A Reduced-Gravity Isopycnal Ocean Model: Hindcasts of El Niño

Abstract

A global isopycnal ocean model is presented for the study of interannual to interdecadal variability in the global ocean. The model treats the primitive equations on a sphere with a generalized vertical coordinate. This coordinate is designed to represent a turbulent well-mixed surface layer and nearly isopycnal deeper layers. Disappearing isopycnics are treated through the quasi-isopycnal technique, in which the coordinate separates from the isopycnic in order to maintain a minimum layer thickness. A reduced gravity treatment is made, with the deepest interface at a mean depth of 2300 m. Coastal topography is represented, but the reduced gravity treatment precludes the use of variable bottom depth. The model is used for hindcast studies of El Niño during the decade from 1982 through 1991 using a combination of climatological wind forcing and wind anomalies derived from various sources. In order to carry out the hindcast experiments, a technique is developed for constructing a mean climatological surface heat flux using the model, climatological wind forcing, and climatological surface temperatures. In the hincast runs, the climatological winds and heat flux are augmented by the wind anomalies and a weak damping of surface temperature anomalies. A series of tests compares different data products for the wind anomalies. The first product is obtained from the Florida State University (FSU) wind analysis. The second and third wind products are obtained from global climate GCM simulations run over observed sea surface temperatures (SST). Although the wind products appear quite similar, the model results show large differences in hindcast skill, reflecting the fact that subtle features of the winds can have large impacts on ocean simulations and can be seen as a primary cause of wide differences in coupled GCM performance. The model maintains a sharp thermocline and a strong equatorial undercurrent in the center of the ocean basin. The heat flux needed to keep the model near the observed temperatures appears consistent with observational studies of the mean heat flux. When measured in terms of the skill in simulating the Niño-3 SST, the NASA Coupled Climate Dynamics Group (CCDG) model and FSU wind products provide the highest skill.