Study about Velocity Index in Actual River during Flooding

Study about Velocity Index in Actual River during Flooding Atsuhiro Yorozuya 1, Kazuhiko Fukami 1 1 International Centre for Water Hazard and Risk Management (ICHARM) under the auspices of UNESCO, Public Works Research Works, Japan E-mail: yorozuya@pwri.go.jp Abstract For the purposes of a development of an automatic water discharge measurement, a field measurement was conducted using non-contact current meters for obtaining Water Surface (WS) velocity, water gauges for estimating WS slope, Acoustic Doppler Current Profiler (ADCP) mounted on tethered ADCP platform, as well as an echo sounder fixed on H-steel for monitoring riverbed during flooding. Observational site was selected aiming understanding of flow as well as river bed characteristic at where riverbed deformation easily occurs. Actually, when two flood waves with magnitude of 1,000 m 3 /s passed through the site involving significant river bed deformation, observation with few mentioned devices were successfully conducted. For example, the echo sounder successfully monitored the bed form with maximum wave height of 1.5m when the water surface elevation increase about 2.5m from normal stage. Based on those observed results, the present study describes mainly about the velocity index with relating to the WS velocity, as well as the Manning's roughness coefficient incorporated with the bed form. The results show that the velocity indices of 0.85±5 were observed by ADCP with none or less bed form height condition; however, the values of 1.1 or 0.7 were also recognized when the bed form predominated. 1. Introduction Water discharge measurements in river, as well as storage of the observed values, are very fundamental in terms of designing the river-infrastructure, such as levee, dam, and etc. For conducting the measurement, the current meter incorporating with a depth sounder has been employed in many countries, especially in continental rivers. In the case of Japan, float-type measurements, which established in 1940s, have employed since then, because character of Japanese rivers is highly unsteadiness, loose boundary, and, above all, rough water surface. Recently many devices have developed with different principles, such as acoustic especially using H-ADCP by Nihei and Kimitu (2008), video images of LSPIV by Muste et al. (2008), or STIV by Fujita et al. (2009), erector-magnetic by Yamaguchi and Niizato (1994), and etc. They have been well developed in terms of durability in the field, as well as reliability of data accuracy, which was confirmed by the direct comparison between those devices and ADCP measurements by Yorozuya et al. (2010), and Hara et al. (2011). However, considering about the water discharge value, additional information such as a water depth, as well as a velocity index are still necessary to be taken into account. Authors have tried to develop an automatic water discharge measurement system employing few devices, such as such as non-contact current meter as fixed type measurement, water

surface slope to determine the velocity index employing the log-law as well as the WS velocity, and ADCP traverse measurement for bathymetry monitoring as well as verification purposes, e.g., Yorozuya at al. (2010). In terms of the velocity index, authors have not satisfied by ourselves as the automatic system; rather, applied constant value as 0.85 as other studies do. Contrary with the authors' previous study, the water discharge measurement was conducted in mountainous area aiming understanding of flow as well as river bed characteristic at where riverbed deformation frequently occurs. In addition to devices listed previously, the automatic river bed monitoring system using echo sounder in the middle of channel was employed. Actually, when two flood waves with magnitude of 1,000 m 3 /s passed through the site involving significant river bed deformation, observation with few mentioned devices were successfully conducted. For example, the echo sounder successfully monitored the bed form with maximum wave height of 1.5m when the water surface elevation increase about 2.5m from normal stage. Based on those observed results, the authors discussed about the timely change of the WS velocity, the Manning's roughness coefficient, as well as a velocity index incorporated with the bed form. Regarding to the velocity indices, the values of 0.85±5 were observed by ADCP with none or less bed form height condition; however, the values of 1.1 or 0.7 were also recognized when the bed form predominated. 2. Methodology The observational site of this study has river characteristics, such as a bed slope of about 1/200, bed material between 30 and 210 mm, and a river width of about 250 m including a main channel of 120m as well as a flood plain of 130m. At upstream of this site, many mountains with slope failure locate, as well as Sabo works have conducted inside of this catchment area. Because of sediment supplies, as well as steep bed slope, frequent river bed elevation change is recognized. Authors experienced flooding with discharge of about 1,000 m 3 /s, whose water surface elevation is about 2.5m higher than normal flow. Usually, this section has normal water discharge of about 50m 3 /s. In following few paragraphs, instrumentation will be explained. Non-contact current meter to measure water surface (WS) velocity has specification of 24.15Ghz ghz with a horn antenna as well as a half power angle of about 12 degree. Eight cross sections were setup with total 120m width of water-flow section during our observation. Single non-contact current meter was employed observing frequency of 1Hz in 5 minutes, whose representative value obtained by averaging them. Thereafter the observation is conducted continuously at another section, until it measured at eight different points within one hour. Using those data, timely interpolated values were employed for determining value at an assigned time. Though this instrument has a characteristic of obtaining falling velocity of rain or snow when it is that weather condition, there was no rain or snow during the observation. Two water gauges were employed to estimate the water surface slope; e.g., the Cera Diver whose specification has a resolution of 0.25cm and an accuracy of 2cm, as well as the one of the Ministry of Land, Infrastructure, Transport and Tourism (MLIT). The Cera Diver located at 360m upstream from the bridge, while the one of MLIT located at 30 upstream.

River-bed elevation change was monitored by two echo sounders fixed with oblique angle on H-type steels located on middle of water flow, whose conceptual diagram is shown in Figure 1. Those three echo sounders were named as observatory B1, B2 and B3 with the angle of 5, 50.4, and 52.8 degree, respectively. Among them, B1 and B2 locate in the main channel, while B3 locates on the floodplain. The monitoring system is located about 30m upstream from a bridge in the Figure 1. The echo sounders have specification of 200 khz, with time interval of 0.2 seconds. 10 seconds measurement was conducted with every 10 minutes, and saved as one single value after averaging them. The beam of echo sounder should be located high enough to illuminate the spot which is outside of wake flow generated by the H-type steel; at the same time, it should be low enough, since measurements cannot start unless a transducer sinks to the flood water. Therefore, the location of the echo sounders should be carefully decided considering designing the magnitude of flooding. For this study, the targeted discharge for B1 and B2 was selected as about 1 time in a year. Logger box H-type steel Solar panel B3 B2 Echo Sounder B1 River bed Location of observation Ring Method Figure 1: Conceptual diagram of the river bed monitoring system with echo sounder Traverse measurements with ADCP mounted on a tethered platform was also conducted to profile horizontal/vertical water velocity distribution as well as bathymetry. Those observed values are employed to obtain the velocity index as well as to estimate WS velocities. For the estimation, the least square method was employed assuming the vertical velocity distribution is the logarithmic profile; thereafter, the velocity at a upper unmeasured zone was extrapolated using the relation. The estimated WS velocities were compared with that obtained by the non-contact current meter for verification purposes. Though this paper are not well explained about the comparison, the WS velocity obtained from both of them were fairly well fitted. To actual measurement, the traversing of the boat with tethering the field workers on the bridge was applied to obtain cross sectional values. The boat path, therefore, was the downstream side of the bridge with the distance of 10 to 20m depend on the water velocity. Peripheral devices, such as a high speed river boat as well as Real Time Kinematic (RTK)- GPS, were employed to the measurement to execute the rough water surface vibration as well as active bedload condition. The Work Horse ADCP with 1,200 khz by Teledyne RD instruments were employed with commands showing in Table1.

3. Observational results Table 1: Commands of ADCP Mode of Bottom truck BM5: 5 normal Pings for Bottom truck BP3: 3 pings Pings for Water truck WP3: 3 pings Type of band WB0: Broad band Distance of first blank WS25: 25 cm Mode of measurement WM12: high speed mode Number of layer WN40: 40 layers Thickness of layer WS25: 25 cm Figure 2 shows time series of the WS elevation, riverbed elevation, the observational time of ADCP traverse measurements, and WS velocity measured by non-contact current meter. As it shows, observation with ADCP as well as non-contact current meter was conducted when two flood waves pass through the observational point. Water surface elevation started to increase at 6:00 of first day, it reached at the first peak around 21:00 with increasing about 2.5m in water depth when discharge of 1,000 m 3 /s and WS velocity of m/s was obtained. After water surface elevation decreases to 118.0m within 24hours, next peak came with slightly less discharge compared with first peak, whose discharge is about 800 m 3 /s and WS velocity of about m/s. The monitoring of the river-bed elevation by observ.b1 started after WS elevation increased about 117.7m, while observ.b2 started about m. At depression time between two flood peaks, the transducer of observ.b2 quit monitoring, since WS elevation reached to below transducer. Thereafter, it started monitoring again with increasing of water depth. Elevation of WS and riverbed, m WS, TP ADCP obser. Observ. B1 Observ. B2 non-contact current meter WS velocity, m/s - 1st Day 2nd Day Figure 2: Time series of river bed elevation, water surface elevation, and timing of ADCP measurements Regarding of observ.b1 starting around 12:00 of 1 st day, sediment deposition prevailed till the peak of flooding. After peak discharge pass, river bed elevation decreased with magnitude of 1m vibrating within the similar range till the WS elevation decrease. At 12:00 of 2nd day, sediment deposition continues resulting about 50cm accumulation in next 12 hours with decreasing of water depth. After next flood came, it shows similar trend with the first one; thereafter, it came back to the initial level. Regarding of observ.b2, the relation between WS 3rd Day 4th Day

and riverbed elevation shows similar trend with that of observ.b1. Slightly different characteristic is that magnitude of the vibration of river bed is higher than that of B1. Common characteristics for both of them are that minimum river bed height appears not at the peak of water discharge but after few hours, as well as timing of the bed vibration. Authors expected that area of flow section have simple-increasing function with WS elevation, such as degradations and aggradations corresponding with increasing and decreasing of water discharge; however, actual phenomenon in this case was not so simple, e.g., that bed forms was developed and pass though the observational section. To illustrate general ideas of the current study as well as observational section, flow pattern, and etc, Figure 3 indicates one of examples of observational results, such as velocity distribution, river bed elevation, and location of bridge pier and echo sounders. The horizontal, vertical, secondary vertical axes indicate distance from the left river bank in m, water velocity in cm/s, and elevation for riverbed and water surface in m, respectively. Among the marks in this figure, the filled circles are for WS velocities converted by the ADCP measurements; vertical-dotted lines for section boarders with section number from 1 to 8; triangles for section averaged values of the filled circles; open-circles for WS velocities measured with noncontact current meters; diamonds for locations of the echo sounder; BP for locations and width of the bridge pier; solid curve for river bed obtained by the ADCP measurements; a broken line for river bed surveyed few years ago; and a dotted line for water surface elevation. velocity, cm/s 50 40 30 20 10 1 2 3 4 5 6 7 8 12 12 12 119.0 117.0 river bed, water surface elevation, m BP BP 11 1 2 3 4 5 6 7 8 9 10 11 12 Distance from left river bank, m Surface velocity by ADCP Averaged velocity by ADCP Surface velocity with non-contact river bed taken by ADCP Original river bed Water surface elevation Location of Echo sounder BP Location of bridge pier Figure 3: Cross sectional distribution of velocity and river bed As this figure indicates, the velocity distributions were widely and locally distributed with affecting by the bridge piers, as well as by local turbulence. Especially, flow as well as sediment deposition behind bridge pier are concaved, though concaved parts is not exactly located at that of bridge pier because of skewed flow. As comparison between triangles as well as open-circles indicates, both of them are not so exactly identical but few difference within acceptable range.

Figure 4 shows time series of water surface elevation as well as water surface slope obtained by water gauges, river bed elevation averaged the observed values in the section 2 obtained by ADCP, WS velocity obtained by the non-contact current meter, and the Manning s roughness coefficient. As figure 3 indicates, the section 2 is one of downstream section from the river bed monitoring B1. It was unfortunate that the riverbed values around 9:00 of the 3 rd day are less dense compared with other time because of unreliable observed values with many missing ensemble. To estimate the Manning s roughness coefficient, the WS velocity with non-contact current meter, WS slope, WS elevation, the river bed elevation which most recently observed by ADCP, as well as the velocity index obtained by ADCP were employed. WS velocity, m/s, Manning's n/100, 1/WS slope 100 7.0 6.0 WS slope Manning's n WS velocity riverbed elevation WS elevation 121.5 elevation, m 1st day 2nd day 3rd day 4th day Figure 4: Time series of river bed elevation, WS elevation, WS slope, WS velocity, and the Manning s roughness coefficient at section 2 Figure 4 indicates that, starting from 18:00 of 2 nd day, as water surface increase, firstly the WS velocity increase until 6:00 of 3 rd day, secondary it start to fluctuate with range about 25cm until 13:00 of 3 rd day, and finally it continuously decreases. On the other hand, river bed elevation changes in this section shows similar trend with obser.b1 indicated in Figure 2, in terms of the shape as well as the magnitude of sand wave. Correspondence between the WS velocity and the riverbed elevation implies that sand wave generated at the bottom of river bed affects to the WS velocity, though WS elevation does not show any similarity, since they are located at the river bank. Regarding of WS slope, it has minimum values about 1/650 in between two flood peaks; thereafter, it reached to the maximum values about 1/300 at 12:00 of 3 rd day, then it continuously decreases. The Manning s coefficient, involving most of variables described in this paragraph, starts around 4 from 18:00 of 2 nd day; thereafter, it decreases the values till 35, probably because of smoothing of roughened river bed which was disturbed by the first peak. With generating of the bed form, the roughness also rapidly increases to more than 5. Thereafter, it gradually decreases with the WS elevation decreasing. Based on this change, authors could estimate the grain roughness is about 35 while form roughness is about additional 2. In this figure, the correspondence among the bed form, the WS velocity, as well as the roughness was clearly described based on the observed values. Figure 5 shows time series of variables alike with Figure 4 adding discharges and velocity index, as well as deducting the WS slope and the Manning's roughness. Similar with the bed elevation change, the velocity index was estimated with averaging the observed values in the section 2 obtained by ADCP. The discharge with 0.85 obtained by applying the velocity index

of 0.85 for obtaining the averaged velocity from the WS velocity by the non-contact current meter, the water depth obtained by the WS elevation as well as the river bed elevation, while the discharge with ADCP index applies the velocity index obtained by ADCP observation. WS velocity, m/s, dishcarge, m 3 /s 20, velocity index/5, 6.0 WS velocity discharge with 0.85 discharge with ADCP index velocity index riverbed elevation WS elevation elevation, m 1st day 2nd day 3rd day 4th day Figure 5: Time series of river bed elevation, WS elevation, WS velocity, velocity index obtained by ADCP, discharge with velocity index of 0.85 and that by ADCP at section 2 As this figure indicates, the velocity index has number in between 0.75 and 0.85 when bed form does not exist. When bed form generated between 6:00 and 15:00 of 3 rd day, the velocity index vibrates between 1.1 and 0.73. Those numbers with the bed from are slightly larger than other study. For example, Polatel (2006) conducted experimental studies as well as numerical computation with different types of roughness. In this study, she obtained the velocity index between 0.85 and 0.908 on roughened bed. On the other hand, Hino and Miyanaga (1977) shows the vertical velocity distribution at the different location of sand wave; e.g., trough, crest and in between. They shows that velocity distribution at the crest is almost vertical or sometimes overshooting type which indicates the velocity index around, while that at the trough is widely distributed including the reverse flow which has possibility of the velocity index of less than 0.8. Correspondingly, two discharges are directly related to differences of velocity index; for example, they are not much different when the velocity indices are around 0.85, while 24% and 16% difference occurs when the velocity index is 1.13 and 0.73, respectively. Figure 6 and 7 indicates the variables obtained at section 7 with the same format and contents of Figure 4 and 5, respectively. As Figure 5 indicates, the riverbed vibration in this section is more severe compared with Figure 4 but somehow similar with the results of the river bed monitoring B2 in Figure 2 in terms of shape of wave as well as magnitude of wave height. Actually, both of them have deposition toward the peak of the WS elevation, and higher oscillation during recession phase within the range of 1.5m. In addition, the WS velocity as well as the Manning's roughness coefficient show similar trend with those in Figure 4, but more dynamically vibrate with corresponding to the river bed fluctuation. Regarding of the roughness coefficient, either grain or form roughness cannot be clearly determined in this section, since riverbed vibrates even low WS elevation; though, the authors can clearly state that the severer river bed vibration makes the higher values of the Manning's roughness coefficients. On the other hand, most of the velocity indices in between 18:00 of 2nd day and

23:00 of 3rd day idles around 0.85 within the range of ±5. With generation of the bed forms, the velocity index correspondingly vibrates. WS velocity, m/s, Manning's n/100, 1/WS slope 100 8.0 7.0 6.0 WS slope Manning's n WS velocity riverbed elevation WS elevation 122.5 121.5 elevation, m 1st day 2nd day 3rd day 4th day Figure 6: Time series of river bed elevation, WS elevation, WS slope, WS velocity, and the Manning s roughness coefficient at section 7 WS velocity, m/s, dishcarge, m 3 /s 20, velocity index/5, 8.0 7.0 6.0 WS velocity discharge with 0.85 discharge with ADCP index velocity index riverbed elevation WS elevation 122.5 121.5 elevation, m 1st day 2nd day 3rd day 4th day Figure 7: Time series of river bed elevation, WS elevation, WS velocity, velocity index obtained by ADCP, discharge with velocity index of 0.85 and that by ADCP at section 7 4. Discussion and Conclusion With aiming to develop an automatic water discharge measurement, the authors discussed about flow properties based on the observed results obtained in the mountain area. With the observed flood event, the bed form was obtained with ADCP traverse measurement as well as the riverbed monitoring system. Based on those results, authors discussed about the velocity index aiming to compile the automatic water discharge measurement system. The significant outputs from above discussion can be listed as follows; 1) most of the velocity index without or with less height of bed form are about 0.85±5, 2) when bed form exist, they varied to 1.1 to 0.7 which is depned on size as well as relative location of the bed form, 3) therefore, existance of the bed form should be carefully confirmed to determine the velocity index, 4) the bed form generated not necessary at the maximum flood peak, 5) time series of the WS velocity maybe one of the indicators 6) the Manning's roughness coefficient was also estimated from observed values, and

7) the authors had expected the riverbed monitoring system as estimating of the flow area, it is also capable to estimating the velocity index. Acknowledgement Observed data presented herein were obtained by River division of National Institute for Land and Infrastructure Management, Japan. Authors appreciated their cooperation for providing us the valuable data. References Ichiro Fujita et al. (2009) Efficient space-time image analysis of river surface pattern using two dimensional fast Fourier transformation. Proceedings of 33rd IAHR Congress, pp.2272-2279. Hiroki Hara et al. (2011) Accuracy of non-intrusive river flow measurements by using space time image velocimetry. 11 th Asian Symposium on Visualization, Niigata, Japan Mikio Hino, and Yoichi Miyanaga (1977) Analysis of two-dimensional flow in a wavy conduit. Proceedings of the Japan Society of Civil Engineers, 264, pp.63-75. (in Japanese) Marian Muste et al. (2008) Large-scale particle image velocimetry for measurements in riverine environments. Water Resour. Res., 44, W00D19, doi:10.1029/2008wr006950. Yasuo Nihei and Akira Kimizu (2008) A new monitoring system for river discharge with H- ADCP measurements and river-flow simulation. Water Resources Research, Vol.44, W00D20, doi:10.1029/2008wr006970. Ceyda Polatel (2006) Large-scale roughness effect of free-surface and bulk flow characteristics in open channel flows. Ph.D. thesis, Univ. of Iowa, Iowa. Takayuki Yamaguchi and Kunio Niizato (1994) Flood discharge observation using radio current meter. Journal of Hydraulic, Coastal and Environmental Engineering(II), 497:28, 41-50. (in Japanese) Atsuhiro Yorozuya et al. (2010) Development of automatic water discharge measurement system. Environmental Hydraulics, Christodoulou & Stamou (eds) @ 2010 Taylor & Francis Group, London, ISBN 978-0-415-58475-3, pp.839-844.