River Modelling for Flood Risk Map Prediction: A Case Study of Kayu Ara River Basin, Malaysia

River Modelling for Flood Risk Map Prediction: A Case Study of Kayu Ara River Basin, Malaysia Sina Alaghmand* 1, Rozi b. Abdullah 2, and Ismail Abustan 2 Abstract This paper presents a simple methodology for river flood risk map prediction in an urbanized area. River flood risk map is a function of hazard, vulnerability and exposure. Hence, in this case, water depth and flow velocity (river flood hazard), land-use type, road accessibility and debris flow (vulnerability and exposure) were incorporated. Furthermore, a systematic methodology was proposed in order to develop and predict river flood risk map for a range of defined scenarios. Therefore, a total of 6 scenarios were identified including three rainfall magnitude (20 year, 50 year and 100 year ARI) and two river basin development conditions (existing and ultimate). The risk components were combined in GIS interface and categorized based on a proposed risk value classification. This case study confirms the efficiency of the proposed method to some extent. However, more detailed analysis need to be undertaken towards a well-developed and applicable framework. Index Terms Flood hazard map, Flood risk map, HEC-RAS, HMS-HMS, Kayu Ara River Basin. I. INTRODUCTION Starting in the year 2000s, extreme rainfall events with high intensity is no longer a new issue in Malaysian urban cities, especially in the West Coast area. This phenomenon is formed mostly through convection process [1]. Hence, flooding is one of the major natural hazards affecting communities across Malaysia and has caused damages worth millions of dollars every year. For instance, the required allocation for flood mitigation projects has increased almost 600% (RM 6000 million) for the 8 th Malaysian Plan compared to RM 1000 million during the 7 th Malaysian Plan [2]. Natural risk can be defined as the probability of harmful consequences or expected loss (of lives, people injured, property, livelihoods, economic activity disrupted or environment damaged) resulting from interactions between natural or human-induced hazards and vulnerable conditions [3]. Risk is sometimes taken as synonymous with hazard, but risk has additional implication of the chance and probability a particular hazard actually occurring. In fact, hazard refers to the probability of a potentially dangerous phenomenon occurring in a given location within a specified period of time [4]. Therefore, risk does not exist if exposure to a harmful situation does not or will not occur [5-6]. River flood mapping is the process of determining inundation extents and depth by comparing river water levels with ground elevation. The process requires the understanding of flow dynamics over the river and the 1 Discipline of Civil Engineering, School of Engineering, Monash University Malaysia, Bandar Sunway, Selangor, Malaysia, Email: sina.alaghmand@monash.edu 2 School of Civil Engineering, Universiti Sains Malaysia, Penang, Malaysia adjacent floodplain, topographic relationships and the sound judgments of the modeler. In fact, river flood mapping is the foundation of river flood risk prediction, which can be produced using water depth, flood extent, flow velocity and flood duration maps [7]. All the existing methods for flood mapping can be grouped into three major categories namely analytical, historical and physiographic methods [8]. These three methods use same procedure to delineate floodplain boundaries by determining the flood elevation at each river cross section. The boundaries are then interpolated between the cross section. The three methods differ only in the way they determine the water surface profile. River flood mapping involves three main components as follows [9]: i. GIS interface as pre-processor (to extract geospatial data) and post-processor (to visualize model outputs) (i.e. HEC-GeoHMS and HEC-GeoRAS). ii. Hydrological model, which develops rainfall-runoff hydrograph from a design rainfall or historic rainfall event (i.e. HEC-HMS). iii. Hydraulic model, which routes the runoff through river channel to determine water profiles and flow velocity (i.e. HEC-RAS). The increasing availability of powerful GIS software packages offer new opportunities for engineers to perform flood mapping incorporated with hydrological and hydraulic models [7]. This is essential as flood modelling is inherently spatial and hydrological and hydraulic models have large spatially distributed data requirements [10]. In recent years, efforts have been made to integrate hydrological and hydraulic models and GIS to enhance the model outputs, which led to the establishment of a new branch of hydraulics and hydrology, namely, hydro-informatics. In general, there are four methods for incorporating of river basin models into GIS. They are classified as stand-alone system, loose coupling system, tight coupling system and embedded system. In this regard, HEC-GeoRAS and HEC-GeoHMS, which were utilized in this study, are among tight coupling system. This paper presents a simple methodology to generate river flood risk map in an urban area. To this aim, water depth and flow velocity (river flood hazard), land-use type, road accessibility and debris flow (vulnerability and exposure) were incorporated. Moreover, the proposed methodology was utilized for a case study in Kuala Lumpur for numbers of defined scenarios. A. Method II. MATERIAL AND METHOD The proposed methodology in this research included five main components; hydrological modelling, hydraulic modelling, river flood visualization, river flood hazard mapping and river flood risk mapping. Note that, in this http://dx.doi.org/10.15242/iae.iae1214510 96

research, flood mapping, flood hazard mapping and flood risk mapping were differentiated. Flood mapping consists of visualization of hydraulic model results in forms of water depth and flow velocity, whereas flood hazard mapping represents combination of water depth and flow velocity to define various level of flood hazard. Finally, flood risk map was the resultant of hazard map, land-use, road accessibility and debris flow risk. This research involves integration of two numerical models: HEC Hydrologic Modelling System (HEC-HMS) as a hydrologic model to simulate rainfall-runoff process and HEC River Analysis System (HEC-RAS) as a hydraulic model to route the runoff through river to determine water surface profiles and flow velocity. In order to assess the effects of rainfall event magnitude (ARI) and also river basin land-use development condition on the river flood hazard maps, six scenarios were defined. These include three different ARI (20, 50 and 100 years) and two land-use development conditions (existing and ultimate). In all the defined scenarios rainfall events with 60 minutes duration are taken into account. Note that, in order to differentiate between the various developments conditions, different percentages of imperviousness were defined for each development conditions (Table I). TABLE I: PERCENTAGE OF IMPERVIOUSNESS AREA IN DIFFERENT DEVELOPMENT CONDITIONS Development Condition Existing Ultimate Sub-river basin 1 26% 90% Sub-river basin 2 26% 90% B. Study area Sub-river basin 3 66% 90% Sub-river basin 4 36% 90% Sub-river basin 5 66% 90% Kayu Ara river basin is the case study in this research, which is located in Kuala Lumpur, Malaysia. The study area covers an area of 23.22 km 2 and is geographically surrounded within N 3 6 to N 3 11 and E 101 35 to E 101 39. Kayu Ara river basin is a well-developed urban area with different land-use and also high population density, and also, 10 rainfall stations and one water level station at the outlet were available (Figure 1a). Fig. 1. Location of rainfall and water level stations in Kayu Ara river basin, Geospatial data extracted using HEC-GeoHMS. III. RESULTS AND DISCUSSION A. Hydrological modelling HEC-HMS was used as the hydrological model in this research, which was linked to GIS using HEC-GeoHMS extension to extract geospatial input data (Figure 1b). To develop a reliable numerical model establishing the credibility of the model is essential. It includes sensitivity analysis, calibration and validation processes. Sensitivity analyses were applied to highlight the most sensitive parameters in the hydrological model. The results of sensitivity analysis showed that imperviousness, lag-time and peaking coefficient were the most sensitive parameters. Moreover, the recorded time series data with 10 minute intervals was available since 1996. Among the recorded rainfall and water level time series, 18 rainfall events were selected for calibration and 18 rainfall events for validation. The established hydrological model for Kayu Ara was used to simulate the rainfall-runoff process based on rainfall design hyetograph extracted from MSMA guideline [11]. Therefore, IDF polynomial equation for, three ARI (20, 50 and 100 years) were used to derive the design rainfall for 60 minutes events as an input to HEC-HMS hydrological model (Figure 2). The modeled runoff hydrograph for each scenario, produced by the validated hydrological model, are shown in Figure 3. ARI 20 year ARI 50 year http://dx.doi.org/10.15242/iae.iae1214510 97

ARI 100 year Fig. 2. Hyetographs of the design rainfall for the defined scenarios. Fig. 4. Water depth and flood extend distribution for ultimate development condition: ARI 20 year, ARI 50 year and ARI 100 years Fig. 3. Simulated runoff hydrographs for rainfall events for the six defined scenario; Existing development condition, Ultimate development condition. B. Hydraulic modelling and flood visualization The modeled runoff hydrographs were used as the main input for hydraulic model. The hydraulic modelling in this research was conducted for the last 5.1 km of Kayu Ara river. The hydraulic model, HEC-RAS, were incorporated with GIS using HEC-GeoRAS extension to prepare geospatial data. Therefore, 25 surveyed cross-sections at 200 m interval were used to create Digital Elevation Model (DEM) of the main channel. The calibration process of HEC-RAS consisted of a total of 20 events. The flood events for calibration were selected from the historical data since 1996 at the water level station, which was located at the outlet of Kayu Ara river basin. In the validation process, HEC-RAS simulated the flood events using the calibrated parameters. A total of 10 events, other than calibration rainfall events, were employed to validate the model. Furthermore, results of the hydraulic model were visualized using HEC-GeoRAS. The generated flood extend maps, flood water depth and flow velocity distribution for the ultimate development condition scenarios are represented in Figures 4 and 5. Fig. 5. Flow velocity distribution for ultimate development condition: ARI 20 year, ARI 50 year and ARI 100 years. C. Flood hazard mapping Flood hazard map covers the geographical areas which could be flooded according to different scenarios [12]. The magnitude of the damage depends on the flood characteristics such as water depth and flow velocity [13]. In this study, water depth and flow velocity are considered as two main http://dx.doi.org/10.15242/iae.iae1214510 98

parameters associate with river flood hazard. In order to produce the flood hazard maps for Kayu Ara river basin NSW flood development manual [14] was applied. With some modification, four river flood hazard categories were determined consisting of low, medium, high and severe. These are shown in Figure 7. Furthermore, the water depth and flow velocity maps which were generated in the previous section were overlaid and analyzed to classify flood hazard for the defined scenarios. These are represented in Figures 7 and 8. Fig. 6. Flood hazard Categories [14] Fig. 8. Flood hazard map for ultimate development condition at Kayu Ara river basin: ARI 20 year, ARI 50 year and ARI 100 years. D. Flood risk mapping Risk can be defined as the probability of a loss, and this depends on three elements including hazard, vulnerability, and exposure. If any of these three elements in risk increases or decreases, then the risk increases or decreases, respectively. Exposure refers in the context of floods only to the question whether people or assets are physically in the path of flood waters or not, vulnerability may be defined as the conditions determined by physical, social, economic, and environmental factors or processes, which increase the susceptibility of a community to the impact of hazards. This concept is demonstrated in Figure 9. Fig. 7. Flood hazard map for existing development condition at Kayu Ara river basin: ARI 20 year, ARI 50 year and ARI 100 years. Fig. 9. River flood risk definition [13] Given Figure 9, to produce flood risk map for the study area four main factors were included, those can fulfil three components of risk definition. The first one was flood hazard map which was created by combining water depth and flow velocity. Then, the flood hazard map was classified in four http://dx.doi.org/10.15242/iae.iae1214510 99

classes, low, medium, high and extreme hazard. For each class one value was defined that shows its level in term of flood risk (Table II). TABLE II: RISK VALUE FOR FLOOD HAZARD CLASSES Flood hazard class Low Medium High Extreme Risk value 1 2 3 4 The second factor which involves in flood risk estimation is land-use type, which contributes to the vulnerability factor. For instance, in land-use map, residential and commercial areas maintain higher risk in comparison with other land-uses such as roads, water bodies, parks, forests and open areas. The land-use map was classified in two classes, residential and commercial and non-residential and non-commercial. Therefore, in land-use map pixels with residential and commercial class are assigned 4 and non-residential and commercial are assigned 1 (figure 10b). The third factor for preparation of flood risk map was accessibility to main road. A vulnerability analysis considers the population and structures at risk within the flood inundated area. In term of vulnerability, the emergency responses may be required, which includes the need for evacuation and emergency services. In river flood event the accessibility to main road is important at least for two purposes; evacuation of people from inundated area and emergency services. In fact, locating far from the main road may lead to slower emergency response during flood inundation and consequently higher risk (Figure 10c). The fourth factor which is included in the proposed methodology is debris flow risk. As the depth of floodwater increases debris begin to float. For instance, if the flood velocity is significant, buildings can be destroyed and cars and caravans can be swept away. In certain areas, the buildup of debris and the impact of floating objects can cause significant structural damage to buildings and bridges at the downstream. Hence, fast moving floodwaters carrying debris expose a greater threat to both people and structures, than those with no debris. In this case, debris risk was considered as a function of distance from the source of the flood. In order to classify the risk value for debris flows for Kayu Ara river basin, the distance from the most upstream is considered and assumed by moving towards downstream amount of debris, and consequently, the debris risk is increasing (Figure 10d). All the above mentioned maps were converted into raster format with 1 m pixel size. In each map, proper values were assigned for each pixel and combined. Then, the result map (flood risk) was created based on the summarized risk values of river flood hazard, land-use type, main road accessibility and debris flow. For instance, minimum value for the flood risk map is 4 which reflects the combination of the pixels with Low class in flood hazard map, non-residential and commercial class in land-use type map, 0-100 m class in main road accessibility and 0-1500 m class in debris flow hazard map. On the other hand, maximum is 16 represents pixels with Extreme class in flood hazard map, residential and commercial class in land-use type map, > 300 m class in main road accessibility and > 4500 m class in debris flow hazard map. Finally, the flood risk map was categorized into four classes based on the value of each pixel; Low (values 4,5 and 6), Medium (values 7, 8 and 9), High (values 10, 11 and 12) and Extreme (values 13, 14, 15 and 16). Figure 12 shows summarized the proposed methodology for flood risk map prediction. Figures 13 and 14 represent the flood risk maps for Kayu Ara river basin. Flood hazard map Land-use map Main road accessibility map (d) Debris flow hazard map (e) Flood risk map Fig. 10. Summary of proposed methodology for flood risk map prediction. Fig. 11. Flood risk map for existing development condition at Kayu Ara river basin: ARI 20 year, ARI 50 year and ARI 100 years. http://dx.doi.org/10.15242/iae.iae1214510 100

Fig. 12. Flood risk map for ultimate development condition at Kayu Ara river basin: ARI 20 year, ARI 50 year and ARI 100 years. IV. CONCLUSION The following conclusions can be considered according to results of this research: i. An increase in river basin land-use development condition leads to increase of imperviousness of the river basin and volume and peak discharge of the generated runoff hydrograph. ii. HEC-GeoRAS and HEC-GeoHMS are powerful tools as pre-processor for preparation of geospatial input data and also as a post-processor for visualization of the hydraulic model results for HEC-RAS and HEC-HMS models, respectively. iii. The generated water level by hydraulic model is significantly sensitive to river basin land-use development condition and magnitude (ARI) of rainfall event. iv. Flood water depth and flow velocity are the most important elements of flood hazard mapping. However, the generated flood hazard pattern distribution is more influenced by water depth in comparison with flow velocity. vi. The proposed methodology for flood risk map prediction in an urban area is a simple technique which can contribute to the research area. This case study confirms the efficiency of the method to some extent. However, more detailed analysis need to be undertaken towards a well-developed and applicable framework. ACKNOWLEDGMENT This work was funded by Universiti Sains Malaysia under Short-term Research Grant. REFERENCES [1] Embi, A.F. and D. N. Mohd, The Changing Microclimate in Urban Areas The Urban Heat Island Connection. 2004. [2] Abdullah, K., Flood Problem and Storm water Management Manual for Malaysia (USMM). Keynote Address:Flood Problem and Storm water Management Manual for Malaysia (USMM), in Seminar on Introduction to Storm water Management Manual for Malaysia. 2000: Kelantan. Malaysia. [3] ACS, Understanding Risk Analysis. 1998, Washington, USA: American Chemical Society. [4] Alexander, D., Natural Disasters. 1993, New York, USA: Chapman & Hall. [5] Omen, G.S., Kessler, A.C., Anderson, N.T., Chiu, P.Y., Doull, J., Goldstein, B., Lederberg, J., McGuire, S., Rall, D. and Weldon, V.V., Framework for Environmental Health Risk Management. 1997, The Presidential/Congressional Commission on Risk Assessment and Risk Management: Washington, USA. [6] Sinnakaudan, S.K., A. Ab Ghani, M.S.S. Ahmad, and N.A. Zakaria, Flood risk mapping for Pari River incorporating sediment transport. Environmental Modelling & Software, 2003. 18(2): p. 119-130. http://dx.doi.org/10.1016/s1364-8152(02)00068-3 [7] Alaghmand, S., R. Abdullah, I. Abustan, M.A.M. Said, and B. Vosoogh, Gis-based river basin flood modelling using HEC-HMS and MIKE11 - Kayu Ara river basin, Malaysia. Journal of Environmental Hydrology, 2012. 20: p. 1-16. [8] Smith, K. and R. Ward, Floods: Physical Processes and Human Impacts. 1998, Chichester, USA: John Wiley & Sons Ltd. [9] Snead, D.B., Development and Application of Unsteady Flood Models Using Geographic Information Systems, in Civil Engineering. 2000, The University of Texas at Austin: Austin, USA. [10] Horritt, M.S. and P.D. Bates, Evaluation of 1D and 2D numerical models for predicting river flood inundation. Journal of Hydrology, 2002. 268(1-4): p. 87-99. http://dx.doi.org/10.1016/s0022-1694(02)00121-x [11] DID, Urban Stormwater Management Manual for Malaysia. 2000, Kuala Lumpur, Malaysia: Department of Irrigation and Drainage Malaysia. [12] EXCIMAP, Atlas of Flood Maps. 2007, European Exchange Circle on Flood Mapping. [13] APFM, Urban Flood Risk Management, A Tool for Integrated Flood Management. 2008, The Associated Program on Flood Management: Australia. [14] NSW, Floodplain Development Manual: the management of flood liable land. 2005, Department of Infrastructure, Planning and Natural Resources. New South Wales Government: Australia Sina Alaghmand was born in July 1983 in Iran. He received his BSc in irrigation engineering from Gorgan University of Agricultural Sciences and Natural Resources in 2006. Then, he moved to School of Civil Engineering of Universiti Sains Malaysia to study river engineering, where he obtained his MSc in 2009. Sina received his PhD in Civil (Water resources) Engineering at University of South Australia in 2014. He has an extensive research experience in wide range of water resources engineering. This includes hydrological and hydraulic modellings for flood map predictions, which he wrote a thesis on this topic during his MSc study (the present paper). Moreover, he has comprehensive research experience on surface-groundwater interactions and floodplain salinity interception, which was his PhD research topic. He has published more than 15 peer-reviewed journal papers and numbers of conference papers. Currently, he is a lecturer at Discipline of Civil Engineering at Monash University Malaysia. Also, he holds an adjunct research fellow position at University of South Australia. Dr. Sina Alaghmand has been recipient of numbers of academic honors and awards during his academic career. These include five scholarships for his BSc, MSc and PhD, Gold medal for his MSc research (UNESCO-IHP Malaysia) and recognition for his PhD research (Goyder Institute for Water Research and Australian Water Association). http://dx.doi.org/10.15242/iae.iae1214510 101