River Improvement Techniques for Mitigating River Bed Degradation and Channel Width Reduction in the Sandy Hii River Where Sediment Transport Occurs at Normal Times

E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018

River improvement techniques for mitigating river bed degradation and channel width reduction in the sandy Hii River where sediment transport occurs at normal times

Takahisa Gotoh1,*, and Shoji Fukuoka1 1Research and Development Initiative, Chuo University, Japan

Abstract. In the sandy Hii River, a large amount of sediment yield from upper river basin had brought developments of braided channels covered with sand waves. In the braided channels, sediment materials on the river beds are capable to move in normal discharge conditions. In recent years, however, the sediment yield decreases due to constructions of check dams and ground sills in the upper river basin. Thus, the river beds downstream of the ground sill have gradually degraded and the main channel widths have been narrowed with the progressing bed degradation. Firstly, we clarified that the effects of non-equilibrium sediment transports around the ground sill during normal discharge conditions on the bed degradation and the channel width reduction by using annual observed data and numerical simulations for bed variations. In addition, we provided the river improvement techniques for mitigating bed degradation and channel width reduction by improving state of non-equilibrium sediment transports passing through the ground sill.

1. Introduction The Hii River is located at Shimane prefecture in Japan and flows into the Sea of Japan through the Lake Sinji and the Lake Naka-umi (see Figure 1). The upper basin of the Hii River had yielded a large amount of sediment composed of fine sand materials. It had caused great aggradation of the river bed in the lower Hii River until 1960s. Due to the bed aggradation, the bed elevations of the main channel in the lower Hii River have raised than the ground level of the urban area along the river. And it has brought flood inundation disasters around this area in the past years. Meanwhile, a large amount of fine sediment yield has formed braided channels which are covered with well developed sand waves. The braided channels in the lower Hii River (around 11.0km) at the normal discharge conditions are shown in Figure 1. In the braided channels, sediment materials on the river bed are capable to move due to normal discharge flows since the river bed materials are almost uniform sizes of fine sands which are approximately 1mm to 2mm in diameter. Figure 2 shows ratio of shear stresses to the critical shear stress in the braided river section (12.4km) during the normal discharge

* Corresponding author: [email protected]

© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018

condition. Here, the normal discharge in this section is approximately 50m3/s by the annual observation records. The ratio of shear stresses to the critical shear stress indicates to be about 3 to 5 times at the cross-section.

Fig. 1. Location of the Hii River and the braided channels in the river.

14 Survey data in 2012 Ratio of shear stress of 7 critical shear stress against mean diameter 6 13.5 Water level in normal 5 discharge flow 13 4

12.5 3

Elevation（T.P.m) 2 12 1 Shear stress / Critical shear stress 11.5 0 0 50 100 150 200 250 300 350 Lateral distance(m) Fig. 2. Ratio of shear stress to critical shear stress in the braded river section. However, in recent years, the sediment yield from the upper river basin has decreased due to constructing a series of check dams and ground sills since 1960s. It has caused rapidly degradation of the river bed from the downstream of the ground sill in the lower Hii River. The bed degradation has brought to reduce the main channel width in the downstream of the ground sill. Moreover, the narrowed main channel led to advance the further river bed degradation. In this study, we investigate mechanism of the bed degradation and the channel width narrowing in the sandy Hii River. In addition, we attempt to provide the idea of river improvement techniques for mitigating bed degradation and channel width narrowing in sandy braided rivers where sediment supplies from upper river basins are almost zero at normal discharge conditions.

2. Mechanism of river bed degradation and channel width narrowing In this section, we investigated mechanism of the bed degradation and the main channel width reduction in the lower Hii River by using annual survey data, hydraulic records and aerial photographs. Figure 3 shows the aerial photographs and the changes in the main channel widths since 1975 until 2011. Here, the main channel widths shown in Figure 3 were determined by using the cross sectional shapes (see Figure 4). Figure 4 shows cross sectional shapes at

2 E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018 condition. Here, the normal discharge in this section is approximately 50m3/s by the annual 17.0km and 18.6km in the periods. In 1975, the braided channels were formed across the observation records. The ratio of shear stresses to the critical shear stress indicates to be full river width defined by widths between the levees. However, since 1992, the reduction about 3 to 5 times at the cross-section. of the main channel widths has begun and gradually progressed toward the downstream from the Igaya ground sill which was installed at 23.3km. Figure 5 indicates the annual changes in the observed mean bed elevations of the main channel. Due to the reductions of the main channel width, the mean bed elevations in the main channel have gradually degraded from the Igaya ground sill toward the downstream. The bed degradation has reached around 14.4km point until 2011.

Flow

Fig. 1. Location of the Hii River and the braided channels in the river. Flow

14 Survey data in 2012 Ratio of shear stress of 7 critical shear stress against mean diameter 6 13.5 Flow Water level in normal 5 discharge flow 13 4

12.5 3 Igaya ground sill

Elevation（T.P.m) 2 12 Main channel width 1 Shear stress / Critical shear stress 11.5 0 0 50 100 150 200 250 300 350 Lateral distance(m) Fig. 3. Annual changes in main channel width in the Hii River. Fig. 2. Ratio of shear stress to critical shear stress in the braded river section. 17k 18.6k 23 25 2012 2006 1998 1992 Channel width until 1985 However, in recent years, the sediment yield from the upper river basin has decreased 22 24 1985 1975 1966 due to constructing a series of check dams and ground sills since 1960s. It has caused 21 23 Channel width since 1998 Channel width until 1985 rapidly degradation of the river bed from the downstream of the ground sill in the lower Hii 20 2012 22 River. The bed degradation has brought to reduce the main channel width in the 19 2010 21 Channel width since 1992 downstream of the ground sill. Moreover, the narrowed main channel led to advance the 18 2006 20 1998 17 19 further river bed degradation. Elevation（T.P.m) 1985 Elevation（T.P.m) In this study, we investigate mechanism of the bed degradation and the channel width 16 1975 18 narrowing in the sandy Hii River. In addition, we attempt to provide the idea of river 15 1966 17 improvement techniques for mitigating bed degradation and channel width narrowing in 0 50 100 150 200 250 300 0 50 100 150 200 250 Lateral distance(m) Lateral distance(m) sandy braided rivers where sediment supplies from upper river basins are almost zero at normal discharge conditions. Fig. 4. Annual changes in cross sectional bed profiles and definition of main channel widths. Figure 6 indicates the annual changes in observed water levels in the normal discharge 2. Mechanism of river bed degradation and channel width at Kamishima observation station (18.6km) which is located about 5km downstream from narrowing the Igaya ground sill. The observed data shows that the water levels in the normal discharge have gradually declined with the river bed degradation. The declining of water levels per In this section, we investigated mechanism of the bed degradation and the main channel year since 1993 until 2011 is approximately -7.2cm. It seems to be almost independent width reduction in the lower Hii River by using annual survey data, hydraulic records and from frequency and scale of the past flood events shown in Table 1. Table 1 indicates the aerial photographs. history of main flood events in the past about 50 years. Figure 7 shows observed cross Figure 3 shows the aerial photographs and the changes in the main channel widths since sectional shapes at upstream and downstream of the Igaya ground sill (23.4km, 23.0km). At 1975 until 2011. Here, the main channel widths shown in Figure 3 were determined by the upstream section of the ground sill, the bed elevations of the main channel in recent using the cross sectional shapes (see Figure 4). Figure 4 shows cross sectional shapes at years became lower than the elevation of the ground sill which was 29.05 T.P.m, T.P. :

3 E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018

Tokyo peil, Japanese datum of leveling. Therefore, the ground sill has prevented sediments transported from the upstream section during the normal discharge conditions and small scale floods. At the cross section of 23.0km downstream of the ground sill, the river bed in the main channel was scoured during the period from 2006 to 2010 although large scale floods did not occur in this period (see Table 1). It means that the river bed degradation has progressed due to effects of non-equilibrium sediment transports passing across the ground sill during the normal discharge conditions and small scale floods. On the other hand, in large scale flood of 2006 flood, flood flows were able to transport sediment materials across the ground sill and the sediment materials were deposited downstream of the ground sill (see Figure 7). Therefore, we found that the sediment passing through the ground sill during large scale floods mitigated the bed degradation and the reduction of main channel widths in the downstream of the ground sill. It was found that improving the state of non-equilibrium sediment transports passing across the ground sill was important to mitigate the bed degradation and the channel width narrowing.

Kamishima

Fig. 5. Annual changes in longitudinal distributions of observed mean bed elevations in the main channel.

30 28 26 Kamishima(18.6k) Ootsu(12.4k) 24 Shinigaya(24.1k) 22 -7.2cm/year 20 18 16 Water level（T.P.m) 14 12 1990/5 1995/5 2000/5 2005/5 2010/5 (year)

Fig.6. Annual changes in observed water levels in normal flow condition.

Table 1. History of flood events in the past 50 years of the Hii River. Year/Month Peak discharge(m3/s) Year/Month Peak discharge(m3/s) 1971/7 1,500 1998/10 1,700 1972/7 2,300 2006/7 2,400 1979/10 1,600 2011/5 1,500 1983/9 1,500 2011/9 1,500 1993/9 1,600

4 E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018

Tokyo peil, Japanese datum of leveling. Therefore, the ground sill has prevented sediments 34 35 2010 Upstream of the 2010 1998 1975 transported from the upstream section during the normal discharge conditions and small 33 2006 2006 1992 1966 ground sill (23.4km) 2005 1985 scale floods. At the cross section of 23.0km downstream of the ground sill, the river bed in 32 2005 30 1998 the main channel was scoured during the period from 2006 to 2010 although large scale 31 1992 1966 25 floods did not occur in this period (see Table 1). It means that the river bed degradation has 30 Bed scouring since progressed due to effects of non-equilibrium sediment transports passing across the ground 2006 until 2010 29 20 Elevation（T.P.m) sill during the normal discharge conditions and small scale floods. Elevation of Elevation（T.P.m) Downstream of the 28 Igaya groundsill On the other hand, in large scale flood of 2006 flood, flood flows were able to transport （29.05 T.P.m) ground sill (23.0km) sediment materials across the ground sill and the sediment materials were deposited 27 15 0 50 100 150 200 250 Deposition0 by 50 100 150 200 250 300 downstream of the ground sill (see Figure 7). Therefore, we found that the sediment passing Lateral distance(m) 2006 flood Lateral distance(m) through the ground sill during large scale floods mitigated the bed degradation and the reduction of main channel widths in the downstream of the ground sill. It was found that Fig. 7. Annual changes in observed cross sectional shapes of upstream and downstream of the Igaya improving the state of non-equilibrium sediment transports passing across the ground sill ground sill (23.4km and 23km). was important to mitigate the bed degradation and the channel width narrowing. 3. River improvement techniques for mitigating river bed degradation and channel width narrowing in the sandy Hii River

3.1. The idea of river improvement technique In this section, we proposed river improvement techniques for mitigating bed degradation Kamishima and channel width narrowing downstream of the Igaya ground sill on the basis of the mechanism clarified in the section 2. The idea of the river improvement technique was composed of lowering the elevation of the ground sill for improving the state of non- equilibrium sediment transports and widening the main channel widths in downstream of Fig. 5. Annual changes in longitudinal distributions of observed mean bed elevations in the main the ground sill. Effects of the proposed river improvement technique were examined by channel. conducting numerical simulations of river bed variations under conditions of a series of

30 flood flows and normal discharge flows. 28 (c) Improvement of the ground sill Installed elevation of the ground sill 26 Kamishima(18.6k) 30 Ootsu(12.4k) Lowered part of the 24 29.5 Shinigaya(24.1k) ground sill 22 29 0.3m -7.2cm/year 28.5 20 28 case1 case2 23.4k 18 Elevation(T.P.m) 27.5 16 0 50 100 150 200 250 300 Water level（T.P.m) Lateral distance(m) 14 12 1990/5 1995/5 2000/5 2005/5 2010/5 (year)

Fig.6. Annual changes in observed water levels in normal flow condition.

Table 1. History of flood events in the past 50 years of the Hii River. Year/Month Peak discharge(m3/s) Year/Month Peak discharge(m3/s) 1971/7 1,500 1998/10 1,700 Fig. 8. River improvement techniques. 1972/7 2,300 2006/7 2,400 1979/10 1,600 2011/5 1,500 Figure 8(a) and (b) show the proposed main channel widths and the longitudinal mean 1983/9 1,500 2011/9 1,500 bed elevations of the main channel. The main channel widths and the mean bed elevations 1993/9 1,600 were determined so as to agree with those in 1992 when the reduction of the main channel was not noticeable problems for the river managements. In addition, for increasing sediment replenishments from the upstream of the Igaya ground sill, we attempted to lower the elevation in the central part of the ground sill as shown in Figure 8(c). The width of the

5 E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018

lowered part of the ground sill was determined based on the width of sand bars in 1975 when the river bed profiles were relatively stable (see Figure 8(d)). The height of the lowered part of the ground sill was 0.3m. It was determined by taking into account elevations of sand bars at the upstream sections of the ground sill and elevations of water intake facilities upstream of it.

3.2. Calculation method and conditions The BVC Method, the Bottom Velocity Computational Method[1], which was able to estimate vertical velocity distributions of flows based on the depth integrated model were employed for simulating flows and bed variations. The vertical velocity distributions in the numerical model were calculated by using depth averaged horizontal vorticity and difference between water surface velocities and bottom velocities. The bed variation was estimated by calculating sediment transport of bed load and suspended load. The bed load transport was calculated by using bed load formula of Ashida and Michiue. The suspended sediment transport was calculated by three dimensional convection–diffusion equation of suspended load concentrations. The entrainment rate of suspended load from river bed was calculated by Kishi & Itakura formula[2]. A series of discharge hydrographs based on the past observation records were given as the boundary conditions of upstream ends (see Figure 9). The normal discharge of rainy seasons and snowmelt seasons were considered in this simulation. The observed water level hydrograph in the Lake Shinji was given as boundary conditions of downstream end. Manning's roughness coefficient in the main channel was 0.027(m-1/3s) which was determined by using temporal changes in observed water surface profiles during 2013 flood in our previous studies[3] in the Hii River. The simulation cases were as follows. In Case1, channel widths and bed elevations downstream of the ground sill were improved as described in Figure 8(a), (b) and the central part of the ground sill was lowered as shown in Figure 8(c). In Case2, channel shapes were same as Case1. But the elevation of the ground sill remained as the current conditions. 10000 Shin Igaya Observation station Flood Flood Flood Flood Flood Flood Flood Flood Flood Flood 1000 /s) 3 Snowmelt season Rainy season 100

Discharge(m 10

1st year 2nd year 3rd year 4th year 5th year 6th year 7th year 8th year 9th year 10th year 1 0 500 1000 1500 2000 2500 3000 3500(day)

Fig.9. A series of discharge hydrographs given at the upstream end.

3.3. Calculation results Figure 10 shows the calculated contour of bed profiles downstream of the ground sill in each case. In Case2, the numerical results show that the bed profile downstream of the ground sill was obviously channelized due to the repetitions of flood flows and normal discharge flows. At the upstream section of the ground sill, the sand bar gradually developed and it gave rise to meandering pattern of the main channel in the downstream of the ground sill. On the other hand, by lowering elevation of the ground sill shown in Case1,

6 E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018 lowered part of the ground sill was determined based on the width of sand bars in 1975 the normal discharge flows were able to transport sediment materials smoothly across the when the river bed profiles were relatively stable (see Figure 8(d)). The height of the ground sill without sand bar developments at the upstream section. Hence, the channelized lowered part of the ground sill was 0.3m. It was determined by taking into account bed morphology did not form clearly in downstream section of the ground sill. Figure 11 elevations of sand bars at the upstream sections of the ground sill and elevations of water shows calculated cross-sectional shape at 20.6km and 22.9km after the repetition of flood intake facilities upstream of it. flows and normal discharge flows for 10 years period. In the calculation results of Case2 which formed channelized river bed morphology, the differences of bed elevations between main channel and sand bars became larger than those of Case1. And the maximum height 3.2. Calculation method and conditions of the differences was more than 2m. The BVC Method, the Bottom Velocity Computational Method[1], which was able to Case2(5th year) Case2(10th year) estimate vertical velocity distributions of flows based on the depth integrated model were Initial bed profile Case1(5th year) Case1(10th year) 24 24 24 24 24 24 24 employed for simulating flows and bed variations. The vertical velocity distributions in the 24 sandbar 24 sandbar 24 Flow numerical model were calculated by using depth averaged horizontal vorticity and Flow 22.9km Flow 23 23 Flow 23 22.9km difference between water surface velocities and bottom velocities. The bed variation was 22.9km 23 23 23 23 22.9km 23 estimated by calculating sediment transport of bed load and suspended load. The bed load 23 23 22 22 22 Elevation(T.P.m)ZB Elevation(T.P.m)ZB ZB 22 Elevation(T.P.m)ZB 22 ZB transport was calculated by using bed load formula of Ashida and Michiue. The suspended Elevation(T.P.m) 22 22 Elevation(T.P.m) 22 30 30 22 30 22 30 30 29 29 29 29 sediment transport was calculated by three dimensional convection–diffusion equation of 29 28 28 28 28 28 27 27 27 27 suspended load concentrations. The entrainment rate of suspended load from river bed was 27 26 26 21 21 26 26 26 21 21 25 21 25 21 25 21 25 21 25 calculated by Kishi & Itakura formula[2]. 24 24 21 21 24 24 24 23 23 23 23 A series of discharge hydrographs based on the past observation records were given as 23 22 22 20.6km 22 22 22 21 21 20.6km 20.6km 20.6km 21 21 the boundary conditions of upstream ends (see Figure 9). The normal discharge of rainy 21 20 20 20 20 20 19 19 20 20 19 19 19 20 seasons and snowmelt seasons were considered in this simulation. The observed water level 20 20 20 20 20 hydrograph in the Lake Shinji was given as boundary conditions of downstream end. 20 20 Manning's roughness coefficient in the main channel was 0.027(m-1/3s) which was determined by using temporal changes in observed water surface profiles during 2013 flood in our previous studies[3] in the Hii River. Fig.10. Contour of the initial and calculated bed profiles of Case1 and Case2. The simulation cases were as follows. In Case1, channel widths and bed elevations downstream of the ground sill were improved as described in Figure 8(a), (b) and the central part of the ground sill was lowered as shown in Figure 8(c). In Case2, channel shapes were same as Case1. But the elevation of the ground sill remained as the current conditions. 10000 Shin Igaya Observation station Flood Flood Flood Flood Flood Flood Flood Flood Flood Flood 1000 /s) 3 Snowmelt season Rainy season 100 Fig.11. Calculated cross sectional bed profiles of Case1 and Case2.

Discharge(m 10 Figure 12 presents the calculated longitudinal distributions of accumulated sediment 1st year 2nd year 3rd year 4th year 5th year 6th year 7th year 8th year 9th year 10th year 1 discharge during a year. The sediment discharge of Case2 rapidly decreased at the 0 500 1000 1500 2000 2500 3000 3500(day) downstream of the Igaya ground sill. However, in the results of Case1, the sediment discharge at the downstream of the ground sill was increased by lowering elevation in the Fig.9. A series of discharge hydrographs given at the upstream end. central part of the ground sill compared with Case2. The calculation results showed that the river improvement techniques of Case1 were able to mitigate the state of non-equilibrium sediment transports around the ground sill and channelization and channel incisions 3.3. Calculation results downstream of the ground sill. Figure 10 shows the calculated contour of bed profiles downstream of the ground sill in As a result, the numerical simulations demonstrated that the lowering elevation in the each case. In Case2, the numerical results show that the bed profile downstream of the central part of the ground sill were effective for mitigating bed degradation with main ground sill was obviously channelized due to the repetitions of flood flows and normal channel width reduction. discharge flows. At the upstream section of the ground sill, the sand bar gradually developed and it gave rise to meandering pattern of the main channel in the downstream of the ground sill. On the other hand, by lowering elevation of the ground sill shown in Case1,

7 E3S Web of Conferences 40, 02033 (2018) https://doi.org/10.1051/e3sconf/20184002033 River Flow 2018

Snowmelt season Rainy season Dry season Flood Total Case1 Igaya ground sill Case2 Igaya ground sill

) 10000 10000 3

1000 1000 discharge (m discharge

accumulated sediment accumulated sediment 100 100 20 21 22 23 24 20 21 22 23 24 Longitudinal distance(km) Longitudinal distance(km) Fig.12. Longitudinal distributions of accumulated sediment discharge for a year in 10th year.

4. Conclusions In this paper, firstly, we investigated mechanism of the bed degradation and the main channel width reduction in the sandy Hii River by using annual observed data. Secondary, we proposed river improvement techniques for mitigating bed degradation and channel width narrowing. The main conclusions in the study are as follows. 1. The normal discharge flows were able to transport sediment materials on river bed in the braided channels in the sandy Hii River. Meanwhile, the ground sill prevented sediment transports from the upstream during the normal discharge flows and small scale floods. These non-equilibrium conditions of sediment transports across the ground sill caused the bed degradation and the main channel width reduction. The channel width reduction induced further bed degradation in the main channels. 2. The proposed river improvement technique was composed of lowering elevation in the central part of the Igaya ground sill and widening of the main channel width based on the past river conditions downstream of the ground sill. The numerical simulations applying the BVC Method demonstrated that lowering elevation in the central part of the ground sill improved the state of non-equilibrium sediment transports over the ground sill. As a result, we were able to mitigate river bed degradation and channel width narrowing in the sandy Hii River where sediment transports occurred at normal discharge flows.

References 1. T. Uchida, S. Fukuoka : Numerical calculation for bed variation in compound- meandering channel using depth integrated model without assumption of shallow water flow, Advances in Water Resources, Vol.72., pp.45-56 (2014) 2. T. Itakura, T. Kishi : Open channel flow with suspended sediments., J. Hydr. Div., 106(8), pp.1325–1343 (1980) 3. T. Gotoh, S. Fukuoka, R. Shibata: Diversion of flood flow associated with a large amount of sediment transport into the Hii River Floodway and measures reducing inflow sediment discharge, Journal of Japan Society of Civil Engineers, Ser.B1(Hydraulic Engineering), Vol.73, No.4, pp.I_895-I_900 (2017)