Abstract

The annual and interannual variability in the subthermocline equatorial Pacific is studied in a simulation of the tropical Pacific for the 1985–94 decade using a primitive equation high-resolution model. The study focuses on temperature variability and vertical energy fluxes. Similar to the observations made by W. S. Kessler and J. P. McCreary, the annual harmonic of vertical isotherm displacements presents phase lines sloping downward from east to west with off-equatorial maxima. Estimates of zonal and vertical phase speeds, the location of the maxima of isotherm displacements, and an analysis of ernergy fluxes suggest the presence of l = 1 Rossby waves. The simulated field is somewhat more trapped toward the equator than the observations. A linear simulation is carried out for the vertical standing modes with characteristics derived from the OGCM simulation. In the linear simulation, high-order meridional mode Rossby waves are more prominent than in the OGCM simulation. However, the (l = 1) Rossby wave in the linear model shares many characteristics with the one in the OGCM simulation. It is the result of both the reflection of surface-forced Kelvin waves on the eastern boundary and of direct forcing at the surface. On interannual timescales the simulation also presents vertical propagation of Rossby waves from the eastern boundary, but not as deep as for the annual cycle and in a much thinner beam. The variability of vertical displacements exhibits an asymmetry with a larger amplitude north than south of the equator. The zone of large interannual variability originating at the eastern boundary extends westward following (l = 1) or (l = 2) WKB ray paths and reaches the western Pacific near 400 m at 3°–4° north but not south of the equator. The 3.3-yr harmonic is prominent in isotherm vertical displacements for this particular simulation. The analysis also suggests the presence of (l = 1) Rossby wave, but the phase line characteristics are also indicative of the contribution of higher-order meridional modes. At this frequency, the solution of the linear model for the (l = 1) Rossby mode agrees well with the OGCM (l = 1) Rossby wave contribution.