Solar Cycle and the Aurora

Presented bilingually. Click to toggle: Scientific Presentation Plain English

(Newell et al., Relation to solar activity of intense aurorae in sunlight and in darkness, Nature, 393, 342, 1998.)

Background:

In an effort to explain why intense electron acceleration events -- i.e., mono-energetic electron spectra known to be associated with discrete aurora [Evans, 1968 JGR] -- are intensely concentrated in the 1800-2400 MLT sector, we have been investigating predictions which we believe follow from the ionospheric conductivity feedback mechanism [Atkinson, 1970, JGR; Lysak, 1991 JGR]. Motivated by this theory we investigated the effect of sunlight on the frequency of electron acceleration events; in the crucial dusk-to-midnight sector there proved to be three times as many such events in darkness as in sunlight. The same ratio also held comparing only (local) summer solstice to winter solstice (Newell et al., 1996; Discrete Aurora Are Suppressed in Sunlight.)

Finding a surprising result is one thing; having it confirmed by others is quite another. Since our results were published there have been several confirming studies, one from our team using auroral imagery instead of particle data (Liou et al., 1997, JGR) and several by completely independent teams (Erlandson and Zanetti, 1998, JGR; Yamagishi et al., 1998 in Polar Cap Boundary Phenomena, J. Moen et al. editors; Collin and Peterson, presented at the Fall 1997 AGU Meeting) with several more in press.

To further test the importance of ionospheric conductivity, we decided to examine the effects of solar cycle on electron acceleration events (and hence auroral arc frequency). Before conducting the research we supposed that the ratio of aurora in sunlight to darkness would be further suppressed at solar maximum compared to minimum, although the total number of aurora would rise at solar maximum.

Data:

Mono-energetic electron spectra were automatically identified in all DMSP satellite SSJ/4 particle data over a 12-year period More information about this procedure can be found in Newell et al. [1996 JGR] and in Newell et al., 1996; Discrete Aurora Are Suppressed in Sunlight.. For each year from 1984 through 1995, the results presented below were constructed in two stages. First, the probability, P(MLAT,MLT) of observing intense electron acceleration event s (>5 ergs/cm²s with a mono-energetic peak) was calculated for each 0.5° x 0.5 hour bin, based simply on the number of spectra fitting the criteria for an acceleration event divided by the total number of satellite spectra taken within that bin. Then the average surface "area" covered by intense arcs was calculated from

_₆₀^⁸⁰

_₁₈^²⁴ P(MLT,MLAT) d(MLAT)d(MLT).
The numbers thus arrived at are a measure of the time-averaged surface area covered with intense discrete aurora.

Here is an example of two such maps, one for 1984, one for 1991, both constructed only for data taken while under sunlit conditions (solar zenith angle less than 85 degrees).

The global frequency of aurora (left) for 1984, around solar minimum and (right) for 1991 shortly after solar maximum. This figure is under sunlit conditions only, when the sun's ultraviolet rays cause increased ionospheric conductivity. (Click on either image to view larger version.)

Results:

First, here are the explicit solar cyle variations under conditions of darkness and sunlight:

The average surface area covered by intense aurora on a yearly basis. For reference, yearly average F10.7 numbers are overplotted. (Left) Solar zenith angle <85°. (Right) Solar zenith angle > 110°. (Click on either image to view larger version.)

Under sunlit conditions, the frequency of intense aurora appears to minimize around solar maximum, while in conditions of darkness, no clear trend appears. However because auroral frequency and even F10.7 number can greatly fluctuate over the course of a year, a better approach is to determine auroral frequency as an explicit function of F10.7 number. Rasmussen et al. [1988] demonstrated that ionospheric conductivity is linearly proportional to F10.7 number, to a reasonably good approximation. We therefore run through the entire 12-year data set repeatedly, each time constructing a map where only data taken while the instantaneous (daily) value of F10.7 lie within a certain bin range (e.g., 70-90, plotted below as F10.7=80).

The global frequency of aurora as an explicit function F10.7 number. Left: Under sunlit conditions (solar zenith angle < 85°), auroral frequency is highly anti-correlated with F10.7 number. Right: In darkness (solar zenith angle > 110°), F10.7 number is uncorrelated with auroral frequency. (Click on either image to view larger version.)

These new results can be reconciled with the traditional expectation of more auroral activity at solar maximum quite easily. We note that during a solar cycle the average values of solar wind parameters to not change significantly; instead what changes is that the solar wind is more variable following solar maximum. Although this result already appears in the literature of solar wind studies, we computed the yearly average values for various solar wind parameters for the 12 years in our study, in order to be sure that the usual result held for our study period.

(Left) Yearly average values of IMF B_z (1984-1996), computed from IMP-8 hourly average values. Only small variations are observed, and these are not appropriate to predict more aurora following solar maximum (1990). (b) The distribution of B_z values at solar minimum (1984) and after solar maximum (1991), normalized to an equal number of observation hours. More extreme values of B_z are observed after solar maximum, even though the centroid of the distribution is essentially unchanged. (Click on either image to view larger version.)

Very similar results also hold for solar wind velocity in this period (and in the usual case): although the average velocity does not increase following solar maximum, more extreme events of very high or low velocity are found.

Since major magnetic storms which bring the aurora to low latitudes occur under extreme conditions (large v and very negative B_z) we can understand why such storms are most common following solar maximum, while still not anticipating the average auroral activity to be any different.

More detail about our understanding of the role of the ionospheric conductivity feedback mechanism on the aurora is given in the next story, Discrete Aurora Are Suppressed in Sunlight..

Acknowledgements:

This work was supported by the National Science Foundation grant ATM--9531489 and Air Force Office of Scientific Research Grant F49620-96-1-0009. Dave Hardy of the Air Force Research Laboratory is the effective P.I. on the DMSP SSJ/4 particle detectors. We thank R. Lepping, J. King, and the Goddard NSSDC for the IMP-8 data.

Send science questions/comments to Dr. P. Newell
Send WWW questions/comments to webmaster