應用卡門濾波法建立洪水演算模式之探討

本文以馬斯金更差分模式（Muskingum differential model）做爲河川洪水流量之演算架構，並配合卡門濾波（Kalman filter）之卽時校正參數，觀察時變參數之穩態代表值，以建立適用之洪水演算模式。另應用定率洪水演算模式一般只引用上、下游之入流與出流量，而忽略了河川側流量對實際洪水量估測之影響，因此，本文特就對未加側流量與加側流量進行討論。本文以八掌溪實測資料做爲探討之樣本，結果顯示不論是否加側流量均有極佳之結果，此點說明了利用卡門濾波法建立之洪水演算模式所得之效果是值得肯定的；另方面從模擬、檢定之結果均可以顯示出考慮側流量後之模式比未考慮之模式有較佳之成果。

關鍵字

卡門濾波；定率洪水演算模式

並列摘要

Based on the Muskingum differential Model, a flood routing method is established. Kalman Filter is utilized to update model parameters on a real-time basis. The stable representation values of time-variant parameters are then chosen to make the proposed method more practical. Lateral inflows are usually neglected in general deterministic flood routing models and only the upstream inflow and downstream outflow are considered. In the proposed routing method, lateral inflows can be included. Field data of flood hydrographs at the upstream, downstream, and a side flow of Pai-chung River is used as a study case. Accurate predictions by the proposed method, with or without lateral inflows, are shown by simulation results. This indicates the usefullness of Kalman Filter correction in the flood routing model. The simulation and calibration results also indicate a better prediction can be obtained by considering the lateral inflows.

並列關鍵字

Kalman filter ； Deterministic flood routing model

主題瀏覽