Evaluating Observation Strategies
Kryeziu, Orial; Haines, Keith
Reading University, UNITED KINGDOM

In the past, the initialization of coupled ocean-atmosphere models has been performed in uncoupled mode, i.e. the ocean and atmosphere components are initialized separately. Complex interacting physics at the air-sea interface and boundary layers and the large biases associated with the coupled model itself lead to rapid shocks at the atmosphere-ocean interface which may affect the skill of the final forecasts. At Reading University, we are investigating the impact of a more balanced assimilation, tuned to the coupled model system, of near surface ESA satellite data such as sea surface temperature, surface salinity and wind and wave data. The improved assimilation of these variables should provide a better understanding of poorly known quantities such as the air sea fluxes of heat, momentum, freshwater. Moreover, it should allow for a better reconstructions of past changes in the ocean-atmosphere conditions, including wave variability, as well as improved initial conditions for coupled forecasting from the short range up to decadal timescales.
After putting together a record of observational data products, the correlated variability of these surface oceanic and atmospheric data-sets is examined. ESA satellite measurements are used to investigate inconsistencies within the product. For example, regions where there are discrepancies in SSTs between atmosphere and ocean products, such as regions where strong (or weak) SST diurnal cycles are detected but where the surface wind speed from ECMWF's numerical prediction model seems too high (or low). Hence, a set of case study periods are identified for sensitivity testing of separate data assimilation schemes and to allow the testing on new coupled assimilation schemes for SST. Also, work has been carried out to develop a visualization software to display comparisons between ocean and atmosphere reanalysis products using the observation operator codes developed for assimilation and observations.