Available online at www.sciencedirect.com

Journal of Hydro-environment Research 2 (2008) 47e59 www.elsevier.com/locate/jher

Sediment transport modeling for Kulim River e A case study

Chang Chun Kiat*, Aminuddin Ab Ghani, Rozi Abdullah, Nor Azazi Zakaria

River Engineering and Urban Drainage Research Centre (REDAC), Universiti Sains , Engineering Campus, Seri Ampangan, 14300 , , Malaysia Received 10 August 2007; revised 12 March 2008; accepted 9 April 2008

Abstract

Rapid urbanization has accelerated impact on the catchment hydrology and geomorphology. This rapid development which takes place in river catchment will result in higher sediment yield and affect river morphology and river channel stability; it also becomes the main cause for serious flooding in urban areas. Therefore, it is necessary to predict and evaluate the river channel stability due to the existing and future developments. This study proceeds at Kulim River in state, a natural stream in Kedah, Malaysia. The FLUVIAL-12 model, an erod- ible-boundary model which simulates inter-related changes in channel-bed profile, width variation and changes in bed topography was selected for this study. Engelund-Hansen formula and roughness coefficient n ¼ 0.030 were found to be the best combination to represent the sediment transport activity in the study reach, where good agreements were obtained for both water level and bed profiles between the measured data and predicted results by FLUVIAL-12 model. The model simulation results for existing conditions, future conditions and long-term modeling show that the sediment size and channel geometry in Kulim River changed significantly. However, modeled results show that future changes in cross sectional geometry will be limited and erosion along the reach will slow down from 2006 to 2016, thus Kulim River was predicted to be stable at most locations. Ó 2008 International Association for Hydraulic Engineering and Research, Asia Pacific Division. Published by Elsevier B.V. All rights reserved.

Keywords: Alluvial river; Sediment transport; FLUVIAL-12 model; River channel stability; Long-term simulation

1. Introduction will not only affect river morphology but also cause instability in the river channel and hence serious damage to hydraulic River is a dynamic system governed by hydraulic and sed- structures along the river and reducing channel capacity to iment transport process. Over time, the river responses by convey the flood water to downstream. changing in channel cross section, increased or decreased sed- Kulim River (Fig. 1) today is also changing, but mostly in iment carrying capacity, erosion and deposition along the response to human activity. These activities include the channel, which affect bank stability and even morphology development to the year 2010 of the based on changes. Rapid urbanization has accelerated impact on the the Kulim Structure Plan, 1990e2010 (MDK, 1993), rapid catchment hydrology and geomorphology. This development urbanization at Kulim River catchment especially construction which takes place in river catchment areas will cause dramatic for housing estate, the on-going 145 km2 Kulim Hi-Tech In- increase in the surface runoff and resulting in higher sediment dustrial Park and sand mining activities. Frequently floods delivery. Sediment delivery is defined as the cumulative that occur in Kulim River Catchment for the past 20 years amount of sediment that has been delivered passing each cross has caused extensive damage and inconvenience to the com- section for a specified period of time. When this happens, it munity especially October 2003 flood which was close to the 100-year average recurrence interval (ARI). Finally, these changes to the river hydrology and sedimentation will in turn * Corresponding author. Tel.: þ604 594 1035; fax: þ604 594 1036; alter the channel morphology, which can include changes to E-mail address: [email protected] (C.K. Chang). channel cross section, stability and capacity (Chang et al.,

1570-6443/$ - see front matter Ó 2008 International Association for Hydraulic Engineering and Research, Asia Pacific Division. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jher.2008.04.002 48 C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59

Fig. 1. Delineated Kulim River Catchment and Study Reach for FLUVIAL-12 Modeling.

2004, 2005). Therefore, river channel behavior often needs to modeling, or mathematical modeling, or both. Physical model- be studied for its natural state and response to human regula- ing has been relied upon traditionally for river projects, but tion. However, studies of river hydraulics, sediment transport, mathematical modeling is becoming more popular as its capa- and river channel changes may be carried out through physical bilities expand rapidly (Chang, 2006b). In Malaysia, mathe- matical modeling has been widely applied for study related with sediment transport such as Sinnakaudan et al. (2003), Da- Table 1 rus et al. (2004) and Ariffin (2004). Similar attempts were also Range of Field Data for Kulim River Catchment (Chang, 2006a; Ab. Ghani made from previous studies at Kulim River (DID, 1996; Ya- et al., 2007) haya, 1999; Lee, 2001; Ibrahim, 2002; Koay, 2004) which Study Site CH 14390 CH 3014 were conducted to determine the river behaviors and the effec- No. of Sample 10 12 tiveness of the flood mitigation projects due to rapid urbaniza- 3 e e Discharge, Q (m /s) 0.73 3.14 3.73 9.98 tion. However, data available from previous studies, including Water surface width, B (m) 9.0e13.0 13.0e19.0 river survey geometry data, sediment data and hydrology data Flow depth, yo (m) 0.20e0.54 0.36e0.58 Hydraulic radius, R (m) 0.23e0.57 0.40e0.63 were limited and up to year 1999. Besides that, the study also Water surface slope, So 0.001 0.001 was limited to a single storm event and river stability could not e e Mean sediment size, d50 (mm) 1.00 2.40 1.10 2.00 be predicted. Hence, the objectives of this study are to exam- e e Manning n 0.029 0.072 0.024 0.037 ine river stability for a long period due to changes made by na- B/y 23.4e44.8 26.0e52.5 o ture or human activities by evaluating Kulim River sediment yo/d50 126.9e369.01 240.0e550.9 R/d50 141.4e406.6 266.5e570.9 transporting capability (Table 1) and determining effect of Bed load, Tb (kg/s) 0.06e0.33 0.11e0.36 flooding due to rapid urbanization at the study area. It is nec- e e Suspended load, Ts (kg/s) 0.02 0.27 0.03 1.21 essary to evaluate and predict the river channel stability for the e e Total load, Tj (kg/s) 0.09 0.56 0.27 1.35 purpose of river rehabilitation due to the existing and future C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59 49

50

45 3 June 1991, 4am (Q = 43.74 m3/s) 40

35 /s) 3 30

25

20

Discharge, Q (m 15

10

5

0 0 1000 2000 3000 4000 5000 6000 7000 8000 Time (Hour)

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Fig. 2. Input Hydrograph for Year 1991. developments in the river catchment. The historical data from headwaters, the Kulim River catchment is hilly and densely 1991 (Fig. 2), up to 2006 (Fig. 3) will be evaluated and used to forested and Kulim River arises on the western slopes of predict river stability for future development and this will al- Gunung Bongsu Range and flowing in a north-westerly direc- low evaluation of river stability over a 16-year period by con- tion. The river slopes are steep and the channel elevation drops sidering the effect of changes in cross section and sediment from 500 m to 20 m average mean sea level over a distance of load. This paper attempts to give an overview of the channel 9 km. The central area of the catchment is undulating with changes and sediment transport phenomena which cause prob- elevations ranging from 100 m down to 18 m average mean lems with river bank and bed stability in Kulim River. sea level. Currently, the catchment area is undergoing rapid urban development with oil palm and rubber plantations being 2. Study area replaced by rapid urbanization. This is likely to increase the magnitude of flood and will also result in discharge and bed Kulim River catchment is located in the southern part of the erosion increment or scouring and deposition. state of Kedah in the northwestern corner of Peninsular Malay- Frequently floods occur in Kulim River catchment and sia with the total catchment area of 130 km2 (Fig. 1). At the cause extensive damage and inconvenience to the community.

100

90

80

70 /s) 3 60

50

40

30 Discharge, Q (m

20

10

0 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000 Time (Hour)

1991 1992 1993 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Fig. 3. Input Hydrograph for Year 1991 to June 1993, 1997 to June 2006. 50 C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59

15.00 1990/1991 1992 14.00 1993/1994 1995 1996 1998 1999 2000 13.00 3 2001 2002 Q = 0.7 x (yo-5.0) Flood Rating 12.00

11.00

10.00

Stage (m) 9.00

8.00

7.00

6.00

5.00 0 50 100 150 200 250 300 350 Discharge (m3/s)

Fig. 4. Flood Rating Curve at Ara Kuda (CH 0).

The effects of flooding are felt most in built-up areas such as flood discharges, high channel roughness, siltation and block- residential and commercial areas within the urban confine ages by debris and refuse. There are several locations within where property damage is more as compared with that in the urban confines which are repeatedly flooded. According agricultural areas (Chang et al., 2004). Flood has been attrib- to the information given by the Department of Irrigation and uted to overbank spill from river and arising from a number of Drainage Malaysia (DID) and DID flood reports, floods have causes, such as undersized river channels and drains to cater occurred during 1 to 5 October 1989, 31 May to 2 June 1991, 24 April 1994, October 1997, 15 to 24 Nov 1998, Jan 2001, April 2001 and October 2003, which was close to the Upstream (CH14390) 100-year ARI. Kulim River and its tributaries are generally a 100.00 undersized relative to flood discharges and in many cases 90.00 channel roughness is high due to a combination of bank irreg- 6 March 2006 80.00 ularity and in-river vegetation growth (Ab. Ghani et al., 2007). 70.00 The problem of flooding related to undersized river channels is 60.00 compounded by:- 50.00 40.00 - Siltation 30.00 - Partial blockage of bridges by debris during flood 20.00 - Increased flood discharges due to on-going urban

Percentage Passing (%) d50=1.50 mm 10.00 development in the catchment area under Kulim Structure Plan, 1990e2010 (MDK, 1993) 0.00 0.01 0.10 1.00 10.00 100.00 Size Particle (mm) 3. Field data collection b Downstream (CH 0) 100.00 River surveys, flow measurement and field data collection 90.00 8 December 2004 provide the basic physical information such as sediment char- 80.00 acteristics, discharge, water surface slope; which is needed for 70.00 the planning and design of river engineering. In addition to the 60.00 data needed for sediment transport studies, use of a sediment 50.00 transport model (Table 2) also requires field data such as chan- 40.00 nel configuration before and after the changes (Table 3), a flow 30.00 record (Fig. 4) and sediment characteristics (Fig. 5), which are 20.00 generally used for test and calibration of a model. Field mea- Percentage Passing (%) d50=0.75 mm 10.00 surements were obtained during October 2004 to January 2007 0.00 along the selected cross sections at Kulim River catchment by 0.01 0.10 1.00 10.00 100.00 using Hydrological Procedure (DID, 1976; DID, 1977) and re- Size Particle (mm) cent manuals (Yuqian, 1989; USACE, 1995; Edwards & Glys- Fig. 5. Initial Bed Material Size Distributions. son, 1999; Lagasse et al., 2001; Richardson et al., 2001). The C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59 51

Table 2 Summary of Input and Output Parameter for FLUVIAL-12 Category Category Parameter Value Source Input Geometry Cross section per section (Total of cross sections ¼ 120) CH 1900 e CH 14390 (DID 1991 Survey) CH 0 parameter (DID 1995 Survey) Reach lengths per section (Total length ¼ 14.4 km) 1991 (DID 1991 Survey) Roughness coefficient Same by cross section (n ¼ 0.020, 0.025, 0.030, Values static at all levels of flow. 0.035, 0.040 were evaluated during the sensitivity analysis) Radius of curvature per section 1991 (DID 1991 Survey) Sediment Sediment samples 2 sediment size distributions of such samples are Data sampling at CH 14390 (Year 2006) and CH required (upstream and downstream section) 0 (Year 2004) Regular non-erodible bank Generally fix at left and right bank, varies by DID 1991 Survey cross section Sediment transport formula Seven sediment transport formulas were Graf’s sediment formula evaluated during the sensitivity analysis Yang’s unit stream power formula Engelund-Hansen sediment formula Parker gravel formula Ackers-White sediment formula Meyer-Peter Muller formula Singer-Dunne formula Specific Gravity 2.65 Default (Soulsby, 1997) Hydrology Discharge hydrograph Varies by hydrograph (Figs. 7 and 8) Historical hydrograph for Kulim River at Ara Kuda streamflow station Design Hydrograph from past study (DID, 1996) Rating Curve Year 1991 to Year 2002 Developed by DID Hydrology Division Output Geometry Width Changes over time in water surface, bed parameter Depth elevation and thalweg profiles. Simulation of Cross-sectional area curvature induced aggradation and deposition. Slope

Sediment Mean sediment size (d50) Changes over time in sediment transport, Bed material size fractions channel scour and fill, aggradation and degradation Sediment concentration Sediment delivery or the total bed material yield Sediment yield during the study period Hydraulic Water surface Simulated water surface based on input hydrograph Mean velocity Flow data sets for representative cross sections Froude number in the study reach data collection includes flow discharge (Q), suspended load selected cross sections (Fig. 6). A summary with ranges for (Ts), bed load (Tb) and water surface slope (So). Besides hydraulics and sediment data collection is shown in Table 1 that, bed elevation, water surface and thalweg, the minimum (Chang, 2006a; Ab. Ghani et al., 2007). Low sediment trans- bed elevation for a cross section were also carried out at the port rates occurred during the field measurements and the

Table 3 Comparison of Simulated Water Level and Bed Profile with Measured Data during 2 Nov 2004 for Roughness Coefficient n ¼ 0.025, 0.030 and 0.035 Roughness Location Water Level (m) Thalweg Level (m) coefficient n Measured Yang fomula Engelund-Hansen Measured Yang fomula Engelund-Hansen fomula fomula Predicted Difference Predicted Difference Predicted Difference Predicted Difference 0.025 CH 0 7.45 7.80 þ0.35 7.80 þ0.35 5.05 5.34 þ0.29 5.29 þ0.24 CH 3014 8.61 8.14 0.47 8.13 0.48 6.66 6.79 þ0.13 6.96 þ0.30 CH 8185 13.55 12.47 1.08 13.67 þ0.12 12.27 11.68 0.59 12.85 þ0.58 CH 14390 25.61 26.00 þ0.39 25.99 þ0.38 23.45 24.58 þ1.13 24.58 þ1.13 0.030 CH 0 7.45 7.80 þ0.35 7.82 þ0.37 5.05 5.40 þ0.35 5.27 þ0.22 CH 3014 8.61 8.36 0.25 8.29 0.32 6.66 6.57 0.09 6.69 þ0.03 CH 8185 13.55 13.00 0.55 13.36 0.19 12.27 11.87 0.40 12.16 0.11 CH 14390 25.61 26.17 þ0.56 26.14 0.53 23.45 24.58 þ1.13 24.58 0.00 0.035 CH 0 7.45 7.80 þ0.35 7.93 þ0.48 5.05 5.33 þ0.28 5.26 þ0.21 CH 3014 8.61 8.59 þ0.02 8.52 0.09 6.66 7.30 þ0.64 7.53 þ0.87 CH 8185 13.55 13.55 0.00 13.45 0.10 12.27 12.26 0.01 12.31 þ0.04 CH 14390 25.61 26.20 þ0.59 26.29 þ0.68 22.85 24.58 þ1.73 24.58 þ1.73 52 C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59

30 30 Initial Bed (1991) Initial Bed (1991) 25 Water Surface (E-H, n = 0.025) 25 Water Surface (E-H, n = 0.030) Bed Profile (E-H, n = 0.025) Bed Profile(E-H, n = 0.030) 20 Measured Water Level (2 Nov 2004) 20 Measured Water Level (2 Nov 2004) Measured Bed Level (2 Nov 2004) Measured Bed Level (2 Nov 2004) 15 15

10 10 Elevation (m) Elevation (m) 5 5

0 0 0 0 0 00 1000 2000 3000 4000 5000 6000 7000 8000 9000 1 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 10000 11000 1200013000 14000 Chainage (m) Chainage (m)

30 30 Initial Bed(1991) Initial Bed (1991) 25 Water Surface (E-H, n = 0.035) 25 Water Surface (Yang, n = 0.025) Bed Profile (E-H, n = 0.035) Bed Profile (Yang, n = 0.025) 20 Measured Water Level (2 Nov 2004) 20 Measured Water Level (2 Nov 2004) Measured Bed Level (2 Nov 2004) Measured Bed Level (2 Nov 2004) 15 15

10 10 Elevation (m) Elevation (m) 5 5

0 0 0 0 0 0 0 0 00 1000 2000 3000 4000 5000 6000 7000 8000 9000 1 2 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 10000 11000 12000 13000 14000 Chainage (m) Chainage (m)

30 30 Initial Bed (1991) Initial Bed (1991) 25 Water Surface (Yang, n = 0.030) 25 Water Surface (Yang, n = 0.035) Bed Profile (Yang, n = 0.030) Bed Profile (Yang, n = 0.035) 20 Measured Water Level (2 Nov 2004) 20 Measured Water Level (2 Nov 2004) Measured Bed Level (2 Nov2004) Measured Bed Level (2 Nov 2004) 15 15

10 10 Elevation (m) Elevation (m) 5 5

0 0 0 0 0 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 1000 20 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 1300014000 10000 11000 12000 13000 14000 Chainage (m) Chainage (m)

Fig. 6. Comparison of Water Level and Bed Profile for Roughness Coefficient n ¼ 0.025, 0.030 and 0.035 (2 Nov 2004).

100 100 90 90 5 October (1am) 80 3 Actual Data Q = 92.90 m /s 80

/s) 70

3 Prediction 70

60 /s) 3 60 50 19 October (10pm) 50 Q = 7.50 m3/s 40 40 30 Discharge, Q(m Discharge,Q (m 30 20 20 10 10 3 October (9am) 0 Q = 8.80 m3/s 0.00 0.20 0.40 0.60 0.80 1.00 0 20 Weibull Probability 80320 80370 80420 80470 80520 80570 80620 80670 80720 80770 80820 80870 80920 80970 810 Time (Hour) Fig. 7. Flood Frequency Analyses Using Gumbel Extreme Value Type-I Distribution. Fig. 8. Hydrograph of the October 2003 Flood at Ara Kuda (CH 0). C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59 53

40000 Peak After Flood

30000

20000

10000 Sediment Delivery (tons)

0 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 Chainage (m)

Fig. 9. Spatial Variations of the Sediment Delivery during the October 2003 Flood.

mean sediment sizes (d50) show that Kulim River is sand-bed Committee on Hydraulics, Bank Mechanics, and Modeling of streams where d50 ranges from 1.00 to 2.00 mm. The aspect River Width Adjustment (ASCE, 1998) and Federal Inter- ratio for Kulim River was between 23 and 53 indicating that agency Stream Restoration Working Group (FISRWG, it was a moderate-size channel. The water-surface slopes of 2001). In addition, applications of the several commonly the study reaches were determined by taking measurements used sediment transport models have been described by Ab. of water levels over a distance of 200 m where the cross sec- Ghani et al. (2003) and Chang (2006a). These applications tion is located (FISRWG, 2001). In this study, the water-sur- illustrate various capabilities of different models and each sed- face slopes were found to be mild, and an average value of iment transport model has its own limitations. The selection of 0.001 is adopted. the right model under certain constraints requires a comprehen- sive knowledge of the capabilities and features of available 4. Software used models. The sediment transport model, FLUVIAL-12 (Chang, Studies of sediment transport, scour and fill, aggradation 1982, 1984, 1988), which was first developed in 1972, has and deposition analyses can be performed by computer model been selected for the Kulim River study. FLUVIAL-12 is simulation. The rapid pace of computer technology has been developed for water and sediment routing in natural and a milestone for mathematical models in sediment transport. man-made channels. The combined effects of flow hydraulics, As a result, the high demand on the models resulted in devel- sediment transport (Fig. 9) and river cross section changes are opment of many models and the selection of the right model simulated for a given flow period. FLUVIAL-12 model is an under certain constraints requires a comprehensive knowledge erodible-boundary model that includes the width adjustment of the capabilities and features of available models. The component, which simulates inter-related changes in chan- review of capabilities and performance of sediment transport nel-bed profile (Fig. 10), width variation (Fig. 11) and changes models has been discussed by the National Research Council in bed topography induced by the curvature effect. Besides, (1983), Fan (1988), American Society of Civil Engineers Task bank erosion, changes in channel curvature and river

30 Peak Water Surface Bed Profile before Flood 25 Bed Profileat Peak Flow Bed Profile after Flood

20

15 Elevation (m)

10

5 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 Chainage (m)

Fig. 10. Prediction of Water surface and Bed Profile Changes during October 2003 Flood. 54 C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59

26 350

3 25 300 306.6 m /s 24

/s) 250 23 3 22 200 21 150 Level (m) 20 CH 12490 100 19 Discharge (m 50 18 0 10203040506070 0 Distance (m) 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84

Initial Bed Level (Year 1991) Bed Level (Peak) Time (hrs) Water Level (Peak) Bed Level (After Flood) Fig. 12. Design Hydrograph for 2010 Landuse (DID, 1996).

15 14 13 meandering can also be modeled. While FLUVIAL-12 model 12 is for erodible channels, physical constraints; such as bank 11 10 protection, grade-control structures and bedrock outcroppings, 9 may also be specified (Chang, 2006b). Applications of FLU- Level (m) 8 VIAL-12 model in several studies (Chang, 1997; Abu Hasan, 7 CH 5306 1998; Abdullah, 2002; Darus, 2002; Chang et al., 2002; DWR, 6 5 2004; Chang, 2004; Dong et al., 2005; Yazdandoost, 2005; 0 10203040506070 Bahadori et al., 2006; Chang 2006a; Chang, 2008) showed Distance (m) that FLUVIAL-12 was capable to predict river changes caused Initial Bed Level (Year 1991) Bed Level (Peak) by nature and human activities, including general scour at Water Level (Peak) Bed Level (After Flood) bridge crossings, sediment delivery, channel responses to sand and gravel mining and channelization. 15 14 13 5. Model simulation 12 11 5.1. Background 10 9 Level (m) 8 The study reach covers approximately 14.5 km of Kulim 7 CH 3014 River (Fig. 1), from the upstream (CH 14390) to the Ara 6 Kuda streamflow station (CH 0). The geometry data consists 5 of existing survey cross-sections in September 1991 between 0 1020304050607080 Distance (m) CH 1900 to CH 14390 at the upstream of Kulim River. These data, consisting of lateral distance and elevations were pro- Initial Bed Level (Year 1991) Bed Level (Peak) Water Level (Peak) Bed Level (After Flood) vided by Department of Irrigation and Drainage (DID) Measured Bed Level (2 Nov 2004) Kulim/Bandar Baharu. However, the survey of CH 0 cross sec- tion in December 1995 was provided by DID Hydrology Divi- sion for the FLUVIAL-12 modeling requirement. In this study, 11 a total of 120 existing survey cross sections were selected 10 along the study reach to define the channel geometry as the 9 input for FLUVIAL-12 model. The hydrograph for this study 8 was measured by DID Hydrological Division. The input 7 hydrograph at Ara Kuda for year 1991 (Fig. 2) was used for

Level (m) 6 model sensitivity analysis whilst model calibration and valida- CH0 5 tion was using the hydrograph from year 1991 to June 2006 4 (Fig. 3). The rating curve which is used to define discharge 05 101520253035404550 variation of stage (water surface elevation) for the downstream Distance (m) boundary condition is shown in Fig. 4; the shifts in stage- Initial Bed Level (Year 1991) Bed Level (Peak) discharge relationships reflect the variability at Ara Kuda Water Level (Peak) Bed Level (After Flood) Measured Bed Level (8 Dec 2004) streamflow station derived from the past 12-year rating curve for Kulim River. The geometric mean of the bed material size Fig. 11. Modeled Cross Section Changes before and after October 2003 Flood. fractions is adequately described from the sediment size C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59 55

35

Peak Water Surface Bed Profile before Flood 30 Bed Profile at Peak Flow Bed Profile after Flood

25

20

15 Elevation (m)

10

5 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 Chainage (m)

Fig. 13. Water Surface and Bed Profile Changes based on Design Hydrograph. distribution. Two sediment size distributions of such samples 5.3. Simulation under existing condition based on sieve analysis are required at the upstream (d50 ¼ 1.50 mm) and downstream (d50 ¼ 0.75 mm) cross sec- Two major floods have occurred in 2001 and 2003 within tions to specify initial bed material compositions in the river the 46-year span. The review of ranking for the flood at Kulim bed (Fig. 5). The summary of the input and output parameter River catchment indicates that the discharge of 92.90 m3/s for model FLUVIAL-12 are shown as Table 2. measured on 5 October 2003 was the highest discharge measured in a 42-year period since 1960. This value, which is based on measurable peak discharges, is obtained from 5.2. Calibration and validation of FLUVIAL-12 a reliable and long period of record at Ara Kuda streamflow station. A flood frequency analysis was carried out for the The accuracy of a model is limited by the quality and quan- 42-year period of data using Normal Distribution, 2-parameter tity of the input data. Therefore, using available geometry, sed- Log-normal, 3-parameter Log-normal, Pearson Type-III, Log- iment and hydrology input parameter including cross section Pearson Type-III, Gumbel Extreme Value Type-I and spacing will affect the model output. Besides, selection of Generalized Extreme Value. It was found that the best result the sediment transport formula, model calibration for rough- was obtained by Gumbel Extreme Value Type-I (Fig. 7) which ness coefficient is also essential. The simulation of the shows the better agreement to the measured streamflow data FLUVIAL-12 was obtained using 1991 cross section survey and the result showed that this flood event is slightly lower and hydrograph. Based on measured water levels, predictions than the 100-year ARI; the peak discharge at 92.90 m3/s of using both roughness coefficients are close to the observed the event is adopted as the design peak discharge. Conse- data during low flow. However, as the field data was not avail- quently, sediment transport modeling was carried out for this able from year 1991 to 2003, a long-term simulation has been flood event (3 to 19 October 2003) as shown in Fig. 8. carried out to calibrate and validate the model based on the Spatial variations of the sediment delivery during the Octo- recent measured water level and bed level data that were ber 2003 flood are shown in Fig. 9. Sediment delivery gener- obtained in 2004 and 2006. Therefore, the calibration of the ally decreased towards downstream especially near to the sand roughness coefficient using measured water level and bed level mining site at CH 5064. This pattern indicated that erosion in November 2004 is done. As a part of the calibration proce- occurred at upstream and more sediment deposited at down- dure, the model was run for 12-year period between 1991 stream of Kulim River. Peak water surface and changes of to1992 and 1997 to 2006. The results of the model simulation the channel geometry due to scour and fill were depicted by during the calibration period agree very well (Table 3 and the simulated changes in channel bed profile as illustrated in Fig. 6), and it can be concluded that prediction using rough- Fig. 10. From the simulation results, flood level was higher ness coefficient n ¼ 0.030 and Engelund-Hansen formula at the downstream compare to the upstream of Kulim River. were in good agreement with measured water levels and bed Whilst, the results also show that scour of the bed occurred profiles and used for model validation. As a part of validation, at upstream and the cross sections near to the sand mining measured water levels and bed profiles, during September area (CH 5064) were subjected to greater changes than other 1991, January 2005 and March 2006 was compared to the pre- cross sections. Commonly, channel degradation was predicted dicted water levels and bed profiles by FLUVIAL-12. Long- at most cross sections at Kulim River after the flood event. term simulations including of the historical flood events Fig. 11 shows the example of cross section changes for several showed very good results for both calibration and validation. locations along Kulim River. In general, the river is stable at Good agreements were obtained for both water level and most locations after October 2003 flood with the exception bed levels between the measured and predicted by FLUVIAL- of CH 5306 and CH 12490 where lateral migration is pre- 12 model. dicted at these two locations. 56 C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59

5.4. Simulation for future condition with time. The pattern of the sediment delivery shows a sharp decrease from upstream to downstream of Kulim River. The The design flood hydrograph for the Kulim River based on selective sediment transport has resulted in the decreasing 2010 landuse (DID, 1996) is shown in Fig. 12. The peak flow trend of the sediment delivery which indicates that long of the event is 306.6 m3/s (18 hour rainfall duration). Simu- term sediment aggradation occurred at upstream and deposi- lated peak water surface and channel bed changes for Kulim tion occurred at downstream of Kulim River. FLUVIAL-12 River based on design hydrograph are shown in Fig. 13. The model was run to predict the channel geometry changes and cross sections especially near to the sand mining area and sediment delivery for the next 10 years. In general, it is found few cross sections especially CH 10000 to CH 14390 were that Kulim River will be in equilibrium conditions with slight subjected to greater changes than other cross sections. In spite degradation or erosion which deepen the river. The modeled of this, channel degradation was predicted at most cross sec- results show that future changes in cross sectional geometry tions after the peak. Fig. 14 shows the cross section changes will generally be limited and erosion along the reach will be for three locations along Kulim River. slowed down in the simulation period from 2006 to 2016. Future changes for the next 10 years were simulated by using hydrograph for year 1991 to 1992 and 1997 to 2006 (Fig. 3). Sediment delivery or the amounts of sediment moving 28 past each cross section predicted for the next 10 years (Year 26 2016) is shown in Fig. 15. The simulation results show that the amount of sediment delivery was twice for year 2016 com- 24 pared to the year 2006, but lesser sediment delivery at the 22 downstream of Kulim River. The decreasing trend of sediment Level (m) 20 delivery indicates long-term sediment deposition at the down- stream of Kulim River. Simulation for Kulim River based on 18 CH 12490 the time series illustrated the changes of the channel geometry 16 010203040506070 as shown in Fig. 16. The cross sections especially CH 10000 to Distance (m) CH 14000 are subjected to change with sediment aggradation, whilst sediment deposition occur at CH 6000 to CH 10000. Bed Level (Before Flood) Bed Level (Peak) Water Surface (Peak) Fig. 17 shows the spatial variations of the predicted median Bed Level (After Flood) grain size in year 2006 and 2016. The model run shows a large decrease in the sediment size at middle reach of Kulim River 16 between years 2006 to 2016; where the reach-average median 14 grain sediment size decrease from 0.77 mm to 0.58 mm. As 12 the channel bed became finer, more sediment was removed by erosion. Fig. 18 shows the example of cross section 10 changes for three locations along Kulim River. In general, Level (m) 8 the modeled results show that future changes in cross sectional 6 CH 5306 geometry will be generally limited and erosion along the reach 4 will be slowed down from 2006 to 2016; the Kulim River was 0 10203040506070 predicted to be stable at most locations. Distance (m)

6. Conclusions Bed Level (Before Flood) Bed Level (Peak) Water Surface (Peak) Bed Level (After Flood) Rapid development in a river catchment will result in high 15 discharge, erosion and deposition which will cause river insta- 14 bility. FLUVIAL-12 has been used to simulate the channel 13 geometry, lateral and vertical elevation changes for the flood 12 events from 1991 to 2006. FLUVIAL-12 was calibrated and 11 10 validated for bed elevation and water surface profile using 9 a number of different sediment transport formulas and rough- Level (m) 8 ness coefficient for several time period. Engelund-Hansen for- 7 CH 3014 mula and roughness coefficient n ¼ 0.030 were found to be the 6 5 best combination to represent the sediment transport activity 0 1020304050607080 in the study reach. Good agreements were obtained for both Distance (m) water level and bed profiles between the measured data and predicted results by FLUVIAL-12 model. The model simula- Bed Level (Before Flood) Bed Level (Peak) Water Level (Peak) Bed Level (After Flood) tion for existing conditions, future condition and long-term- modeling show the amount of sediment delivery will decrease Fig. 14. Modeled Cross Section Changes before and after Design Flood. C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59 57

3000000

2500000 Year 2006 Year 2016

2000000

1500000

1000000

Sediment Delivery (tons) 500000

0 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 1000011000120001300014000 Chainage (m)

Fig. 15. Spatial Variations of the Predicted Sediment Delivery.

30 Initial Bed (Year 1991) Predicted Water Surface (Year 2016) 25 Predicted Bed Profile (Year 2006) Predicted Bed Profile (Year 2016) 20

15 Elevation (m)

10

5 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 Chainage (m)

Fig. 16. Water Surface and Bed Profile Changes based on Design Hydrograph.

Thus Kulim River was predicted to be stable at most on the water surface profile simulated from the three scenar- locations. ios, it should also be considered that the proposed bund level In summary, flooding in Kulim River is found to affect and bank protection should stay above the predicted water sur- channel geometry, cross sectional geometry, sediment size face to avoid overtopping and reduce the flooding impact. The and sediment delivery, which consists of scour and fill. Based present study provides an estimate of sediment transport in

1.5 Year2006 Year2016

1 (mm) 50 d

0.5

0 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 Chainage (m)

Fig. 17. Spatial Variations of the Predicted Median Grain Size for Year 2006 and 2016. 58 C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59

26 and REDAC’s staff for their involvement in completion of 25 this paper. 24 23 22 Notation 21 Level (m) 20 The following symbols are used in this paper: CH 12490 19 18 B Water surface width (m) 0 10203040506070 d50 Mean sediment size (mm) Distance (m) n Manning’s roughness coefficient 3 Initial Bed Level (Year 1991) Predicted Bed Level (Year 2006) Q Discharge (m /s) Predicted Bed Level (Year 2016) R Hydraulic radius (m) So Water-surface slope 16 15 Tb Bed load (kg/s) 14 Ts Suspended load (kg/s) 13 Tj Total bed material load (kg/s) 12 yo Flow depth (m) 11 10 Level (m) 9 8 CH 5306 References 7 6 Ab. Ghani, A., Zakaria, N.A., Abdullah, R., Chang, C.K., Sinnakaudan, S.K., 0 10203040506070 Distance (m) Mohd Sidek, L., 2003. River Sediment Data Collection and Analysis Study, Contract Research No. JPS (PP)/SG/2/2000. Department of Initial Bed Level (Year 1991) Predicted Bed Level (Year 2006) Irrigation and Drainage, Malaysia, . Predicted Bed Level (Year 2016) Ab. Ghani, A., Zakaria, N.A., Chang, C.K., Ariffin, J., Abu Hasan, Z., Abdul Ghaffar, A.B., 2007. Revised equations for Manning’s coefficient for sand- 15 bed rivers. International Journal River Basin Management 5 (4), 329e346. 14 Abu Hasan, Z., 1998. Evaluation of Scour and Deposition In Malaysian Rivers 13 Undergoing Training Works: Case Studies of Pari and Kerayong Rivers. 12 MSc. Thesis. Penang : Universiti Sains Malaysia. 11 Abdullah, S., 2002. Simulation of River Bed Changes Due to Sand Mining 10 Operation at Muda River. MSc. Thesis. Penang : Universiti Sains Malaysia. 9 American Society of Civil Engineers or ASCE, 1998. River width Adjustment Level (m) 8 II: Modeling, by the ASCE Task Committee on Hydraulics, Bank 7 CH 3014 Mechanics, and Modeling of River Width Adjustment. Journal of 6 Hydraulic Engineering 124 (9), 903e917. 5 Ariffin, J., 2004. Development of Sediment Transport Models for Selected 0 1020304050607080 Rivers in Malaysia Using Regression Analysis and Artificial Neural Distance (m) Network. Ph.D. Thesis. Penang : Universiti Sains Malaysia. Initial Bed Level (Year 1991) Predicted Bed Level (Year 2006) Bahadori, F., Ardeshir, A., Tahershamsi, A., 2006. Prediction of alluvial river bed Predicted Bed Level (Year 2016) variation by SDAR model. Journal of Hydraulic Research 44 (5), 614e623. Chang, C.K., Ab. Ghani, A. & Abdullah, R., 2004. Effect of a 100-year Flood Fig. 18. Predicted Cross Section Changes for Year 2006 and 2016. on River Stability: Case Study of Kulim River. 1st International Conference on Managing Rivers in the 21st Century: Issues & Challenges, 21ste23rd September, Penang, Malaysia, pp. 473e478. moderate sandy stream and serves as a reference for sediment Chang, C.K., Ab. Ghani, A., Zakaria, N.A., & Abdullah, R., 2005. Sediment transport modeling of sandy streams in Malaysia and Transport in Kulim River, Malaysia. XXXI IAHR Congress: Water e overseas. Engineering for the Future - Choice and Challenges, 11th 16th September, Seoul, Seoul, Korea, pp. 1154e1162. Chang, C.K., 2006a. Sediment Transport in Kulim River, Kedah. M.Sc. Thesis. Acknowledgements Penang : Universiti Sains Malaysia. Chang, H.H., 1982. Mathematical Model for Erodible Channels. Journal of the Hydraulics Division 108 (HY5), 678e689. ASCE. The authors gratefully acknowledge Department of Drain- Chang, H.H., 1984. Modeling of River Channel Changes. Journal of Hydraulic age and Irrigation (DID) Kulim/Bandar Baharu and Hydrology Engineering 110 (2), 265e267. ASCE157-172. Closure in 113(2), 1987. Division for providing river survey data, hydrological data and Chang, H.H., 1988. Fluvial Processes in River Engineering. John Wiley and relevant information for this research. Special thanks go to Sons, New York, NY. 432pp. Prof. Howard H. Chang from San Diego State University, Chang, H.H., 1997. Modeling Fluvial Processes in Tidal Inlet. Journal of Hydraulic Engineering 123 (12), 1161e1165. ASCE. USA and Prof. Pierre Y. Julien from the University of Colo- Chang, H.H., Pearson, D., Tanious, S., 2002. Lagoon Restoration near rado, USA for their advice and help. The authors would also Ephemeral River Mouth. Journal of Waterway, Port, Coastal, and Ocean like to thank all undergraduate and postgraduate students Engineering 128 (2), 79e87. ASCE. C.K. Chang et al. / Journal of Hydro-environment Research 2 (2008) 47e59 59

Chang H.H., 2004. Fluvial Modeling Study of Feather River Responses to Interagency Stream Restoration Working Group (15 Federal agencies of Oroville Dam and Related Issues. Task 7- Hydraulic and Sediment the US Government), Portland. Transport Modeling of the Feather River. Draft Final Report prepared Ibrahim, N.A., 2002. Evaluation and Development of Sediment Transport for the Dept. Water Resources. Equations for Kinta River Basins, Kulim River and Kerayong River. Chang, H.H., 2006b. Generalized computer program: FLUVIAL-12 MSc. Thesis. Penang : Universiti Sains Malaysia. Mathematical Model for Erodible Channel Users Manual. San Deigo State Koay, B.C., 2004. River Equilibium: Case Study of Kulim River Final Year University, San Deigo, California. Project Thesis. Penang : Universiti Sains Malaysia. Chang, H.H., 2008. Case Study of Fluvial Modeling of River Responses to Lee, C.B., 2001. Application of River Modeling (Fluvial-12): Case Studies of Dam Removal. Journal of Hydraulic Engineering 134 (3), 295e302. Kulim River and Melaka River. Final Year Project Thesis. Penang : Darus, A., 2002. Conservation and Restoration of Urban Rivers: Case Studies Universiti Sains Malaysia. of Raia River and Pari River. MSc. Thesis, Universiti Sains Malaysia. Lagasse, P.F., Schall, J.D., Richardson, E.V., 2001. Stream Stability at Darus, A., Ab. Ghani, A., Zakaria, N.A., Abdullah, R., 2004. Natural Stable Highway Structures, US Department of Transportation, Federal Highway River Section Design using FLUVIAL-12 (May 2005). Bulletin Institution Administration. Publication No. FHWA NHI 01-002 (Hydraulic Engineer- of Engineers Malaysia, 24e27. ing Circular No. 20), 3rd Edition, Washington, D.C. Department of Irrigation and Drainage Malaysia., 1976. River Discharge Majlis Daerah Kulim or MDK, 1993. Kulim Structure Plan, 1990e2010, Measurement by Current Meter e Hydrological Procedure No. 15. Kuala Lumpur. Department of Irrigation and Drainage Malaysia., 1977. The Determination of National Research Council, 1983. An Evaluation of Flood-level prediction Suspended Sediment Discharge e Hydrological Procedure No. 19. using Alluvial River Model. National Academy Press, Washington, D.C. Department of Irrigation and Drainage Malaysia or DID, 1996. Study on Flood Richardson, E.V.,Simons, D.B., Lagasse, P.F., 2001. River Engineering for High- Mitigation and Drainage Master Plan for Kulim and its Surroundings, way Encroachments e Highways in The River Enviroment, US Department Kedah Darul Aman. of Transportation, Federal Highway Administration. Publication No. FHWA Department of Water Resources or DWR, 2004. Effects of Project Operations NHI 01e004 (Hydraulic Design Series Number 6), Washington, D.C. on Geomorphic Processes Downstream of Oroville Dam: Task 7 e Sinnakaudan, S., Ab Ghani, A., S Ahmad, M.S., Zakaria, N.A., 2003. Flood risk Hydraulic And Sediment Transport Modeling With Fluvial-12. Draft Final mapping for pari river incorporating sediment transport. Elsevier Science. Report. Oroville Facilities Relicensing FERC Project No. 2100, The Journal of Environmental Modeling and Software 18 (2), 119e130. Resources Agency, State of California. Soulsby, R.L., 1997. Dynamics of Marine Sands: A Manual for Practical Dong, Y.H.,Lu, J.Y.,Fan, B.L., Huang, Y., 2005. Research on Variations of Typ- Applications. Thomas Telford, London. ical Lower Jingjiang River Reach after Three Gorges Project Operation. United States Army Corps of Engineers or USACE, 1995. Sedimentation Journal of Yangtze River Scientific Research Institute 22 (2), 9e12. Investigations of Rivers and Reservoirs. USACE Engineering and Design Edwards, T.K., Glysson, G.D., 1999. Field Methods for Measurement of Manual. Publication No. EM 1110-2-4000, Washington, D.C. Fluvial Sediment. U.S. Geological Survey Techniques of Water-Resources Yazdandoost, F., 2005. Numerical Simulation of Sediment Transport in Investigations. Book, Chapter C2. Dynamic Flow Conditions. International Journal of Sediment Research Fan, S.S., 1988. Twelve selected computer stream sedimentation models 20 (4), 358e365. developed in the United State. In: Proceeding of the Interagency Yahaya, N.K., 1999. Development of Sediment Rating Curves for Rivers In Symposium on Computer Stream Sedimentation Model, Denver, Malaysia: Case Studies of Pari, Kerayong and Kulim Rivers. MSc. Thesis. Colorado. Published by the Federal Energy Regulatory Commission, Penang: Universiti Sains Malaysia. Washington, DC. Yuqian, L., 1989. Manual on Operational Methods for the Measurement of Federal Interagency Stream Restoration Working Group or FISRWG, 2001. Sediment Transport. Operational Hydrology Report No. 29, World Meteo- Stream Corridor Restoration: Principles, Processes, and Practices. Federal rological Organisation, Geneva, Switzerland, 169 p.