Hindcasting Ocean Climate Variability Using Time-Dependent Surface Data to Drive a Model: An Idealized Study

Abstract

This paper investigates the hindcasting of interdecadal climate events using an ocean circulation model driven by different combinations of time-varying surface flux, sea surface temperature (SST), and sea surface salinity (SSS) data. Data are generated from a control run, against which the subsequent model experiments are compared. The most robust results are obtained using flux boundary conditions on both surface temperature and salinity. For these boundary conditions, model results am relatively insensitive to noise in the surface data and take about 20 years to overcome the imposition of an incorrect initial condition. Model results are much more sensitive to noisy inputs when run using SST and SSS data. To obtain meaningful results, SST data alone are not sufficient; SSS data are also required. This is related to the well-known instability of ocean climate models upon a switch to mixed boundary conditions. Time-varying SSS data cannot be replaced by climatology; using a best-fit T?S relation, to calculate anomalies in SSS from those in SST is also found to give disappointing results. The difficulty of trying to correct for inaccuracies in surface heat flux using SST data, while at the same time using a flux boundary condition on surface salinity, is demonstrated.