Development and Application of a Hybrid Downscaling Method over Hokkaido Region, Japan

The warming of the Earth's atmosphere system is likely to change temperature and precipitation, which may affect the climate, hydrology and water resources at the river basins over the world. The importance of temperature change becomes even greater in snow dominated basins where it controls the snowmelt processes during the late-winter, spring and summer months. Snow cover is one of the major water resources in Hokkaido region, Japan, so that it is important to examine what future changes could happen to hydrological cycles in the snow dominated basins as a result of climate change. General Circulation Models (GCMs) provide projections of future atmospheric and hydrologic conditions. However, the spatial resolutions of GCM Projections which are publicly available are approximately 100-200 km. These resolutions are too coarse for the climate change impact assessments at the scale of the local basins. Therefore, downscaling methods play a critical role in extracting finer-scale information from GCM-based coarse projections of the future climate conditions. Methods for downscaling the climate projections onto a geographical region are classified broadly into two categories, namely, statistical downscaling and dynamical downscaling. The dynamical downscaling method is based on the hydrodynamic/thermodynamic simulation of a regional hydro-climate system under the initial and boundary conditions that are provided from the GCM simulation data. Because of the large inter annual variability of snowfall in Japan, long-term dynamical downscaling is necessary in order to determine the change in snow cover due to global warming. However, the dynamical downscaling is computationally expensive and time consuming. From this point of view, a hybrid downscaling method which combines dynamical and statistical techniques was developed in this study in order to generate climate projections that are appropriately scaled for use in a hydrologic evaluation in the snow dominated basins. The main objective of this study is to assess the regional climate change impact on the climate, hydrology and water resources at the snow covered river basins in Hokkaido using the hybrid downscaling method. For the dynamic downscaling, a Regional Spectral Model was employed as a regional climate model at a 10 km spatial grid. A linear regression model using ground observation data was applied to further downscale gridded simulated output to a 1 km spatial grid. For the first step, Japan Reanalysis atmospheric data provided from Japan Meteorological Agency was used as initial and boundary conditions for the hybrid downscaling in order to explore its utility. The performance of the hybrid downscaling was evaluated by the comparison of model-reconstructed precipitation and air temperature against ground observations in time and space with satisfactory results. These results lead to the conclusion that the hybrid downscaling method presented in this study may be useful for the regional climate change impact study at the scale of the local basins in the snow dominated region.