Downscaling is necessary method of scale transformation in studying impact of climate change on regions with GCM output. Impact of precipitation variation in spatial on some hydrologic processes is prominent, for example, evaportranspiration, runoff generation and concentration, which are interrelated with underlying surface obviously. A new method, Delta-DCSI downscaling method is developed here to improve the ability of future precipitation scenarios to describe its spatial distribution in sub-grids. Its is based on the Delta method, spatial distribution of future precipitation in GCM grids is taken into account with one interpolation method of dualistic cubic surface. That is to say, precipitation of GCM output in each grid is interpolated into each meteorologic station firstly. Then, change of future precipitation relative to mean simulated value of baseline period is calculated in each station. Finally, future precipitation of each station is gotten by adding the change to its observed value in baseline period. Theoretically, spatial distribution of future precipitation by this new method should be more obvious than that gotten from Delta method. Contrasts between them showed that change of future precipitation gotten from the new method is more smooth and reasonable than that of Delta for describing of spatial variation in sub-grids is strengthened and continuity of GCM output is recurred in the new method. Conclusions of future precipitation gotten from different methods showed that the effect of new method is obvious in areas with dense meteorologic stations. And it is very different from that of the Delta method in those areas.
