Indian Ocean mean state biases and IOD behaviour in CMIP6 multimodel ensemble

The Indian Ocean Dipole (IOD) is the main coupled mode of interannual variability in the equatorial Indian Ocean. The largest IOD event in 2019 is thought to have influenced the strong Indian monsoon precipitation, widespread Australian bushfires, and extreme rainfall and flooding in East Africa during that year. Despite its socio-economic importance, the region suffers large biases in weather and climate models used for seasonal forecasts and climate projections.

In this study, the performance of 42 models from the sixth phase of the Coupled Model Intercomparison Project (CMIP6) in reproducing the observed climate over the Indian Ocean is examined. Model simulations of precipitation and 850 hPa winds in the Atmospheric Model Intercomparison Project (AMIP) experiments for the period 1979-2014 are compared to observational and reanalysis data. Biases in the mean state during boreal summer (JJA) in the AMIP models are analysed to determine whether biases in the seasonal cycle established in JJA impact the IOD behaviour. Skill metrics are calculated to quantify the model performance in reproducing the observed JJA mean state and cluster analysis on the mean state biases is performed to characterise bias patterns in summer that may affect the Indian Ocean seasonal cycle and IOD. Results show that AMIP models simulate varying bias patterns in JJA and that the AMIP multi-model mean outperforms all individual models in reproducing the observed JJA mean state. For comparison, the Indian Ocean mean state biases are investigated in coupled models from the 20th-century all-forcings (CMIP) experiments to determine the impact of ocean-atmosphere coupling and coupled sea surface temperature biases on model performance. The IOD behaviour in the AMIP and CMIP models is assessed and the response of the atmospheric circulation to IOD forcing is examined by performing regression analysis. We investigate whether the ability of a model to capture characteristics of the IOD and simulate IOD teleconnection patterns is related to its representation of the mean state. We expand this work to investigate the variability in the Indian Ocean in the Met Office Global Seasonal Forecasting System version 6, GloSea6, with a focus on examining the systematic errors that develop in the region. The work will contribute to our understanding of Indian Ocean biases in weather and climate models, and their likely sources, and thus the wider implications for predictability of the IOD.