Subsurface Ocean Biases in Climate Models and Its Implications in the Simulated Interannual Variability: A Case Study for Indian Ocean

Coupled ocean atmosphere general circulation models (CGCMs) are known to have deep ocean biases when simulated over long term climate timescales. Here, an analysis has been done for investigating the impacts of these deep ocean biases, especially those occur in temperature and salinity, ensuing biases in ocean dynamics and large scale air-sea interactions in state of the art CGCMs. The outputs from historical runs of 20 Coupled Model Intercomparison Project Phase 5 (CMIP-5) models have been analyzed. All candidate models develop internal warm and saline biases approximately between a depth range of 100 and 800 m in long term simulations. These internal biases are found to have implications in large scale ocean dynamics via their linkage through baroclinicity of the ocean. The role of internal biases in the ensuing dynamics is correlated via relations between Brunt Väisälä frequency (N2) and baroclinic wave speeds using both Sturm-Loiuville theorem and WKBJ approximation. The CMIP-5 models analysed here have a higher baroclinic speed compared to that of observations. Annual propagating modes in the ocean reveal higher speeds in most of the models, and that the phase reversal is taking place at lead or lag by a month or more as compared to observations. The study suggests that faster wave dispersions in climate models due to subsurface biases in temperature, salinity, N2 and baroclinicity have potential to impact simulated planetary scale events in terms of their life cycle, periodicity and seasonality. A corollary being a cautionary outlook on climate projections made by coupled models as long as the biases are persistent.