Tidal influence in the determination of long-term trends in the mesosphere-lower thermosphere from LIDAR observations

Abstract Atmospheric temperature has both variations with time constants of order ½ year or longer, such as trends, annual and semiannual oscillation as well as variations with shorter time constants such as two-day waves, diurnal and semidiurnal tides, and gravity waves. Long-term amplitudes, such as trends and solar flux variations, have been determined from the time series of observed sodium lidar temperatures averaged over a night or a fixed 2-h interval by ignoring the short time-scale variations. Since during the twenty-eight years of the CSU/USU sodium lidar experiments diurnal and semi-diurnal temperature tides could be as much as 9 K and 18 K, respectively, it is particularly important to estimate their effects. We accept the Climatological Tidal Model of the Thermosphere (CTMT) and show that, for linear models, the tidal shift of trends depends upon (1) the times of observations, (2) averages of the short-time variations, and (3) the basis functions of the long-time variations, such as t, sin(2πt/1yr) and solar flux, but (4) not on observed temperatures. This allows planning experimental campaigns and analyses to reduce the tidal effects. We compare the tidal shifts of temperature trends and tidal shifts in the annual and semiannual amplitudes for CSU/USU experiments using 2-h averages (_2 MN) and nightly averages (_Ngt) using data from all times during the year and those restricted to different seasons. For example, the maximum tidal shifts in trends in temperatures, winds, and air density are predicted to occur at altitudes around 100 km. For the CSU/USU experiments the tidal shifts of temperature trends vary between 0 and 1.5 K/decade depending on altitude and annual/seasonal data used; between 85 and 100 km, they are much smaller for 2-h means (_2 MN) than for nightly means (_Ngt).