Workshop Abstracts

Previous Abstract | Back to Abstracts Page | Next Abstract

An eddy resolving tidal-driven model of the South China Sea assimilating along-track SLA data using the EnOI

Francois Counillon, Jiping Xie, Laurent Bertino
Mohn Sverdrup Center/NERSC
(Abstract received 12/17/2010 for session X)

The upper ocean circulation in the South China Sea (SCS) is driven by the Asian monsoon, the Kuroshio intrusion through the Luzon Strait, strong tidal currents, and a complex topography. Here, we demonstrate the benefit of assimilating along-track altimeter data into a nested configuration of the HYbrid Coordinate Ocean Model that includes tides. Including tides in models are important because they interact with the main circulation. However, assimilation of altimetry data into a model including tides is challenging because tides and mesoscales features contribute to the elevation of ocean surface at different time scales and require different corrections. To address this issue, tides are filtered out of the model output and only the mesoscale variability is corrected with a computationally cheap data assimilation method: the Ensemble Optimal Interpolation (EnOI). The EnOI uses a static ensemble and allows for multivariate updates. The data assimilative system is tested for the period 1994-1995, during which time a large number of validation data are available. Data assimilation reduces the Root Mean Square Error of Sea Level Anomalies from 9.3 cm to 6.9 cm and improves the representation of the mesoscale features. With respect to the vertical temperature profiles, the data assimilation scheme improves the results at intermediate depth, but a slight degradation of the results at the surface is noted. The comparison to surface drifters shows an improvement of surface current by approximately 8.3%, with largest improvements in the northern SCS and east of Vietnam. In addition, some new features to the original EnOI scheme are proposed: a static ensemble composed of a running selection of members to handle the strong seasonal variability and a formulation for the ageing of the observation error that is dependent on space and time. These two features lead to slight improvements of the multivariate properties of the method.

Previous Abstract | Back to Abstracts Page | Next Abstract

2011 LOM Workshop, Miami, Florida February 7 - 9, 2011