The authors investigate the relationship between bias in simulated sea surface temperature （SST） in the equatorial eastern Pacific cold tongue during the boreal spring as simulated by an oceanic general circulation model （OGCM） and minimal wind mixing （MWM） at the surface. The cold bias of simulated SST is the greatest during the boreal spring, at approximately 3℃. A sensi- tivity experiment reducing MWM by one order of magnitude greatly alleviates cold biases, especially in March-April. The decrease in bias is primarily due to weakened vertical mixing, which preserves heat in the uppermost layer and results in warmer simulated SST. The reduction in vertical mixing also leads to a weak westward current in the upper layer, which further contributes to SST warming. These findings imply that there are large uncertainties about simple model parameters such as MWM at the oceanic surface.
The northern Indian Ocean （NIO） experienced a decadal-scale persistent warming from 1950 to 2000, which has influenced both regional and global climate. Because the NIO is a region susceptible to aerosols emis- sion changes, and there are still large uncertainties in the representation of the aerosol indirect effect （ALE） in CMIP5 （Coupled Model Intercomparison Project Phase 5） models, it is necessary to investigate the role of the AIE in the NIO warming simulated by these models. In this study, the authors select seven CMIP5 models with both the aerosol direct and indirect effects to investigate their performance in simulating the basin-wide decadal-scale NIO warming. The results show that the decreasing trend of the downwelling shortwave flux （FSDS） at the surface has the major damping effect on the SST increasing trend, which counteracts the warming effect of greenhouse gases （GHGs）. The FSDS decreasing trend is mostly contrib- uted by the decreasing trend of cloudy-sky surface downwelling shortwave flux （FSDSCL）, a metric used to measure the strength of the AIE, and partly by the clear-sky surface downwelling shortwave flux （FSDSC）. Models with a relatively weaker AIE can simulate well the SST increasing trend, as compared to observation. In contrast, models with a relatively stronger AIE produce a much smaller magnitude of the increasing trend, indicat- ing that the strength of the AIE in these models may be overestimated in the NIO.
The performance of the eddy-resolving LICOM2.0 in simulating the Indonesian Throughflow has been evaluated against the INSTANT data in the present study.The mean vertical structures of the along strait velocities are simulated well in LICOM2.0,but the large velocities at the bottom of the Lifamatola Passage and the Timor Passage cannot be reproduced by LICOM2.0.The causes are considered to be both the errors in the topography and the tidal mixing at the bottom.Despite several biases in the mean velocities,the mean inflow and outflow volume transports in LICOM2.0 are almost identical to the INSTANT data.Compared with the lower resolution LICOM,the most significant improvement is the better simulation of the partitions of the inflow and outflow transports in individual straits.The outflow for low-resolution LICOM is mainly through the Ombai and Lombok Strait,whereas that for LICOM2.0 is mainly through the Timor Passage.The variability of the vertical structure of velocities and the volume transport are also investigated.LICOM2.0 overestimates the magnitude of the upper-layer currents and the amplitude of the variation.We also found that the largest correlation coefficient occurs in the shallowest strait,the Lombok,whereas the lowest occurs in the Timor Passage,especially in the upper layer.The latter may be caused by the unrealistic transport through the Torres Strait in LICOM2.0.
