Interannual Biases Induced by Freshwater Flux and Coupled Feedback in the Tropical Pacific

Abstract

Abstract Freshwater flux (FWF) forcing–induced feedback has not been represented adequately in many coupled ocean–atmosphere models of the tropical Pacific. Previously, various approximations have been made in representing the FWF forcing in climate modeling. In this article, using a hybrid coupled model (HCM), sensitivity experiments are performed to examine the extent to which this forcing and related feedback effects can contribute to tropical biases in interannual simulations of the tropical Pacific. The total FWF into the ocean, represented by precipitation (P) minus evaporation (E), (P − E), is separated into its climatological part and interannual anomaly part: FWFTotal = (P − E)clim + FWFinter. The former can be prescribed (seasonally varying); the latter can be captured using an empirical model linking with large-scale sea surface temperature (SST) variability. Four cases are considered with different FWFinter specifications: interannual (P − E) forcing [FWFinter = (P − E)inter], interannual P forcing (FWFinter = Pinter), interannual E forcing (FWFinter = −Einter), and climatological (P − E) forcing (FWFinter = 0.0), respectively. The HCM-based experiments indicate that different FWFinter approximations can modulate interannual variability in a substantial way. The HCM with the interannual (P − E) forcing, in which a positive SST − (P − E)inter feedback is included explicitly, has a reasonably realistic simulation of interannual variability. When FWFinter is approximated in some ways, the simulated interannual variability can be modulated significantly: it is weakened with the climatological (P − E) forcing and is even more damped with the interannual E forcing, but is exaggerated with the interannual P forcing. Quantitatively, taking the interannual (P − E) forcing run as a reference, the Niño-3 SST variance can be reduced by about 12% and 26% in the climatological (P − E) forcing run and interannual E forcing run, respectively, but overestimated by 11% in the Pinter forcing run. It is demonstrated that FWF can be a clear bias source for coupled model simulations in the tropical Pacific.