Winter/Spring 2002

By further prescribing time dependent forcing only in the tropics (30^oS-30^oN), we show that the recent upward trend in the NAO is related to tropical forcing. Further specification of the tropical forcing into sub-sectors leads us to believe this upward trend is connected with the effect of diabatic forcing in both the Eastern Pacific and Indian Oceans. High frequency, interannual variability of the NAO, however, seems to be more closely related to extra-tropical forcing.

The model, with the full global diabatic forcing, is also able to reproduce the recent eastward shift of the sea level pressure pattern associated with the NAO. By investigating the behavior of the model using forcing that has been linearly regressed against the NAO index we show that this eastward shift is associated with non-linear dynamical processes.

I.) We demonstrate the dramatic changes in the deep and intermediate water masses of the northwest Atlantic since the late 1950s by compiling time series of water mass properties for the various waters at different locations along the deep western boundary current of the sub-polar gyre. That analysis demonstrated the sustained freshening of the water masses over the past 4 decades. We grouped the data within the time frames representing extreme conditions and mapped volume, temperature and salinity of each water mass of the study and show the sources of this freshening and how it is carried around the sub-polar gyre by the deep western boundary current and then into the interior by the North Atlantic Current.

II.) Labrador Sea Water (LSW) scored more citation than any other water mass formation. However, the published estimates of its spreading time across the North Atlantic were biased by the desire of authors to see LSW there ASAP. Applying a wide range of tools to the most complete collection of the hydrographic data across the North Atlantic, we identify LSW in different basins of the North Atlantic, quantify its volume and property and show that the spreading time of LSW is, at least, twice slower than reported by Sy at el (1997). It takes 1 to 2 years for the bulk of LSW to arrive to the Irminger Sea (0.5 in Sy at el.) and 5 years to the Iceland Basin (2 in Sy at el.). We also suggest that the transport of LSW to the Iceland Basin is mostly associated with the North Atlantic Current and oppose the existence of "shortcut" by which LSW reaches the Iceland Basin directly from the Labrador Sea.

III.) The changes in the stratification of the deep water are of great dynamical consequences. The thickness of the density surfaces associated with the Labrador Sea waters have increased significantly between the 60s and the 90s over the entire sub-polar gyre. These means that the waters in the upper 1.5 km have become denser while those deeper than 2 km have become less dense. These changes will impact the strength and deep structure of the North Atlantic current.

Mesoscale eddies in the ocean mix fluid parcels in a way that is highly constrained by the stratified nature of the fluid:- so much so that much of our intuition about ocean mixing comes from thinking in density coordinates. Temporal-residual-mean (TRM) theory provides the link between the different views that are apparent from averaging turbulent flow in height coordinates and in density coordinates. The TRM theory reduces the parameterization problem from three dimensions to two dimensions and it shows how the divergent part of the relevant eddy density flux is skew-symmetric in height coordinates and that the total advection velocity can be adiabatic. The Gent-McWilliams scheme is best interpreted as a scheme for implementing the temporal residual mean. The TRM theory has very specific things to say about how the tracers in coarse-resolution models should be interpreted and how the skew diffusion streamfunction should approach zero at boundaries. It is emphasized that the extra advection in the Gent-McWilliams scheme the TRM theory is not the bolus transport because the extra TRM advection is non-divergent and diapycnal in character while the bolus velocity is divergent and adiabatic.

This seminar will bring together the results of a program of evaluation of ocean temperature analyses, nowcasts, and hindcasts, with the object of establishing definite guidelines for the next generation of ocean model data assimilation. We begin with a review of the theoretical framework for ocean model data assimilation; and describe three ocean models and one analysis system, providing different applications of the principles. Comparative evaluations of surface and subsurface temperature fields produced by the models and analyses, using observations from moored buoys and bathy profiles, are presented for the Northwest Atlantic.

The evaluations of comparative outputs of model and analysis systems have been made in the following ways. Time series plots of sea surface temperature records from the moored buoys in the Atlantic, are compared with time series plots for collocated, model and analysis output SSTS. Temperature sounding records from bathythermographs deployed in the region by commercial, research, and Coast Guard and Navy ships, are compared with collocated "virtual soundings" created from the multi-level model and analysis output files.

Together, the results of the comparisons show that both the construction of analysis algorithms and the timeliness of the data have significant impacts on product evaluations; and lead to a set of clear guidelines for the next generation of ocean model data assimilation. Primary among these are the importance of using nearest real?time data in a multivariate algorithm based upon covariance structures that represent the spatial relationships of analysis increments as closely as possible.