ABSTRACT

Reduce space Four Dimensional Data Assimilation (R4Dvar) is the efficient way to assimilate data without developing tangent linear and adjoint model. As it was shown before, the R4Dvar outperforms the convention adjoint based 4Dvar data assimilation approach in case of the assimilation into strongly non-linear model. We illustrate the performance of the R4Dvar approach in several twin-data experience with state of the art models and and discuss it potential advantages of the new data similation tool