Through the groundscatter process the Super Dual Auroral Radar (Super‐DARN) has become a powerful tool for studying F region gravity waves. However, the measurement of the gravity wave position is not direct and relies on an assumption relating ground scatter distance to reflection distance. In previous studies it has been assumed that the tilting of the ionospheric reflecting layer was negligible. Hence the gravity wave distance has been calculated as if the reflecting layer was strictly horizontal. Using virtual height data from an ionosonde and ray tracing, we show that this assumption leads to a systematic error of about 16% in the positioning of the ionospheric reflection point, with the error more than 30% on occasion. Using ray tracing, we obtained an improved relation between ionospheric reflection and ground scatter distances. With this improved distance calculation, we have found the direction and velocity for a number of gravity waves. These waves were found to be traveling equatorward, usually, with velocities between 50 and 280 m/s, in agreement with previous gravity wave observations and with the notion of filtering by the thermospheric wind. In some cases the source locations were determined by using gravity wave dispersion. These locations were found to be on the poleward side of the auroral oval during periods of weak, but observable, magnetic disturbance. Our ray‐tracing studies found that the strongest features were due to gravity waves of 3–20 km amplitude.

Measurements in the Mesosphere and Lower Thermosphere (MLT), from the Saskatoon MF (medium frequency) radar (52°N, 107°W), are used for the years 1992 and 1993 to study gravity waves (GW) and their variability. GW-band time series (10–100 min; 2–6 h), their spectra and their 12, 24, 48 h oscillations are used. Fifty days of data from three seasons are selected, which are themselves marked by strong differences in middle atmospheric winds and solar tides. Encouragingly consistent modulations of the GW variances at 12, 24, 48 h periods are noted during the summer months, for both short events (a few days) and for the entire 50 days (for the 2–6 h). The inferred GW propagation directions (toward the NE) are internally consistent with calculations of GW propagation-directions using a new correlation method and consistent with independent measurements of gravity waves discussed in earlier articles from Saskatoon. In this season, all three oscillations are comparable and moderately large in the hourly-mean winds. During the winter, when the 12 h tide is dominant, there is apparent modulation of the GW variances for short events only, despite the size of the tidal wind oscillations (larger than summer). The inferred GW propagation directions are closest to eastward, although the inference is not strong, due to the weakness and variability of the modulations. Surprisingly, in the autumn months, when the 12 h tide is at its annual maximum, the modulation is very weak. However, an event of a few days was identified when some consistency was identified. Spectra from a Lomb–Scargle spectral analysis of the variances for nine years are also used to provide a climatology. There is dominance of peaks near 24, 12 and even 48 h during summer months with a much weaker tendency for peaks to occur in winter and autumn. The existence of peaks at 6 h, although possible when the propagation directions of the GW fluxes are isotropic, is actually a rare event. This also confirms earlier results from Saskatoon, that indicate anisotropy of the GW directions is usually quite strong. The intermittent nature of the GW modulation at tidal periods suggests strongly that the wave sources are intermittent in strength and direction and that the background wind at lower heights also contributes variability.

Polar cap F region digital ionosonde height measurements occasionally show a type of height variation that we call a “downward rippling” variation because of its appearance. The variation typically extends over about 2 hours, and the overall height change is of the order of 100 km. We can simulate similar appearing height variations from vertical perturbations caused by gravity waves if the gravity waves have limited vertical wavelengths (a few hundred kilometers) and if there is high‐speed convective flow.

Radar echoes from ranges less than 500 km are routinely observed by the Super Dual Auroral Radar Network (SuperDARN) on most days. Many of these echoes have properties which are markedly different from what one would expect from E or F region irregularities. We show that these unusual short-range HF echoes are due to scattering off meteor trails. This explains why, among other things, the Doppler shift from the short-range echoes taken from the SuperDARN Saskatoon antenna are consistent with the mesospheric winds observed by the Saskatoon MF radar. This means that the SuperDARN radars can be used to study neutral winds at meteor heights, a result which is especially interesting since it opens up the capability for a global coverage of mesospheric winds using the worldwide distribution of SuperDARN radars.

This study identifies polar cap gravity waves from the perturbations that they produce in ionospheric vertical velocity. In the December 1994 to February 1995 interval, there were 15 wave groups that were defined well enough for us to study their properties. Most of the properties of these waves are similar to the gravity waves that were measured by the DE 2 satellite [Johnson et al., 1995]. Our waves had typical gravity wave speeds (75–225 m s−1) and produced vertical velocity fluctuations >20 m s−1. The typical wave period was 30 min. Source locations were mostly west of Eureka at distances ranging from 300 to > 1500 km. Source times and wave propagation directions were such that the gravity waves would not have been excluded by critical speed effects. Most of the wave groups showed large spatial resonance effects, the average vertical ionospheric displacement being 61 km pp. Spatial resonance occurs when the plasma convection velocity matches the wave speed. Our polar cap gravity waves were also associated with relatively large perturbations in ionospheric convective velocity and magnetic perturbations having typical magnitudes of 300 m s−1 and 20 nT (both peak to peak), respectively. The magnetic perturbations appeared to be due to E region currents driven by the electric fields associated with the convection perturbations.

Fifteen months of daily mean winds, semi-diurnal tides, and gravity waves observed from the Canadian Prairies MF radar network were examined to find correlations in their amplitude fluctuations. In all cases, the correlations of the wave/winds between the three sites were greater than 95% significant. Generally, while the correlations between the various wave/winds types were much weaker, there was evidence of gravity wave/tidal and tidal/mean wind interactions.

A new method of hodograph analysis is introduced and applied to studies of long period wind oscillations (2 to 10‐hour period) in the upper mesosphere. observed by a triangle of medium frequency radars (500 km spacing) in the Canadian Prairies. This is compared with a method of spatial Fourier analysis which is also employed to study these oscillations. The results of these analyses are shown to be consistent with the behavior of atmospheric gravity waves: in particular, the gravity wave propagation directions are found to vary with season and with latitude. However, the directions do not seem to be consistent with blocking by the mean wind field, probably due to the gravity waves having phase velocities much higher than typical mean winds.

A new interferometry analysis of MF mesospheric radar scatter has been attempted. The experiment was run during local noon times for 17 days during the summer of 1989. The analysis yielded consistent values for the amplitude and direction of gravity wave fluxes. Also, the flux magnitudes found are in general agreement with that found in other studies. It is concluded that this interferometry analysis is very promising for gravity wave flux measurements.

Observations of the NW‐SE component of F region convection obtained with a scintillation drift experiment have been compared with power and Doppler velocity measurements of auroral E region coherent backscatter at 50 MHz made with the Bistatic Auroral Radar System (BARS), which is able to observe only at large magnetic aspect angles. It was found that E region backscatter was observed only when the NW–SE component of the F region drift was in the SE direction. This and other observations are shown to be consistent with a recently proposed explanation for these large aspect angle VHF backscatter observations, based on refraction through auroral ionization structures in the E region. In most cases the vector velocity derived from BARS observations had a magnitude substantially below that inferred from the scintillation measurements. Observations during one period were noticeably different from the others, with unusually small Doppler velocities.