Alternative Paper for Alfven conference: or otherwise
Misleading aspects of the use of the oxygen red line for optical studies of the cusp and cleft aurora.
Roger Smith and Joe Minow
Contents:
1. Introduction. Past uses of the redline and their problems
2. Tracing precipitation for cases of soft particle precipitation
3. Remodelling and revision of previously published work.
4. Conclusions.
Introduction
It has become traditional to trace the dayside aurora using the 6300A emission of atomic oxygen. In many ways this appears to make good sense and it is attractive in practise since the intensity of this emission in the dayside aurora is normally relatively high, making observations cheaper and easier. However, the research community working in this area has ceased to be concerned about the aeronomy associated with this emission which has been well known for decades and which sounds alarm bells for those who would use the imaging and photometric data without due consideration for the warnings.
Rees and Luckey (1974) demonstrated the utility of the red line emission rate when used in ratio with 4278A N2+ first negative (0,1) band, or less reliably with the 5577A emission of atomic oxygen. A high ratio with either of these emissions indicates the low energy precipitating flux which is well known to characterize the cusp region at ionospheric heights. Hence, for many years, at least since the ISIS satellite photometric plots (Shepherd and ??), we have been inclined to identify cusp phenomena by plotting the ratios of emissions.
As Rees and Luckey pointed out, but we have forgotten, the theory which justifies the use of these emission ratios and provides the formulation for their interpretation is one dimensional (ie assumes that the flux tube in the ionosphere maintains itself without loss or gain of contents, be they ionized or neutral). The forgotten factor is the long lifetime of the 1D atom when it is excited in the upper thermosphere. The effective lifetime at the upper reaches of an auroral display can be over 60 seconds. The problem, which is obvious, is that the excitation point of the 1D oxygen atom is most likely to be far from the emission point because of the effects of the wind. Also, by the time the emission takes place, the precipitation region has moved on to some other place well away from the observed emission point. Considering some typical numbers, the neutral wind, which removes the neutral excited oxygen atom, is typically 300 m/s. An air parcel will then travel 18km in that time. Also the source of the precipitation, which is expected to move with ionospheric convection in the cusp aurora, has a typical speed of 1 km/s and a travel distance of 60kmn in 1 minute.
The inescapable conclusion from these elementary calculations is that the point of emission of the 6300A aurora in the dayside aurora is neither where it was excited nor with the auroral form which was, at one time, conjugate with it. This emission will neither show where the excitation takes place nor provide reliable spectral-ratio quantities when the other component is a prompt emission or whose upper state is only slightly metastable. Thus, in the well-known typical cusp aurora situation of poleward moving forms, there will be 6300 emission between the forms which was excited in them. In contrast, the emission from 5577A or 4278A will track the excitation region much more closely. Therefore, the cusp aurora seems to have a large diffuse component in the red line which is likely to be no more than the expected aeronomical result of drifting clouds of 1D oxygen carried on the wind away from a quite sharply defined source. Also the region of largest red-to-green or red-to-blue ratio will be between the poleward moving forms and the ratios on the forms will be systematically reduced from the expectation of 1-D geometry.
At this point it is convenient to introduce the term "fossil aurora" to refer to aurorally generated emissions which occur sufficiently delayed from excitation that they are well separated from position inferred from a particle detector. With this understanding, one can readily classify the red line emissions in the dayside as fossil aurora and the structural forms as fossil arcs. We who study the dynamics of the cusp and other dayside auroral phenomena need to take full cognizance of the fact that our 6300A auroral observations are of fossil aurora and make due allowance for this in drawing conclusions from our observations.
Other, more subtle, effects are likely to occur because the effective lifetime of the 1D state is a function of altitude. Therefore the lower edge of a fossil arc will decay faster than the upper edge and the mean altitude of such a form will rise with time. The biggest problem with this is the apparent horizontal motion inferred for such a form,viewed from one side, when it is assumed that the lower border is at a steady height. Hence there can be assumed horizontal motions of red line aurora forms which are actually caused mainly or wholly by upward motions of the lower border.
Finally, a word about the projection of 6300A observations using an assumed constant height. For the dayside aurora, where the precipitation is soft and the field-aligned extent is hundreds of km (ref!), an arc or ray viewed from the side projects to a large horizontal distance which is greatly different than the dimension perpendicular to the L-shell. While it is clear that for a simple projection of an auroral image, this is the only option, the errors introduced into the scientific process are unacceptable. To give an example of the type of problem which arises, consider Figure 1 which is reproduced from a recent paper by Moen et al., 1996. The large patch of projected aurora is shown to be conjugate with a satellite pass. The original auroral image can be recognized to show tall rays converging towards the magnetic zenith. If these rays had been projected along the field line, the result would have been a curvy line of dots almost negligible thickness at the scale of the plot and satellite path will miss the aurora. The science discussed in the Moen et al. paper is highly dependent on the conjugacy argument, associating measurements made on the ground with those from the satellite, and hence one can readily see that, without proper consideration of the correct method of projection, the results of this paper will be misleading.
2. Tracing precipitation for cases of very soft precipitation.
Despite the foregoing difficulties, the 6300A emission is a very efficient indicator of soft precipitation because the secondary electrons produced in the precipitation process are heavily concentrated in the upper thermospheric region. The cross-section for 1D production is x times higher than for 1S and the fraction radiating is close to unity. Nevertheless some 1S will still be produced, along with A state N2+ which will cause emissions at 5577 and 4278A. Of these the 5577A emission is usually the brighter. We consider here the use of 5577A emission as a racer for cusp precipitation.
It is often stated that the cusp is characterized by the lack of 5577A emission (Karandikar, Meng, etc). This led to the identification of the cusp region by means of the so-called "midday gap" seen in many of the earlier satellite observations, particularly DMSP images. Observers from the ground also see 5577A emission getting weaker near the cusp ionosphere, however it does not go to zero. There remains a small but measurable signal above the airglow level. Figure 2 shows two panels of false-colored plots of meridian scanning photometer data from Longyearbyen in Svalbard. The upper panel shows a cusp encounter in 5577A emission. The upper part of this figure (5577A) indicates weak but well-defined regions of auroral precipitation seen in the poleward-moving forms. The lower part of the 5577A emission originates from much harder precipitation, as we can tell by using the 6300/5577 ratio (which works well in diffuse auroral conditions where the emission is spatially continous and the effective lifetime of 1D is much lower). In the lower panel of Figure 2 the emission rate of the 6300A aurora is shown. It is clear that the 6300A emission rates are low in the region of the diffuse aurora; a condition which results from the low altitude of excitation and the consequent heavy collisional quenching. It is also clear that the 6300 emission covers a much broader area in the region of soft particle transients than the 5577A. The 6300 emission in the vicinity of transients joins them together together making a continuous bright region. Although the green emission describes the precipitation regions well, the red emission is a relatively poor indicator of the position and time of occurrence.
It is clear that the red line emission has limited uses in the study of cusp aurora. Consideration has to be given to the adoption of new wavelength(s) to be used as standards. As has been outlined above, the functions to be fulfilled for optical observations of the cusp include the mapping of precipitation zones and the inference of the characteristic energy of the precipitation. Lummerzheim et al have suggested the use of 8446 instead of 6300 along with 4278. This combination has the advantage that both are relatively prompt and can map the precipitation zones. This also means that they will not drift with the wind and give false readings of the ratio on or off the auroral form. There may be other useful ratios, but it is preferable to use atomic emissions where available since molecular emissions are necessarily weak in the cusp region. Also, the 4278A emission has a resonance scattering component under sunlit conditions, such as is common for observations made in Svalbard at cusp passage. It would be preferable to substitute an emission like 5577A for that reason even though it (5577) is not completely understood and there is a substantial airglow component. The best conclusion that can be reached at present, without much more extensive calculations is that 8446 and 5577 emissions should be used.
3. Remodelling and revision of previous work using red line emission.
Previous published work needs to be reviewed with the aim of removing conclusions which are based on an incomplete understanding of the 6300A emission. In particular, all works on dayside aurora which make quantitative use of the 6300/5577 ratio for the determination of characteristic energy. It will also be necessary to re-evaluate any conclusions based on small movements of redline arcs where the upward movement of the lower border could have been mistaken for a horizontal displacement.