Evaluation of mixed layer depth in the northern Indian Ocean simulated by CMIP5 climate models

Abstract

We examine the mixed layer depth (MLD) in the northern Indian Ocean (NIO) simulated by 19 CMIP5 models to find the underlying bias dynamics that might lead to inaccurate climate prediction. All models show a systematic deep MLD bias in winter in the northern Arabian Sea (NAS). This deep bias can be largely attributed to a cold surface air temperature (SAT) bias accompanied by north-easterly winds from the landmass. In addition, the MLD bias appears to be associated with a bias in the upper salinity stratification in the NAS where the Arabian Sea high salinity water (ASHSW) is formed due to the combination of higher salinity water and lower temperature in winter. The models with large MLD biases show a deeper formation of the ASHSW with high salinity compared to other models in the northernmost of the NAS. Our finding suggests that improvements in atmospheric forcing, especially the SAT bias, and salinity stratification may have a significant contribution to reducing the deep MLD biases in the NAS.