| Summary: | This study investigated the effectiveness of new flood risk index that has been created for the purposed of flood risk control in Malaysia. Sixteen monitoring stations from [om river basins namely Muda River Basin, Kuantan River Basin, Johor River Basin and Langat River Basin were being selected as study areas in this study. This study applied secondary data which was obtained from Department of Drainage and Irrigation (DID) for hydrological data from year 1982 to 2012 and from Town and Country Planning Peninsular of Malaysia for land usc data from 1990 to 2012. All selected data were being tested and evaluated by using integrated Chemometric techniques including Descriptive Statistics (OS). Spearman correlation test, Principal Component Analysis (PCA), Multiple Linear Regression (MLR), Statistical Process Control (SPC), Artificial Neural Network (ANN) and Hierarchical Agglomerative Cluster Analysis (HACA). The Spearman con-elation test confirmed that rainfall variables was categorized as a weak correlation compared to otbcr variables such as stream flow, suspended solid and water levels that have higher levels of relationship with ignificant probability value (p < 0.01). Based on PCA analysis result, it was confirmed that all variables were significant to be selected as variables in developing the new flood risk model due to high correlation coefficient of factor loading with value greater than 0.6. Multiple Linear Regression (MLR) analysis was proved that different location with different direction of development by State Government has different types of contributors for suspended solid that directly flow into the study area. Likewise, SPC was applied to determine the control limits for all hydrological variables involved in this study. Three classes of control limit were calculated based on the time series analysis data. Those classes arc Upper Control Limits (UCL), Center Limit Value (CL V), and Lower Control Limit (L L). Based on the control limit values, the new flood risk index (FRI) was designed to determine the risk level for flood and also able to become as a new reference for flood early warning system in Malaysia. The prediction performance given by ANN proved the capability of FRI due to its accuracy with result greater than 90%. The application of HACA in this study was successfully clustered all year involved in this study based on its risk level. There were three different risk levels of flood (low, medium and high) in this analysis. It was proved that the temporal classification results from this analysis was imilar with the actual flood record obtained from DID (1982 - 2012). This study successfully shows that, the FRI was a capable tool to be applied as flood risk warning system in Malaysia. This new intervention of FRI also able to provide a broader assistance for local authority in developing a better flood risk management for this country.
|
|---|