The colocation of the Canadian Auroral Network for the OPEN Program Unified Study (CANOPUS) magnetometers and the Dartmouth PSFRs allows the relationship between auroral currents and auroral roar emissions to be studied continuously. For this study we selected 5 days (3 days in 1995 and one each in 1997 and 1998) during which all five PSFRs were operational, auroral roar was seen at several stations, and the emissions either displayed significant time and latitude variations or occurred during a FAST conjunction. The range in magnetic local time (MLT) that is included in this study for each of the five selected days is shown in Figure 1.
The solid line in Figures 3f, 4f, 5f, 6f, and 7f displays the latitude of the peak emission intensity as measured by integrating the PSFR power over a range of frequencies which includes the auroral roar emission but excludes certain frequencies which are dominated by anthropogenic signals or interference lines. The invariant latitude of each PSFR site is indicated on the right side of Figures 3f, 4f, 5f, 6f, and 7f by the appropriate site abbreviation. The latitude of the peak emission intensity is determined by setting a threshold at the measured power level of the site with the highest or second highest noise level (see discussion below) and discerning the site with the largest emission power above this threshold. Gaps in the line indicate times when no emissions occur above the threshold level, and vertical bars indicate transitions when the peak intensity changes rapidly from one site to another. The coarse spacing of the PSFRs is evident in the large jumps in latitude of this line, but in reality the latitude variations are probably much smoother.
The electrojet boundaries, determined every 5 min from magnetometer data using the method described above, are also shown in Figures 3f, 4f, 5f, 6f, and 7f as dotted lines. Along the bottom of Figures 3-7 a grayscale strip represents the strength of the electrojet currents with darker gray indicating stronger currents. The two grayscale colorbars above Figures 3a, 4a, 5a, 6a, and 7a indicate the intensity of auroral roar emissions in Figures 3a-3e, 4a-4e, 5a-5e, 6a-6e, and 7a-7e and the strength of the electrojet at the bottom of Figures 3f, 4f, 5f, 6f, and 7f, respectively.
The large variation in the intensity of the background noise detected at the five sites requires that the signals be compared carefully in order to assure that the latitude of maximum wave intensity is not determined by the background noise at any station and that the site of maximum wave intensity is not identified in cases where another site, with higher background noise, could reasonably have harbored more intense but undetectable emission signals. For two days, May 2, 1997, and February 17, 1998, the threshold level for determining the latitude of maximum auroral roar emission was set at the background level of the Churchill receiver (4.5 10 V/m), which during this time had the least sensitivity because of the installation of a smaller antenna there in 1997-1998 for polarization measurements [see]Shepherd:97a. For the other 3 days the threshold level was set at the background level of the Arviat receiver (2.0 10 V/m), in effect ignoring the data from the Gillam receiver at the southern extreme of the chain, which had a significantly higher background noise level, except for a short time during the October 4, 1995, event when the emissions are clearly most intense at Gillam. By choosing the lower threshold and ignoring the Gillam data when the auroral roar activity was clearly centered at the poleward stations, a much clearer picture of the latitude variations of the emissions is obtained than would be possible if the only emissions studied were those that exceeded the rather high noise level of the Gillam receiver. In all 5 days some data points for the solid line, representing the latitude of the peak emissions, have been removed to eliminate spurious peaks due to intermittent anthropogenic signals, atmospherics, or other radio noise.