STRATEGIES FOR PARALLEL VARIATIONAL DATA ASSIMILATION

The prospects to parallelize a variational data assimilation scheme fr om the starting point of an existing parallel forecast model and its a djoint equations have been investigated. Three parallelization strateg ies to do the implementation and two partitioning algorithms to comput e data distributions have been suggested. Numerical simulations of the parallelizations show that the strategies and the partitioning algori thms can be combined to perform well on both shared memory and distrib uted memory machines and that they give better results than direct use of standard parallelization methods. (C) 1997 Elsevier Science B.V.