Comprehensive model simulation of thermal tides in the Martian atmosphere

This paper discusses the thermotidal oscillations in simulations performed with a newly developed comprehensive general circulation model of the Martian atmosphere. With reasonable assumptions about the effective thermal inertia of the planetary surface and about the distribution of radiatively active atmospheric aerosol, the model produces both realistic zonal-mean temperature distributions and a diurnal surface pressure oscillation of at least roughly realistic amplitude. With any reasonable aerosol distribution, the simulated diurnal pressure oscillation has a very strong zonal variation, in particular a very pronounced zonal wavenumber-2 modulation. This results from a combination of the prominent wave-2 component in the important boundary forcings (topography and surface thermal inertia) and from the fact that the eastward-propagating zonal wave-1 Kelvin normal mode has a period near 1 sol (a Martian mean solar day of 88 775 s). The importance of global resonance is explicitly demonstrated with a series of calculations in which the global mean temperature is arbitrarily altered. The resonant enhancement of the diurnal wave-1 Kelvin mode is predicted to be strongest in the northern summer season. In the model simulations there is also a strong contribution to the semidiurnal tide from a near-resonant eastward-propagating wave-2 Kelvin mode. It is shown that this is significantly forced by a nonlinear steepening of the diurnal Kelvin wave. The daily variations of near-surface winds in the model are also examined. The results show that the daily march of wind at any location depends strongly on the topography, even on the smallest horizontal scales resolved in the model (similar to few hundred km). The global tides also play an important role in determining the near-surface winds, especially so in very dusty atmospheric conditions. The results for the diurnal and semidiurnal surface pressure oscillations in seasonal integrations of the model are compared in detail with the observations at the two Viking Lander sites (22 degrees N and 48 degrees N). The observations over much of the year can be reasonably reproduced in simulations with a globally uniform aerosol mixing ratio (and assuming more total aerosol in the northern winter season, when the largest dust storms are generally observed). There are features of the Viking observations that do not seem to be explainable in this way, however. In particular, in early northern summer, the model predicts amplitudes for the diurnal pressure oscillation at both lander sites that are at least a factor of 2 larger than observed. Results are presented showing that the low amplitudes observed could be explained in the dust distribution tended to be concentrated over the highlands, rather than being uniformly mixed. Annual cycle simulations with a version of the model with an interactive dust transport do in fact reveal the tendency of the circulation to organize so that larger dust mixing ratios occur over highlands, particularly near subsolar latitudes. When the model includes globally uniform surface dust injection and parameterized dust sedimentation, the annual cycle of the diurnal and semidiurnal tides at both lander sites can be rather well reproduced, except for the periods of global dust storms. The attempts to simulate the observed rapid evolution of the tidal pressure oscillations during the onset of a global dust storm also demonstrate the importance of a nonuniform dust concentration. Simulations with the version of the model incorporating interactive dust are able to roughly reproduce the Viking observations when a strong zonally uniform dust injection is prescribed in the Southern Hemisphere Tropics and subtropics.