Generalized Inversion of Tropical Atmosphere–Ocean Data and a Coupled Model of the Tropical Pacific

Abstract

It is hypothesized that the circulation of the tropical Pacific Ocean and atmosphere satisfies the equations of a simple coupled model to within errors having specified covariances, and that the Tropical Atmosphere–Ocean array (TAO) measures the circulation to within errors also having specified covariances. This hypothesis is tested by finding the circulation that is the weighted least squares best fit to the dynamics of the simple model, to its initial and boundary conditions, and to a year of monthly mean TAO data for sea surface temperature, for the depth of the 20°C isotherm, and for surface winds. The fit is defined over the entire tropical Pacific and from 1 April 1994 to 31 March 1995. The best-fit circulation or state estimate is calculated using variational methods. Posterior error covariances are estimated using statistical simulation. The best fit is also subjected to a significance test. It is found that, although the fit to data is largely within standard errors, the misfit to dynamics is significantly larger than standard errors. These standard errors are estimated a priori, and with difficulty, from published scale analyses of tropical circulation. An examination of the dynamical residuals indicates that the simple coupled model should be improved by including momentum advection both in the oceanic and atmospheric components.