Flood forecasting using semi-distributed hydrological model coupled with weather research and forecasting model

• Main Author: Zaidi, Syeda Maria
• Format: Thesis
• Language: English
• Published: 2019
• Subjects: TA Engineering (General). Civil engineering (General); TC Hydraulic engineering. Ocean engineering
• Online Access: http://umpir.ump.edu.my/id/eprint/31123/1/Flood%20forecasting%20using%20semi-distributed%20hydrological%20model%20coupled%20with%20weather%20research.pdf; http://umpir.ump.edu.my/id/eprint/31123/

Kuantan River Basin (KRB), is an important watershed of Kuantan District which has been experiencing floods since decades. The incomplete information of hydro-meteorological data, and insufficient rainfall and streamflow gauging stations remain the key factors influenced on flood forecasting accuracy. This study aimed to cope with the problem by bridging the gap of missing hydro-meteorological data using Weather Research and Forecasting (WRF) model coupled with Hydrologic Engineering Center-Hydrologic Modeling System (HEC-HMS) model. Three rainfall event categories (extreme, heavy, and moderate) were used to evaluate the model’s capability in simulating flood events. The research was covered 4 objectives; (i) to evaluate the performance of microphysics (MP) and cumulus (CU) schemes parameterization for WRF model,(ii) to identify the best physical schemes combination of WRF for precipitation forecasting at KRB iii) to determine GIS-based hydrological model parameters for KRB, and (iv) to quantify the ability and accuracy of proposed flood forecasting framework based on a coupled semi-distributed hydro-meteorological model. Performance of 48 combinations of WRF schemes including 8 MP and 6 CU schemes were first evaluated to simulate single rainfall event. Then selected top 5 best WRF schemes combinations were further investigated to determine the highest performance scheme to simulate events for all categories in KRB. All the obtained results were validated against the observed rainfall data. HEC-HMS model integrated with ArcGIS was used estimate flood hydrographs. Statistical indices include Percentage Error (PE), Nash-Sutcliff Efficiency (NSE), Root Mean Square Error (RMSE), Hit Rate (HR), False Alarm Ratio (FAR), Proportion of Correction (PC), Threat Score (TS) and Bias (B) were applied to evaluate the model performances. The results of the 48 schemes simulations revealed that all the parametrized schemes were found less sensitive to HR and FAR. an average range of PC (0.61 to 0.67), TS (0.55 to 0.67), and RMSE (41.8 to 54.4) indicated the parametrization of WSM6GF, SBUBMJ, LinGF, MDMBMJ, and MDMGF performed relatively better to simulate the event Comparison results of objective (ii) identified SBUBMJ as the most suitable schemes to capture spatial and temporal rainfall in KRB with mean average PE of ±5.1%, ±20.2%, ±23.7% for extreme, heavy, and moderate rainfall, respectively. In HEC-HMS streamflow calibration and validation processes showed that the parameters Soil Conservation Service- Curve Number (SCS-CN) and Storage Coefficient (R) were found to be sensitive to the model performance. Validation results of the coupled WRF and HEC-HMS simulation revealed satisfactory performance in simulating heavy rainfall events with NSE ranges from 0.59 to 0.65 and 0.73 to 0.83, PE for peak discharge ranges from -23.30% to -36.37%, and peak-volume ranges from -20.8% to -28.9%. Good agreement between the models was identified in moderate rainfall events with NSE ranges 0.73 to 0.83, PE for peak discharge ranges from -6.89% to 14.48%, and peak volume range from 4.7% to 4.9%. For the extreme events, the models indicated low performance with NSE ranges from 0.40 to 0.06, PE of peak discharge from -15.74% to 17.23%, and peak volume from -14.65% to -26.06%. From the overall analysis, the study has determined that WRF model can be applied as the best alternative meteorological input to be used for sparse rainfall gauge areas or areas where rainfall observation stations fail to function. Hence the model framework is significant in providing reliable information on flood forecasting by considering about average percentage error of about ±16% to ±25% flow discharge values.