On the problem of data assimilation by means of synchronization

Abstract

The potential use of synchronization as a method for data assimilation is investigated in a Lorenz96 model. Data representing the reality are obtained from a Lorenz96 model with added noise. We study the assimilation scheme by means of synchronization for different noise intensities. We use a novel plot representation of the synchronization error in a phase diagram consisting of two variables: the amplitude and the width of the error after a suitable logarithmic transformation (the so-called mean-variance of logarithms diagram). Our main result concerns the existence of an >optimal> coupling for which the synchronization is maximal. We finally show how this allows us to quantify the degree of assimilation, providing a criterion for the selection of optimal couplings and validity of models. Copyright 2009 by the American Geophysical Union.