Bayesian Projections of the Amur and Selenga River Runoff Changes in the 21st Century Based on CMIP6 Model Ensemble Simulations.

The analysis is carried out for changes in runoff of the Amur and Selenga rivers in the 21st century according to the CMIP6 (Coupled Model Intercomparison Project, Phase 6) climate model ensemble simulations using the Bayesian approach versus stream gage data on annual runoff and GPCP-2.3 dataset on annual precipitation over catchments on different timescales. For both catchments, significant intermodel differences are associated with the projections of multiyear mean runoff and interannual variability. The intermodel distribution of Bayesian weights indicates a high role of uncertainty related to initial conditions for model simulations. There is a positive trend in total runoff in the Amur River basin in the 21st century under all analyzed anthropogenic forcing scenarios. For total runoff of the Selenga River, there are no trends in the 21st century for all analyzed scenarios. No significant trends for the Amur and Selenga surface runoff were revealed for all algorithms for considering Bayesian weights and all anthropogenic forcing scenarios. At the same time, significant interdecadal variations in the interannual variability of runoff were found. [ABSTRACT FROM AUTHOR]