Model Uncertainty in the Projections of Sub-Daily Extreme Precipitation

Abstract:

The risk of flood and drought is strengthening due to future intensification of precipitation extremes and such intensification is more rapid for sub-daily precipitation extremes which can cause flash flood and prompt landslides. Despite this fact, the credibility of projections from general circulation models (GCMs) in sub-daily precipitation extremes is least investigated in previous studies. Therefore, in our study, we tried to explore the reliability of GCMs in the projections of such extremes. We have used 20 Coupled Model Intercomparison Project Phase 5 (CMIP5) models under RCP8.5 (Representative Concentration Pathway). Model uncertainty is estimated by giving weightage based on their inter-dependency which arises due to the sharing of common codes, and parameterization processes. The uncertainty is estimated in the projection of intensity-based index i.e. monthly maximum 3-hour precipitation (R×3hr) in four different meteorological seasons (DJF, MAM, JJA, and SON) for both near (2026-45) and far future (2081-99) time period. We observed higher spatial contribution and magnitude of uncertainty in the southern hemisphere for DJF and MAM in contrast northern hemisphere for JJA and SON seasons. The quartile ranges of model uncertainty are found to be more in JJA among all seasons. While moving from the polar region to equatorial regions, the reliability of climate models gradually decreases with lowest agreement in the mid-latitude and equatorial regions. The uncertainty is increasing with increase in time period and the decrease in confidence is found to be more in JJA as compared to other seasons.