We have applied a full hierarchical Baysian (HB) model to simulate streamflow at the Coweeta Basin that drains western North Carolina, USA under a doubled CO₂ climate scenario. The full HB model coherently assimilated multiple data sources and accounted for uncertainties from data, parameters and model structures. Full predictive distributions for streamflow from the Bayesian analysis indicate not only increasing drought, with substantial decrease in fall and summer flows, and soil moisture content, but also increase in the frequency of flood events when they were fit with Generalized Extreme Value (GEV) distribution and Generalized Pareto Distribution (GPD) under this doubled CO₂ climate scenario compared to the current climate scenario. Full predictive distributions based on the hierarchical Bayesian model, compared to deterministic point estimates, is capable of providing richer information to facilitate development of adaptation strategy to changing climate for a sustainable water resource management.