Origins of Coupled Model Biases in the Arabian Sea Climatological State

This study investigates biases of the climatological mean state of the northern Arabian Sea (NAS) in 31 coupled ocean–atmosphere models. The focus is to understand the cause of the large biases in the depth of the 20°C isotherm [Formula: see text] that occur in many of them. Other prominent biases are the depth [Formula: see text] and temperature [Formula: see text] of Persian Gulf water (PGW) and the wintertime mixed-layer thickness (MLT) along the northern boundary. For models that lack a Persian Gulf (group 1), [Formula: see text] is determined by the wintertime MLT bias [Formula: see text] through the formation of an Arabian Sea high-salinity water mass (ASHSW) that is too deep. For models with a Persian Gulf (group 2), if [Formula: see text] > MLT (group 2B), PGW remains mostly trapped to the western boundary and, again, [Formula: see text] directly controls [Formula: see text]. If [Formula: see text] MLT (group 2A), PGW spreads into the NAS and impacts [Formula: see text] because [Formula: see text] > 20°C; nevertheless [Formula: see text] still influences [Formula: see text] indirectly through its impact on [Formula: see text]. The thick wintertime mixed layer is driven primarily by surface cooling [Formula: see text] during the fall. Nevertheless, variations in ΔMLT among the models are more strongly linked to biases in the density stratification (jump) across the bottom of the mixed layer than to [Formula: see text] biases. The jump is in turn determined primarily by sea surface salinity biases (ΔSSS) advected into the NAS by the West India Coastal Current, and the source of ΔSSS is the rainfall deficit associated with the models’ weak summer monsoon. Ultimately, then, ΔD20 is linked to this deficit.