next up previous
Next: References Up: Latitudinal dynamics of auroral Previous: February 17, 1998

Discussion and Summary

Hughes:98 surveyed auroral roar at various latitudes to show that the peak occurrence rate of 2$f_{ce}$ emissions is surprisingly near the statistical poleward edge of the auroral zone (75$^\circ$-76$^\circ$ invariant) rather than the middle of the auroral zone. The five events described in section 4 are the first auroral roar emissions studied in detail by observing the emissions simultaneously with receivers spread over a range of latitudes that sample the polar cap and auroral zone and by comparing these observations with ground magnetometer and satellite data. The work presented in this study suggests why auroral roar peaks in occurrence at the poleward edge of the auroral zone as suggested by Hughes:98.

In four of the 5 days studied, the peak intensity of auroral roar emissions is located at or poleward of the poleward electrojet boundary inferred from the magnetometers, often by many degrees or many hundreds of kilometers. During the October 3 event (Figure 3) the peak intensity remained up to 9$^\circ$ poleward of the electrojet boundary for all but a few brief periods. The same behavior occurs during the October 4 event (Figure 4) except for brief periods when the electrojet boundary is moving rapidly (e.g., 0045 UT). Again, during the November 5 event (Figure 5) the intensity peaks poleward of the electrojet boundary by up to 7$^\circ$, except for a few brief moments. During the February 17 event (Figure 7) the location of the peak intensity is again nearly always poleward of the electrojet boundary by up to 6$^\circ$, except for one brief burst of auroral roar detected at Gillam at 0405 UT and after $\sim$0600 UT when the latitude of the electrojet boundary changes rapidly.

The May 2 event (Figure 6) is strikingly in contrast to the other 4 days. The intensity most of the time peaks many degrees, often up to 10$^\circ$, equatorward of the poleward electrojet boundary, closer to the middle of the auroral zone, and for a short time (0428 UT) equatorward of the equatorward electrojet boundary. This unexpected result could be explained by the presence of double or multiple auroral arcs. Perhaps emissions can penetrate to the ground between the poleward and equatorward emission regions. There is evidence during some of this period ($\sim$0500-0645 UT) for double electrojet structures, inverted-V type structures with latitudinal scale sizes of $\sim$100 km, in the magnetometer and meridian scanning photometer data. Furthermore, the auroral roar recorded between the two substorms from $\sim$0425 to 0700 UT sometimes has a very different character at adjacent sites or is detected only at a single site, indicating that several sources are involved during this period. However, the strong auroral roar recorded after 0230 UT which occurs at all five sites and peaks $\sim$7$^\circ$ equatorward of the poleward electrojet boundary, is clearly generated at a single source.

It is not clear whether all of the emissions observed within the auroral zone can be explained by double electrojet structures. Another possibility is that auroral roar sometimes occurs when the aurora has severe longitudinal structure, so that the emission is poleward of its causative arc but equatorward of the average boundary defined by the the magnetometer data. During periods of high activity, such as substorms, current structures can become wrapped up or kinked and located at different latitudes on closely spaced longitudes. The magnetometer array may not resolve such kinks if they are $<\sim$ 10$^\circ$ wide. This explanation is consistent with the observations of periods when the electrojet boundary is moving rapidly poleward, that is, during a substorm, and the peak intensity of emissions appears to lag, as in the November 5 event at 0050 UT when the electrojet boundary moves from 69.9$^\circ$ to 76.2$^\circ$ and the peak intensity remains fixed near Arviat at $\sim$71$^\circ$. This idea does not apply to the May 2 event, where the peak intensity of the emissions is consistently equatorward of the poleward electrojet boundary for many hours.

Kellogg:79 observed that auroral roar emissions were screened when auroral arcs moved directly overhead. The explanation given is that the electron precipitation responsible for the auroral arcs, and most likely for the auroral roar emissions, ionizes the lower layers of the ionosphere ($E$ and $D$), raising the plasma frequency ($f_{pe}$) cutoff in these regions above the frequency of the emissions, preventing them from propagating to the ground. Kellogg:84 speculate that no auroral roar emissions were detected at Cleary, Alaska ($\sim$59 $^\circ \, \Lambda$) because diffuse aurora was always present, screening them from the source region. Further evidence of screening was given by Weatherwax:95 using a meridian scanning photometer colocated with a PSFR and also given by Shepherd:98b using electron density profiles from the incoherent scatter radar at the Sondrestrom Radar Facility.

Evidence of screening can be seen in many of the examples presented in this study. For example, during the October 3 event from 0200 to 0300 UT (Figure 3), auroral roar is detected very strongly at Arviat and Baker Lake, falls off in intensity at Taloyoak, occurs only intermittently at Churchill, the location of the electrojet boundary, and is not detected at Gillam, which is inside the auroral zone. Between 0300 and 0310 UT, emissions are briefly detected at Gillam only after the poleward boundary of the electrojet moves equatorward of the latitude at this station. These data suggest that the auroral roar occur continuously, but the stations at which they are observed are controlled by the screening ionization generated by auroral precipitation.

It is curious that in some cases the most intense auroral roar emission are up to 7$^\circ$ poleward of the auroral zone, inside the polar cap where there are no apparent sources of free energy to generate these waves. In these cases, auroral roar emissions must propagate poleward from lower latitudes to explain the observations. If the wave intensity falls off as the distance from the source increases, one would expect the emissions to peak at the station in the polar cap which is nearest the source, not at a station which is even farther poleward. One possible explanation is that the emissions are not generated isotropically and preferentially propagate at larger angles to the background magnetic field.

Another possibility is suggested by Figure 11a, which shows a relatively narrow ($\sim$2$^\circ$) region of low-energy ($\sim$1 keV) electron precipitation poleward of the most northern auroral arc. These low-energy precipitating electrons do not produce significant conductivity necessary to significantly affect ionospheric currents, nor are they energetic enough to generate auroral roar. However, they may be energetic enough to produce ionization capable of partially screening emissions. Waves generated at the lower latitude inverted-V near Gillam (67$^\circ$) traveling along paths to Churchill (69$^\circ$) pass
through this somewhat tenuous layer and are partially absorbed, while those directed to higher latitudes at Arviat (71$^\circ$) pass over this layer and are unaffected. The result is an apparent peak of emission intensity at the higher latitude of Arviat, 5$^\circ$ poleward of the auroral zone, and weaker emissions at Churchill, 3$^\circ$ poleward.

Partial screening of auroral emissions suggests another possible explanation for why emissions are sometimes detected inside the auroral oval, which is that they may not be entirely screened. During the three periods when auroral roar is detected at Gillam during the May 02, 1997, event (0230, 0415, and 0645 UT), the ground-level magnetic field fluctuations due to the electrojet currents are $<$150 nT, weaker than periods when emission are not seen at Gillam. It is possible that the weaker precipitation-induced ionization associated with this weaker electrojet allows emissions to propagate where they would normally be completely reabsorbed or reflected by the plasma.

In all of the 5 days studied, periods exist during which the peak intensity of auroral roar emissions appears to track with the poleward edge of the electrojet. This effect probably results from a combination of the auroral roar source region moving as the oval expands or contracts and the emissions being screened from sites equatorward of the most poleward region of precipitation. It is also possible that ionospheric cavities, such as those described by Doe:93, which tend to occur poleward of the statistical auroral oval, play a role in auroral roar generation and thus explain why the emissions follow the movement of the poleward boundary Shepherd:98b.

The observations during the five study days presented in this paper are consistent with auroral roar being generated by the precipitating electrons at the latitudes of inverted-V structures. Electrons capable of generating auroral roar are observed at altitudes above the probable source region by the FAST satellite in all eight conjunctions studied. The emissions are free to propagate into the tenuous plasma of the polar cap and are reflected and/or reabsorbed by the precipitation-induced ionospheric layers in the auroral zone. The resulting intensity of emissions detected on the ground appears to peak at or near the latitude of the most poleward inverted-V structure or, on occasions, poleward of that location. Multiple inverted-V structures and dynamic precipitation boundaries can produce conditions such that auroral emissions are seen and peak inside the auroral zone.

It is unknown whether auroral roar is generated over a wide range of latitudes spanning the auroral zone and only emissions from a narrow region near the poleward edge propagate to the ground or whether emissions are only generated in a narrow region near the poleward edge at the most poleward inverted-V structure. A phased antenna array is being deployed at the Sondrestrom Radar Facility near Kangerlussuaq, Greenland, this year by Dartmouth College researchers. The direction finding capabilities of this system will allow the source location to be better determined.

The authors thank M. L. Trimpi for designing the Dartmouth College radio receivers and R. Brittain for software contributions. David St. Jacques, the Churchill Northern Studies Centre, Ralph King, the Canadian Geological Survey, and Rodger Mannilaq maintained the radio receivers at remote sites. E. J. Lund, W. Peria, and C. C. Chaston provided assistance with analyzing the FAST data, and H. al Nashi prepared the CANOPUS border position estimates. This research was supported by National Science Foundation grant ATM-9713119 to Dartmouth College. The CANOPUS array was constructed and is operated by the Canadian Space Agency. The work at the University of Alberta was supported by the Natural Sciences and Engineering Research Council of Canada under grant OGP5420.

Hiroshi Matsumoto thanks N. Sato and another referee for their assistance in evaluating this paper.

next up previous
Next: References Up: Latitudinal dynamics of auroral Previous: February 17, 1998

Simon Shepherd 2002-06-05