Both GIS-based areal precipitation estimation and gridded precipitation data from GCM outputs have the problem of poor resolution,and to some extent neglect the influence of small or medium landform on precipitation distribution.A variety of statistical downscaling methods are briefly reviewed in this paper.Taking into account of the dynamic effect of geographic factors,a new downscaling scheme for areal precipitation estimation is proposed using 6 hour temporal resolution re-analysis data of NCEP/NCAR from April to September of 2010,and observational precipitation data from more than 20 weather stations in Jiangsu.The scheme chooses and constructs appropriate large-scale predictor,and retrieves the small-scale geographic factors from high resolution DEM,parameterizes the dynamic effect of geographic factors,and integrates the regression analysis,Partial Least Squares (PLS) and spatial interpolation.The actual precipitation series in the researched weather stations are successfully retrieved by the proposed scheme and the high-resolution spatial distribution of areal precipitation are drawn.