Future change of water cycle could cause serious impacts to human society. Outputs from climate models are often used to make future climate scenarios, which are used as input for impact assessment model. In addition, many impact studies carry out bias correction of climate model outputs so that input for impact model is quantitatively consistent with observation. This study employs three bias correction methods against daily precipitation outputs from three regional climate models, in order to evaluate uncertainties of future scenarios stemming from choice of regional climate models and bias correction methods. The result indicated precipitation scenario's uncertainty from regional climate models and bias correction methods is higher in heavy rainfall season. In addition, we developed impact model for Sameura river basin, and utilized those precipitation scenarios as input for our impact model. The result suggested water storage in Sameura reservoir during non-irrigation season might be higher in the future period of 2091-2100. It should be noted that there are some limitations in simulation design of this study (i.e. limited number of Global Climate models, bias correction methods, etc), and thus, further improvement of simulation conditions would be necessary for more reliable estimations.