A Hybrid Global Ocean Data Assimilation System at UMD and NCEP

The talk discusses implementation of the Hybrid Local Ensemble Transform Kalman Filter (Hybrid-LETKF) at the National Centers for Environmental Prediction (NCEP) within the Global Ocean Data Assimilation System (GODAS) and at University of Maryland within the Simple Ocean Data Assimilation (SODA). This hybrid ocean data assimilation system is designed to accompany developments of hybrid atmosphere assimilation systems. Here we present a comparison with current reanalysis approaches through a set of observing system simulation experiments. These experiments have been carried out using the historical observation sampling applied to an ocean simulation for the eight year period 8-year period 1991-1998, with meteorological errors determined from the NOAA 20th Century Reanalysis surface forcing.

Relative to the current operational 3DVar-GODAS, Hybrid-GODAS reduces analysis errors, in some cases dramatically. Some derived variables such as sea surface height are also examined and they show improvement, which is encouraging for our ongoing development of sea level assimilation. The reasons for the improvement will be discussed, which seem to be related to improvements in spatially and temporally varying background error covariances, and results from experiments with real data should also be available to show.