DIAGNOSTICS OF WATER FORMATION DURING OCEAN DATA ASSIMILATION

It has long been a goal to use data assimilatin to confront physical models with data sets in such a way as to learn about model failings, as well as to learn about climatic trends in the time evolving ocean state.

In the ocean the timescales for changes to water properties such as potential temperature, salinity and dissolved tracers, are long when the water is out of contact with the ocean surface. This should allow data assimilation to be used over long periods to accumulate information from sparse ocean data sets. In this presentation results from the global high resolution OCCAM model will be shown. Altimeter sea level data and Temperature profile data have been assimilated into the model over a period of several years during the 1990s and ocean profiles have been compared with independent data from WOCE cruises, Fox et al. (2000). The assimilation methods used are designed to preserve the ocean water properties and volumes as far as possible. The Cooper and Haines (1996) assimilation method for altimeter data uses a vertical displacement of water columns to avoid changing water properties or volumes. The Troccoli and Haines (1999) assimilation method for T profile data, allows water volumes to change but preserves the T/S properties of the waters (given that salinity data is not available).

New diagnostics are introduced in which water transformations brought about by the data assimilation are analysed and interpreted using the methods developed by Walin (1982) and others. The role of assimilation in maintaining N Atlantic mode water volume against model dissipation and outflow across the Equator is used as an example. By careful treatment of water properties and the volume of different water masses during assimilation it should be possible in future to extract information about changes in surface forcing over time, and hence about climate change, from sparse oceanographic data sets.