SDO; Pesnell et al. 2012) as part of its Living With A Star program to help understand the effects of solar EUV variability on the earth. One of the three instruments on board is the EUV Variability Experiment (EVE; Woods et al. 2012). Its prime objective is to improve upon previous measurements by recording changes in the solar EUV irradiance on flaring timescales (10 s). It achieves this through its MEGS-A (Multiple EUV Grating Spectrograph) and -B components, which cover the 60–370 Å and 370–1050 Å wavelength ranges, respectively. It also comprises a 106 Å wide, broadband photometric diode, MEGS-P, which is placed at the minus first order of the MEGS-B grating centred on the Lyα line.

Flare-related enhancements in Lyα from EVE data were first reported by Milligan et al. (2012) for the 15 February 2011 X2.2 flare, and a follow-up study claimed that Lyα emission made up 68% of the total measured radiated losses from the chromosphere for that event (Milligan et al. 2014). In contrast, Kretzschmar et al. (2013) found only a 0.6% increase in Lyα emission during an M2 flare using the LYRA radiometer (12001230 Å; Benmoussa et al. 2009) on board PROBA2. More strikingly, in both cases the temporal behaviour of Lyα appeared to mimic that of the SXR emission, rather than of the impulsive hard X-ray (HXR) emission as one might expect for intrinsic chromospheric emission. Kretzschmar et al. (2013) also found that by taking the time derivative of the Lyα profile, the resulting peaks closely resembled those of the derivative of the SXR lightcurve. The authors cited this as evidence for pre-flare heating prior to the onset of accelerated particles. While the possibility of gradual heating by thermal conduction cannot be ruled out, it is more likely that other instrumental factors might have been contributing to this anomalous behaviour. This research note aims to highlight the inconsistencies between modern Lyα flare observations, and to urge users to exercise caution when interpreting data from EVE and other broadband photometric instruments. The data from MEGS-P are the focus of this paper and are discussed in greater detail in Sect. 2. Spectrally resolved Lyα flare observations from SORCE/SOLSTICE are presented in Sect. 3 as a benchmark, while a comparison between EVE and GOES/EUVS-E Lyα measurements is given in Sect. 4. A summary and conclusions are given in Sect. 5.

thumbnail Fig. 1

Top panel: GOES SXR lightcurves during the X2.2 flare that occurred on 15 February 2011 in 1–8 Å (solid curve) and 0.5–4 Å (dotted curve). Middle panel: Lyα lightcurve taken using SDO/EVE MEGS-P for the same event. Bottom panel: LyC lightcurve from SDO/EVE MEGS-B data. The vertical dotted and dashed lines in each panel mark the start and peak times of the GOES event, respectively. From Milligan (2015).

2. SDO/EVE MEGS-P flare observations

The SDO/EVE instrument now provides routine, broadband (100 Å; Hock et al. 2012) Sun-as-a-star observations in Lyα through its MEGS-P diode at 10 s cadence, albeit with a reduced duty cycle. Until recently the MEGS-B and -P components were often only exposed to the Sun for 4 h per day because of unforeseen instrumental degradation, although earlier in the SDO mission 24–48 h MEGS-B flare campaigns were run during periods of very high solar activity. In October 2015, the EVE flight software was updated so that MEGS-B and -P will now respond autonomously to M-class flares or greater based on the SXR flux detected by the ESP 1–7 Å channel. Multiple three-hour flare campaigns are permitted each day, and a three-hour observation is made each day if a flare campaign has not been triggered. Since the loss of MEGS-A and SAM in May 2014, the decision to only expose MEGS-B and -P during periods of increased activity, essentially makes EVE a dedicated flare instrument. Although the primary motivation behind this is to be able to quantify the total EUV flux, including Lyα, incident on the Earth’s atmosphere during the largest solar flares for inclusion in atmospheric models, these data are also useful in terms of quantifying the composition and energy budget of radiative losses in the chromosphere.

However, in order to utilize the EVE MEGS-P data1 correctly, it is crucial to understand what it is depicting. Figure 1 shows the time profiles for the 15 February 2011 X-class flare (SOL2011-02-15T01:44) using SXR (top) from the X-Ray Sensor (XRS) on board the Geostationary Orbiting Environmental Satellites (GOES); Lyα (middle); and the free-bound Lyman continuum2 (LyC; bottom) with a recombination edge at 912 Å. Lyα and LyC both result from transitions to the ground state of neutral hydrogen, although from different upper levels. However, the lightcurves of Lyα appear to show a gradual, slowly-rising, GOES-like profile with a rise time of 10–20 min, like that seen in SXR; a similar behaviour was found by Kretzschmar et al. (2013) using LYRA data. The LyC profile, on the other hand, appears impulsive and bursty with a rise time of just a few minutes, and peaks during the rise in SXR. The latter case is what one might expect for a chromospheric plasma heated via Coulomb collisions (e.g. Brown 1971). So why do these two profiles behave so differently given that they are formed in the same layer of the Sun’s atmosphere? Is the peculiar behaviour of Lyα merely due to the broad response function of MEGS-P, or is it indicative of how Lyα responds to heating during large explosive events? Possible instrumental issues may include contamination of the passband by lines and/or continua from other ion species, or the core and the wings of the Lyα line itself – which are formed at distinctly different depths in the solar atmosphere – responding to different heating mechanisms, or on different timescales. In these cases the broadband nature of the observations would therefore appear to smooth out the time profiles. If it is a genuine solar effect, then maybe the excitation/decay timescales for the 2p-1s transition of neutral hydrogen are longer than for other transitions or elements; or perhaps the opacity of the line is causing photons from the core to be scattered into the wings; or the Lyα emission is actually peaking as the plasma cools, rather than as it is being heated.

thumbnail Fig. 2

a) Lyα (blue) and SXR (black) emission and their derivatives (cyan and grey, respectively) during the 15 February 2011 flare. b) LyC lightcurve for the same event (black). Overplotted in cyan is the time derivative of the Lyα lightcurve normalized to LyC. c) Lightcurve of Lyβ (1026 Å; black curve). Overplotted in cyan is the time derivative of the Lyα lightcurve normalized to Lyβ. d) Lightcurve of C III (977 Å; black curve). Overplotted in cyan is the time derivative of the Lyα lightcurve normalized to C III. The vertical dotted and dashed lines in each panel mark the start and peak times of the GOES event, respectively.

Figure 1 shows that there is a distinct difference between the temporal behaviour of Lyα and LyC. However, by taking the time derivative of the Lyα profile and normalizing it to LyC there appears to be a remarkable agreement, at least on the rise phase of the 15 February 2011 flare as shown in Fig. 2b. Both the LyC time profiles and the derivative of Lyα now peak in concert with the derivative of the SXR emission, which is often taken as a proxy for the hard X-ray (HXR) emission under the assumption of the Neupert effect (Neupert 1968). This is also in agreement with Kretzschmar et al. (2013). This empirical relationship assumes that the energetic electrons that generate HXR emission through the thick-target bremsstrahlung process are responsible for the heating and mass supply (through chromospheric evaporation) of the SXR emitted by the hot coronal plasma. In other words, SXR are an indirect consequence of the initial chromospheric heating. The fact that the Lyα profiles exhibit a similar behaviour might suggest that the 2p–1s transition of H I is not directly affected by collisional excitation, but perhaps by an alternative mechanism such as thermal conduction.

The correlation between d(Lyα)/dt and LyC is not unique. In Figs. 2c and d, the time derivative of Lyα is also normalized to Lyβ (1026 Å) and C III (977 Å), respectively; when two arbitrarily chosen, intrinsically chromospheric emission lines from MEGS-B data are used, the same agreement still holds. In some other flares, however, the derivative of Lyα precedes the LyC (or Lyβ or C III) emission by 1–2 min. The reason for this is unclear.

thumbnail Fig. 3

Top panel: Lyα line profile observed by SORCE/SOLSTICE before (dashed curve) and during (solid curve) the 28 October 2003 flare. Also visible is the nearby Si III line. The vertical coloured lines denote the wavelengths of the lightcurves in the bottom panel. Bottom panel: lightcurves at different wavelengths of the Lyα line during the 28 October 2003 flare. The time profiles of the neighbouring Si III line is also shown as a grey histogram. The vertical dashed and solid lines denote the approximate times of the pre-flare and flare spectra in the top panel, respectively. The GOES SXR lightcurve (black) is also shown for context.

3. SORCE/SOLSTICE flare observations

The SOLSTICE instrument (McClintock et al. 2005a,b) on board the SORCE satellite typically takes spectral irradiance measurements across the EUV range once per orbit. However, during the X17 flare that occurred on 28 October 2003 (SOL2003-10-28T11:10) it was fortuitously scanning through the Lyα line in steps of 0.35 Å at ~1 min cadence (~1 s exposure per step). Woods et al. (2004) only reported a 20% increase in the core of the line, but the wings of the line were found to increase by a factor of two. This is significantly higher than the 6% increase reported by Brekke et al. (1996) during the end of the impulsive phase of the X3 flare on 27 February 1992 using the SOLSTICE instrument on the Upper Atmosphere Research Satellite. The line profiles from before and during the 28 October 2003 flare are shown in the top panel of Fig. 3. The coloured lightcurves in the bottom panel are from the wavelength bins denoted by the same vertical coloured lines in the top panel.

In agreement with Woods et al. (2004), the time profiles for each wavelength bin peaked during the rise in SXR, confirming an impulsively heated atmosphere and the blue wing enhancement was found to be greater than that of the red wing, although this may be due to the presence of a blended emission line around 1211 Å (possibly S X; log T = 6.2). The larger enhancements in both wings relative to the core are most likely due to opacity effects as photons generated in the core of the line are redistributed (see Woods et al. 1995). Generating a separate lightcurve for the nearby Si III line at 1206.5 Å (log T = 4.7) shows an impulsive behaviour peaking in the rise phase of the SXR (grey histogram; bottom panel of Fig. 3). Therefore, all emission within the ~22 Å wide wavelength range observed by SOLSTICE is clearly indicative of a rapidly responding chromosphere to a sudden deposition of energy. This implies that neither opacity effects nor the nearby Si III line are responsible for the peculiar behaviour of the MEGS-P data.

4. GOES/EUVS-E flare observations

The three most recent GOES Satellites, GOES-13, -14, and, -15, each have an onboard EUV Sensor3 (EUVS; Viereck et al. 2007; Evans et al. 2010) in addition to the more familiar X-Ray Sensor. GOES-13 has been taking sporadic measurements in the EUV since mid-2006, while GOES-14 data are available from mid-2009 to mid-2010, and again for a few months in late 2012. GOES-15 has been much more consistent, observing the Sun continuously since early 2010. The A and B channels on each EUVS instrument are centred around 100 Å and 304 Å (the resonance line of He II), respectively, while the E channel has a width of ~100 Å (1180–1270 Å) spanning the Lyα line in a broadband manner similar to MEGS-P. The calibrated GOES/EUVS Version 3 data are available at 1 min cadence, while the raw data are available every 10 s.

Fortunately, GOES-15 also observed the 15 February 2011 flare. The normalized lightcurves from both GOES/EUVS-E (red and green curves) and EVE MEGS-P (blue curve) are plotted in Fig. 4 along with the SXR lightcurve (black) for context. There are obvious discrepancies between the changes in relative intensity quoted by each instrument, which can most likely be attributed to calibration issues. However, the differences in temporal behaviour are much more intriguing. The MEGS-P curve shows a distinct pre-flare increase beginning at ~01:40 UT, and continues to increase over the subsequent ~20 min, peaking close to – or possibly after – the GOES SXR peak. The EUVS-E emission, on the other hand, increases abruptly around 01:48 UT and peaks around 5 min later, close to the peak of the GOES SXR derivative (01:52:36 UT; vertical dashed line). The EUVS-E emission then begins to decay, while the MEGS-P emission continues to rise. Fluxes from both instruments return to their pre-flare values around an hour later. Therefore, although both GOES/EUVS-E and SDO/EVE MEGS-P are full-Sun, broadband, high-cadence instruments centred on the Lyα line, they exhibit distinctly different time profiles for the same event.

thumbnail Fig. 4

Comparison between normalized irradiance as observed by SDO/EVE MEGS-P (blue curve) and GOES/EUVS-E at 10 s (red curve) and 60 s (green curve) cadence during the 15 February 2011 flare. The GOES SXR light curve is also plotted in black for context. The vertical dashed line denotes the time of the peak of the GOES SXR derivative.

5. Summary and conclusions

In the 5 years since its launch, SDO/EVE has detected increased Lyα emission in dozens of flares over a range of magnitudes at 10 s cadence with its MEGS-P broadband diode. In all cases, the Lyα time profile exhibits a slowly varying, gradual behaviour similar to that seen by thermal plasma emitting in SXR. By contrast, LyC and other intrinsically chromospheric emission appears much more impulsive, mimicking that of the HXR emission, as would be expected from an electron-beam-heated atmosphere. In comparison, spectrally and temporally resolved Lyα emission from SOLSTICE appears impulsive during at least one previously observed event. The Lyα emission detected by GOES/EUVS-E also appears to peak during the impulsive phase of the 15 February 2011 flare and is distinctly different from that displayed by EVE which peaks some 5–10 min later. This suggests that the broadband nature of the EVE (and LYRA) measurements are in some way smoothing out this intrinsic, bursty nature. Perhaps other, higher temperature lines or continua are contributing to the passband, or a data processing algorithm in the EVE pipeline is responsible. While these subtle differences may not significantly affect research into effects on changes in the solar irradiance on the upper terrestrial atmosphere, they could dramatically affect how we interpret such changes in Lyα in the context of solar flares themselves.

One way to resolve this issue would be to convolve synthetic spectra from the RADYN radiative hydrodynamic code (Allred et al. 2005, 2015) with the broadband response functions to establish what other lines and continua might be contributing to the passbands during large events. Spectrally resolved Lyα flare observations from other instruments are not suited to this task for a number of reasons: SORCE/SOLSTICE data only span 22 Å while the MEGS-P passband is 100 Å wide; SOHO/SUMER never observed solar flares due to the sensitivity of its optics; and Skylab data were often saturated during large events. The spectral response of the MEGS-P diode could also be mapped by calibrating the EVE rocket instrument using a monochrometer with a deuterium lamp. Once this issue is resolved, simultaneous Lyα and LyC observations from EVE (along with Lyβ, Lyγ, etc., and other chromospheric diagnostics) will be a valuable tool for investigating the heating of the lower solar atmosphere. Understanding the broadband nature of Lyα observations from EVE will also help prepare for the influx of flare data from future instruments. For example, the Mars Atmosphere Variation and Evolution (MAVEN; Eparvier et al. 2015) features a Lyα radiometer (1210–1220 Å) for measuring the effects of changes in solar EUV irradiance on the martian atmosphere. The Solar Orbiter mission – due for launch in 2018 – also includes a Lyα channel as part of its Extreme Ultraviolet Imager instrument.


1

Version 5 of EVE data were used in the preparation of this paper.

2

The Lyman continuum was fit over the 800–912 Å range in EVE MEGS-B data using the techniques described in Milligan et al. (2014).

Acknowledgments

The authors would like to thank the anonymous referee for the constructive comments that greatly enhanced this paper. They also thank Marty Snow (LASP) for access to the high-cadence SOLSTICE data, Janet Machol (NOAA) for help with the GOES EUVS-E data, and Mihalis Mathioudakis (QUB) for many stimulating discussions on this issue. This work was supported by NASA LWS/TR&T grant NNX11AQ53G and LWS/SDO Data Analysis grant NNX14AE07G.

References

  1. Allred, J. C., Hawley, S. L., Abbett, W. P., & Carlsson, M. 2005, ApJ, 630, 573 [NASA ADS] [CrossRef] [Google Scholar]
  2. Allred, J. C., Kowalski, A. F., & Carlsson, M. 2015, ApJ, 809, 104 [NASA ADS] [CrossRef] [Google Scholar]
  3. Benmoussa, A., Dammasch, I. E., Hochedez, J.-F., et al. 2009, A&A, 508, 1085 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  4. Brekke, P., Rottman, G. J., Fontenla, J., & Judge, P. G. 1996, ApJ, 468, 418 [NASA ADS] [CrossRef] [Google Scholar]
  5. Brown, J. C. 1971, Sol. Phys., 18, 489 [NASA ADS] [CrossRef] [Google Scholar]
  6. Eparvier, F. G., Chamberlin, P. C., Woods, T. N., & Thiemann, E. M. B. 2015, Space Sci. Rev., 195, 293 [NASA ADS] [CrossRef] [Google Scholar]
  7. Evans, J. S., Strickland, D. J., Woo, W. K., et al. 2010, Sol. Phys., 262, 71 [Google Scholar]
  8. Hock, R. A., Chamberlin, P. C., Woods, T. N., et al. 2012, Sol. Phys., 275, 145 [Google Scholar]
  9. Kretzschmar, M., Dominique, M., & Dammasch, I. E. 2013, Sol. Phys., 286, 221 [NASA ADS] [CrossRef] [Google Scholar]
  10. McClintock, W. E., Rottman, G. J., & Woods, T. N. 2005a, Sol. Phys., 230, 225 [NASA ADS] [CrossRef] [Google Scholar]
  11. McClintock, W. E., Snow, M., & Woods, T. N. 2005b, Sol. Phys., 230, 259 [NASA ADS] [CrossRef] [Google Scholar]
  12. Milligan, R. O. 2015, Sol. Phys., 290, 3399 [NASA ADS] [CrossRef] [Google Scholar]
  13. Milligan, R. O., Chamberlin, P. C., Hudson, H. S., et al. 2012, ApJ, 748, L14 [NASA ADS] [CrossRef] [Google Scholar]
  14. Milligan, R. O., Kerr, G. S., Dennis, B. R., et al. 2014, ApJ, 793, 70 [NASA ADS] [CrossRef] [Google Scholar]
  15. Neupert, W. M. 1968, ApJ, 153, L59 [NASA ADS] [CrossRef] [Google Scholar]
  16. Nusinov, A. A., & Kazachevskaya, T. V. 2006, Sol. Syst. Res., 40, 111 [Google Scholar]
  17. Pesnell, W. D., Thompson, B. J., & Chamberlin, P. C. 2012, Sol. Phys., 275, 3 [NASA ADS] [CrossRef] [Google Scholar]
  18. Qian, L., Burns, A. G., Chamberlin, P. C., & Solomon, S. C. 2010, J. Geophys. Res., 115, 9311 [CrossRef] [Google Scholar]
  19. Rubio da Costa, F., Fletcher, L., Labrosse, N., & Zuccarello, F. 2009, A&A, 507, 1005 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  20. Tobiska, W. K., Woods, T., Eparvier, F., et al. 2000, J. Atmospheric and Solar-Terrestrial Physics, 62, 1233 [Google Scholar]
  21. Viereck, R., Hanser, F., Wise, J., et al. 2007, in Solar Physics and Space Weather Instrumentation II, Proc. SPIE, ed. Fineschi, 6689 [Google Scholar]
  22. Woods, T. N., Rottman, G. J., White, O. R., Fontenla, J., & Avrett, E. H. 1995, ApJ, 442, 898 [NASA ADS] [CrossRef] [Google Scholar]
  23. Woods, T. N., Eparvier, F. G., Fontenla, J., et al. 2004, Geophys. Res. Lett., 31, 10802 [Google Scholar]
  24. Woods, T. N., Kopp, G., & Chamberlin, P. C. 2006, J. Geophys. Res., 111, 10 [CrossRef] [PubMed] [Google Scholar]
  25. Woods, T. N., Eparvier, F. G., Hock, R., et al. 2012, Sol. Phys., 275, 115 [Google Scholar]

All Figures

thumbnail Fig. 1

Top panel: GOES SXR lightcurves during the X2.2 flare that occurred on 15 February 2011 in 1–8 Å (solid curve) and 0.5–4 Å (dotted curve). Middle panel: Lyα lightcurve taken using SDO/EVE MEGS-P for the same event. Bottom panel: LyC lightcurve from SDO/EVE MEGS-B data. The vertical dotted and dashed lines in each panel mark the start and peak times of the GOES event, respectively. From Milligan (2015).

In the text
thumbnail Fig. 2

a) Lyα (blue) and SXR (black) emission and their derivatives (cyan and grey, respectively) during the 15 February 2011 flare. b) LyC lightcurve for the same event (black). Overplotted in cyan is the time derivative of the Lyα lightcurve normalized to LyC. c) Lightcurve of Lyβ (1026 Å; black curve). Overplotted in cyan is the time derivative of the Lyα lightcurve normalized to Lyβ. d) Lightcurve of C III (977 Å; black curve). Overplotted in cyan is the time derivative of the Lyα lightcurve normalized to C III. The vertical dotted and dashed lines in each panel mark the start and peak times of the GOES event, respectively.

In the text
thumbnail Fig. 3

Top panel: Lyα line profile observed by SORCE/SOLSTICE before (dashed curve) and during (solid curve) the 28 October 2003 flare. Also visible is the nearby Si III line. The vertical coloured lines denote the wavelengths of the lightcurves in the bottom panel. Bottom panel: lightcurves at different wavelengths of the Lyα line during the 28 October 2003 flare. The time profiles of the neighbouring Si III line is also shown as a grey histogram. The vertical dashed and solid lines denote the approximate times of the pre-flare and flare spectra in the top panel, respectively. The GOES SXR lightcurve (black) is also shown for context.

In the text
thumbnail Fig. 4

Comparison between normalized irradiance as observed by SDO/EVE MEGS-P (blue curve) and GOES/EUVS-E at 10 s (red curve) and 60 s (green curve) cadence during the 15 February 2011 flare. The GOES SXR light curve is also plotted in black for context. The vertical dashed line denotes the time of the peak of the GOES SXR derivative.

In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.