圖書典藏及影音數位平台

由於全球氣候變遷效應影響，台灣近年之颱洪極端事件日益嚴重，降雨量及降雨強度亦呈增加趨勢。面對此問題，需認知防洪工程有其保護極限，水文降雨條件超過保護標準時，仍有面臨淹水的可能。水利署目前已經由雨量、流量、水位、潮位等即時監測進行洪水預報作業，於洪水期間發布淹水預警報至防災體系之相關單位及告知民眾，以期降低或避免洪水期間可能發生之災害。惟目前水利署之洪水預警系統多屬一維模擬模式，如今隨著模式演算法則精進，以及電腦運算速度不斷提升，具有穩定排程作業能力與快速平行計算架構的高效能計算整合平台，將可以縮短大量的資料準備時間與數值模擬計算時間，故即時二維淹水模式實仍未來防災工作的下一個努力的方向。水利署之「水文氣象規劃整合平台」多年來已整合氣海象、雷達、雨量水位等水文資料，亦已完成整合多個水文水理模式。而本所目前已陸續推動全臺之二維淹水模擬，為將本所模擬之模組納入水利署之平台以期能完成二維即時淹水模擬，增加未來防災之應變能力，爰辦理本計畫。

For 1D water level forecast, the integration platform of hydrology and meteorology in WRA can provides the forecast data of 3 to 24 hours lead time within one hour of operation during typhoon. This result is very good, but for 2D inundation forecast, it is still difficult because the computation time is too long. Therefore, the real-time inundation area cannot be provided, instead of warning information such as potential flood map, warning rainfall, vulnerability of flood.To keep going on real-time flood forecast, this study focuses on technology of integration platform, flood inundation model and high-performance cloud computing environment. The boundary conditions for flood inundation model can be provided through the integration platform by multi-data source importing, data checking, data addendum, interpolation in space and time, data format transform. Furthermore, the flood inundation model must be adjusted for the purpose of computing efficiency and accuracy. By parallel computing and automatic architecture, the simulation time and mistakes can be reduced. Finally, the cloud computing and virtual machine can provide the good computing environment of high-performance and stability.The experience of FEWS_Taiwan is as a reference to develop the architecture of high-performance flood inundation computing and to enhance the computing power by automation and parallel. For the analysis of different mesh number of model, the optimized allocation of simulation can be found to reach the goal of operation model. The power of real-time forecast can provide a lot of information for early warning and emergency response.This study is a two-year project. The first year focuses on study of the interface between SOBEK model and FEWS_Taiwan integrated platform. Taichung city is used as a study area which is a high land drainage problem. Furthermore, the test for simulation of domain-decomposition based on computation of space resolution 20m is completed. The second year will keep going on developing FEWS_Taiwan integrated platform towards “Operation”. Tainan city will be used as a study area that is a low land drainage problem.The results of the first year are summarized as follows：1. Literature review on flood inundation models, disaster prevention, interface between model and integrated platform.2. Observed and predicted data such as rainfall, water level and discharged are linked to integrated platform.3. Development and correction of 2D flood inundation model for Taichung city.4. Domain decomposition and multi-grid allocation.5. Integration between 2D flood inundation model and platform.6. Plan for high-performance issue on integrated platform of hydrology and meteorology.7. Efficiency analysis and evaluation for 2D flood inundation computing.8. Educations for operation of FEWS_Taiwan integrated platform.