LOW ORDER MODELS OF STORM TRACK VARIABILITY

The storm tracks in the Earth's atmosphere are the main locus of midlatitude weather systems. Their geometric structure is determined by local surface boundary conditions interacting with the mean flow which leads to local variations in the growth rate of baroclinic instability. Baroclinic instability is the relevant hydrodynamic instability which gives rise to midlatitude weather systems; its growth time scales are set by the vertical shear of the mean wind.

The observed storm track shows variability on a broad range of timescales. The power spectrum of modes of variability is wide with substantial slow variability. Given that baroclinic instability works on relatively fast timescales, of around 3-5 days, it is the non-linear properties of the storm tracks that produce the slow timescales of 10 days or more, up to climate time scales.

The aim of this project is to better understand the underlying non-linear dynamics leading to slow variability. In fact, the response of the storm tracks to slow climatic forcings is one of the key unknowns in future climate predictions, and it is one of the aims of this project to understand better how the underlying nonlinear dynamics of the storm track would react to changing external forcings.

The main paradigm of understanding non-linear scale interactions in the storm tracks revolves around the idea of wave-mean flow interaction. Recent work, building on a long line of relevant progress, has demonstrated how low-order predator-prey models appear to capture important properties of the slow wave-mean flow interactions. There is substantial, but circumstantial evidence to show that the variability of storm tracks is determined by low-order dynamical systems. The derivations of the underlying predator-prey models has been mostly heuristic up to now, and first-principles derivations from the Navier- Stokes equations in the past have not been able to incorporate current understanding of storm-track variability.

The initial approach to our study of the low-order dynamics of the storm track will come from two different sides:

Firstly, using what we have learned from observed predator-prey style behaviour in the storm tracks we will apply model reduction techniques to reduce the full hydrodynamic models to low order versions that include known important dynamical features of the storm track (such as downstream jet latitude shifts).

Secondly, starting from the published predator-prey models, we will attempt to extend those systems in a geophysically relevant way to capture more complex (likely chaotic) dynamics. Candidates for such extensions are: interactions with the humidity field (ther- modynamically the key source of energy in the storm track), jet latitude (a "known unknown" in climate dynamics), interactions with imposed external forcings (periodic, seasonal forcings, or stochastic forcings), but there may be others.

There are several additional avenues available around those two directions and following on from them for approximately the final year of the project. Such avenues include substantial data analysis (where novel phase-space projections have recently been used), and idealised modeling (such models exist and are widely used to address fundamental GFD problems, typically around jet stream formation and variability).