Experimental Implementation of an Ensemble Adjustment Filter for an Intermediate ENSO Model

Abstract

The assimilation of sea surface temperature (SST) anomalies into a coupled ocean?atmosphere model of the tropical Pacific is investigated using an ensemble adjustment Kalman filter (EAKF). The intermediate coupled model used here is the operational version of the Zebiak?Cane model, called LDEO5. The assimilation is applied as a means of estimating the true state of the system in the presence of incomplete observations of the state. In the first part of this study assimilation is performed under the ?perfect model? assumption, where SST observations are synthetically derived from a trajectory of the model. The focus is on how and why changes in the filter parameters (ensemble size, covariance localization, and covariance inflation) affect the quality of the analysis. It is shown that isotropic covariance localization does not benefit the analysis even when a small number of ensemble members are used. These results suggest that destruction of the ?balance? between variables caused by localization is more detrimental than spurious correlation due to small ensemble size. In the second part of this study the EAKF is used to assimilate an independent dataset of SST observations. The EAKF/Zebiak?Cane assimilation system is able to correctly estimate the phase and intensity of ENSO, as measured by the average SST anomaly in the eastern equatorial Pacific. A comparison of the analysis herein to independent wind stress and thermocline depth datasets suggests that even with the assimilation of only SST observations it is possible to reproduce over 70% of the interannual variability of thermocline depth in the eastern equatorial Pacific and off the coast of the Philippine Islands. The interannual variability of zonal wind stress in the central and western equatorial Pacific is also well correlated with independent observations (R > 0.75).