DA is the objective analysis technique that combines imperfect, incomplete and/or irregular observations with our knowledge about the physical laws (manifested in the form of numerical weather prediction models) to provide the regular and physically consistent estimate of the atmospheric state.

Due to the non-linear, chaotic nature of the atmosphere and inevitable errors in observations and prediction models, numerical simulations of weather contain errors that can grow rapidly with time. Since this error growth can vary with the weather phenomena and larger-scale flow regime, greater practical use can be made of weather prediction products if the uncertainty in a forecast can be accurately characterized. EF techniques are meant to represent this uncertainty through using multiple forecast simulations each with slightly different initial conditions and variations in model.

MAP has contributed to the advancement of the theory of Ensemble Transform Kalman filter (ETKF) and hybrid ensemble-variational data assimilation algorithm and strives to further develop and advance DA and EF theory and algorithm.