When performance of models until today published IPCC 5nd WG1 assessment report compared to the global, the more repeated generation the more improved the performance of models and appears closer to the observation. but uncertainty of prediction in climate change scenarios is tends to be relatively large in greenhouse gases scenario for high concentration and appear larger than the temperature in the precipitation. so, this paper is comparative analysis for reduce of uncertainty about precipitation blending method of climate change scenario based on stationary probability precipitation.
In this paper, is study about method of to mix using nonstationary probability of rainfall and climate change scenarios. For this purpose, This study is to estimate the Nonstationary probability precipitation applied method of probability weighted moments and Gumbel distribution using observed AWS data and blend with stationary probability precipitation using RCP 4.5 data and RCP 8.5 data. Result of each weight was calculated by location and return period, target year.
In this study used out the AWS rainfall data between 1973 year and 2012 year. Also, nonstationary probability precipitation be calculated by applying linear regression model, each power regression model, method of probability weighted moments , Gumbel distribution. Power1 regression model among each regression model was judged most appropriate and carried to blend between climate scenario and result using Power1 regression model. About RCP daily data of climate scenario between 1956 year and 2099 year, carry out to corrected based on observed AWS data with same period and calculate stationary probability precipitation using corrected RCP data. Finally, carried out to blend between nonstationary probability precipitation of Power1 regression model and stationary probability precipitation of corrected RCP data. Methods of blending is to calculate weight using variation each target year and period using stationary probability precipitation of observed data.