Evaluation of CMIP6 precipitation simulations across different climatic zones: Uncertainty and model intercomparison

This study analyzes the performance of precipitation estimates from historical runs of the CMIP6 (Climate Model Intercomparison Project Phase 6) over the climatic regions of Iran. In order to capture the spatio-temporal precipitation patterns, using a set of evaluation metrics, 12 GCMs (General Circulation Models) are compared to the observation data from the GPCC (Global Precipitation Climatology Centre) at the common 1° spatial resolution for 1950–2014. A comprehensive assessment is performed at different temporal scales including monthly, seasonal and annual. Results indicate that the reliability of precipitation estimates varies significantly across space and time. The CMIP6 models best reproduce the climatological features of precipitation and its spatio-temporal changes over the arid and hyper arid areas of the country. The outputs of the models exhibit less systematic biases in the arid zone. In addition, a strong underestimation is detected throughout the rainiest zone, indicating high uncertainty in wet regions. All models tend to show some level of underestimation in summer months with the lowest rainfall. The findings illustrate substantial inter-model variability over different climatic zones. Each of the CMIP6 models appears to be more suitable in a specific climatic zone. The models that performed reasonably well in the humid zone (CNRM-CM6-1 and MRI-ESM2-0), did not perform well in the hyper arid and arid zones. Similarly, models (HadGEM3-GC31-LL, BCC-CSM2-MR, and CanESM5) that performed well in the arid and hyper arid zones did not perform as well in the humid zone. Results inform what types of models are suitable for different climate zones. •The performance of precipitation estimates from 12 CMIP6 models is evaluated.•The assessment is done at different temporal scales over Iran's climatic regions.•A strong underestimation is detected throughout the rainiest zone.•All models tend to show some level of underestimation in summer months.•Results inform what types of models are proper for regional climate impact studies.