Construction of Hydrological Drought Severity-Area-Duration Curves Using Cluster Analysis

In this study, cluster analysis is applied to classify the whole of South Korea into several groups for characterizing spatially drought events. The standard precipitation index (SPI) is selected as the drought parameter. Using hierarchical cluster analysis and complete linkage method, South Korea is divided into 6 groups. The SPIs of 58 gage station's precipitation data provided in Korea Meteorological Administration are computed and their spatial informations are extracted using EOF analysis. Spatially extended EOF values through kriging theory are used to construct the severity-area-duration (SAD) curve in each group. As a whole, drought severity decreases as drought duration increases. However, it is found that the spatial reduction rate of severity by cluster-based SAD curves is less than that of non-cluster based SAD curves. The analysis of clustered SAD curves reveals that the drought severity is the deepest in the east coast of South Korea.