Evaluation of Downscaled <scp>CMIP6</scp> Performance for Rainfall and Wind Over the Maritime Continent

ABSTRACT High‐resolution climate change projections over the Maritime Continent (MC) are increasingly critical to develop climate policies and adaptation planning at the regional level. However, modelling rainfall and wind over the MC still leads to substantial errors in capturing climate processes over the most complex land–atmosphere–ocean interaction. This study evaluated the performance of the 14 downscaled CMIP6 models from the NASA Earth Exchange Global Daily Downscaled Projections (NEX‐GDDP) at 0.25° resolution in simulating spatial and temporal variability of rainfall and wind over the MC. The simulated rainfall and wind speed over the period 1995–2014 are evaluated against meteorological station data and gridded datasets, including the Climate Hazards Group InfraRed Rainfall with Station data (CHIRPS) for rainfall and the European Centre for Medium‐Range Weather Forecasts Reanalysis 5 (ERA5) for wind speed. Results demonstrated that the models are capable of representing the regional characteristics of the annual and seasonal variation of rainfall and wind. Nevertheless, systematic biases remain consistently overestimated for rainfall when compared with observations (35%–94%) and slightly underestimated compared with CHIRPS (−4% to −5.6%). The model overestimated wind for both mean and extreme wind (i.e., 34%–120% for observations and 11%–48% for ERA5), except western MC showed a slight underestimation relative to ERA5 (−6.9%). The largest bias of extreme rainfall intensity is mainly depicted over region C (Maluku) compared to regions A (Sumatra) and B (Java). This result suggests that the models are capable of simulating monsoonal and equatorial rainfall patterns compared to anti‐monsoonal rainfall, which may be dominated by anvil rain type. Our study also highlighted that model rankings vary according to the variable, reference dataset and metric used, and that no single model consistently outperforms.