Downscaling methods have been proposed for establishing the linkages between the large-scale climate variables given by GCMs and the observed characteristics of hydrologic variables at a local site. These downscaling methods, however, are not suitable for dealing with cases where the hydrologic data at the location of interest are limited (a partially-gauged site) or not available (an ungauged site). The downscaling of a hydrologic process such as the daily precipitation for such cases remains still a crucial challenge for water resources planning and management in practice. The present study proposes therefore a statistical downscaling (SD) approach to establishing accurately the linkages between the climate variables given by GCM outputs and the "estimated" daily precipitation characteristics at a location of interest where the precipitation data are limited or unavailable. More specifically, the suggested SD procedure is based on a combination of two components: (ⅰ) a regional stochastic method for reconstructing the unmeasured daily precipitation series at the ungauged site; and (ⅱ) a SD model for describing the linkage between the constructed daily precipitation series and the climatic predictors given by GCM outputs. The feasibility and accuracy of the proposed SD procedure was assessed based on the NCEP re-analysis and observed daily precipitation data available from a network of 62 raingages in South Korea. Results of this assessment have indicated that the proposed procedure could provide comparable results as those given by the downscaling using real observed precipitation data at the same local site.
展开▼
