Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 The Sun's Bastille Day flare on July 14, 2000 produced a variety of geoeffective events. This solar eruption consisted of an X-class flare followed by a coronal mass ejection that produced a major geomagnetic storm. We have undertaken a study of this event beginning with an analysis of the effects of the radiative phase of the flare on the dayglow and the ionosphere. The key new enabling work is a novel method of evaluating the X-ray and extreme ultraviolet (EUV) solar spectral irradiance changes associated with the flare. We find that the solar radiative output enhancements modeled during the flare are consistent with measurements of both solar EUV irradiance and far UV Earth thermospheric dayglow. We use the SAMI2 model to predict global ionospheric changes along a magnetic meridian that show significantly different northern and southern effects, suggesting that flares can be used to study ionospheric dynamics.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Solar flares, which are large eruptions of radiation and particles on the Sun, have long fascinated the space scientist as well as the general public. In his book, Ionospheric Effects of Solar Flares, Mitra  pointed out that the short radiative impulse of a solar flare (typically less than an hour) offers an ideal way to test our understanding of photoionization, plasma dynamics, and recombination in the ionosphere because “the bulk of ionospheric measurements are too slow to allow any insight into the nature of these ionospheric reactions”. A superb opportunity to do this occurred with the Bastille Day (July 14, 2000) flare. The radiative phase of this X-class flare began at about 1000 Universal Time (UT), followed by a coronal mass ejection (CME) and the arrival 40 min later of relativistic particles at Earth. Some 38 hours after the flare, a strong geomagnetic storm struck Earth, sending the 3-hour ap index to its maximum value of 400. Figure 1 shows the solar X-ray and EUV behavior during the event.
 Much of our knowledge of the magnitude and spectral content of the solar EUV irradiance comes from a 20-year old database [Hinteregger et al., 1981] of limited duration (∼4 years), of inadequate cadence, and with serious reliability questions. For example it is widely believed that the solar irradiance below 20 nm should be increased by at least a factor of 2 to account for the production of photoelectrons, E-region electron densities, dayglow levels, and direct solar measurements (see the references cited by Solomon et al. ). As well, total EUV energy between 5 and 105 nm differs among irradiance models by a factor of two. Compounding these issues is a serious lack of understanding of the variability of the solar spectral irradiance on the shortest time scales, i. e., during solar flares for which observations are essentially absent. With such fundamental uncertainties in a major driver of space weather, any occasion to probe solar radiative forcing and atmospheric response is welcome.
 In this paper, we report initial findings of the ionospheric effect of the radiative (only) phase of the solar eruption. A cornerstone of the effort is new theoretical knowledge of the changing solar EUV and X-ray spectral irradiance. Unfortunately, the only full disk solar observations with sufficiently high cadence to resolve flares are the Geostationary Environmental Operational Satellite (GOES) measurements of X-rays in two bands (0.1–0.8 and 0.05–0.4 nm) and the Solar EUV Monitor (SEM) observations of two EUV bands (0.1–50 nm and 26–34 nm) from the Solar Heliospheric Observatory (SOHO). Neither of these instruments has sufficient spectral resolution for use in models of the ionospheric responses. Fortunately, these observations can be combined with other observations of the flare and a new approach for modeling solar EUV irradiance [Warren et al., 1998, 2001a] to estimate the flare EUV spectral irradiance enhancement. We then predict the flare-induced changes in the FUV dayglow, and compare these predictions with observations from the Advanced Research and Global Observation (ARGOS) satellite. We also provide estimates of the changes in the heating and photoionization rates due to the flare. Finally, we predict the ionospheric response to the flare radiation using the SAMI2 model [Huba et al., 2000]. A quantitative analysis of the event, including the effects of energetic particles and the ensuing geomagnetic storm on the atmosphere and ionosphere will be treated in a future investigation.
2. Solar Flare EUV Spectral Irradiance
Warren et al. [2001a] have developed a qualitatively new approach for modeling solar EUV irradiance variability. In their model (called NRLEUV), irradiances of optically thin emission lines are calculated from emission measure distributions, adopted solar abundances, a comprehensive data base of atomic physics parameters, a simple model of limb-brightening, and areas of solar features derived from full-disk solar images. The emission measure is the integral of the square of the electron density in the solar atmosphere along the line of sight. The differential emission measure is a function of temperature [Warren et al., 2001a] and can be determined from a relatively small number of emission lines. Once the emission measure has been determined it can be used to calculate the intensity of any optically thin emission line, regardless of wavelength. For the NRLEUV model, quiet Sun, coronal hole, and active region emission measures were constructed from spatially and spectrally resolved solar observations. Unlike the approach taken with empirical models, this method does not use the Hinteregger et al.  EUV irradiance measurements, nor does it rely on any direct EUV irradiance observations to determine variability.
 Because the temperature distribution observed in a flare is significantly different than that of the three solar regions currently included in the NRLEUV model, it was necessary to derive a new emission measure. Because the spectrally resolved data available for this flare are rather limited, we make a number of approximations in order to derive the flare emission measure. We use the Dere and Cook  emission measure derived from Skylab Spectroheliograph observations as the basis for the flare emission measure at temperatures above T = 3.16 × 106 K. In order to match the soft X-ray fluxes observed in both GOES channels we not only need to increase the magnitude of the emission measure at all temperatures but we also add a component at temperatures greater than T = 1.58 × 107 K. The emission measure at lower temperatures is determined from Transition Region and Coronal Explorer (TRACE; Handy et al. ) observations of the flare. From the TRACE 17.1 nm and 160 nm images we calculate the contrast between the pre-flare active region and flare emission, that we then use to scale the NRLEUV model active region emission measure, as well as to estimate the area of the flare. The TRACE 17.1 nm images determine the flare emission measure at 106 K, while the 160 nm images determine the emission measure at 105 and 104 K. We assume that 90% of the 160 nm emission during the flare arises from C IV. This is consistent with spectrally resolved irradiance observations of an X3 flare by Brekke et al. . To ensure continuity of the emission measure distribution, we spline the lower temperature points to the higher temperature part of the distribution derived from the Skylab data and the GOES fluxes.
 For this work we have computed irradiances for the period before the flare and at the peak of the flare. Since the present version of the NRLEUV model does not extend below 5 nm, to model the pre-flare solar irradiance at these wavelengths we have added a hot active region component derived from pre-flare GOES fluxes to the three components already included in the model. The component derived from the GOES fluxes accounts for the highest temperature active region emission, which has a negligible contribution to the solar irradiance above 5 nm but is an important contributor at the shortest wavelengths. We are currently planning to compute time-dependent irradiance spectra for this flare and these results will be presented in a future paper [Warren et al., 2001b].
 Plotted in Figure 2 are model estimates of the pre-flare and peak flare (at 1024 UT) solar spectra (upper panel) and their ratio (lower panel). The irradiances were calculated at 0.1 nm intervals, but the plots are displayed in 1 nm histograms to illustrate the primary differences more clearly. Note that near 30.4 nm (the He I line), the 1 nm irradiance value increases by a factor of about 2.5. However, when the bin size is adjusted to cover the SEM range from 26–34 nm, the peak-flare to pre-flare ratio is only 1.5, in good agreement with the SEM measurement of 1.4. This underscores the need to have an accurate solar flare irradiance model and measurements at higher spectral as well as temporal resolution to evaluate aeronomical consequences. Simply scaling the quiet or active solar EUV spectral irradiance is inappropriate because the enhanced temperatures produce quite different emission features.
3. Flare Effects on Atmospheric Energy Inputs and Dayglow
 We used the Atmospheric Ultraviolet Radiance Integrated Code (AURIC; Strickland et al. ) to calculate atmospheric energy inputs and dayglow emissions prior to and at the peak of the flare. The solar EUV irradiance spectrum is directly input to AURIC which calculates energy deposition rates, photoelectron fluxes, excitation rates, and column emission rates of a variety of atmospheric species. The default neutral atmospheric specification is MSISE-90 (abbreviated herein as (MSIS) [Hedin, 1991] and the default irradiance range is 1–105 nm.
 The example of AURIC output shown in Figure 3 typifies the atmospheric region observed from the ARGOS satellite just after the peak of the flare (about 10.6 UT). Inputs to AURIC include the solar 10.7 cm radio flux on the previous day (239), the 81-day average (192), and the daily Ap (51) needed to evaluate MSIS. The left panel in Figure 3 shows the (approximate) heating and photoionization rates with and without the flare, using the solar irradiances from Figure 2 (in 0.1 nm bins). The heating rate was estimated by subtracting from the total EUV energy deposition rate the energy used for ionization, dissociation, and radiation. Above 130 km, the heating rate at the peak of the flare is nearly 1.5 times larger than the pre-flare value. The ratio of peak- to pre-flare heating rates increases with decreasing altitude, reaching a factor of 3.5 at 90 km. The ratio would be even larger, were AURIC to include X-rays below 1 nm. The photoionization rate exhibits a qualitatively similar change during the flare, as do the excitation rates of N2 (a1Πg) and O(3d 3D°), (the parent state of the 98.9 nm emission) shown in the right panel of Figure 3. Both of these excited states are produced in the dayglow by photoelectron impact excitation. The excitation rates in Figure 3 are used to obtain column emission rates for comparison with ARGOS observations (see Meier  for discussion of the properties of these features).
 The LORAAS (Low Resolution Airglow/Auroral Spectrograph) instrument on board the ARGOS satellite, orbiting at 840 km and 0230/1430 local time, observed the FUV dayglow (80–170 nm) on the limb during the flare (Dymond et al.  describe the mission and observations). Because of extinction by O2 in the FUV, the limb-viewing (100–700 km tangent altitude) LORAAS can only observe emissions originating from above about 140 km. At the peak of the flare, the ARGOS satellite had just entered sunlight. We compared LORAAS observations with AURIC calculations between 83° and 40° solar zenith angle (10.45 and 10.65 UT). When viewing above 200 km the LORAAS instrument observed column emission rate enhancements of 50–70% above previous orbits, in both OI 98.9 nm and the (4, 2) and (1, 0) N2 LBH bands (near 141.5 nm). For comparison, the AURIC model predicted an increase of 55%. Another spectral feature with N2 parentage is NII 108.5 nm, produced by photoionization-dissociation excitation. The AURIC model predicts an increase of 40% for 108.5 nm, while the LORAAS measurements were similar to the LBH band observations. When viewing between 150 and 175 km, LORAAS observed increases of some 70–100% for the nitrogen emissions and 50–70% for 98.9 nm; AURIC calculations gave similar enhancements. Thus, our analysis demonstrates that the observed EUV dayglow increases are generally consistent with the AURIC estimates using the NRLEUV solar spectral irradiance model of the flare.
4. F-Region Ionospheric Flare Responses
 We are now in a position to examine the response of the ionosphere to the flare. We used the SAMI2 model [Huba et al., 2000] to examine the change in ionospheric conditions caused by the flare. SAMI2 treats the dynamic plasma and chemical evolution of seven ion species (H+, He+, N+, O+, N2+, NO+, and O2+) in the altitude range 85 km to several thousand km. The ion continuity and momentum equations are solved for all 7 species; the temperature equation is solved for H+, He+, O+, and electrons. SAMI2 models the plasma along the Earth's dipole field from hemisphere to hemisphere, includes the E × B drift of a flux tube, and ion inertia in the ion momentum equation for motion along the dipole field line. The neutral atmosphere is specified by MSIS and the HWM-90 empirical wind model [Hedin et al., 1991b]. Although the SAMI2 lower boundary is currently set at 85 km, the model does not as yet have a D-region description suitable to study the X-ray energy deposition.
 We made two runs for the day of the flare: one that did not include the flare irradiance (i.e., a benchmark run) and one that included the flare irradiance. The latter case assumed that the flare began at 10:00 UT, linearly increased to the peak around 10:25, and linearly decreased to the pre-flare value at 10:45. We assumed that the solar X-ray irradiance tracked the EUV irradiance; this assumption, while not exact, is adequate for our initial estimates (Figure 1, except for the particle contamination in SEM data).
Figure 4 shows the relative change in the electron density, ne, on the Greenwich meridian. The change is defined by the ratio: (ne [flare peak] − ne [no flare])/ne [no flare]. The maximum F-region enhancement of about 40% takes place in the northern hemisphere in the altitude range 200–300 km. At the time of the flare, neutral winds unrelated to the solar irradiance increase, cause an upward plasma flow in the northern hemisphere and downward in the southern hemisphere. This causes the north/south asymmetry in the electron density enhancement. The lesser degree of enhancement in the south is due to the downward flow to altitudes of enhanced molecular chemistry, which reduces the electron density. The peak increase in the electron density is followed by a decay of order hours.
 Although ionospheric observations are made routinely from the ground with ionosondes and other sensors, their cadence is typically 15 to 60 min, and therefore too poor to capture details of the flare effect. So far, we have been unable to locate other electron density observations during the flare for comparison with SAMI2.
 We have carried out a study of the responses of the F-region ionosphere to flare radiation. The key advances that have permitted this study of short time scale space weather phenomenon include the ability to estimate the solar spectral irradiance from X-rays through the EUV for input to physics-based ionosphere and dayglow models and concurrent solar and dayglow observations for validation. We have concentrated our study on the ionosphere not only because it responds quickly to a large impulse of radiation but also because radiative forcing during a flare has been largely uncertain and hence ionospheric responses are largely unspecified. On the other hand, the neutral thermospheric response is expected to be much more sluggish and muted.
 While the solar X-radiation increased by more than a factor of 200 during the Bastille Day flare, the predicted EUV increase was a more modest 50%. The modeled EUV enhancement between 26 and 34 nm is in good agreement with the SEM measurements. To put these results in context, the EUV spectral irradiance increase during this flare appears to be comparable to its increase from solar cycle minimum to maximum, according to the NRLEUV model [Warren et al., 2001a]. The increases in the photoelectron-excited FUV dayglow predicted by the AURIC model are consistent with those measured by the LORAAS instrument on board the ARGOS satellite, and are similar to those reported by Opal . This supports the approach used in the NRLEUV model to determine the solar EUV spectral irradiance. Complete validation of the model with error analysis awaits measurements of the spectral irradiance during a flare.
 The SAMI2 ionospheric model predicts an increase of up to 40% in the F-region electron density in response to ∼50% increase in EUV radiation. The response depends on geographic location because dynamical effects produce upward or downward motions of the plasma that influence the magnitude of the electron density change as well as its time dependence. Because E × B drifts and winds control the electron density changes, different longitude-local time combinations are likely to produce different morphological effects. So far, we have not found ionospheric observations for comparison with SAMI2 that have been made with sufficiently high cadence or made at mid- and low-latitude locations in the sunlit hemisphere. We expect that important insights into ionospheric dynamics will be forthcoming if sufficient numbers of measurements, especially on a global scale, were available. Such measurements are clearly crucial to understanding the ionospheric impact of radiative forcing by flares.
 This work was supported in part by the Office of Naval Research.