P3.2
Low-frequency oscilllations in the atmosphere induced by a mid-latitude SST front
Michael Ghil, University of California, Los Angeles, CA
Mid-oceanic thermal fronts, such as the Gulf Stream (GS) and Kuroshio Extension, are permanent features of the mid-latitude ocean circulation. What impact do they have on low-frequency atmospheric oscillations? To answer this question, one must separate the effects on the atmosphere due to a purely steady oceanic front and those due directly to the variability of that front. Warner et al. [1] tested the sensitivity of the atmospheric marine boundary layer (AMBL) to sea surface temperature (SST) using a smooth and a high-resolution version of the SST forcing. Their AMBL develops a front near the GS's north wall that is much stronger with the high-resolution forcing. In both simulations, responses to the surface forcing extended upward to about 1200 m. Feliks et al. [2,3] studied the long-term behavior of the free atmosphere as a function of SST anomalies. They used a hierarchy of models: from a highly idealized, linear AMBL to a quasi-geostrophic free atmosphere, first barotropic [2] and then baroclinic [3]. They linked the SST with the vertical wind w at the top He of the troposphere. Over cold SSTs, air descends and there is a cyclonic flow in the free atmosphere, while over warm SSTs, air ascends and the free-atmospheric flow is anticyclonic. Their SST front spins up an eastward jet in the free atmosphere. Three kinds of unstable oscillatory modes are obtained: one symmetric due to barotropic instability, with a period of 30 days; one antisymmetric due to baroclinic instability, with a period of 6–8 months, and one with a northward-propagating mode; the latter has an antisymmetric and a symmetric component, and a period of 2–3 months. These effects depend on the atmospheric model's having a sufficiently high resolution of at least 50 km x 50 km. Very recently, Minobe et al. [4] used satellite observations of rain and clouds, operational atmospheric analyses and a high-resolution GCM to illustrate the impact of the GS on the upper troposphere. The narrow rain band above the part of the GS that is warmer than 16°C disappears when the SST is smoothed in the model. These authors explain empirically the mechanism by which the SST Laplacian acts on the wind convergence via the effect of the former on the Laplacian of sea level pressure (SLP) [5]. The previous work [1–3] showed that high resolution was crucial in simulating the effect of an SST front on the atmosphere. We therefore choose to study this effect in a global circulation model (GCM), LMD-Z, which allows a high-resolution zoom over a selected area; hence the ‘Z' in LMD-Z. The GCM's dynamics corresponds to mainly horizontal exchanges on a 3D-grid, while its physics is a juxtaposition of atmospheric columns without interactions [6]. The model has 19 levels and the horizontal resolution we use is of 3° outside the zoomed area and of 0.5° inside it; this zoom allows us to resolve correctly the effects of the GS front. The zoomed area is (20° lat. x 40° long.) and it is centered at (65°W, 40°N). In order to filter out the seasonal signal, the model is forced with a perpetual day, the 15th of February. The simulations are 872 days long and we study the time average over the entire period. Two simulations are performed: a control simulation with the climatological SST field, and one where the SST front has been modified. To have a stronger gradient, we add a theoretical SST centered at (60°W, 40°N) and the following formulation: f(T)= 2cos(x)*(-8)sin(y). The GS has an axis inclination of 25°. The Figure Vertical wind anomalies w' in (Pa s–1). The altitude units are 10² mb. We study the anomalies that correspond to the differences between the GS simulation and the control simulation. We notice a strong near-surface southward flow for the mean meridional wind anomalies v' of 2 m/s above the SST front, between the isotherms of 288 and 294°K (not shown). Figure 1 displays an average of the vertical wind anomalies over the longitude band (75°W, 40°W), between the latitudes 35°N and 47°N, for the 19 model levels. An upward flow can be seen on the warm side of the front (blue shading), with values that reach 0.02 Pa/s, and a downward flow over the cold side (red shading), with values that reach 0.01 Pa/s. These slanting, predominantly vertical flows develop at the top of the AMBL, as expected by [2,3] and [4]. An asymmetry with altitude is found over the SST front, as in [4], with a stronger ascendant than descendant flow. We note also the northward slope of the vertical wind with altitude. We found encouraging results that are similar to those predicted by the idealized models of [2,3] and encountered in the observations of [4]. However, it is difficult to find clear oscillation periods, as in [2,3], because of the complexity of the GCM flow fields. For a better understanding, we need to separate the theoretically expected results from those due to model drifts. In future work, we will change the perpetual day that could induce an accumulation of snow in different regions and so could be at the origin of a temperature drift in the model. In parallel, we will carry out similar studies with a GCM model that is simplified in terms of both dynamics and physics (PUMA [7]), to bridge the gap between the full LMD-Z and the idealized models of [2,3].
Poster Session 3, POSTERS: Large-Scale Studies
Tuesday, 13 January 2009, 9:45 AM-11:00 AM, Hall 5
Previous paper Next paper
Extended Abstract (708K)