Data assimilation in the FOAM operational short-range ocean forecasting system: A description of the scheme and its impact

A detailed description of the data assimilation scheme used in the Forecasting Ocean Assimilation Model (FOAM) operational ocean forecasting system is presented. The theoretical basis for the scheme is an improved version of the analysis correction scheme, which includes information from previously assimilated data. The scheme requires the a priori specification of error covariance information for the background model field and the observations. The way in which these error covariances have been estimated is described and some examples are given. The FOAM system assimilates sea surface temperature, sea-level anomaly, temperature profile, salinity profile and sea-ice concentration data. Aspects of the scheme that are specific to each of these observation types are described. Two sets of experiments demonstrating the impact of the data assimilation are presented. The first set are in an idealized identical-twin setting, using the 1/9 degrees-resolution North Atlantic FOAM configuration in which the state of the true ocean is assumed to be known. These experiments show that the analyses and forecasts are improved by assimilating the altimeter sea-level-anomaly data. The second set of experiments comprise data impact studies in a realistic hindcast setting. These experiments show a positive impact on the analyses from the Argo temperature- and salinity-profile data. (c) Crown Copyright 2007.