J. Stadsnes, K. Aarsnes, and J. Bjordal
Department of Physics, University of Bergen, Allegaten 55, N-5007 Bergen, Norway,
X-ray imaging has proved to be a powerful technique for remote mapping of the precipitation of energetic electrons, providing spatial, temporal and spectral information on the precipitating electrons both at day and night. The strengths and weaknesses of the X-ray imaging technique compared to auroral imaging in the visible and ultra-violet spectral ranges are discussed. A short review is given of the development in this field, including some future plans. Some of the results obtained from auroral X-ray studies are discussed.
Energetic electrons precipitating into the Earth's upper atmosphere cause the auroral emissions from exited atoms and molecules, but X-ray bremsstrahlung is also generated when the electrons are decelerated. The X-rays, like the aurora, can be used for remote mapping of the electron precipitation. There is a clear connection between the energy spectrum of the precipitating energetic electrons and the spectrum of the emitted X-rays. Thus, by spectral imaging of the auroral X-ray sources, it is possible to gain information about the spatial distribution as well as the energy spectrum of the precipitating electrons. One can then derive maps of the energy deposition in the upper atmosphere, and height profiles of the resulting ionization and conductivities.
The particle excitations causing aurora in the visible and UV spectral range can be produced by primary and secondary electrons with kinetic energy greater than the excitation energy, which is in the range from a few eV to about 10 eV. Thus the auroral emissions are basically an integral response to the precipitation of electrons at all energies above the excitation energy and therefore not strongly dependent on the electron energy spectrum. This is to a large extent true for the prompt emissions, but not for emissions from so-called metastable states, involving a time constant and the possibility of non-radiative quenching of the excited state when the exitation occurs deep in the atmosphere. Thus Eather and Mende (1972) found that the intensity ratio between the auroral emissions at 630 nm and 427.8 nm contains information on the energy spectrum of the precipitating electrons. This information is however limited to electrons of less than about 5 keV.
Since the X-ray measurements are not contaminated by sunlight, the X-ray technique can be used to study energetic electron precipitation on the dayside of the Earth as well as on the nightside.Visible auroral studies are limited to the dark hemisphere, whereas it is possible with some limitations, to study the aurora also on the sunlit side of the Earth at vacuum UV wavelengths (Frank et al., 1982, Anger et al., 1987). Compared to direct satellite measurements of the precipitating electrons, which are limited to the magnetic field lines close to the satellite, the X-ray imaging technique can provide instantaneously large area maps of the flux and energy of the particles.
The X-ray technique for studying the energetic electron precipitation from below the X-ray source region was used on stratospheric balloons already in the late 1950s. For an overview of the early balloon measurements see Anderson (1965). Because of the absorption of the X-rays in the air between the source and the balloon altitudes, these studies were limited to >20 keV X-rays. Soon, the X-ray technique was also adopted to sounding rocket measurements which were not limited by the atmospheric absorption of the X-rays but could measure all the way down to the instrumental detection limit, typically a few keV. The first crude satellite-based mappings of the X-ray aurora were done already in the late sixties. From a balloon one can see only a limited region, of the order of 100 - 150 km in diameter, of the X-ray source region. From sounding rockets one can cover wider areas of the atmospheric source, and with satellites one can image the whole auroral oval if the orbit is high enough above the polar region.
CONNECTION BETWEEN PRECIPITATING ENERGETIC ELECTRONS AND X-RAYS
Several theoretical studies have been done on the fluxes of X-rays generated from precipitating energetic electrons (Rees, 1964, Berger and Seltzer, 1972, Seltzer and Berger, 1974, Walt et al., 1979). Figure 1 shows the differential energy X-ray fluxes at balloon altitudes and at satellite altitudes as extracted from graphs given in Berger and Seltzer (1972) for typical balloon altitudes, and Seltzer and Berger (1974) for satellite altitudes. The spectra shown are the X-rays resulting from precipitation of electrons with an exponential energy spectrum with 30 keV e-folding energy. Isotropic electron fluxes spread homogeneously over the horizontal plane have been assumed in their model. As seen, at balloon altitudes the X-ray flux falls off very fast below about 30 kev, because of the photoelectric absorption in the air between the balloon and the X-ray production layer. It is also seen that the flux at higher energies is reduced with a factor of 5-10 at balloon altitudes. This is an effect of the Compton scattering. The Compton scattering also limits the sharpness of X-ray images obtained from balloon borne X-ray cameras. The situation can be improved by going to higher altitudes, 2-3 mb pressure, which corresponds to about one half scattering mean free path for 50 keV X-rays propagating in the nadir direction.
Fig. 1. Differential energy X-ray fluxes normalized to incident integral electron flux. Isotropic electron fluxes with an exponential energy spectrum with 30 keV e-folding energy and spread homogeneously over the horizontal plane. The upper solid line represents the calculated X-ray fluxes at satellite altitudes from Seltzer and Berger (1974). Lower solid line and dashed line give the X-ray fluxes at balloon altitudes, 6-7 g/cm2 and 9-10 g/cm2 respectively, from Berger and Seltzer (1972).
BALLOON BASED OBSERVATIONS
The greatest strength of balloon based observations is the long time series, typically a day or longer, which can be obtained. In addition, the balloon simply drifts with the air at stratospheric altitudes, and this drift is very slow compared to the velocity of rockets and satellites, so for shorter time intervals the balloon can be considered to be stationary. Balloon measurements of auroral X-rays have yielded valuable information about the temporal development of energetic electron precipitation at different time scales from hours to sub-seconds.
Figure 2 shows the gross features of the X-ray events observed during the night of 5-6 August 1964 from two balloons launched from Alta and Andenes in Northern Norway, both sites at L=6 and separated about 280 km in longitude (Trefall et al.,1966). The magnetogram from Andenes shows a magnetic bay around geomagnetic midnight, but else it was fairly magnetically quiet locally, and the Kp index was between 3+ and 1+ for the 12 hours interval covered by this figure. More or less continuous X-ray events from the midnight to the pre-noon sector were observed.
Fig. 2. Gross features of X-ray events observed during the night of 5-6 August 1964 by two balloons launched from Alta and Andenes in Northern Norway, and variation of the horizontal magnetic intensity at Andenes (from Trefall et al., 1966).
We will focus on the time interval marked with an arrow, where microbusts occurred as shown in Figure 3 (Trefall et al., 1966). Microbursts are short bursts of energetic electron precipitation lasting a fraction of a second. The upper panel shows examples of isolated microbursts seen most clearly at the easternmost balloon (Alta). In the lower panel the westernmost balloon (Andenes) shows the clearest microbust signatures, and we here see examples of precipitation pulses of about 5 s duration which are made up of a series of microbursts. Later on we have densely packed microbursts without a clear slow modulation of the precipitation. Note the great difference in the appearance of microbursts at the two balloons located less than 300 km apart from each other. A study by Parks (1967) based on four collimated X-ray telescopes flown on a balloon to 2-3 mb floating level, gave a typical spatial extension of 80 km for microbursts. Rosenberg et al. (1981) studied the conjugacy of X-ray microbursts and VLF chorus. They got results which were consistent with near-equatorial cyclotron resonance interactions being responsible for the microbursts in the energetic electron precipitation.
Fig. 3. Details of typical microburst events (from Trefall et al., 1966)
The temporal relations between substorms on the nightside and X-ray events on the morning and dayside were studied by Sletten et al. (1971). They concluded that the precipitating electrons in most cases had been injected on the night side during substorms and then gradient-drifted over to the morning and dayside where they were acted on by some precipitation mechanism.
During the decades of 1960 and 1970, large scale balloon programs involving simultaneous launches of balloons from several sites across the auroral oval were carried out from Northern Scandinavia. By launching in the summer time when the stratospheric winds were blowing westward, the balloons would drift all the way to Greenland during several days. By launching new balloons at one day intervals, one could establish a grid of observation points covering both a substantial latitude- as well as a longitude-range. In this way one could gain information about the spatial morphology of the electron precipitation during different kinds of magnetospheric activity. Especially magnetospheric substorm related precipitation in different magnetic local time sectors was studied in detail by combining the balloon observations with ground based measurements from riometers and magnetometers (Bjordal et al.,1971, Kremser et al.,1973, Maral et al.,1973, and Kangas et al.,1975). Later on the emphasis was put on coordinated balloon and geostationary satellite based measurements (see e.g. Kremser et al. 1982, Torkar et al.1987, Ullaland et al. 1993).
Some attempts of imaging the auroral X-rays by balloon borne devices have been done. The first, crude information about the spatial source distribution was obtained by using several collimated detectors pointing in different directions (Parks, 1967, Bjordal et al.,1976). Later, more genuine X-ray cameras were used (Werden and Parks,1987, Yamagami et al. 1990). As pointed out earlier, because of the Compton scattering there are limitations on the sharpness one can obtain from such balloon-based images, if not using very large balloons bringing the payload up to the 2-3 mb floating level.
IMAGING OF AURORAL X-RAYS FROM ROCKETS
This technique was implemented in the 1970s (Vij et al.,1975, Aarsnes et al.,1976, Goldberg et al.,1982). By using a well collimated X-ray spectrometer pointing at a slant angle with the axis of a spinning rocket, one can scan the atmospheric X-ray source regions and thereby build up an image of the X-ray aurora. At the same rocket one can do in-situ spectral measurements of the energetic electron fluxes. This enables an intercomparison of the parent electrons and the X-rays generated in the atmosphere around the footpoint of the geomagnetic field line of the rocket. Results of such comparisons have been published e.g. by Vij et al. (1975). These authors calculated the X-ray flux expected from the measured electron flux using thick target bremsstrahlung theory and a Monte Carlo calculation of the transport of X-rays through the atmosphere. They found that the deduced X-ray spectrum agreed quite well with the X-ray observations. Rocket experiments are especially suited for making targeted studies of specific phenomena as the rocket can be launched just when the right kind of auroral conditions occur at the launch site.
Simultaneous mappings of the X-rays and visible aurora were obtained on a rocket flight in 1983 (Stadsnes et al., 1986). An auroral photometer looking at the 427.8 nm N2+ emission line, was oriented in parallel with a collimated X-ray spectrometer. By combining the rocket's spinning and coning motion, images of the auroral sources were obtained as illustrated in Figure 4. There are clear similarities between the spatial distribution of the X-rays and the aurora, e.g. the maximum to the East and the minimum in the North are seen both in the X-rays and in the aurora. On the other hand, especially in the South-West quadrant the aurora is more pronounced than the X-rays, indicating precipitation of soft electrons which generates aurora but do not have high enough energy to generate the >2.9 keV X-rays.
Fig. 4. Simultaneous mappings of >2.9 keV X-rays and visible aurora at the 427.8 nm N2+ emission line as obtained by the rocket F63 POLEWARD LEAP launched from Andya rocket range. A source height of 110 km has been assumed both for the X-rays and the aurora.
X-RAY IMAGING FROM SATELLITES
The first satellite observations of auroral X-rays were done in 1972 (Imhof et al., 1974). This first experiment used a collimated Ge spectrometer (Ex > 50 keV) which was mounted on a satellite spinning in a cart-wheel mode. Another experiment was flown in 1977 by Mizera et al. (1978). They used a combination of a collimated proportional counter and several CdTe detectors to measure X-rays from 1.4 keV to about 90 keV around the nadir direction. In this way information about the spatial distribution of the X-ray sources was obtained for regions near the satellite orbit plane.
By using a highly collimated, cross track scanning X-ray detector, one image was generated per satellite pass on the polar orbiting DMSP F6 satellite (Mizera et al., 1985). On the P78-1 satellite Imhof et al. (1980) flew eight collimated CdTe detectors which were looking out to each side of the satellite's orbit plane. As the satellite was spinning in a cartwheel mode, these X-ray detectors were scanning the X-ray source regions in the atmosphere from horizon to horizon.
An X-ray Imaging Spectrometer (XRIS) was flown on the S81-1 satellite as a part of the SEEP Experiment (Calvert et al, 1985). XRIS employed a one-dimensional pinhole camera oriented crosstrack, to obtain a two-dimensional X- ray image as the satellite passed over the auroral scene (push broom technique). It had a large area proportional counter that detected X-rays from about 4 keV to 40 keV in 16 pixels in the cross track direction. Figure 5 shows some examples of images from XRIS (Imhof et al, 1985). At an orbit height of 250 km, XRIS yielded a strip image of the upcoming X-rays about 500 km wide and with a spatial resolution of about 30 km. In May and June 1982 XRIS provided the first detailed satellite-based images of the auroral X-rays. In Figure 5 one clearly sees isolated patches of energetic electron precipitation occurring both on the dayside and on the evening side, and at latitudes corresponding to poleward of the auroral oval as well as inside the trapping boundary. Auroral arc-like structures are also seen.
Fig. 5. Mappings of X-ray intensities (4-40 keV) observed at high latitudes by XRIS on the S81-1 satellite (from Imhof et al., 1985).
Another X-ray imaging experiment, AXIS (Atmospheric X-ray Imaging Spectrometer), was flown on the UARS satellite (Chenette et al., 1992). AXIS consisted of 16 cooled Si detectors mounted in an array and looking cross track from horizon to horizon (push broom technique). The spatial resolution was better than 100 km near the orbit plane. The energy covered was 3 keV to 100 keV. The UARS satellite launched September 12 1991 has a 585 km by 57 degrees inclination orbit. This inclination is not perfect for studies of auroral zone related electron precipitation, but for certain longitudes, e.g. over North America, it is fairly good. Figure 6 shows an image of the auroral X-rays obtained during one pass of the UARS satellite during a magnetically active period in November 1991 (Sharber et al., 1993). Chenette et al.( 1993) have also shown a series of images obtained in the Northern and Southern auroral oval during three consecutive passes of UARS during this magnetic storm. With AXIS on UARS one has to wait for the next pass, about 1 1/2 hours later, before getting another picture of the same part of the auroral oval. One may imagine what could be obtained from an X-ray camera on a satellite at an orbit height sufficient to image the whole auroral zone at the same time, and dwelling at high altitudes long enough to cover this area for periods comparable to e.g. the duration of substorms.
Fig. 6. Mapping of X-ray intensities (3 to 100 keV) as obtained by AXIS on the UARS satellite during a magnetic storm periode in November 1991. The Q=3 auroral oval (dotted), circles of constant magnetic latitude at 60° and 65°, and the terminator (dashed) are shown for reference (from Sharber et al., 1993)
The first genuine 2 dimensional X-ray detector, PIXIE, is now flown on the POLAR satellite in the NASA Global Geospace Science Programme (Imhof et al.,1995, Chenette et al., 1996). PIXIE - the Polar Ionospheric X-ray Imaging Experiment- consists of a pinhole collimator and a multiwire proportional chamber. Because of the high apogee altitude of POLAR, PIXIE is able to image the whole auroral oval with fairly coarse spatial resolution for long time intervals at the apogee part of the satellite orbit. Whereas more detailed images of large parts of the oval can be obtained near perigee. Thus it is possible, for the first time to study the spatial and temporal distribution of the energetic electron precipitation e.g. during substorm injections.
DERIVATION OF ELECTRON SPECTRUM AND FLUX FROM MEASURED X-RAYS
A number of methods have been developed to invert the X-ray spectra to get the precipitating electron fluxes (Luhmann 1977, Walt et al., 1979, Gorney 1987, Robinson et al., 1989). For a discussion of the different inversion techniques see Robinson and Vondrak (1994). The accuracy of the deconvolution technique is dependent on the enegy range and resolution of the X-ray detector as well as on the count rates. In order to minimize the statistical fluctuations in the measurements the count rates must be high, and that means fairly large X-ray detectors, because of the low efficiency of bremsstrahlung X-ray production from energetic electrons. The auroral X-ray detectors flown until now have all, to some extent, had some shortcomings with respect to fulfilment of the "ideal" requirements.
In order to proceed with an inversion process with limited information content in the X-ray measurements, one has to make some simplifying assumption about the shape of the electron spectrum. This is illustrated in the following example from inversion of XRIS data. Datlowe et al. (1988) derived electron spectra for different types of electron precipitation imaged by XRIS. The approach used in their analysis was to assume a functional form of the electron spectrum and then model the bremsstrahlung produced. They then modeled the detector response to this X-ray flux. By comparing the calculated pulse height distribution with the observed one, an improved estimate of the spectral parameters was derived. They used an iterative process until very small changes were seen from one iteration to the next. They used a two-parameter fit to the electron spectrum, the precipitation rate in electrons per cm2 s, and the characteristic spectral slope. Figure 7 shows two examples of such fits to exponential electron spectra for isolated patches.
Fig. 7. Spectra from isolated patches of X-ray emission in the polar cap. Crosses are observed X-ray fluxes, and the lines give the X-ray flux produced by electron spectra of exponential form, E0 is the e-folding energy. (From Datlowe et al., 1988).
After performing the inversion from X-ray spectra to electron spectra, one can derive maps and generate height profiles of the associated energy deposition into the upper atmosphere. In Figure 8 are shown contours of the horizontal distribution of the X-ray fluxes as measured by XRIS during one crossing of the auroral oval. Figure 9 shows height contours of ion production rate calculated from these X-ray measurements along the center pixel line of Figure 8. Knowing the height profile of the energy deposition it is also possible to model the height profiles of the Pedersen and Hall conductivities caused by the energetic electron precipitation.
Fig. 8. XRIS X-ray image (5.4 - 11.6 keV) from June 26, 1982. The minimum flux contour is 50 photons /cm2 s sr, and subsequent contours are spaced logaritmically at 4 contours per decade (from Datlowe et al., 1988).
Fig. 9. Contours of ion production rate calculated from the x-ray measurements on June 26 1982 (Figure 8). The lowest contour corresponds to 10 ions/cm2 s, and the most intense inner contour is at 105 ions/cm2 s. (From Datlowe et al., 1988)
CONCLUSIONS AND FUTURE PLANS
Imaging of auroral X-rays has proved to be a powerful technique for studying the precipitation of energetic electrons. From space one can map the distribution of the auroral X-ray sources, thereby getting the spatial distribution of the electron precipitation. This can be done in the daylit hemisphere as well as on the nightside. From the X-ray spectra it is possible to infer the electron energy spectrum in each picture element. But in order to do this with good precision, one needs X-ray spectral cameras with sufficiently large geometric factors to give high count rates and thereby good statistics.
Balloon-based X-ray measurements have contributed greatly to our knowlegde and understanding of the temporal development of energetic electron precipitation events on different time scales, from substorms to pulsations and subsecond microbursts. The balloon based studies have also provided much of the knowledge we have on the spatial morphology of the energetic electron precipitation at all local times.
Rocket-flown collimated X-ray spectrometers have provided pictures of the spatial distribution of the energetic electrons in wide regions around the rocket thereby also providing a spatial reference frame for interpreting the in-situ measurements on the rocket. By comparing the X-ray mapping with simultaneous rocket-based photometric mapping of the aurora one has gained information on the differences in the spatial and temporal distribution of the precipitation of the energetic electrons (>5 keV) and the less energetic electrons causing most of the aurora (<5 keV).
Satellite borne X-ray imagers have developed gradually from fairly simple collimated detectors scanning the atmosphere to truly two dimensional cameras providing instantaneous images of the X-ray aurora. From the X-ray camera PIXIE on the high altitude polar orbiting satellite POLAR it is thus for the first time possible to obtain global pictures of the energetic electron precipitation and study the development of the precipitation, e.g. during a magnetospheric substorm. From the X-ray spectral images one can derive maps of the characteristic energy spectra and dynamics of the precipitating energetic electrons, covering at the same time the auroral as well as the polar and subauroral regions of space. Combining with the images from the visible and UV cameras on POLAR, one should be in a good position to determine the boundaries between different magnetospheric source regions for the precipitating electrons, according to spectral characteristics of the different electron populations. From the PIXIE measurements one can also derive the global energy input to the upper atmosphere from precipitating energetic electrons during different solar wind/magnetosphere coupling conditions.
In the near future, balloon based X-ray mappings are scheduled in conjunction with auroral mappings from the POLAR satellite. An X-ray imager will also be flown on one of the first rockets to be launched from Svalbard to study the dayside cusp and the boundary layer. The X-ray imager will provide information about the position and shape of the dayside boundary layer by mapping the associated energetic electron precipitation.
Planning of a multi-spectral auroral imaging observatory (AURIO) has been on its way for some years now in the ESA Programme (Stadsnes et al. 1987). In 1990 ESA selected AURIO together with an advanced particle and fields observatory, APAFO, to constitute the Space Science Elements in the Polar Platform Programme, but without accomodating these experiments on the first mission called ENVISAT-1. As of today it is not yet clear when AURIO will be flown. AURIO would provide images of the aurora in the IR, visible and UV-range, and of the X-ray aurora (from few keV to >100 keV) with much better spatial and temporal resolution than yet obtained. Together with the in-situ particle and fields measurements, this would enable studies of fine and medium-scale coupling processes between the magnetosphere and the upper atmosphere.
The authors thank their colleagues in the Space Physics Section at University of Bergen, and the PIXIE and the AURIO science teams for many inspiring discussions on different aspects of auroral X-ray imaging. This work was supported by the Research Council of Norway.