| Recently the Meteorological Research Institute of Japan (MRI) has developed the Atmospheric General Circulation Model (MRI-AGCM) which spatial resolution is roughly 20km. Because this spatial resolution is much finer compared with other GCMs、this simulation outputs is highly expected to be used as a basis of adaptation measurements to the increase of flood risk under climate change for interested regions. MRI-AGCM has two simulation results named MRI-AGCM3.1S (3.1) and MRI-AGCM3.2S (3.2). For the purpose of examining the performance of MRI-AGCM、we compared these two simulation results with observed precipitation data in the Monsoon Asia (APHRODITE Monsoon Asia: APHRO_MA). According to the comparison result、both 3.1 and 3.2 shows relatively good agreement with observation in terms of annual and monthly precipitation in many areas in Asia. Especially、it is remarkable that they well reproduced the precipitation distribution which is affected by orographic effect in Asian region. However、they show some difficulties in reproducing extreme precipitation which is important for flood risk management. While 3.1 underestimates heavy precipitation、3.2 overestimates. The large difference between the two simulations results implies that the hydrological simulation results in which these precipitation data are input will differ each other. This might make the discussion on the change of flood magnitude due to climate change difficult. Considering this background、we proposed a simple and straightforward statistical bias correction method for MRI-AGCM precipitation to serve as a basis for flood impact assessment study. Not only extreme precipitation but also seasonal precipitation should be considered in flood assessment because soil water content、which depends on antecedent precipitation conditions、will affect flood magnitude. Thus、the proposed method in this research primarily aimed to correct intensity of MRI-AGCM daily precipitation samples to express both seasonal patterns and extreme values appropriately by fitting the frequency distribution of MRI-AGCM precipitation output to that of observation. The method was applied to the Monsoon Asia region. There was a large discrepancy between original precipitation output of 3.1 and 3.2 especially for extreme precipitation in Monsoon Asia region、but it reduced very much by introducing the bias correction method. It is clear that it becomes easier to use the precipitation output of MRI-AGCM for hydrological simulation by introducing the bias correction method. |
