Fall 2003

The evolution of the mixed patch formed after a deep convection event is studied for the winters of three years: 1987, 1992, and 1999. It is demonstrated that the mesoscale may play a major role for the spreading of the Western Mediterranean water masses out of the Gulf of Lions and towards the Gibraltar. Equivalently we may argue that the intermediate and deep waters conveyor belt of the WMED is eddy driven.

Winds in this area are equatorward throughout the year, and weak in SBC due to sheltering by a coastal mountain range. In spring, the near-surface currents are equatorward everywhere except along the northern shelf of SBC. In summer this poleward flow grows stronger, while the flow at SBCs eastern mouth turns poleward too. SMB currents are only intermittently poleward. SVD analysis of currents and wind suggests that large-scale changes in winds and currents are related, as are poleward flow along the northern shelf of SBC and a sharp wind stress gradient across SBC.

A three-dimensional numerical ocean circulation model is used to simulate the 1998 flow. The model is forced by a daily wind stress pattern derived from buoy winds. The model is run with and without assimilation of observed temperatures. The no-assimilation case tends to have colder water temperatures, which implies that surface heat flux (which is not included in the model, but whose effects can be incorporated into the model by temperature assimilation) is important. The surface heat flux estimated from the model is most similar to the observation-derived values of Winant and Dorman [1997] in summer, and least similar in winter, possibly due to the 1997-1998 ENSO event. They are more similar in SMB than in SBC, which suggests that the model transport is more realistic in the former. In EOF analysis of the Coriolis and pressure gradient terms for modeled near-surface currents, the first mode suggests that the non-geostrophic part of the flow is driven by wind stress. In the second mode, the non-geostrophic term has a relatively large magnitude at the northern shelf of SBC, which is not balanced by wind stress.

The modeled and observed currents are similar in direction, but the former tends to have smaller annual mean and variance values. SVD analysis identifies modes of variability between the two that have similar spatial patterns and temporal evolutions. Thus, in addition to the annual means and variances, the modeled and observed flows are also comparable in terms of their variabilities.

In my previous POMSS presentation, I talked about the nature of the eastward shift in the spatial pattern of the interannual variability of the NAO, as unraveled by using an atmospheric model with dry dynamics driven by diabatic forcing diagnosed from observations. The model results reveal the nonlinear dependence of the spatial pattern of the NAO on the NAO index, the pattern being shifted to the east (west) for high (low) NAO index. General agreement is found between the model and observations. Therefore, it can be argued that the recent eastward shift in the NAO pattern is a consequence of the trend towards higher NAO index during the last several decades of the 20th century.

This time, I am going to be focusing on the upward trend of the NAO, as well as the trend in the troposphere over the whole Northern Hemisphere. The hemispheric trend is associated with a deepening of both the Aleutian and Icelandic lows, a pattern that bears close resemblence to the "Cold Ocean Warm Land" pattern with a positive projection on the NAO (PNA) over the Atlantic (Pacific) sector. Experimentation using the same Hall model show that the observed simultaneous deepening trend in both (Aleutian and Icelandic) lows can be largely attributed to the diabatic forcing from the tropics. While the contribution from the extratropical forcing is rather trivial. The mechanism by which the tropical forcing drives the extratropical circulation trend is a planetary wave train emanating from the Indo-Pacific region of the tropical oceans. In the model, the extra-tropical storm tracks also play an important role in amplifying the original wave signal over the North Atlantic sector, resulting a resonant NAO-like dipolar response there. These model results underscore the importance of studying the tropical diabatic processes and the associated teleconnection in understanding the recent climate change during the last half century.

I will discuss implicit and explicit filtering in LES models, but the focus of the talk will be the ADM and how I hope to extend this model. The original ADM was implemented using a spectral solver while Gullbrand has recently adapted the ADM for staggered, finite difference grids. I will highlight my efforts to extend the ADM to a collocated, finite volume formulation.

The ADM solver I am developing uses an unstructured Cartesian grid with anisotropic adaptation developed by Ham et al. An adaptive grid is very attractive when solving turbulent flows; adaptive grids refine only where necessary, significantly reducing the number of cells for a given accuracy.

Early stages of the project have involved testing the FLAME (Family of Linked Atlantic Model Experiments) model using prescribed windstress forcing that mimics the lifecycle of hurricanes undergoing ET. The FLAME model is a 1/3 deg x 1/3 deg (lat/lon) z-level, eddy-permitting model of the North Atlantic Ocean. The windstress specification is that of a Rankine vortex. Two simulations have been run for the cases of hurricanes Fabian and Juan of the 2003 hurricane season. The vortex is fit to data from the National Hurricane Center and is moved along the final storm track.

Preliminary results for ET-like wind forcing using an ocean - which has been spun-up by climatological windstress forcing - show the typical right-of-track cooling in SST in the wake of a moving hurricane. The magnitude and breadth of the cooling in the hurricane Fabian simulation are much too large in the original run owing to unrealistically shallow mixed layers in the model. While work was being done to solve the mixed layer representation, we ran a simulation of Hurricane Juan using late-September climatology for the upper ocean over the Scotian Shelf - the area traversed by Hurricane Juan. The location and magnitude of ocean cooling agreed well in overall structure when compared with actual SST data. This talk will highlight data and results from this event, which lend confidence to the model.s mixing scheme.