Interannual Salinity Variability in the Tropical Pacific in CMIP5 Simulations

Abstract

Salinity variability and its causes in the tropical Pacific are analyzed using observations, reanalysis products and model simulations. The mixed-layer salinity (MLS) budget analyses from observations and reanalysis products indicate that its interannual evolution is closely related to ENSO and is predominantly governed by surface forcing and surface advection in the western-central equatorial Pacific. It is found that the observed MLS tendency leads Ni˜no3.4 by about 12 months due to the effect of negative freshwater flux (evaporation minus precipitation). These observation-based analyses are used to evaluate the corresponding simulation using GFDL-ESM2M. It is evident that the model can simulate the spatiotemporal variations of MLS with some discrepancies compared to observations. In the warm pool of the equatorial Pacific the MLS tendency in the model is sensitive to ocean dynamics, however model biases cause the tendency to be underestimated. In particular, the freshwater flux is overestimated while the ocean surface zonal current and vertical velocity at the base of the mixed layer are underestimated. Due to model biases in representing the related physics, the effects of surface forcing on the simulated MLS budget are overestimated and those of subsurface and surface advection are relatively weak. Due to weaker surface advection and subsurface forcing than observed, the simulated compensations for surface forcing are suppressed, and the simulated MLS tendency that leads Ni˜no3.4 by 8–10 months, which is shorter than the observed lead time. These results are useful for the interpretation of observational analyses and other model simulations in the tropical Pacific.