Developments in Hydrometric Technology: New and Emerging Instruments for Mapping River Hydrodynamics

by Marian Muste¹, Won Kim² and Janice M. Fulford³

Introduction

New demands on surface-water resources from an increasing world population and rising global living standards are requiring water managers to improve river flow measurements. Water managers are requiring flow instrumentation to measure those resources more accurately, in more detail and at lesser cost. Fortunately, new and emerging developments in flow instrumentation are significantly improving our capabilities to measure surface-water discharge and flow dynamics of rivers.

Until recently, the way in which instrumentation measured discharge and flow dynamics had not changed. Flow measurements relied on mechanical velocity meters that used the force of water to rotate a propeller, a method that had been used since the beginning of the previous century. During the past 20 years, the availability of inexpensive computing power, electronics and improved batteries has led to the development of electronic velocity instruments for mapping river hydrodynamics that previously would have been impossible.

Today, acoustic-, radar-, and image-based electronic velocity meters are revolutionizing the measurement of surface-water discharge and flow dynamics. In most measurement locations, they are replacing mechanical instruments and are becoming the velocity-measurement instruments of choice. These instruments can offer superior efficiency, performance and safety. Moreover, electronic instrumentation can measure velocities faster over larger areas, at higher spatial resolutions and at a more reasonable cost than previous mechanical instruments.

The new and emerging instruments are capable of measuring spatially distributed two-and three-dimensional kinematic features that can be related to important morphologic and hydrodynamic aspects of the natural rivers. In some cases, such as the emerging radar and image-based instruments, the instrument does not touch the water during the measurement. Unfortunately, none of the existing WMO guidance, which largely predates the new technologies being used, addresses the newer instrumentation. However, many groups, including the authors of this article, are looking at the performance and abilities of the new and emerging instruments.

Two electronic velocity instruments, acoustic Doppler current profilers (ADCPs) and large-scale particle image velocimetry (LSPIV), are examples of the new and emerging velocity instruments that are changing the way surface-water resources are measured. These instruments can efficiently measure river velocities that are needed to better understand complex geomorphic, hydrologic, and ecologic river channel processes and their interaction under normal and extreme conditions. Recent comparison studies with older methods and instrumentation are presented herein which demonstrate the utility of ADCPs and the potential of LSPIV in measuring discharge.

Acoustic Doppler current profilers: operation

ADCP is a new instrument that is typically mounted on boats (downward looking), but can be moored on the bottom (upward looking) or on the bank (sideways looking). ADCPs require the sensor to be in contact with the water to transmit and measure sound pulses (pings) directed through the water column. The sound pulses reflect or echo from small, suspended particles or bubbles moving in the acoustic beams (Figure 1), producing a shift in the transmitted sound from which the velocity is calculated. This phenomenon, the Doppler shift, is the same as the change in pitch heard by a person when a train blowing a whistle passes by. The pulses sent in different directions or beams (typically 3 or 4) from the ADCP sense different velocity components parallel to each of the beams. Under the assumption that the currents are uniform (homogeneous) within layers of constant depth, a trigonometric transformation is used to convert the velocity along the beams into three velocity components associated with a Cartesian coordinate system oriented to the instrument. Each acoustic pulse from a boat-mounted ADCP produces velocity measurements throughout the flow depth (Figure 1).

(a)	(b)
Figure 1 — Operating principle of an acoustic Doppler current profiler (Teledyne RDI ADCP configuration): (a) beam arrangement; (b) measurement output

ADCPs mounted on a moving boat can measure with relative ease the multi-component velocity profile below the boat, automatically providing velocity, depth and location information wherever the boat travels. Instrument manufacturers (e.g. RDI, 1996) state 0.25 per cent accuracy for velocity measurements made during ideal conditions of uniform horizontal velocities that are rarely, if ever, achieved. Poor measurement conditions where the water has few if any suspended particles to reflect the sound pulses or excessive sediment concentrations that absorb the sound pulses can prevent ADCPs from being used. Also, in some measurement situations, such as next to a vertical wall, the assumption of uniform horizontal velocities is invalid and measurement errors can result. Velocity errors can also result from sediments moving along the channel bottom and not using GPS. A vast amount of literature describing the underlying principles, configuration and operational aspects of ADCPs (e.g. RDI, 1996; SonTek, 2000) is available for reference.

Acoustic Doppler current profilers: measurement capabilities

Although acoustic velocity meters were used initially to measure velocity in oceanographic environments, the development of ADCPs for shallower water conditions led to their use for river discharge. The United States Geological Survey (USGS) first used ADCPs in 1985 and published a description of a system for measuring discharge in real-time using an ADCP in 1993 (Simpson and Oltmann, 1993). Comparison discharge measurements with proven mechanical meters have supported the use of ADCPs for discharge measurements (Mueller, 2003).

ADCPs are an established tool for river measurements that are manufactured by several companies (RDI, 1996; Sontek, 2000). Currently, both point velocity acoustic meters and ADCPs are used in approximately 30 per cent of the discharge measurements made by the USGS (Oberg et al., 2005). ADCPs have largely replaced the use of mechanical meters from boats in the USGS. ADCPs can accurately measure discharge in rivers with bi-directional flow in the water column without using special techniques, because the instruments measure the speed and direction of the flow. Mechanical velocity instruments typically measure only the speed of the flow, unless special techniques are used.

Most ADCP measurements in rivers have been, and still are, conducted from moving boats to obtain discharges. Boats (manned or tethered) with ADCPs attached are navigated across the river between opposite locations on the river bank (a transect) to make the flow measurement. Discharges can be obtained by using either customized algorithms developed for ADCP measurements conducted from moving boats (RDI, 1996) or conventional algorithms that use fixed-boat positioning with the velocity-area method. Consequently, the instrument software is tailored for discharge measurements with little attention to the additional information that can be extracted from the ADCP raw velocities. The voluminous data produced during a discharge measurement is rapidly and efficiently produced by an ADCP in comparison with older mechanical velocity instruments. For example, the operational settings used for Kissimmee River (Florida) discharge measurements (Merwade et al., 2008) used about 800 individual acoustic pulses (pings) to produce measurements at 8 000 locations (or acoustic bins) throughout the measurement transect.

With increased interest in streamwater quality and surface-water habitats, additional information is needed for monitoring, modelling and investigating sediment transport, scour, habitat restoration and hydraulic structures. Much of the needed information, such as computing the forces exerted by water and possibly measuring sediment concentrations, is actually available in the ADCP raw files but the appropriate extraction and processing tools are not provided by instrument manufacturers. Customized algorithms have been developed for these purposes by various user groups.

Measurements by Merwade et al. [ibid.] in the Kissimmee River illustrate the capabilities of ADCPs to provide data other than discharges using customized software such as AdcpXP developed at the Iowa Institute of Hydraulic Research (Kim et al., 2005). The raw files can be processed to represent river velocity and velocity-derived information either as bulk or averaged values for a cross section (one dimension) or as local values at a specific location in the flow (such as two or three dimensions). Examples of one-dimensional information that can be determined from ADCP raw files are: average cross-sectional depth, mean cross-sectional velocity, and Froude number.

	Figure 2 — Illustration of the ADCP capabilities to provide multi-dimensional information in rivers: (a) instantaneous velocity vector field acquired by ADCP operated from moving boats; (b) mean velocity distribution at selected verticals; (c) visualization of the cross-section circulation

Visualization software can be used with ADCP data to show how local velocity varies within a river cross-section (see Figures 2(b), 2(c) and 2(d)). It is possible to compute mean and turbulence characteristics in a river cross-section from time-series velocities collected by tethering or anchoring the ADCP boat at a fixed location (Szupiany et al., 2007). Examples of two-and three-dimensional information that can be determined from ADCP raw files are: vector plots of velocity, contour plots of velocity magnitudes, eddies, upwelling, and mean velocities and turbulence at fixed points in the flow.

Large-scale particle image velocimetry: operation

LSPIV is an emerging instrument that is based on an image-based technique, namely particle image velocimetry (PIV), used in fluids laboratories. Over the last 30 years, fast-paced developments in optics, lasers, electronics and computer hardware and software have triggered dramatic increases in the use of image-based techniques for flow visualization and quantitative measurements in laboratories. However, LSPIV has yet to be validated in the field for the same range of discharge measurement conditions as ADCPs.

Particle image velocimetry has greatly improved our capabilities to measure instantaneous velocity vectors in a variety of flows produced in the controlled environment of laboratories (e.g. Adrian, 1991). An appealing aspect of image velocimetry is its inherent simplicity; i.e. using images instead of transducer output such as signals, which make the technique more user-friendly than its predecessors. The technique records the image as raw digital information and can be reprocessed as needed with different spatial and temporal resolutions to obtain flow details. These capabilities have rapidly established image velocimetry as the method of choice for detailed turbulence measurements of two-and three-dimensional laboratory flows. Despite this popularity, image velocimetry techniques have not been widely applied outside fluids laboratories.

The first image velocimetry measurements for riverine environments were made in Japan by Fujita and Komura (1994). These measurements required imaging of large river surface areas, hence, the technique was dubbed Large-Scale PIV. LSPIV entails all four typical components of conventional PIV: illumination (by the Sun), flow seeding, image recording and image processing. As the images for LSPIV are typically recorded from an oblique angle, there is an additional correction to be applied to the images.

The measurement process is initiated by taking images of the water surface from a strategically chosen position (Figure 3). The movement of the water surface is perceptible only if it contains visible elements moving with the flow. In many situations, there are naturally occurring floating patterns (such as foam, boils, small debris and free-surface waviness) in the river that very efficiently provides visible elements that act as flow tracers. If they are missing, artificial seeding of the flow area to be measured can be done.

As mentioned above, the recorded images are geometrically distorted due to the perspective effect embedded in the imaging from an oblique angle. The photo images are transformed to their undistorted appearance and then processed to obtain velocities at the water surface. The movement of the flow is estimated from pairs of consecutive images through statistical inference conducted on the image patterns floating on the free surface. Velocities are then calculated over the entire image by dividing the estimated displacements by the time interval between successive images.

Discharge is computed using velocity-area methods. The water-surface velocity measured by LSPIV is adjusted to give a better estimate of the mean velocity in the water column and is multiplied by the appropriate sub area of the cross-section bathymetry, as sketched in Figure 3(d). Channel bathymetry can be obtained from direct surveys using specialized instruments (e.g. sonars or ADCPs). The channel bathymetry can be surveyed at the time of the LSPIV measurements or prior to them under the assumption that bathymetry is not changing in the time interval between the bed and water-surface measurements.

Figure 3 — Large-scale particle image velocimetry (LSPIV) principle and operational components: (a) illumination and seeding; (b) image recording; (c) image reconstruction to obtain ortho-rectified images and image processing; (d) algorithm for estimation of stream discharge using the free surface LSPIV measurements

For field measurements with poor water-surface illumination, scarce seeding, or other adverse measurement conditions acting on the water surface, measurements can have drastically reduced accuracy or be impossible to perform. Usually, sunlight is needed for LSPIV measurements, making night-time measurements difficult. For flows with a lack of pattern or tracers, erroneous velocities may be calculated and the resolution of the velocity map reduced. Poor camera angle to the flow can also result in reduced resolution. Accuracy of discharge measurement with LSPIV is affected by the bathymetry used, the assumption of velocity variation with depth and the measurement of water elevation during the measurement. Previously measured bathymetry may be different from that during the measurement and the adjustment of the surface-water velocities to an average velocity in the water column may not be very accurate. In slow flows, especially in combination with wind, LSPIV-measured surface velocities are not reliable for discharge measurements.

LSPIV configurations have been continuously improved and include a fixed real-time LSPIV system continuously measuring the discharge in Iowa River (Hauet et al., 2008) and a mobile truck-based system which can be deployed near practically any stream measurement site (Kim, 2008). Essential features of these configurations are sketched in Figure 4.

Figure 4 — LSPIV alternative configurations

Large-scale particle image velocimetry: measurement capabilities

The key advantage of LSPIV is that it simultaneously and remotely measures flow velocities over the entire imaged flow surface at higher resolutions than HF radar systems. Typical HF radar systems are limited to 300-m resolution. LSPIV can have resolutions of one metre or less. This feature is unique to LSPIV among velocity instruments. Areas from 100 to 5 000 m2 have been mapped non-intrusively with LSPIV to provide instantaneous velocity vector fields, document flow patterns and measure river discharges (Muste et al., 2008). Because of the ability of LSPIV to remotely measure velocity, it can be ideal for situations when floating debris can damage instruments in the water and place field personnel at risk. This is critically important during high flow periods that can be potentially hazardous for instruments and for the technical staff conducting measurements.

Image-based techniques are being investigated and used for non-intrusive (remotely sensed) discharge measurement and comprehensive characterization of river hydrodynamics in the USA and the Republic of Korea (Muste et al., 2008), Japan and France (Hauet et al., 2008). The raw LSPIV measurements are instantaneous vector fields over the imaged area, as illustrated in Figure 5(a). From the LSPIV vector field, spatial and temporal flow features such as mean velocities, streamlines and vorticity, as well as other velocity-derived quantities, such as flow, can be determined. In situations where LSPIV requirements are met, the technique can efficiently measure surface-flow velocities at numerous locations with considerably less effort than is required with point-based and profiling instruments. In some situations, such as measurements during extreme flow events (floods, hurricanes) or for very slow and shallow flow, LSPIV may be the only measurement alternative. Because LSPIV is an emerging instrument, published work is limited and covers a small sample of the river conditions in which discharge measurements are routinely made.

Figure 5 — LSPIV measurements in a small creek: (a) raw image; (b) transformed image indicating the grid for velocity vector computation; (c) resultant mean vector field

Comparison measurements

Recently, work was conducted comparing the measurement capabilities of LSPIV with ADCPs, mechanical meters and the index-velocity method over a range of flows. The test location is a 3.3 km long, stable cobble-bed river reach downstream of the Goesan hydroelectric power plant in Korea. There is an official gauging station in the reach and the dam upstream of the test reach supplies a wide range of discharges, from 6 m3/s to 1 400 m3/s, depending on the season.

Two different discharge measurement methods were used with the ADCP: moving boat; and tethered fixed-boat. The moving-boat method used a Rio Grande 1 200 kHz ADCP mounted on an inflated boat and made at least four transects of the river. The tethered fixed-boat method used a River Cat ADP at 20 or more locations along the measurement transect in accordance with ISO 748 (ISO, 2007) guidance. Discharge measurement by LSPIV technique used a digital camera but fewer measurements were made than with the ADCP.

Two methods were simultaneously used for a comparison: a conventional velocity-area technique using a mechanical instrument; and an index-velocity method with a fixed acoustic velocity instrument. The mechanical instrument, a standard Price AA current meter, was used either by wading at shallow flow or by using a bridge crane at faster and deeper flow. Velocity-area measurements were made at 20 or more locations in the measurement transect in accordance with ISO 748 guidance.

The index-velocity method is a continuous discharge measurement that uses a side-looking acoustic Doppler velocity meter (model: Argonaut-SL 1.5 MHz) to measure a reference velocity in the measurement section. This reference velocity and water level can then be used to calculate a discharge using a relationship between discharge, the reference velocity and the water level at (or near) the measurement section. The relationship is derived from many measurements made using the conventional velocity-area technique with a Price AA current meter.

Simultaneous discharge measurements were made during the period 2005-2007 with ADCP, LSPIV and the comparison methods. Measured discharge by ADCP, LSPIV and the comparison methods was compared with dam discharge. Dam discharge was determined from gate setting. The relationship between gate setting and discharge was found previously by laboratory modelling. Figure 6 shows the non-dimensional discharge measured by ADCP, LSPIV and comparison methods. The non-dimensional discharge is the discharge measured with an instrument and method divided by the dam discharge. Discharge measured by ADCP fixed‑boat method is within +/- 10 per cent of dam discharge. Discharge by moving-boat ADCP method also has good agreement with dam discharge, but with a slight positive bias. Though there are not many cases measured by LSPIV, the resultant discharge also is fairly similar to the dam discharge.

Figure 6 — Intercomparison of ADCP, LSPIV and other methods
Summary statistics of discharge measurement techniques

The table below lists summarized statistics of non-dimensional discharges by type of measurement method. The non-dimensionalized average discharges that are greater or less than one indicate a positive or negative bias with the dam discharge. Smaller values of standard deviation and root mean squared error (RMSE) are also indicators of closer agreement with the dam discharge. A smaller confidence interval by t-distribution (Student’s t for 95 per cent confidence interval) indicates the significance of the non-dimensionalized average discharge (and associated bias).

Statistical index	Discharge measurements techniques
	ADCP moving-vessel	ADCP fixed-vessel	LSPIV	Velocity-area	Index-velocity
Non-dimensionalized average discharge	1.055	1.013	0.987	1.022	0.983
Standard deviation of relative differences	5.7	4.4	8.1	6.5	9.3
Root mean squared error of relative differences	2.2	1.3	4.8	1.7	2.2
Confidence interval by t-distribution	±3.29	±2.54	±12.89	±3.23	±4.78

Because of the small number of measurements, none of the biases indicated by any of the non-dimensionalized average discharge are statistically significant. The velocity-area and ADCP fixed-boat techniques have similar statistics and those measurements are most similar to the dam discharge. The ADCP moving-boat technique measurements are more like the dam discharge than either LSPIV or the index-velocity technique. LSPIV measurements had the largest variability (large values for standard deviation and RMSE) of all the techniques. However, the small number of LSPIV measurements limits the significance of the LSPIV statistics.

Outlook

Proper application of any instrument in a measurement situation requires an understanding of its underlying principles of operation. In addition, for new and emerging instruments, careful evaluations of the instruments’ measurement capabilities should be conducted before the instruments are used routinely. ADCPs have become mature tools for measuring riverine environments, with well-documented capabilities. LSPIV is a technique still under scrutiny and considerable efforts are being made to develop the technique and to make it robust in various measurement conditions. Overall, LSPIV cannot be considered as the all-in-one instrument, but one which efficiently complements other instruments and supports well a variety of measurement purposes.

The mobility, autonomy and speed of the measurement may make ADCPs and LSPIV appropriate for intensive measurements during normal and extreme hydrological events. These new technologies have initiated a new era in river measurement by reducing costs and effort while increasing the number of measurements and allowing measurements in hazardous situations. The use of the new generation of instrumentation may shed light on critically important processes, such as the interaction of the main channel and overbank (floodplain) flows during floods, the impact of floodplain flows on riparian vegetation and habitat, evolution of meandering streams and the effect of river structures on the river ecosystem. The higher-dimensional flow components estimates provided by the new instruments may lead to advances in river monitoring of channel stabilization, bathymetry change due to dam removal, bank erosion, stream and wetland ecology, stream corridor restoration and environmental impact. Currently, ADCPs and LSPIV, along with many of the new instruments from the non-intrusive category, are the subjects of an overall assessment by the WMO project being carried out through the Commission for Hydrology: Assessment of the Performance of Flow Measurements Instruments and Techniques (Fulford et al., 2007).

References

Adrian, R.J., 1991: Particle-imaging techniques for experimental fluid mechanics, Ann Rev Fluid Mech, (23), 261-304.

Fujita, I. and S. Komura, 1994: On the accuracy of the correlation method, Proceedings of the 6th International Symposium on Flow Visualization, 858-862.

Fulford, J., P. Pilon, Z. Kopaliani, P. McCurry and C. Caponi, 2007: Call for collaboration in WMO project for the Assessment of the Performance of Flow Measurement Instruments and Techniques, J. Hydr. Engrg., 133 (12), 1439-1440.

Hauet, A., A. Kruger, W. Krajewski, A. Bradley, M. Muste, J.-D. Creutin and M. Wilson, 2008: Experimental system for real-time discharge estimation using an image-based method, Journal of Hydrologic Engineering, 13(2), 105-110.

ISO, 2007: Hydrometry—Measurement of Liquid Flow in Open Channels Using Current-Meters or Floats; ISO 748, 46 pp.

Kim, D., M. Muste and L. Weber, 2005: Development of new ADCP post-processing and visualization capabilities, Proceedings, XXXI IAHR Congress, Seoul, Republic of Korea.

Kim, W., 2008: Intercomparison of new hydrometric techniques in an experimental river reach, Proceedings, Experiences and Advancements in Hydrometry, Korea Institute of Construction Technology, Seoul, Republic of Korea.

Merwade, V., D. Kim and M. Muste, 2008: Characterization of river morphology and hydrodynamics with acoustic methods, Water Resources Research (in review).

Muste, M., I. Fujita and A. Hauet, 2008: Large-scale particle image velocimetry for measurements in riverine environments, special issue on hydrologic measurements, Water Resources Research, (submitted).

Mueller, D.S., 2003: Field evaluation of boat-mounted acoustic Doppler instruments used to measure streamflow, Proceedings IEEE/OES Seventh Working Conference on Current Measurement Technology, 30-34.

Oberg, K.A., S.E. Morlock and W.S. Caldwell, 2005: Quality-Assurance Plan for Discharge Measurements Using Acoustic Doppler Current Profilers, US Geological Survey Scientific Investigations Report 2005-5183.

RD Instruments, 1996: Acoustic Doppler Current Profilers—Principle of Operation, A Practical Primer, San Diego, CA, USA.

Simpson, M.R. and R.N. Oltmann, 1993: Discharge-Measurement Systems Using an Acoustic Doppler Current Profile with Applications to Large Rivers and Estuaries, US Geological Survey Water-Supply Paper 2395.

SonTek, 2000): Doppler velocity log for ROV/AUV applications, SonTek Newsletter, 6 (1), SonTek, San Diego, CA, USA.

Szupiany, R.N., M.L. Amsler, J.L. Best and D.R. Parson, 2007: A comparison of fixed- and moving-vessel flow measurements with an ADCP in a large river, J. Hydr. Engrg., 133 (12), 1299-1309.

________

¹ Iowa Institute of Hydraulic Research-Hydroscience & Engineering, University of Iowa, Iowa City, USA
² Korea Institute of Construction Technology, Seoul, Republic of Korea
³ US Geological Survey, Stennis Space Center, Mississippi 39529, USA