A relocatable EnKF ocean data assimilation tool for heterogeneous observational networks

Abstract

This study investigates the performance of a multivariate Ensemble Kalman Filter coupled with a relocatable limited-area configuration of the Regional Ocean Modeling System to predict ocean states by assimilating a heterogeneous data set involving underwater gliders and ship observations. In particular, two different ensemble initialization techniques are exploited and evaluated with the dataset collected during the REP13-MED experiment conducted by CMRE on 5-20 August 2013 in the Ligurian Sea. Results show that the forecast skill is significantly improved when the free ensemble is initialized from a long term climatology of the Mediterranean Forecast System. In particular the results obtained reveal significant increased skills in salinity forecasting in comparison with the previous ensemble initialization technique [6].