On the simulations of latent heat flux over the Indian Ocean in CMIP6 models

Abstract

The historical simulations of latent heat flux (LHF) over the Indian Ocean (IO) region in 46 climate models participating in the phase 6 of Coupled Model Intercomparison Project (CMIP6) have been evaluated against the ERA5 dataset for the period 1975–2014. The multimodel mean (MMM) captures the observed spatial pattern of LHF climatological mean state to a great extent, but there are large biases in magnitude, especially over the region north of 20 °S and south of 40 °S, and these biases are common to more than 70% of the individual CMIP6 models. The origins of these biases are due to overestimated surface winds and vertical humidity gradients in the models. The simulated trend values of LHF are largely underestimated and are highly inconsistent in individual models. The basin-wide increase of surface saturation specific humidity (Qs) is largely compensated by a similar increase in near-surface saturation specific humidity (Qa) trend, which leads to reduced vertical humidity gradient for the latent heat transfer to take place at the air-sea interface. The long-term trend of sea surface temperature (SST) and Qs are found to be reasonable in CMIP6 models. One of the reasons for the underestimated LHF trend in CMIP6 models is the faster rate of increase of Qa and very slow rate of increase of surface wind speed compared to the ERA5 dataset.