Verification of the Observed Discharge Data by Using Tank Model and Flood Forecasting by Gumbel’s Method in Kathmandu Valley of Bagmati Basin at Khokana

Abstract

Flow estimation at a point in a river is vital for a number of hydrologic applications including flood forecasts. This paper presents the results of a basin scale rainfall-runoff and hydraulic modeling in Bagmati river basin in Nepal. A distributed lumped conceptual flood forecasting model, namely Tank Model was calibrated in this study for the Bagmati River Basin of Kathmandu valley at Khokana. Ten years rainfall, evaporation and discharge data were collected from DHM and compiled as an input toTank Model. The simulation was done for a period of ten years data, eight years during calibration and two years during validation. Statistical method and objective functions were applied to evaluate the verification capability of the Tank Model. Eight years of flood data were used to calibrate the Tank Model and the performance of the model was verified by using 2007 and 2008 data. A set of tank coefficients that suit tank configuration selected for Bagmati River Basin were determined by trial and error calibrations. The predicted peak discharge was close to the observed value and the smaller discharges followed the observed trend. A mean of 15m 3 /s, 11m 3 /s was calculated for the observed discharge while the mean of the simulated discharge was 16m 3 /s, 12m 3 /s, during the validation period of 2007 and 2008 respectively.The standard deviation was 25m /s for the observed discharge and 17m 3 /s, 10m 3 /s for the simulated discharge during the validation period of 2007 and 2008 respectively. The overall correlation coefficient between observed and simulated discharge was 0.84, 0.90, 0.85, 0.91, 0.91, 0.91, 0.83,and 0.79 during calibration period.The maximum instantaneous flow of 942m /s was recorded while the lowest flood flow of 103 m 3 /s was recorded during observation. The17-year mean instantaneous flood flow is 420.76m 3 /s with a skewness of 0.98 and Confidence Level (95%) of 126.31.The maximum instantaneous flow of 820.41 m /s was recorded while the lowest flood flow of 59.25m 3 /s was recorded during simulation. The10-year mean instantaneous flood flow is 221.26m 3 /s with a skewness of 2.64 and Confidence Level (95%) of 158.08. Measured and predicted flood flows show no significant differences hence, a goodness of fit of the Gumbel distribution. 3 /s, 13m 3 3 3