New acoustic system for continuous measurement of river discharge and water temperature

Kiyosi KAWANISI*, Arata KANEKO, Shinya NIGO, Mohammad SOLTANIASL, Mahmoud F. MAGHREBI

1. Graduate School of Engineering, Hiroshima University, 1-4-1, Kagamiyama, Higashi-Hiroshima 739-8527, Japan

2. Ministry of Land, Infrastructure, Transport and Tourism, 2-4-36 Kitaku Sikadachou, Okayama 700-0914, Japan

3. Civil Engineering Department, Ferdowsi University of Mashhad, P. O. Box 91775-1111, Mashhad, Iran

New acoustic system for continuous measurement of river discharge and water temperature

Kiyosi KAWANISI*1, Arata KANEKO1, Shinya NIGO2, Mohammad SOLTANIASL3, Mahmoud F. MAGHREBI3

1. Graduate School of Engineering, Hiroshima University, 1-4-1, Kagamiyama, Higashi-Hiroshima 739-8527, Japan

2. Ministry of Land, Infrastructure, Transport and Tourism, 2-4-36 Kitaku Sikadachou, Okayama 700-0914, Japan

3. Civil Engineering Department, Ferdowsi University of Mashhad, P. O. Box 91775-1111, Mashhad, Iran

In many cases, river discharge is indirectly estimated from water level or streamflow velocity near the water surface. However, these methods have limited applicability. In this study, an innovative system, the fluvial acoustic tomography system (FATS), was used for continuous discharge measurement. Transducers with a central frequency of 30 kHz were installed diagonally across the river. The system’s significant functions include accurate measurement of the travel time of the transmission signal using a GPS clock and the attainment of a high signal-to-noise ratio as a result of modulation of the signal by the 10th order M-sequence. In addition, FATS is small and lightweight, and its power consumption is low. Operating in unsteady streamflow, FATS successfully measured the cross-sectional average velocity. The agreement between FATS and acoustic Doppler current profilers (ADCPs) on water discharge was satisfactory. Moreover, the temporal variation of the cross-sectional average temperature deduced from the sound speed of FATS was similar to that measured by a temperature sensor near the bank.

streamflow; fluvial acoustic tomography; cross-sectional average velocity; unsteady flow; water temperature

1 Introduction

River discharge is an important hydrological factor in river and coastal planning/management, control of water resources, and environmental conservation. Therefore, establishing the method and technology for streamflow measurement is a crucial issue. However, it is very difficult to measure cross-sectional average velocity in unsteady flows or during extreme hydrologic events, such as flooding.

For continuous measurement of water discharge, a few different pieces of equipment are available, e.g., acoustic velocity meters (AVMs) and horizontal acoustic Doppler currentprofilers (H-ADCPs) (Ruhl and DeRose 2004; Wang and Huang 2005). The main drawback of previously presented methods is that the number of velocity sample points in the cross-section of a stream is often insufficient for estimating cross-sectional average velocity. H-ADCPs can measure a horizontal profile of velocity over a range with sufficiently strong acoustical backscatter. However, H-ADCPs do not provide any information for vertical velocity profiles. Moreover, the horizontal profile range of H-ADCPs decreases with increasing suspended sediment concentration (SSC). In addition, H-ADCPs do not work well in estuaries because of the sound inflection.

Although several methods have been introduced to estimate velocity distribution (Chiu and Hsu 2006; Maghrebi and Ball 2006), the results are disputable in complex flow fields such as stratified tidal flows or unsteady flows. Thus, innovative methods and/or equipment for continuous measurement of river discharge are needed.

In this study, the fluvial acoustic tomography system (FATS) was developed and utilized to measure outflow rates from an estuary weir. FATS has advantages over competing techniques: namely, the accurate measurement of the travel time of the transmission signal using a GPS clock, and the attainment of a high signal-to-noise ratio (SNR) of signals due to modulation by the 10th order M-sequence. As a result, FATS works well even during flood events in which SSC and acoustic noise are very high (Kawanisi et al. 2010b). FATS also works well in estuaries with saltwater intrusion (Kawanisi et al. 2009; Kawanisi et al. 2010a).

2 Measurement principles and error analysis

The basic principle of FATS is similar to what is used in an AVM. In other words, the cross-sectional average velocity is calculated using the travel time method (Sloat and Gain 1995). The authors have tentatively called FATS a next-generation AVM in a previous paper (Kawanisi et al. 2008). An old-fashioned type of AVM measures average velocity along a transverse line. Therefore, the AVM requires different strategies, the index velocity method and the velocity profile method, for computing discharge. FATS is able to estimate cross-sectional average velocity using ray paths that cover the section, unlike an old-fashioned type of AVM.

The travel time along theith reciprocal ray pathbetween a pair of transducers in the flowing medium is formulated as

where +/− represents the positive/negative direction from one transducer to another,cis the sound speed,dsis the increment of arc length along the ray path,uis the flow velocity,nis the unit vector along the ray path, andNis the number of ray paths. The path integrals are taken along rays. We assume that the two-way path geometry is reciprocal andThe two-way travel time difference may be expressed as

whereLiis the length of theith ray path, andandare the range average water velocity and the sound speed along theith ray path, respectively:

is calculated from

The cross-sectional average water velocity in the flow direction,, is defined as

whereis the component of the mean water velocity along the ray paths, andθis the angle between the ray path and streamline.

In order to estimate the cross-sectional average velocity, it is preferable that the ray paths cover the cross-section as much as possible. The ray paths of FATS probably cover the cross-section in a freshwater environment. However, a salt wedge under the transducer causes ray paths to be reflected, so those ray paths are not able to penetrate bottom layers (Kawanisi et al. 2009; Kawanisi et al. 2010a).

In order to accurately identify the arrival time of a traveling sound mixed with noises, the transmission signal is phase-modulated by applying the pseudo-random sequence (Simon et al. 1985; Zheng et al. 1998). Fig. 1 shows the transmission signal modulated with the 3rd order M-sequence as a typical example. The carrier signal is phase-modulated by taking a product with the M-sequence. By transmitting this modulated signal, the SNR is increased significantly. In this study, the higher order (10th order) M-sequence is applied to get higher SNRs, increasing by2n−1, wherendenotes the order of the M-sequence.

Fig. 1 Phase-modulation of carrier signal by 3rd order M-sequence

The transmission signal with a phase-modulation is expressed as

whereM(t) is the 10th order M-sequence, andA(t) andψ(t) are the amplitude and phase functions, respectively. The angular frequencyωwas set to 30 kHz in this study. The received signalS(t) was processed using the cross-correlation between the received signal and the 10th order M-sequence. This process serves to identify the precise arrival time.

Based on the total differential of Eqs. (5) and (6), the relative errors ofandare estimated with Eqs. (8) and (9), respectively:

The average travel time errorand the error of travel time differenceδ(Δti) are negligible when the pair of transducers is synchronized precisely with the GPS clock. As a result, the relative error of the average sound speedand the relative error of the average flow velocityare equated with the relative error of the ray’s length. The termon the right-hand side of Eq. (9) can be neglected ifδ(Δti) is insignificant.

3 Experimental site and methodology

An experiment with FATS was carried out from June 8 to 25, 2009 on the Hyakken River. Fig. 2 shows an aerial photograph of the experimental site. The array of sluice gates at the mouth intermittently opens to discharge fresh water at low tides, so that saline water does not enter the river. The direct distance between S1 and S2 was 418.5 m. The cross-section along the ray path is shown in Fig. 3.

A couple of broadband transducers were installed diagonally across the channel, as shown in Fig. 2. The central frequency of the transducers was 30 kHz. The transducers were mounted at a height of 0.45 m above the bottom, as shown in Fig. 3, whereZis the elevation relative to the mean sea level. The acoustic pulses of FATS were simultaneously transmitted from the two omni-directional transducers triggered every minute by a GPS clock. The angle between the ray path and the stream direction,θ, was assumed to be 35° when water was discharged through the gates.

An upward-facing ADCP was located on the bottom in front of the sluice gates (Fig. 2). The ADCP was operated at 2 MHz and the bin length was set to 0.1 m. The profile interval, which describes how often the instrument collected the current profile data, and the average interval, which specifies how long the instrument would be actively collecting data within each profile interval, were both 600 s.

Water level and temperature were measured every 600 s with a pressure-temperaturesensor attached to the frame of a downstream transducer. The pressure-temperature sensor was located at a height of 0.4 m above the bottom.

Fig. 2 Aerial view of study area and experimental setup

Fig. 3 Bathymetry along sound transmission line and locations of two transducers

4 Results and discussion

4.1 Cross-correlation between modulated signal and M-sequence

The typical wave forms of cross-correlation and the amplitude functionA(t) of the signal received by the upstream transducer are shown in Fig. 4. The cross-correlation between the modulated signal and the 10th order M-sequence has a sharp peak. Thus, we can determine the accurate arrival time of sound from the sharp peak of the cross-correlation wave. As a result, the cross-sectional average velocity was deduced from the arrival time of the transmitted signal.

Fig. 4 Typical time plots of amplitude function and cross-correlation wave form

4.2 Temporal variation of outflow rate

The river discharge is calculated by FATS as follows:

whereAis the cross-sectional area in which sound paths travel,His the water level, andθis the angle between the ray path and stream direction. The outflow rate from the gatesQ0is deduced from the following equation:

whereVis the water volume between the gates and the river cross-section along the sound transmission line (ray path) of FATS.

The temporal variation of the water level and outflow rate are illustrated in Fig. 5, whereHis the water level relative to the mean sea level. The water level varies due to intermittent discharge, and the variation of the water level reflects the duration of gate opening each day (Fig. 5(a)). Unfortunately, the measurement fell into abeyance from June 12 to 16 owing to some troubles with the downstream system (Fig. 5(b)). Specifically, we attached a new Bluetooth device in order to improve the system on June 12, and its operation damaged a DC-DC converter. The DC-DC converter was fixed on June 16. As a result, the system recovered and the improvement enabled us to access the system without wires.

Fig. 5 Time plots of water level and outflow rate for June, 2009

The water discharge estimated by FATS indicates a considerable magnitude even when the gate is closed. This is probably caused by the change in flow direction. In order to estimate the flow direction, a four-station system with two crossing transmission lines is required.

4.3 Relationship between ADCP and FATS data

In this subsection, the outflow rates deduced from the ADCP and FATS are compared in order to evaluate the performance of FATS. The relationship between the FATS velocity () and depth-averaged velocity of the ADCP (vADCP) is shown in Fig. 6. With low velocity (vADCP＜0.1 m/s ), the correlation betweenandvADCPis low, as shown in Fig. 6(a).

Conversely, there is high correlation in the higher velocity range, though there is large scatter whenvADCP＞0.8 m/s . Fig. 6(b) presents the data whenvADCP≥0.1 m/s; the solid line denotes a regression line,=0.083 4+0.237vADCP. The standard deviations of the residuals from the regression line forvADCPandare 0.096 m/s and 0.026 m/s, respectively. The two instruments were installed in different places. Moreover, the definitions of average are different:vADCPis depth-averaged andis the cross-sectional average velocity. Thus, the low correlation betweenvADCPanddoes not indicate poor performance of FATS.

Fig. 6 Relationship between cross-sectional average velocity of FATS and depth-averaged velocity of ADCP

From the regression line, we can see that the decay of the FATS velocity is more moderate than that of the ADCP. The decay of the FATS velocity after the closing of the gates is delayed because the cross-section of FATS is somewhere other than the gates.

The relationship between outflow rates of the ADCP and FATS is shown in Fig. 7. The ADCP discharge (QADCP) is estimated as a product of the depth-averaged velocity and the stream cross-sectional area of the gates. At a low flow rate, the FATS discharge (Q0) has a poor correlation withQADCP, like the poor correlation between the average velocities shown in Fig. 6(a). As shown in Fig. 7(b), however, correlation between both outflow rates is high with larger flow rates, both outflow rates are comparable, and the regression equation for the outflow rates isQ0=41.4+0.5QADCP. The standard deviations of residuals from the regression lines forQADCPandQ0are 19.3 m3/s and 11.8 m3/s, respectively.

Fig. 7 Relationship between outflow rates of FATS discharge and ADCP discharge

4.4 Temporal variation of water temperature

The sound speedcis estimated by Medwin’s formula (Eq. (12)) as a function oftemperatureT(°C), salinityS, and depthD(m) (Medwin 1975) in the ranges of 0≤T≤35° C, 0≤S≤45, and0≤D≤1 000m:

Since there is no saltwater intrusion at the experimental site, the water temperature can be estimated from the sound speed.

Fig. 8 shows temporal variation of water temperatures obtained from the temperature sensor at S2 and FATS. Both temperatures indicate diurnal variation, except on June 10, when the diurnal variation was not determined because of the rain. The temperature at S2 measured by the temperature sensor is higher than the cross-sectional average temperature estimated by FATS because the location of the temperature sensor is shallow.

Fig. 8 Temporal variation of water temperature for June, 2009

5 Conclusions

In order to conduct continuous measurement of river discharge, an FATS that utilizes a GPS clock and M-sequence modulation was developed and applied to shallow unsteady flow. Using a pair of transducers installed diagonally across the river, FATS was able to measure the cross-sectional average velocity. The cross-sectional average velocity of the river stream was estimated from the travel time of the transmission signal obtained along the ray paths, which cover the channel cross-section. The sufficiently high signal-to-noise ratio was obtained owing to the 10th order M-sequence modulation of the transmission signal.

In addition to measurement of flow rate, the cross-sectional average water temperature was deduced from the mean acoustic speed measured by FATS. This means that all of these environmental factors are observational targets of FATS.

Acknowledgements

We would like to thank Dr. Noriaki Gohda of Hiroshima University/Aqua Environmental Monitoring Limited Liability Partnership (AEM-LLP) for strong support in field work and data processing.

Chiu, C. L., and Hsu, S. M. 2006. Probabilistic approach to modeling of velocity distributions in fluid flows.Journal of Hydrology, 316(1-4), 28-42. [doi:10.1016/j.jhydrol.2005.04. 011]

Kawanisi, K., Kaneko, A., Razaz, M., and Abe, T. 2008. Measurement of cross-sectional average velocity in a shallow tidal river with a next-generation acoustic velocity meter.Proceedings of 16thIAHR-APD Congress and 3rd Symposium of IAHR-ISHS, Vol. V: Hydraulic Structures for Water Projects, 1973-1977.. Beijing: Tsinghua University.

Kawanisi, K., Watanabe, S., Kaneko, A., and Abe, T. 2009. River acoustic tomography for continuous measurement of water discharge.Proceedings of 3rd International Conference and Exhibition on Underwater Acoustic Measurements: Technologies and Results, 2, 613-620. Nafplion: Hellas Foundation for Research and Technology.

Kawanisi, K., Razaz, M., Kaneko, A., and Watanabe, S. 2010a. Long-term measurement of stream flow and salinity in a tidal river by the use of the fluvial acoustic tomography system.Journal of Hydrology, 380(1-2), 74-81. [doi:10.1016/j.jhydrol.2009.10.024]

Kawanisi, K., Watanabe, S., Kaneko, A., and Abe, T. 2010b. Continuous measurement of flood flow and cross-sectional average salinity in the Ota diversion channel with fluvial acoustic tomography.Annual Journal of Hydraulic Engineering-JSCE, 54, 1081-1086. (in Japanese)

Maghrebi, M. F., and Ball, J. F. 2006. New method for estimation of discharge.Journal of Hydraulic Engineering, 132(10), 1044-1015. [doi:10.1061/(ASCE)0733-9429(2006)132: 10(1044)]

Medwin, H. 1975. Speed of sound in water: A simple equation for realistic parameters.TheJournal of the Acoustical Society of America, 58, 1318-1319. [doi:10.1121/1.380790]

Ruhl, C. A., and DeRose, J. B. 2004.Investigation of Hydroacoustic Flow-Monitoring Alternatives at the Sacramento River at Freeport, California: Results of the 2002-2004 Pilot Study,Scientific Investigation Report (2004-5172). Reston: U. S. Department of the Interior, U. S. Geological Survey.

Simon, M. K., Omura, J. K., and Levitt, B. K. 1985.Spread Spectrum Communications Handbook.New York: McGraw-Hill.

Sloat, J. V., and Gain, W. S. 1995.Application of Acoustic Velocity Meters for Gaging Discharge of Three Low-Velocity Tidal Streams in the St. John River Basin, Northeast Florida, Water-Resources Investigations Report(95-4230). Tallahassee: U. S. Department of the Interior, U. S. Geological Survey.

Wang, F., and Huang, H. 2005. Horizontal acoustic Doppler current profiler (H-ADCP) for real-time open channel flow measurement: Flow calculation model and field validation.Proceedings of 31st IAHR Congress, 319-328. Seoul: International Association for Hydro-Environment Engineering and Research.

Zheng, H., Yamaoka, H., Gohda, N., Noguchi, H., and Kaneko, A. 1998. Design of the acoustic tomography system for velocity measurement with an application to the coastal sea.Journal of Acoustic Society of Japan (E), 19, 199-210.

This work was supported by the Construction Technology Research and Development Program of the Ministry of Land, Infrastructure, Transport and Tourism of Japan (No. 31), and the River Fund (N0.19-1212-005, 21-1212-009).

*Corresponding author (e-mail:kiyosi@hiroshima-u.ac.jp)

Received Sep. 17, 2009; Accepted Dec. 2, 2009