• Home
• What is GODAE
• Activities
• Working Groups
• Science
  • Observing the ocean
  • Forecasting the ocean
  • Data Assimilation
  • Evaluating the predictions
• Data & Products
• Special Issues
• Related Websites
• What's New
• Calendar
• Documents

GODAE is sponsored by

• 
• 
• 
• 
• 
• More Sponsors >

Your location: Home / Science / Data Assimilation

Data Assimilation

The quality of an ocean forecast depends very much on the initial condidtions the model is started up with. Initial conditions, which come as a set of dependent variables such as temperature, salinity, density etc., are obtained from the analysis of the current ocean state. The available ocean observations are one component of this analysis. However, this is only one piece of information used to create the model intital conditions.

The observations are supplemented by a "first guess" field or background, which is mostly a short term (~6 hour) forecast from a previous model run, called the "assimilation cycle". While the direct observations are unevenly spread over the model domain, the first guess field ensures that there are reasonable data over all regions. The process of creating model initial conditions using observations in combination with the "first guess" field is called Data Assimilation.

The process in which observations and the first guess field are combined is called objective analysis. There are different methods through which this is achieved. While some (successive or nudging) methods use emperical information, others (optimal interpolation or OI) use statistical information about the observations and first guess field errors to successfully create the initial conditions. The optimal interpolation technique provides a more scientific and flexible way of including observations from differnt sources, as it allows individual weights to be applied to the observations depending on their error.

Variational methods are based on the minimization of a cost-function measuring the discrepancy between a solution of the model and the observation. Variational methods also have the advantage of allowing one to eliminate the initialisation step.

The use of data assimilation has improved the quality of ocean forecast greatly. GODAE scientists make sure that data assimilation techniques and methods are further developed to ensure a constant improvement of ocean forecasts.

(Last Updated: 12-02-2008)