Evaluation of Coupled Model Intercomparison Project Phase 6 model-simulated extreme precipitation over Indonesia

The ability of 42 global climate models from the Coupled Model Intercomparison Project Phase 6 (CMIP6), consisting of 20 low resolution (LR) and 22 medium resolution (MR), are evaluated for their performance in simulating mean and extreme precipitation over Indonesia. Compared to Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS), the model climatologies and interannual variability are investigated individually and as multimodel ensemble means (MME-mean) at monthly and seasonal time scales for the historical simulation over the period 1988-2014. Overall, results show that both LR and MR CMIP6 model skills in simulating mean and extreme precipitation indices vary across specific Indonesian regions and seasons. The individual and MME-mean tend to overestimate the observed climatology, being largest over drier regions, yet MR models perform better compared to the LR regarding the mean bias presumably due to increased resolution. CMIP6 models tend to simulate extreme precipitation better in the dry seasons compared to the wet season. The MME-means of the LR and MR groups mostly outperform the individual models of each group in simulating wet extremes (R95p and Rx5d) but not for the dry extremes (CDD). Among the 42 CMIP6 models, three models consistently perform poorly in simulating Rx5d and R95p, namely FGOALS-g3, IPSL-CM6A-LR, and IPSL-CM6A-LR-INCA, and one model in consecutive dry day (CDD) simulation, MPI-ESM-1-2-HAM, and caution is warranted. Given the knowledge of such biases, the LR and MR CMIP6 climate models can be suitably applied to assist policy makers in their decision on climate change adaptation and mitigation action.

Automated summary

The ability of 42 global climate models from the CMIP6 project in simulating mean and extreme precipitation over Indonesia is evaluated. The results show that the performance of the models varies across different regions and seasons, with medium resolution models generally outperforming low resolution models. The models tend to overestimate observed precipitation, especially in drier regions. They also simulate extreme precipitation better during the dry season compared to the wet season. Ensemble mean models perform better in simulating wet extremes, but not for dry extremes.