2. Method and results

The 2.78 MHz medium frequency radar MFR) at Ramfjordmoen (70°N, 19°E, Hall, Reference Hall2001) transmits through a 4x4 crossed dipole array, and therefore illuminates structures in refractive index, dominated above the radar by corresponding structures in electron density (i.e. the ionospheric D-region). The geographic location is beneath the auroral oval and therefore a degree of ionisation extending to as low as 70 km altitude is not uncommon. Reflections from these structures give rise to an interference pattern on the ground which is subsequently recorded by simple equi-spaced inverted-V antennas connected to the receivers. Cross-correlations between signals from these receivers provide information on horizontal winds that cause the reflecting structures to drift through the radar beam. Autocorrelations indicate backscattered power. The backscattered power, however, fades with time as the scatterers change shape and/or become less distinct as they drift through the radar beam. Several mechanisms might cause this fading, but a common prime one is considered to be neutral turbulence (e.g. Briggs, Reference Briggs1980). Thus, by determining the characteristic fading times of the autocorrelations, it is possible to estimate turbulent intensity - or in fact an upper limit for it (since other mechanisms may be present) – which is parametrised by the energy dissipation rate, ε´ (the prime being used to differentiate between this determination and the actual dissipation rate, ε). At the same time, kinematic viscosity increases with altitude due to decreasing density, and at a certain height, energy dissipation due to viscosity, ε_min, equals that due to neutral turbulence; this can be thought of when inertial forces equal viscous forces, i.e. when the Reynolds number R_e is unity. This condition, the intersection of the ε_min, and ε, profiles defines the turbopause. The reader is referred to Fukao et al. (Reference Fukao, Yamanaka, Ao, Hocking, Sato, Yamamoto, Nakamura, Tsuda and Kato1994), Hall et al. (Reference Hall, Holmen, Meek, Manson and Nozawa2016), Weinstock (Reference Weinstock1984) and references therein for an in-depth description of turbulent energy dissipation rate estimation and thereafter turbopause identification.

The results are shown in Figure 1. Turbulent intensity is indicated below the turbopause in the vicinity of 105 km. The distinct seasonal variation is due to gravity wave saturation and breaking being more evenly distributed though the mesosphere in winter, whereas breaking is more concentrated immediately above the mesopause in summer when temperature gradients are steep. Again the atmospheric physics is fully described in references to be found in Hall et al. (Reference Hall, Holmen, Meek, Manson and Nozawa2016). Figure 2. Illustrates the turbopause altitude alone, here with a 30-day smoothing applied to enhance clarity. Earlier analyses of this dataset had identified possible trends; two possible candidates, identified by simple inspection are highlighted (1996–2001 and 2001–2005), but if we consider the entire series from 1996 to 2021, there is no convincing trend to be seen, although the determined value of 0.68 km decade⁻¹ does indeed exceed the 2-σ uncertainty. For the individual segments and the entire series, the 95% confidence limits (Working & Hotelling, Reference Working and Hotelling1929) are indicated by dotted hyperbolae.

Figure 1. Turbulent intensity from Northern Norway (70°N, 19°E) (upper limits – see main text). Thick line shows turbopause.

Figure. 2. Turbopause estimate over Northern Norway (70°N, 19°E). Purely from visual evaluation, two periods of change are identified: 1996–2001 and 2001 to 2005, and trendlines are fitted to these and to the entire period. Trends are indicated in the annotation on the plot and dotted hyperbolae indicate the 95% confidence limits.