Optimized flood inundation mapping using integrated gis and hydraulic model for ungauged river basin

• Main Author: Mokhtar, Ernieza Suhana
• Format: Thesis
• Language: English
• Published: 2018
• Subjects: Hydrology; Floods
• Online Access: http://psasir.upm.edu.my/id/eprint/71382/1/FK%202018%2083%20IR.pdf; http://psasir.upm.edu.my/id/eprint/71382/
• Institution: Universiti Putra Malaysia

Flood inundation mapping is one of the efficient methods for prediction of flood hazard and risk areas for emergency response and city development planning. River discharge and flood depth are critical parameters in hydraulic modelling for accurate flood hazard estimation. However, limited availability of observed discharge and river morphologies data results in the erroneous calculation and imprecise flood simulation and forecasting. Several empirical equations have been developed in order to predict the discharge. But, the impact of flood inundation mapping via minimum hydraulic variables has not been widely investigated. In addition, resampling techniques have been applied in order to increase the flood prediction; however, studies on the effect of resampled data with respect to the elevation of different land-use categories are limited. This study attempts to determine a suitable discharge equation and assess the errors of flood inundation mapping at the ungauged station. The study was carried out along Padang Terap River, Kedah Malaysia using Interferometric Synthetic Aperture Radar (IFSAR) and light detection and ranging (LiDAR) DEMs. Through utilisation of this dataset water surface elevation (WSE) was delineated via Manning, Dingman and Sharma, and Bjerklie’s equations. The Dingman and Sharma’s equation which employs observed data presented a significantly noble agreement with measured and predicted WSE, followed by Manning and Bjerklie equations with the similarity of 80%, RMSE value of 2% and relative error of around 13%. Next, the uncertainty of hydraulic variables was investigated via Bjerklie’s equation, while the sensitivity analysis was evaluated through Monte Carlo simulation. Furthermore, a method for calculation of discharge without ground data via GIS technique was proposed. In addition, the effect of applying normal depth and known water surface (W.S.) boundary conditions were examined and the flood extent was verified with TerraSAR-X and historical flood marks. The F-statistics value was found to be 0.64-0.66 for normal depth and known W.S. boundary condition, respectively. By utilising modified IFSAR and known W.S. boundary condition, the mean absolute error (MAE), root mean square error (RMSE) and Nash–Sutcliffe efficiency (NSE) were found to be 0.261, 0.365 and 0.808. Quality of the IFSAR elevation data was assessed by comparing the output with observed Global Positioning System (GPS) and 15 cm resolution LiDAR on different land-use types. Results indicated that the LiDAR, original and the resampled IFSAR DEMs are correlated in elevation value about 90%. The equation was interpolated on the original and resampled IFSAR DEMs to improve the medium-resolution data for WSE delineation. Then, an additional sensitivity analysis was carried out at 95% confidence interval. The findings revealed that the optimize IFSAR5m is superior to the original DEM based on the MAE and RMSE values of 0.785 m and 1.071 m, respectively. WSE generated in HEC-RAS via different cross-section intervals (50, 100, 150 and 200 m) revealed that 100 m cross-section of the modified IFSAR DEM (MID1m) is the most suitable for flood extent mapping with MAE of 1.053 m. Overall, the novelty of this is attributed to its evaluation of the performance discharge equations for flood mapping, especially in a spatial context. Furthermore, the uncertainties obtained from the hydraulic variables utilised in the discharge equation should be recognized. Consequently, by considering that these errors contributed to the flood hazard maps, the prediction of the inundated area and water depth could be produced accurately, especially at the data-scarce areas. Moreover, this study contributes to development of novel methodological approach by estimating discharge without ground data observation on optimized DEM with limited data available. The outcome of this research may support the current flood modelling in mitigation planning and strategies.