Summary: | 碩士 === 逢甲大學 === 土木及水利工程研究所 === 88 === The distribution of water resources supply in Taiwan has a dramatic variation in temporal and spatial scale within a year. Besides, due to not only the unique geophysical characteristics of the Island but also the limited storage of reservoirs in the long low stream flow period, the water shortage caused by drought occurs once a while. As a result, there are limitations in many aspects for the development and utilization of water resources. This highlights the scarcity of water resources in Taiwan.
In order to reduce the losses due to droughts, the government may adopt risk management process and issues drought warning by using advanced monitoring system, as well as prepare contingency plans to deal with different drought situations in advance. However, whether to reduce the water supply to the agricultural users and transfer the water to domestic and industrial uses is a crucial decision in drought management.
The functions of an early warning system for nature hazards usually can be found in the framework of a drought management. The announcement of a drought period made by decision-makers can be viewed as a warning to the public. Therefore, the decision-makers should improve the forecast accuracy and be ware of the adverse effects to the credibility of warning system by issuing false alarms.
The main purpose of this study is to develop a drought warning model. The conceptual framework for a drought warning system contains three subsystems, including “threshold decision system", "forecast subsystem", and “response subsystem". The "threshold decision system" selects the optimal threshold for issuing a warning, and the "forecast subsystem" provides forecasted values of hazard indicators to the threshold decision system. The "response subsystem" is a model that evaluates the response of organizations and individuals to a warning issued as an output from the threshold decision system. It is expected that the drought warning system can issue a warning early enough for implementing contingency plans to mitigate the drought impacts.
By using stochastic dynamic programming approaches, an optimization model for selecting the optimal decision thresholds for fallow rice paddies has been developed. The drought management of Tsengwen reservoir system of the Chia-Nan area is the case study. The water shortage rate is the major drought indicator in this study. The computation results show that the optimal decision thresholds are 26%, 29% and 35%, when the critical water shortage rates are 15%, 17% and 20%, respectively. The sensitivity analysis also shows that the performance of the optimization model is satisfactory. However, many parameters of the model are estimated or assumed values in current stage. Therefore, both the forecast system and the loss function are suggested to be developed based on actual field data in further studies.
|