Introduction

Auroral roar is a relatively narrow-band radio-wave emission ( $\delta f/f < 0.1$) generated in the auroral ionosphere at frequencies near 2 and 3 times the electron cyclotron frequency ($f_{ce}$) [Kellogg and Monson, 1979, 1984; Weatherwax et. al., 1993, 1995]. Recent evidence strongly suggests that the 2$f_{ce}$ emissions originate where the wave frequency $f$ matches 2$f_{ce}$ in the F region of the ionosphere [Hughes and LaBelle, 1998]. Seasonal, diurnal, and spectral characteristics of auroral roar have been documented [LaBelle and Weatherwax, 1992; Weatherwax et. al., 1993, 1995; Weatherwax, 1994; LaBelle et al., 1995; Shepherd et al., 1998; Hughes and LaBelle, 1998], and several theories of generation mechanisms have been presented [Weatherwax, 1994; Weatherwax et. al., 1995; Yoon et al., 1996, 1998; Willes et al., 1998]. Simultaneous measurement of ionospheric structure provides a test of such generation mechanisms, yet no electron density measurements have previously been reported during auroral roar observations. An earlier attempt using ionosonde data in Alaska proved unfruitful because the ionograms were too disturbed at times of auroral emissions [Weatherwax, 1994, p. 48].

Several theories of auroral roar generation require the presence of localized F region electron density cavities of the type first described by Doe et al. [1993] to provide regions for enhanced wave growth resulting from reflections at the cavity boundaries, to allow wave mode conversion on the density gradients of the cavity or from coalescence of reflecting waves [Yoon et al., 1998; Willes et al., 1998], and to provide for ducting of O or X mode waves to the ground [Yoon et al., 1996]. The presence or absence of these $N_e$ cavity structures during roar observations can be determined with the electron density measurements from the ISR. Other theories predict that auroral emissions occur when the upper hybrid frequency (defined by $f_{uh}^2$ = $f_{ce}^2$ + $f_{pe}^2$) is equal to a harmonic of the electron cyclotron frequency ($f_{ce}$) [e.g., Kaufmann, 1980; Gough and Urban, 1983; Weatherwax et al., 1995]. Using ISR density measurements, this upper hybrid matching condition can be tested in the F region of the ionosphere during observations of auroral roar. The ISR electron density measurements provide the opportunity to determine whether these theories can still be considered viable candidates for explaining auroral roar.

In 1995, Dartmouth researchers installed a programmable stepped frequency receiver (PSFR) at the Sondrestrom radar facility ($66.987^\circ$ N latitude, $309.051^\circ$ E longitude, 73.3 invariant). By operating the PSFR on a continuous basis, we achieved significant overlap with ongoing ISR operations. A number of coincidences did occur during auroral roar events and are analyzed in the following sections. Section 2 presents the data for these periods, and sections 2.1 and 2.2 address ionospheric structure features and the upper hybrid matching condition, respectively.