Projection and uncertainty analysis of global precipitation-related extremes using CMIP5 models

摘要

Climate change is expected to influence the occurrence and magnitude of precipitation-related extremes and to increase drought and flood risk. Thus, future changes in dryness and wetness over global land areas are analysed using future climate simulations from the World Climate Research Programme’s (WCRP) Coupled Model Intercomparison Project Phase 5 (CMIP5) under RCP4.5 forcing scenario. Model reproducibility is evaluated first, and it is shown that high performance can be achieved in present-day climate simulations by models, particularly in multi-model ensemble (MME) results. For future climate simulations, the highest reliability regarding changes in precipitation and its related extremes is found over Northern high latitudes, while the lowest confidence levels are mainly localized over the tropics. The projections indicate a high likelihood that there will be a shift to fewer dryness but to more extreme precipitation events or/and flood events in future over Northern high latitudes. Among populated areas, Mediterranean basin is highlighted as displaying a relatively high reliability of increases in both dryness and wetness indicators, implying increased probabilities of both drought and flood events, despite the fact that there would be less precipitation. In North America and Asian monsoon areas, dryness indictors show no obvious changes, while markedly increases are found in wetness indicators, concurrent with a high model agreement. In contrast, southern Africa, Australia, and the Amazon basin show relatively high reliability regarding increases in dryness, but a low confidence level in wetness. The severity of these changes is not uniform across annual and seasonal scales and is region dependent. Two sources of uncertainty in projections are investigated in this study: internal and inter-model variability. The analysis indicates that internal and inter-model variability are the dominant sources of uncertainty in extreme climate projections, and inter-model variability is much larger and increases with time. Further analysis shows that both sources of uncertainty generally perform lower on annual and global scales than on seasonal and regional ones.