Forecasting Outlet Water Level with Yuansantze Flood Diversion Tunnel for Keelung River Upstream Watershed

Yuansantze Flood Diversion Tunnel is located at the upstream of Keelung River. The Tunnel diverts part of flood from upstream basin discharge during typhoon periods. It results in over-estimation of forecasted water levels at the Dahua Bridge station (Wu. et al., 2004). This paper takes into accounts of flood diversion to extend capability of previous forecast model. The accurately predicted results at the Dahua Bridge Station can provide with better forecasting of downstream flood water levels. This improvement will allow inhabitants residing along river to be evacuated timely. Stochastic relation includes flood diversion, present and the antecedent records at the river gage, and rainfalls are primarily concerned in the model. This paper assumes that water stage at the outlet of watershed at any time is linear combination of all the affecting factors. Based on historical data during typhoons events, a recursive relationship is developed by employing the least squares method and simulated annealing algorithms. With the recursive relationship formula, water levels at the Dahua Bridge Station can be predicted more accurately. Present model is applied to predict water levels at the Dahua Bridge Station for the Typhoon events of Krosa and Jangmi. Good agreement between forecasted and measured results is observed.