Atmospheric Predictability of the Martian Atmosphere: from Low-dimensional Dynamics to Operational Forecasting?

Since the work of Barnes (1980, 1981), it has been evident that Martian mid-latitude meteorology is commonly dominated by large-scale baroclinically unstable transient waves. Viking Lander data showed that such waves are strongly modulated by the seasonal cycle, apparently disappearing altogether during northern summer. During seasons around northern winter, however, they are typically very active and dominated by relatively low wavenumbers (m = 1-3). Unlike their terrestrial counterparts, however, Barnes (1980, 1981) found that Martian baroclinic transients were typically highly regular in their behaviour in time, with near-periodic signals at the VIking Lander sites persisting for months and with temporal frequency spectra (see Fig. 1) exhibiting strong peaks at periods of 2-8 days. The impression was therefore that Mars’s atmosphere is typically more coherent, and therefore perhaps intrinsically more predictable, than that of the Earth. Subsequent modelling and observations have largely confirmed this impression. In particular, Collins et al. (1996) showed in simple model experiments that, in the absence of thermal tidal forcing, baroclinic waves in an otherwise reasonably realistic circulation would develop into extremely regular, spatially monochromatic wave patterns progressively moving around the winter pole, with occasional jumps in wavenumber as the seasonal cycle advanced. Comparison of successive years of such simulations even gave the impression of hysteretic behaviour in wavenumber transitions, much as found in the regular regime of laboratory experiments on baroclinic waves in rotating, cylindrical containers (e.g. Hide and Mason (1975)). Such extremely regular behaviour in the non-diurnal Mars general circulation model (GCM) simulations was significantly more coherent than the observations of Barnes (1980, 1981), although more realistic episodic variability was restored when simulations were re-run with the diurnal cycle included. Such results suggest that much of Mars’ synoptic variability may owe much to an interaction between a relatively small number of freely evolving, global, baroclinically unstable atmospheric modes and the periodically-forced diurnal tide. The simple spatial structures associated with such a simple spectrum of near-normal modes of the atmosphere with a periodic forcing would strongly suggest that synoptic-scale meteorology on Mars may be governed by a relatively low-dimensional attractor. This is in significant contrast to the Earth, for which the likely attractor dimension is probably quite high. The consequences for atmospheric predictability in a regime dominated by a low-dimensional attractor (chaotic or not) could be profound, with correspondingly important implications for the practical ability of atmospheric models to forecast Martian meteorology from a given initial state. In this presentation, we will review various studies exploring the possible nature of the low-dimensional Martian meteorological ‘attractor’ in simplified models, and recent attempts to characterize more realistic atmospheric circulation systems in Mars GCMs and assimilated data (see Lewis et al., this volume). Most recently, attempts have been made by Newman et al. (2004) to determine quantitatively the growth of perturbations in realistic Martian circulation patterns, based on the ‘breeding vector’ method (Kalnay, 2004). This leads on naturally to the possibility of ensemble forecasting experiments, and the presentation will conclude by giving an overview of recent results and future prospects of making detailed synoptic forecasts of Martian meteorology.