Drought early warning system in reservoir operation: Theory and practice

This paper revises the drought early warning system (DEWS) for reservoir operation, originally developed by Huang and Yuan (2004). The changes are made in response to requirements raised by the Water Resources Agency in Taiwan. We extend the DEWS to include a number of new factors: in particular, to take into account meteorological drought and the determination of alert levels in policy making based on information entropy. The goals of this paper are critical: (1) revising the DEWS based on experience, (2) examining a specific case study of how the system works, (3) involving users in decision making, and (4) comparing theory with practice. The Shihmen reservoir in northern Taiwan was picked by the agency as a real-case study. In hindsight, the revised drought early warning system performed well over a simulation of five consecutive years (1999-2003). These results verify the applicability of this renewed DEWS in reservoir operation for lowering potential drought threats.