Simulation of tropical Pacific and Atlantic Oceans using a HYbrid Coordinate Ocean Model

The climatological annual cycle of the tropical Pacific and Atlantic Oceans is simulated using the HYbrid Coordinate Ocean Model (HYCOM) configured in a non-uniform horizontal grid spanning 30°S–65.5°N and 102°E–15°E. The model is initialized with climatological summer temperature and salinity and is forced by climatological atmospheric fields derived from the COADS and ECMWF ERA-15 reanalysis. The model is spun up for 20 years to reach a reasonable steady state in the primary region of interest from 20°S to 20°N, and year 20 is analyzed. The COADS simulation is primarily analyzed because it is slightly better in more respects than the ECMWF simulation, particularly in the representation of upper-ocean thermal structure. The model generally reproduces the seasonal variability of major circulation features in both oceans reasonably well when compared to climatologies derived from several observational datasets (surface drifters, TAO mooring array, COADS, Levitus, Pathfinder SST), and when compared to other model simulations. Model evaluation is complicated by the fact that the different climatologies, including the atmospheric reanalysis climatologies that drive the model, are averaged over different time intervals. In the tropics, the model thermocline reproduces the observed zonal slopes and meridional ridges/troughs in the thermocline. The simulated Equatorial Undercurrent compares favorably to observations, but is slightly deeper than observed. The model overestimates temperature in the Pacific warm pool regions, both west and east, by more than 1°C when compared to all observed climatologies. The model also tends to overestimate temperature in the eastern equatorial cold tongues in both the Atlantic and Pacific, with this overestimate being confined to a very small region of the far eastern Pacific during winter. This overestimate varies substantially depending on which observed climatology is used for the comparison, so model limitations are only partly responsible for the simulated–observed temperature differences in the cold tongues.