Screening the utility of climate information for watershed applications in Korea

Predictions regarding extremes of surface precipitation and temperature from climate models are potentially useful for developing water resources management decisions. The present paper uses a probabilistic utility index based on Kolmogorov-Smirnov formulae for discriminating high versus low variability in watershed mean areal precipitation and temperature and watershed-outlet discharge in the Korean Peninsula on the basis of climate model simulations of high versus low nodal surface precipitation and temperature. The theory is based on the Georgakakos [Georgakakos, K.P., 2003. Probabilistic climate-model diagnostics for hydrologic and water resources impact studies. J. Hydrometeorol. 4, 92-105] general development that allows for ensemble climate model simulations and variable-size analysis windows centered on the watershed of interest. In this study, for the first time, single and ensemble simulations from four atmospheric general circulation models with grid box areas ranging from 200 x 200 km2 to 400 x 500 km2 and with varying physical parameterizations and sea surface temperature forcing (observed versus persisted) are used to discriminate concurrent extreme values of the observed watershed variables in watersheds with largest area of approximately 26,000 km2. The analysis is done for a variety of periods and temporal resolutions. Monte Carlo simulations are employed to identify values of the utility index that are significant at the 5% level in each case. The utility index is computed for standardized anomalies of the variables of interest. The utility analysis shows that climate simulations have significant potential value for all watersheds in the Korean Peninsula especially for water resources management applications that use watershed mean areal temperature. The study also shows that several climate models are potentially useful in various seasons for discriminating extremes in watershed mean areal precipitation and watershed-outlet discharge for the Korean watersheds. Although single climate model simulations would be useful, the character of the results suggests that multi-model ensembles would be particularly useful for water resources management applications.