Evaluation of a method to derive ionospheric conductivities using two auroral emissions (428 and 630 nm) measured with a photometer at Tromsø (69.6°N)
- 673 Downloads
KeywordsAuroral emission Conductivity Polar ionosphere Photometer EISCAT Tromsø
auroral ionospheric remote sensor
Common Program One
European Incoherent Scatter
International Geomagnetic Reference Field
International Reference Ionosphere
National Institute of Polar Research
The ionized component in the polar upper atmosphere (i.e., ionosphere) couples with both the magnetosphere and the thermosphere/mesosphere. One of the manifestations of the polar upper atmosphere is the three-dimensional (3D) current system: The magnetospheric currents, which occasionally flow along the earth’s magnetic field lines into (or out of) the ionosphere, are connected to ionospheric currents in the ionospheric conducting layer (i.e., E region); consequently, the thermosphere is heated by Joule heating. Furthermore, thermospheric wind is affected by ion motion through collisions between ions and neutrals (i.e., ion drag). Modeling of the physical interaction between the magnetosphere, ionosphere, and thermosphere therefore requires reliable realistic spatial and temporal distributions of the electric field, the energy of particle precipitation, and ionospheric conductivities. In particular, it is of vital importance to obtain the electrical properties of the ionosphere characterized by the Hall and Pedersen conductivities, because they affect the 3D current system, Joule heating, and so on.
Spatial and temporal distributions of the ionospheric electric field in the polar ionosphere can be obtained by using coherent radar systems (SuperDARN) (e.g., Greenwald et al. 1995). SuperDARN radars can provide the ionospheric electric field with temporal and spatial resolutions of about 2 min and about 45 km (down to 15 km) over the polar ionosphere, respectively. On the other hand, it is generally difficult to obtain observationally spatial and temporal distributions of the ionospheric conductivities. The European Incoherent Scatter (EISCAT) UHF radar (Folkestad et al. 1983; Rishbeth and van Eyken 1993) can make height-resolved estimates of the ionospheric (Hall and Pedersen) conductivities and the ionospheric currents that were first derived from incoherent scatter (IS) radar measurements by the Chatanika IS radar (Brekke et al. 1974). Such studies have been carried through by measurements obtained by the EISCAT UHF radar (e.g., Brekke and Hall 1988; Fontaine and Peymirat 1996; Sugino et al. 2002a, b). However, IS radar measurements are usually limited in space and time. Imagers onboard satellites can provide ionospheric physical parameters with a good spatial coverage. A suitable multi-wavelength observation of auroral emissions allows one to derive the energy distribution of the auroral particles (see a review by Robinson and Vondrak 1994). The global auroral conductance distribution has been estimated with auroral imagers onboard satellites (e.g., Coumans et al. 2004). The spatial resolution of the auroral imagers on satellites is typically about a hundred kilometers and is insufficient for determining the distribution of the ionospheric currents, because the spatial scale of a large-scale field-aligned current (FAC) covers a few hundred kilometers or less. Measurements with several tens of kilometers of spatial resolution of the ionospheric conductivity are needed to determine the FAC distribution from the spatial gradient of the ionospheric currents.
On the other hand, if ground-based optical observations are made at multiple stations, they are capable of providing spatial and temporal distributions of auroras with a high spatial and temporal resolution, although successful observations require a dark clear sky. Nocturnal ionospheric conductance can be inferred from optical data sets. There are several combinations of auroral wavelengths to infer characteristic energies and energy fluxes of the precipitating auroral electrons (e.g., Lanchester et al. 2009). Among them, the emission intensities at wavelengths of 427.8 nm (N2 + first negative band) and 630.0 nm [metastable atomic state O(1D)] can be used to infer the characteristic energies and energy fluxes of them (Niciejewski et al. 1989; Rees and Luckey 1974). The emission intensity at 427.8 nm is related directly to the energy flux, and the intensity ratio of 427.8 to 630.0 nm is related to the characteristic energy (e.g., Vondrak and Sears 1978).
Some comparative studies of the ionospheric conductance derived with an optical instrument and an IS radar have been conducted. Mende et al. (1984) compared conductances (i.e., height-integrated conductivities) derived from auroral spectroscopic measurements with those from the Chatanika IS radar (64.9° geomagnetic north). Although, in general, fairly good agreement was found between values derived by these different methods, disagreements between them indicate limitations of each experimental technique. Mende et al. (1984) summarized that the IS radar yielded more accurate values of the conductance with a limited temporal and spatial resolution, while the optical measurements provided less accurate values with higher temporal resolution and wider spatial region. Robinson et al. (1989) confirmed qualitatively the empirical relationship between the auroral luminosities observed with an auroral imager onboard the Dynamics Explorer 1 (DE 1) satellite and the Hall and Pedersen conductance derived from the Chatanika IS radar. Vondrak et al. (1985) used simultaneous and common-volume observations from the auroral scanning photometer onboard ISIS 2 and from the Chatanika IS radar and showed that ionospheric electron density profiles were determined from the emissions of N2 + and neutral atomic oxygen at 391.4 and 630.0 nm. Robinson et al. (1992) compared electron density measurements from the Søndre Strømfjord (67°N, 51°W) IS radar with electron densities inferred from OI (135.6 nm) and N2 + (391.4 nm) emissions observed by the auroral ionospheric remote sensor (AIRS) onboard the Polar Bear satellite and found a good agreement between the measured and inferred profiles. These studies show that the multi-wavelength optical method has an ability to provide ionospheric conductivities. However, the methods have not yet been well evaluated qualitatively in detail.
To evaluate a multi-wavelength optical method for derivation of ionospheric conductivities in more detail, simultaneous observations are needed that last for several hours or longer with optical instruments and an IS radar. Lanchester et al. (2009) have estimated average energy and flux of precipitation electrons using simultaneous images of aurora and EISCAT radar data obtained at Tromsø (geographic coordinates of 69.6°N, 19.2°E) and found a relatively good agreement between electron density modeled by the N2 1P band emissions (670 nm) and density profiles measured directly with the EISCAT UHF radar with a time resolution of 0.44 s. Peak electron densities derived from the two methods, however, were sometimes different, and density profiles observed with the EISCAT UHF radar showed large fluctuations in a height range above 120 km (see Fig. 7 in Lanchester et al. 2009). The differences and fluctuations would cause uncertainty of the Hall and Pedersen conductivities. Ion recombination time constant would also cause an uncertainty when the average energy and flux of precipitation electrons are calculated from electron density profiles. The time constant under electron density of a few 1011 m−3 is about 10 s (see Fig. 3 in Semeter and Kamalabadi 2005). To validate a multi-wavelength method for derivation of ionospheric conductivities qualitatively, we conducted photometric observations continuously during darkness intervals from October 2002 to March 2003 at the EISCAT Tromsø site in Norway. The photometer was pointed toward the local magnetic field-aligned direction at the Tromsø EISCAT site. In this paper, we aim at evaluating the method for derivation of conductivities by photometer using two wavelengths (427.8 and 630 nm). We present a comparison of ionospheric conductivities derived from the EISCAT UHF radar and the four-wavelength photometer for two nights on October 10 and 11, 2002. It should be pointed out that although usage of 844.6 nm emission would be better (cf. Ono 1993), this paper aims at evaluation and presenting capabilities of the usage of the two emission lines (427.8 and 630.0 nm). This is because the two lines are principal auroral emissions as well as easy to use, and few evaluation studies have been conducted.
The electron density data analyzed in this study were obtained with the tristatic EISCAT UHF radar, the so-called Kiruna–Sodankylä–Tromsø (KST) UHF radar system whose field of view is ~0.7°. We analyzed Common Program One (CP1) mode (Collis 1995) data. In the CP1 mode, the transmitting antenna at Tromsø is fixed along the local geomagnetic field direction. The physical parameters directly measured are electron density, electron/ion temperatures, and ion drift velocity. Electron density profiles were calibrated by ionosonde (Hall and Hansen 2003) data obtained simultaneously at the same observational field. Post-integration of the EISCAT radar data is made to reduce the noise level, and EISCAT data with post-integrated time of 10 s were used in this study.
Optical data used in this study were obtained with a four-wavelength photometer and a digital camera colocated with the EISCAT UHF radar at Tromsø. The four-wavelength photometer has a field of view of about 1.2° and was pointed along the local magnetic field line as well; the EISCAT radar and the photometer conducted common-volume measurements. Auroral lights with different wavelengths were fed into four channels by beam splitters and dichroic mirrors. The photometer has two dichroic mirrors used for separating blue and infrared wavelengths. Interference filters at 427.8, 630.0, 670.5, and 844.6 nm were selected based on Ono (1993) to obtain characteristic auroral emissions. Auroral emissions after the interference filters were detected by four photo-multipliers. Two wavelengths of emissions at 427.8 and 630.0 nm were utilized in this paper. Intensity of the photometer was calibrated by a calibration system at the National Institute of Polar Research (NIPR) in Japan. The transmission curves of the filters were measured by illuminating a diffuse screen with a scanning grating monochromator at NIPR. The filter efficiency for 427.8 nm was obtained by integrating, in wavelength, the product of the normalized filter transmission curve and the normalized synthetic spectrum at 300 K calculated following Kurihara (2004). Photometer data were obtained with an A/D board, and their time resolution reached 0.1 s. Images of the digital camera were used for monitoring auroral structures as well as weather conditions. The field of view of the digital camera was about 90° × 60° (about 240 × 140 km at 120 km height) with a fish-eye lens. We obtained sky images every minute with an exposure time of 8 s.
It should be pointed out that we did not use low intensity data where the intensity of 630 nm emissions was lower than the criterion described below. This is because we cannot distinguish airglow (usually less than about 100 R) or town light from aurora light, and the ambiguities of such cases become significantly larger. The criterion value was determined as follows: (1) we calculated a mean value (called, “first mean value”) with a standard deviation using all the data sets throughout the night, (2) we calculated a mean value (“second mean value”) using data whose values are less than the first mean value plus the standard deviation, and (3) the second mean value is used as a background level value and consequently is used as a criterion value. For example, from 18:00 to 23:30 UT on October 10, 2002, the criterion value was about 0.7 kR.
Case of October 10, 2002
Between 23:20 and 23:30 UT, both the conductivity values are in good agreement and the ratio is scattered about 1. However, the derived average energy was about 4–5 keV as shown in Fig. 2c. In Fig. 2a, the EISCAT UHF radar captured high electron density below 100 km, indicating particle precipitation with the higher energy (>30 keV) occurred. In this case, our derived average energy would be wrong. We will back to this later in this chapter.
Case of October 11, 2002
At last, it should be pointed out that the temporal scattering of the ratios (as shown in Figs. 3b, 5b, 8b, 10b) is probably due to the different sizes of the views of the two instruments, and thus, the scattering does not indicate a problem in the photometric method.
As presented in the previous session, the temporal variations of the conductivities derived from the two methods correspond well to each other, but the conductivity values from the photometer observations using 427.8 and 630.0 nm are generally lower than those derived from the EISCAT observations. In the case of the Pedersen conductivity, the inclinations of the liner fitting are about 0.6–0.7, while those of the Hall conductivity are 0.4–0.5, meaning that the Hall conductivity values from the photometer are almost half the values of those from EISCAT measurements. Here we discuss possible causes of the difference, such as (1) unsuitable assumption of the shape of the energy spectrum of the precipitating particles, (2) the long lifetime (~110 s) of the O(1D) state (630 nm line), (3) contamination/scattering of the emission line due to the atmosphere, (4) possible contributions from proton precipitations, (5) collision cross section between atomic oxygen and electron, and (6) composition of the atmosphere predicted by NRLMSISE-00. Finally, we have also examined effects of the cross section and the atomic oxygen density.
Shape of the energy spectrum
Based on rocket observations, Ogasawara et al. (2006) reported that most of the observed energy spectra of precipitating electrons were well expressed by kappa distributions with the thermal energy of a few hundreds of eV, while the spectrum inside a strong arc was better fitted by the sum of a Maxwellian distribution on the lower energy side and a power law at higher energies. Lanchester et al. (1997) also reported that a combined energy distribution of the Maxwellian and Gaussian distributions can reproduce the altitude profile of the electron density for an aurora with a sharp structure well. Furthermore, Miyoshi et al. (2015) showed that electrons with a wide energy range (tens of keV and a few hundred keV) simultaneously precipitated into the ionosphere during pulsating aurora in the morning side. It is quite difficult to reproduce such high-energy precipitation flux using only the two aurora emissions with assumption of Maxwellian distribution. These previous results can (at least partly) explain the reason why the Hall conductivity derived with the photometric method was underestimated in almost the entire interval. A similar trend can be found for the Pedersen conductivity during the same intervals. These results would suggest inadequate assumption of the energy spectrum shape that would cause underestimation of the Pedersen and Hall conductivities from the photometric method.
Long lifetime of the emission line at 630 nm
The emission at 630.0 nm is a forbidden transition of the electronic state of oxygen atoms from O(1D) to O(3P) with a long lifetime (~110 s). Ono and Hirasawa (1992) reported that the emission intensity at 630.0 nm has a tendency to gradually decrease with a time constant of 35–40 s. We utilized 10-s post-integrated data in this study. Considering the emission at 630 nm lasts for approximately 40 s, it would introduce a significant contamination to the derived average energy when the amount of precipitating aurora significantly varies with time and space. In particular, the validity of the derived average energy is a strong concern when the intensity of aurora light varies quickly with time. When a discrete arc passes through the field of view of the photometer, it would cause underestimation of the average energy of the particle precipitation.
Contamination/scattering of the emission line due to the atmosphere
Contributions from proton precipitations
In this analysis, energetic electrons have been taken into account as the only source of ionization. Several studies (Basu et al. 1987; Senior 1991; Galand and Richmond 2001) have, however, shown that energetic protons play a major role in producing ionospheric electrons and conductance at given locations and times, particularly in the cusp and at the equatorward boundary of the auroral oval before geomagnetic midnight. Senior (1991) compared the height-integrated conductivities derived from EISCAT data with a statistical model of conductance obtained from precipitating electron characteristics measured by the DMSP satellites. She found that the EISCAT-derived conductance agreed well with that from the DMSP model in the morning sector but was systematically larger than that of the model in the evening sector. She suggested that this difference is due to the E region electron production from energetic proton precipitations, which occurs preferentially in the evening sector. This idea could partly explain the facts shown here that demonstrate that the Pedersen and Hall conductivities seem to be underestimated by the photometric method. We could not distinguish the auroral emissions due to electrons and/or protons, because observations of proton auroral emissions such as H α or H β were not conducted in October 2002. In order to distinguish the effect of the energetic protons as well as to derive conductivities more accurately, observations of proton auroras should be carried out together, and the contributions of protons to the ionospheric conductivities should be taken into account properly.
Collision cross section between atom oxygen and electron
To calculate the auroral emission ratio needed for deducing the characteristic electron energy, we have applied the model presented in Ono (1993), where for the population of O(1D), three production processes and two loss processes were taken into account. A dominant mechanism of the O(1D) state in his model is a direct electron impact (cf. Solomon et al. 1988). Ono (1993) used the theoretical electron impact excitation cross section of the O(1D) state from Banks et al. (1974) that was originally presented by Green and Stolarski (1972). Johnson et al. (2005) reviewed the comprehensive analyses of experimental and theoretical cross-sectional values, and then, they showed that there are significant differences between both values of theoretical and experimental cross sections of O(1D) state excitation (see Fig. 3 of Johnson et al. 2005). Doering (1992) presented that although his experimental values were in excellent agreement with theoretic values at incident energies greater than 9 eV, at lower incident energies the experimental values had a sharp peak nearby 6 eV, approximately by factor 2 larger than theoretic values by Lan et al. (1972). It seems like that the cross-sectional values used in Ono (1993) would be lower than the experimental values in factor 2 or so. This would be another cause for discrepancy between photometric and EISCAT values; the Hall conductivities would be underestimated.
Composition of the atmosphere predicted by NRLMSISE-00
The neutral atmosphere density would contribute to the discrepancy in the conductivities calculated by the two methods. The models of the auroral red (630.0 nm) line predict strong dependence of its intensity on the ratio of the thermospheric atomic oxygen (O) to the molecular nitrogen (N2) concentration. Hecht et al. (2012) compared NRLMSISE-00 predictions over the year 2002–2006 for column O/N2 with TIMED/GUVI daytime observations as well as photometer (nighttime) observations made at two high latitude stations at Poker Flat (65.1°N, 212.5°E) and Fort Yukon (66.6°N, 214.7°E). Hecht et al. (2012) showed that while the nighttime observations showed considerable agreement with NRLMSISE-00 predictions, they were much more variable (+/− about 35%) than the predictions, suggesting that there are significant local effects in the auroral zone that are not captured by the model. This is also another cause (but we believe the usage of NRLMSISE-00 is the best option).
Dependencies of the conductivities on the cross section and the atomic oxygen density
Dependences of the Hall and Pedersen height-integrated conductivities on the cross section and oxygen atom density
Oxygen atom density
Simultaneous observations were conducted by using a multi-wave photometer and the EISCAT UHF radar at Tromsø for the two nights of October 10 and 11, 2002. By using these data sets, we compared electron density profiles as well as Pedersen and Hall height-integrated conductivities in order to evaluate the photometric method using two emission lines at 427.8 and 630.0 nm. In general, good agreement was found between height-integrated conductivities derived by the EISCAT UHF radar and photometer in terms of temporal variations, indicating that the optical method using the two emission intensities of 427.8 and 630.0 nm basically has the ability to derive conductivities for (diffuse) auroras. In cases of auroral arcs coming into the field of view of the photometer, the difference between the two methods is significant. We have discussed possible causes of the differences: incorrect assumption of the shape of the energy spectrum of the precipitating particles, the long lifetime (~110 s) of the emission line at 630 nm, contaminations/scattering of the emission line by the atmosphere, energetic proton precipitations, collision cross section between atom oxygen and electron, and composition of the atmosphere predicted by NRLMSISE-00. In the cases of discrete arc auroras, the energy spectrum of the precipitating particles can no longer be assumed to be the Maxwellian distribution. Also owing to the long lifetime of the 630 nm line, the derived average energy of particle precipitations can be overestimated and underestimated. Underestimation of the average energy causes electron density to be overestimated above ~160 km and underestimated below ~140 km. Therefore, care is necessary to analyze photometric observational data in cases of a discrete arc aurora, and it is of vital importance that simultaneous monitoring of auroras be carried out together with photometric observations. Furthermore, usage of incorrect collision cross section between atom oxygen and electron as well as composition change during the high auroral activity should also increase ambiguities for derivation of conductivities. Finally, we conclude that, even though photometer measurements are weakened by these limitations, the photometric method using the two wavelengths (427.8 and 630.0 nm) is very valuable for deriving ionospheric conductances (at least for the Pedersen conductance) due to its easy installation/operation.
KA developed/operated the photometer system, analyzed data, and provided scientific results. SN analyzed data and organized the manuscript. YO contributed to the operation of the photometer and analysis of data, and science. AB contributed to the operation and science. CH contributed to the operation and science. RF designed the science plan first and contributed to science. All authors read and approved the final manuscript.
We are indebted to the director and staff of EISCAT for operating the facility and supplying the data. EISCAT is an International Association supported by China (CRIRP), Finland (SA), Japan (ISEE and NIPR), Norway (NFR), Sweden (VR), and the UK (NERC). The four-wavelength photometer was calibrated using the optical facility at the National Institute of Polar Research, Japan. This study was partly supported by Grants-in-Aid for Scientific Research (15H05747 and 17H02968) of Japan Society for the Promotion of Science (JSPS) and by JSPS Program for Advancing Strategic International Networks to Accelerate the Circulation of Talented Researchers under Grant G2602.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- Brekke A, Hall C (1988) Auroral ionospheric quiet summer time conductances. Ann Geophys 6:361–375Google Scholar
- Doering JP (1992) Absolute differential and integral electron excitation cross sections for atomic oxygen: 9. Improved cross section for the 3 P → 1 D transition from 4.0 to 30 eV . J Geophys Res 97(A12):19531–19534Google Scholar
- Gattinger RL, Vallance Jones A, Hecht JH, Strickland DJ, Kelley J (1991) Comparison of ground-based optical observations of N2 second positive to N2 + first negative emission ratios with electron precipitation energies inferred from the Sondre Stromfjord radar. J Geophys Res 96:11341–11351CrossRefGoogle Scholar
- Hall CM, Hansen TL (2003) 20th century operation of the Tromsø Ionosonde. Adv Polar Upper Atmos Res 17:155–166Google Scholar
- Hecht JH, Christensen AB, Strickland DJ, Majeed T, Gattinger RL, Vallance Jones A (1999) A comparison between auroral particle characteristics and atmospheric composition inferred from analyzing optical emission measurements alone and in combination with incoherent scatter radar measurements. J Geophys Res 104:33–44CrossRefGoogle Scholar
- Kurihara J (2004) Energetics and structure of the lower thermosphere observed by sounding rocket experiment. Dissertation, Tokyo UniversityGoogle Scholar
- Miyoshi Y, Oyama S, Saito S, Kurita S, Fujiwara H, Kataoka R, Ebihara Y, Kletzing C, Reeves G, Santolik O, Clilverd M, Rodger CJ, Turunen E, Tsuchiya F (2015) Energetic electron precipitation associated with pulsating aurora: EISCAT and Van Allen Probe observations. J Geophys Res 120:2754–2766. doi: 10.1002/2014JA020690 CrossRefGoogle Scholar
- Senior C (1991) Solar and particle contributions to auroral height-integrated conductivities from EISCAT data—a statistical study. Ann Geophys 9:449–460Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.