Climatological screening of climate model output with observations for Korean water resources applications

摘要

This study focuses on the characteristic features of the surface hydroclimatology of seven watersheds on the South Korean Peninsula and the degree to which four modern-day climate models represent it at nodal and watershed scales. Long-term means and standard deviations of watershed mean areal precipitation and surface air temperature, and watershed-outlet discharge are computed for each month of the year using daily observed data (1979-2002) aggregated to monthly averages. Cumulative frequency distributions of standardized anomalies by season and watershed characterize hydroclimatic variability. The results show that the surface climate variables examined exhibit strong seasonal cycles (monsoonal behavior) with substantial differences in mean and variability across watersheds and seasons. Single simulations from two high resolution (200 x 200 km(2)) and 10-member ensemble simulations from two coarser resolution (280 x 280 km(2) and 400 x 500 km(2)) models are also used to produce model hydroclimatologies of long-term monthly means, standard deviations, and seasonal cumulative frequency distributions of standardized anomalies for nodes on and around the South Korean Peninsula. The models differ in physical and numerical formulations and some utilize observed and others utilize persisted sea surface temperature forcing. A comparison of the unadjusted model-produced and observed hydroclimatology on peninsular scales showed that several models skillfully reproduce features of the hydroclimatology for certain seasons, but with substantial deficiencies in other features shown as well. The results of this paper form a basis for the judicious use of climate model simulations for developing hydroclimatology for means and variability of surface watershed variables for water resources planning in the South Korean Peninsula. Copyright (C) 2007 Royal Meteorological Society.