Jae-Won Choi, Yumi Cha, Jeoung-Yun Kim
This study investigated a drought prediction method on the basis of similarity of spatiotemporal patterns of past droughts in Korea. The method was implemented in the following steps: First, drought areas in Korea were divided into four drought subregions by means of hierarchical clustering analysis. Second, spatiotemporal drought statistics for each subregion for the period from 1926 to 2008 were established. Temporal statistics involve the drought onset, end dates, duration, and regional drought intensities, measured by the Effective Drought Index (EDI). These statistics were collected over the four subregions, and spatial patterns of drought were examined. Third, the analogous drought events that had spatiotemporal patterns similar to those of the current or subject drought were selected. Fourth, the progress of the subject drought and the selected drought were compared. Finally, the progress pattern of the subject drought was predicted on the basis of the hypothesis that it will progress in a way similar to the selected analogous case. We applied this predicted method to several previous drought cases and evaluated the prediction efficiency. The results showed that this method was efficient in predicting droughts for about 1 year.