Jake Badger P.I.s Brian Hoskins, Mike Blackburn, Paul Valdes
Understanding the basic processes important to storm tracks is the principle aim of this project. Storm tracks are regions of high synoptic activity. In the northern hemisphere these are located in the Atlantic and Pacific ocean basins. Knowledge of what fundamental dynamical mechanisms are responsible for such organization of activity, including the seeding of synoptic events, is required. Recent idealized work has shown that disturbances with structure, confined in the horizontal and vertical, and located in the low troposphere are most efficient in leading to the formation of deep sustained midlatitude systems. This can be understood in terms of the shear leading to the unshielding of the PV distribution and the growth by untilting (Venetian blind mechanism) with an associated upward propagation. An idealized storm track model set-up, such as that developed by Valdes and Hoskins, will be used to determine the characteristics of seeding events and the impact of physical parameterizations on the seeding mechanisms.
Experiments will be performed to investigate the effect of various parameterizations schemes of radiation, surface fluxes, turbulence and moist convection on the model storm evolutions. Introducing localized variations of friction, moisture and sea surface temperature should indicate the relative importance of these processes in organizing the storm tracks. The basic understanding sought is highly relevant to the North Atlantic Oscillation, future European climate change, European seasonal forecasting and palaeoclimate studies.
In tandem, the tool used to achieve this scientific objective, a climate circulation model is being developed in collaboration with others at Reading. This model is an intermediate general circulation model (IGCM), so called because it lies in terms of complexity and sophistication somewhere between idealized dynamic models and full general circulation models. It will include modular code for various parameterizations of radiation, convection and surfaces processes. The model will be executable on parallel machines, facilitating very fast high resolution integration, and also be very portable between different machines, thus making it available to a large base of users.
Motiviation: Mid-latitude weather systems are often considered to be the result of the instabilities of a basic westerly wind increasing with height. However their growth can far exceed the growth rates given by linear unstable modes. Forecast centres are also very concerned to know what sort of errors in their initial data could grow most rapidly. Here the aim is to examine the mechanisms for rapid growth in idealized models of the atmosphere.
Results: In a 2-D Eady set-up ( a westerly wind linearly increasing with height in a uniform atmosphere bounded above and below by horizontal boundaries) linearized numerical model, simple vertically confined initial perturbations can exhibit very rapid growth, Fig. 1. The peak kinetic energy growth rate increases as the vertical confinement increases and as the horizontal scale decreases, Fig. 2.
Figure 1: Perturbation meridional
wind at day 0 (upper plot) and day 2 (lower plot) for a vertically
confined perturbation; length scale, L=4000km and half-width height,
H_w=2km.
Red contours indicate northerly wind and blue contours indicate
southerly wind. The contour interval is 0.5ms^{-1} for both
plots.
Figure 2: Growth rate for
differently configured vertically confined
perturbations. The control is a perturbation of the form L=4000km:
H_w=4km. The appropriate growth rate for the fastest growing normal
mode is also shown.
The maximum kinetic energy growth rate can for a transient period exceed that of the fastest growing normal mode without exploiting exaggerated vertical tilt. The rapid initial growth is maximized by minimizing the involvement of the boundaries; the initial growth takes place even if the separation of the boundaries is increased by an arbitrary amount.
This can be understood in terms of potential vorticity unshielding. In the Eady set-up interior quasigeostrophic pseudo-potential vorticity, q', is simply advected by the basic shear flow. The initial distribution of q' is such that the associated flow is confined, Figs. 3 and 4, due to the shielding effect of the regions of -q' above and below the central region of +q'.
Figure 3: Perturbation quasigeostrophic
pseudo-potential vorticity, q', (solid line) and perturbation
meridional velocity, v', (dotted line) at day 0
for the vertically confined perturbation in Fig. 1.
The contour interval for q' and v' is
10^{-5}s^{-1} and 10^{-1}ms^{-1} respectively. Red indicates
+q',+v'; blue -q',-v'.
Figure 4: As in Fig. 3 but at 0.5 days.
The basic flow dismantles this configuration, reducing the shielding, giving rise to flow away from the initial perturbation. Figure 5 shows how the difference in flow can be attributed to the difference in the potential vorticity.
Figure 5: The difference in q' and v' between day 0.5 and day 0.
Perturbations more confined vertically and horizontally will have larger magnitude +q' and -q' values for the same initial kinetic energy. Therefore the effect of unshielding and hence the growth rate for these perturbations is greater.
The inclusion of meridional potential vorticity gradient, a step towards realism, slightly reduces the initial growth. The principle impact is to allow propagation of interior Rossby waves. Upward propagation is associated with growth of the waves; downward propagation with decay. Such asymmetry plays a crucial role in determining the low level as the optimal location of perturbations for significant growth after the initial growth shown here.
Conclusions: Rapid initial growth is maximized by increasing the vertical and horizontal confinement of the perturbation and minimizing the involvement of the boundaries. Growth after this initial growth period requires coupling of upper and lower boundary waves. With the inclusion of meridional gradient of potential vorticity, the relative ease of upward propagation of disturbance compared to downward propagation means that, for efficient excitation of both boundaries, perturbations are best sited in the lower troposphere. These characteristics are in agreement with forecast centre optimal perturbations arrived at by sophisticated singular vector analysis.
EGS Conference 1998 abstract
Return to DYNAMICAL PROCESSES home page.