Data assimilation methods in oceanography developed in the past 30 years in support of short term ocean predictions and fundamental studies on optimal estimation of state fields for ocean hydrodynamics and biogeochemistry.
The methodology consists of correcting the ocean model simulation or hindcast state fields with observations coming from satellites and in situ sensors in order to minimize the combined errors of the numerical model and the observations (the latter mainly due to representativity errors). The theoretical background is connected to Bayesian statistics and Kalman filtering.
At INGV, a three dimensional variational data assimilation scheme (3DVAR) was developed by Dobricic and Pinardi (2008) Paper download and it is used to produce analyses and reanalyses for the Mediterranean Sea.
This basic method is used also to carry out Observing System Experiments (OSE) and Observing System Simulation Experiments (OSSE) as shown in these slides (open slides to be linked).
Operational Oceanography Group @ INGV – Follow us