An indirect approach for discharge estimation: A combination among micro-genetic algorithm, hydraulic model, and in situ measurement

Abstract

To develop a flood forecasting system, estimating the discharge hydrograph is essential. In general, discharges at gauged river sites are calculated by applying simple methods such using the relationship of measured stages to discharges, namely rating curves, or multiplying mean velocity with flow cross-sectional area. The flow cross-sectional area can be determined using measured stages from river geometry surveys. The mean velocity is considered to be the measured surface velocity multiplied by a conversion factor. The conversion factor can be estimated by using the regression approach given a known discharge. However, to obtain discharge for extreme events is difficult. Extrapolation was necessarily made among known discharges to “guess” the discharge hydrograph during floods. Therefore, a novel approach which combines micro-genetic algorithm (µGA), a one-dimensional (1-D) flood routing model, and onsite instrumentation is being proposed to obtain the optimal conversion factor, and therefore the discharge hydrograph. This approach was validated using two events: one synthetic test and one recorded event at Yilan River. The results showed that µGA efficiently converged to an optimal conversion factor which showed a less than five percent difference when comparing with synthetic versus observed values. A sensitivity analysis was also conducted to assess the impact of the quantity of selected gauged stations on the value of optimal factor in the optimization process.