A path integral method for data assimilation

摘要: Abstract Described here is a path integral, sampling-based approach for data assimilation, of sequential and evolutionary models. Since it makes no assumptions on linearity in the dynamics, or Gaussianity statistics, permits consideration very general estimation problems. The method can be used such tasks as computing smoother solution, parameter estimation, data/model initialization. Speedup Monte Carlo sampling process essential if integral has any chance being viable estimator moderately large Here variety strategies are proposed compared their relative ability to improve efficiency resulting estimator. Provided well details useful its implementation testing. applied problem which standard methods known fail, an idealized flow/drifter problem, been testbed assimilation involving Lagrangian data. It this kind context that may prove tool oceanic studies.