On the use of EOF analysis in dynamical model testing

Abstract

The legitimacy of using the statistical technique of empirical orthogonal function (EOF) analysis to test the dynamical relationships between oceanic forcing and response has generally been assumed. Here it is proven that using the coherence relationships between selected forcing time series (such as wind stresses) and EOFs representing oceanic response (such as currents) to drive linear, frequency-domain models simulating the response is a dynamically consistent procedure. In other words, if the model to be tested is an accurate representation of the true dynamics, then each model-predicted response will converge to the EOF as all the coherent forcing is added. Consequently, the accuracy of the dynamics in a chosen model can be tested using each EOF taken separately. The procedure of using a coherence analysis between forcing and EOF to determine model forcing also facilitates the diagnosis and uncertainty analysis of model solutions. The circulation on the Labrador Shelf is examined as an example of its application.