Flow Measurement and Instrumentation - vliz.be · Flow Measurement and Instrumentation 19 (2008) 29–40 Flow Measurement and Instrumentation ¬‚owmeasinst Accuracy of discharge

Flow Measurement and Instrumentation 19 (2008) 29–40

Flow Measurementand Instrumentation

www.elsevier.com/locate/flowmeasinst

Accuracy of discharge measurements in a vegetated river

L. De Doncker∗, P. Troch, R. Verhoeven

Ghent University, Hydraulics Laboratory, Department of Civil Engineering, Sint-Pietersnieuwstraat 41, 9000 Ghent, Belgium

Received 27 April 2007; received in revised form 29 July 2007; accepted 15 August 2007

Abstract

Hydraulic measurements are necessary to calibrate numerical models and to collect information about the situation under field and laboratoryconditions. The accuracy of the measurements depends on several factors.

Laboratory research is carried out to evaluate the accuracy of different techniques and methods under well-controlled conditions. Fieldmeasurements are performed to evaluate the impact of natural conditions on the accuracy of the measurements.

The better the quality of the discharge measurements, the more possible is the reliable determination of the roughness coefficient, leading toan improved accuracy of hydraulic modelling.

Several measurement techniques and instruments (as hydrometric propellers and electromagnetic devices) are discussed in this paper.Furthermore, the possibility of continuous measurements is researched, next to the influence of the macrophytes on discharge measurementin vegetated rivers.c© 2007 Elsevier Ltd. All rights reserved.

Keywords: Accuracy of hydraulic measurements; Discharge measurements; Hydraulic laboratory and field measurements; Vegetated rivers

1. Introduction

A numerical model, describing the exchange processesbetween surface water, groundwater and vegetation, needscalibration data to provide accurate results and to allowconclusions about interaction processes in the study area arounda vegetated river. As water flow is the driving force behindmost of these processes, accurate information on dischargesand velocities is of major importance. In the research field ofenvironmental engineering, hydraulic measurements are carriedout on a regular basis.

For that, measurements of surface water flow have to becarried out. Chow [2], ISO 748 [8] and Herschy et al. [6]describe classical flow velocity measurements with a propellercurrent meter. More advanced techniques use Doppler andelectromagnetic devices. Discharges are easily computed byintegration of the measured velocities over the cross-section.

Although, over the years, advantages and disadvantages ofthe traditional techniques have been encountered, they all havedemonstrated their value, both in the lab and in the field.

∗ Corresponding author.E-mail address: [email protected] (L. De Doncker).

0955-5986/$ - see front matter c© 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.flowmeasinst.2007.08.007

To avoid the disadvantages, Bass [1] presents the constanttemperature anemometry (CTA) systems. The device comprisesthree main parts. The temperature probe allows continuousmonitoring of the water temperature. Five velocity probesconsist of a hot-wire film encased within a resin attachedto a steel rod. The hot-wire film records voltages, and aconversion to velocities is provided. This system can be usedin vegetation without disturbing the environmental conditions.A high sampling frequency is possible, so that variations invelocity can be registered.

Fenton [3] presents some applications of mathematical andcomputational methods on the practice of flow measurement,resulting in more accurate and possibly simpler hydrographicalprocedures, whereas Thomas [10] discusses the benefits andpractical advantages in introducing a standard programmein the field of open channel flow measurement. Ward andTepper [14], demonstrated the possibilities of the use of theDoppler technology for velocity measurements.

Voet [13] indicates the limitations of measurementinstruments. Continuous registration asks for the possibilityto measure a wide variation of water depth or water velocity.While carrying out continuous measurements, the results haveto be checked to eliminate deviations due to different disturbing

http://www.elsevier.com/locate/flowmeasinst

mailto:[email protected]

http://dx.doi.org/10.1016/j.flowmeasinst.2007.08.007

30 L. De Doncker et al. / Flow Measurement and Instrumentation 19 (2008) 29–40

Fig. 1. Study area of the river Aa, Poederlee, province of Antwerp, Belgium.

influences such as obstacles in the river. Most of the time,discharges are calculated from registered water levels using a‘stage/discharge’-curve. A stage/discharge-curve is determinedby measurements of discharge and water level, the relationbetween these parameters is based on a polynomial functiondetermined by the use of the least square method. Velocitymeasurements, which are carried out with the most care andprecision, can add accuracy to this curve. Problems are detectedfor small rivers where the manual discharge measurements haveto be carried out fast due to the short reaction time of thesystem. Furthermore, for heavy rainfall, a continuous dischargeduring the measurements is not realistic. Increase and decreaseof water levels occurs very fast [12].

However, field measurements are difficult to perform dueto the continuously changing conditions. As a consequence,repetition of a test is not possible. Accuracy of themeasurements is therefore a responsibility of the operator.

In order to upgrade the overall quality of the measurements,‘International Standards’ describing correct measuring tech-niques have been elaborated. ISO748 consists of the descrip-tion of ‘Velocity-Area methods’ for the measurement of openchannel flow, while ISO2537 may be used for propeller mea-surements in open channels, ISO6416 for the use of acousticmeasurement instruments and ISO15768 for the electromag-netic devices.

In this study, two different techniques have been used tomeasure discharges in a vegetated river and the results are com-pared and discussed. On the one side, these techniques includethree methods based on velocity measurements. First of all, cur-rent meters are used. Second, velocities were measured with anelectromagnetic device and third, the ADCP was tested. On theother side, a technique based on a calibrated weir is considered.

2. Measuring locations

2.1. Laboratory measurements

First of all, the instruments under concern were calibratedin the Hydraulics Laboratory (Ghent University). Furthermore,

the measurement results of the devices were compared to eachother under well-defined and -controlled flow conditions andfinally, the influence of the measurement time on the results waschecked. These tests were performed in order to investigate theinfluence of the ‘unsteadiness’ of steady flow on the gaugingvalues of each instrument. Also, the control measurementsallow us to obtain a realistic view on the performance of eachinstrument. Results and discussion can be found in paragraph 5.

2.2. Field measurements

The field measurements were performed in the catchmentbasin of the river Aa, which is situated in the region of Antwerpin Belgium and is hydrographically part of the Nete basin.More than 40% of the water in the Nete basin is dischargedby the river Aa, which is therefore an important river. The riverAa flows into the Kleine Nete near the city of Grobbendonk.The origin of the river Aa is found near the communitiesof Merksplas and Turnhout and the river streams throughTurnhout, Gierle, Gielen, Poederlee and Vorselaar. The riverAa has a total length of 36.8 km and the drainage area is about23,700 ha. The study area is focused on the downstream part ofthe Aa, on a reach controlled by 2 weirs over a distance of 1.4km (Fig. 1). Fig. 2 shows the measuring locations upstream anddownstream.

3. Measuring methods

3.1. Introduction

Three different methods to measure the discharge in theriver Aa are applied and compared. First, methods makinguse of velocity measurements are applied. The techniquesused are current meters, electromagnetic devices and aDoppler instrument. Furthermore, the calibration formula ofthe downstream weir, obtained from a laboratory modelstudy is used. Also the limnimetric dataset acquired by HIC(Hydrologisch Informatie Centrum, Hydrologic Information

L. De Doncker et al. / Flow Measurement and Instrumentation 19 (2008) 29–40 31

Fig. 2. Upstream cross-section (use of cable from boat) and downstream cross-section (use of cable from bridge) for velocity measurements.

Centre, Borgerhout) mentioned further in this paper is discussedand compared to the measurements.

It can be added that different measurement techniques aresuitable for different goals. On the one hand, calibrated weirsare necessary for long term registration and deliver informationfor longer periods. The setup of mass balances in river stretches,on the other hand, is important on a shorter term and needsdirect measurements.

The techniques presented here are classified into two groups:the ‘global’ measurements and the ‘local’ measurements. As anexample, (chemical) tracer measurements are part of the firstgroup. Here, a time-averaged velocity over a certain distance inthe river is measured. Other techniques mentioned are basedon the local measurement of the velocity in one point andintegration of these velocities over the section.

A continuous registration of the water level and the positionof the flap of the weir allows for a continuous calculationof the discharge. Discontinuous measurements with the othertechniques only deliver good results if the discharge is constantduring the measurement period.

3.2. Methods making use of velocity measurements

3.2.1. In generalKnowing the dimensions of the river section and the

measured velocities, the flow is calculated using the principalof integration of the velocities over the cross-section (1) [2,8]:

Q =

n∑i=1

m∑j=1

Ui j∆Ai j (1)

with Q = discharge in river section (m3/s), Uij = local watervelocity (m/s),∆Aij = part of the wetted cross-section A towhich Ui j is attributed (m2).

The section is divided into several (m) vertical parts. Foreach of these vertical parts, the velocity of the water is measuredon different (n) depths, so that the velocity profile is knownover the entire water depth. The result of the measurements isan overview of velocities over the section. Integration of thevalues over the wetted area yields the discharge, from which thespatially averaged velocity over the cross-section is calculatedby dividing the discharge by the wetted cross-section area.

In general, a limited number (1 or 2) of measurements oneach vertical are carried out according to standards, supposinga Prandtl Von Karman velocity profile. However, this profile isnot seen in vegetated rivers [9], so more intensive gauging in alarger number of measurement points per vertical is needed.

Good performance of the measurements with electromag-netic and acoustic devices is described by the ‘InternationalStandards Organisation’ [8]. For measurement of liquid flows inopen channels, ISO 2537 reports the use of current meters. ISO9213 describes the electromagnetic method and ISO 6416 theultrasonic or acoustic method. Doppler devices are discussed inISO/TS 24154.

3.2.2. Current metersFirst, flow measurements have been carried out using

hydrometric propellers (Type: OTT, C31 Universal CurrentMeter). This instrument is designed for flow metering incombination with a cable suspended from a bridge or boat.The rotation rate of the calibrated propeller is proportional tothe water velocity. The velocity is determined by registrationof the time period over a pre-selected number of propellerrevolutions with a counter unit. Knowing the dimensions of theriver channel and the measured velocities, the flow is calculatedusing the continuity equation (Eq. (1)).

3.2.3. Electromagnetic current devicesFurthermore, an electromagnetic instrument (Type: OTT,

Nautilus C2000/SENSA Z300 and Valeport, Type 801) is usedto measure the velocities. The advantage of this device is that itcan be used in between vegetation. There are no moving partswhich can interfere with the macrophytes, the device is wear-resistant and maintenance-free. The range of measurementsis preferably between 0.0 and 2.5 m/s. The instrument canbe used in shallow water. The measurements are independentof any parameters, such as temperature, suspended sedimentconcentration and salinity.

The electromagnetic flow meter is based on Faraday’sLaw that a conductor (water) moving in a magnetic field,produced by a coil in the sensor, produces a voltage. Thisvoltage is perpendicular to the movement of the conductor andperpendicular to the direction of the magnetic field. The voltage


Table 1Calibration parameters

Angle of weir (◦) hkr (m TAW) a b n m

0 8.830 9.247 1.662 1.457 59.0210 9.200 10.250 1.705 1.501 9.44320 9.564 9.980 1.712 1.441 6.54030 9.913 9.851 1.706 1.612 7.31740 10.237 9.663 1.719 1.950 6.18750 10.532 9.498 1.761 2.120 4.44360 10.800 9.559 1.776 – –

is proportional to the velocity of the water. The samplingvolume is measured above the surface of the sensor (OTT),around the cylindrical volume of the sensor (Valeport) or abovethe flat volume of the sensor (Valeport).

3.2.4. Acoustic Doppler current profiler (ADCP)Third, the Streampro ADCP is used to measure the

velocities. This instrument is applied for shallow water (max.2 m) discharge measurement. Doppler technology is used tomeasure the vertical velocity profile in a chosen number ofsections across the river. The device is pulled over the cross-section with a constant velocity perpendicular to the flow.The velocity is measured at ‘every’ point of the cross-section,i.e. the slower the movement of the device, the more are theverticals that can be sampled. Next, the velocity is measuredover the entire water depth, at different heights. As a result, avertical surface of velocities in the cross-section is composed.This faster measurement method allows more measurementsand more data collection. Velocity and depth information arecombined to estimate the discharge. Due to the risk of falsereflections, measurements start and end at about 1 m from thebanks.

3.3. Methods making use of a calibrated weir

The calibration formula, for different positions of thedownstream weir in the measurement stretch (Fig. 4),mentioned above, was determined in 1995 by the HydraulicsLaboratory (UGent) [11], using a scale model (scale: 1/5) ofthe weir in the lab. Water levels were measured with electronicbalances (accuracy 0.2 mm + / − 0.1 mm), discharge wasregistered with an electromagnetic current meter (accuracy0.2% of the exact value). The angle of the flap of the weirvaried over 0◦, 10◦, 20◦, 30◦, 40◦ and 50◦, for submerged andfree overflow. For an angle of 60◦, only measurements for freeoverflow were performed.

Expressions for the discharge over the weir are given inEqs. (2) and (3). Coefficients of these equations are gathered inTable 1. The scale model is presented in Fig. 3 together with theparameters of the equations (hopw = upstream water level (mTAW), ha f w = downstream water level (m TAW), hkr = crestlevel (m TAW), a, b, m and n = parameter values in Eqs. (2)and (3)).

Submerged flow

Q = a(hopw − hkr )b(

1 −

(ha f w − hkr

hopw − hkr

)m)(1n

). (2)

Fig. 3. The river Aa: scale model of the downstream weir and parameters forcalibration of the downstream weir.

Free flow

Q = a(hopw − hkr )b. (3)

For coefficients a and b, following expressions are obtained,in which θ is the angle of the flap (◦):

a = 9.565 + 0.02675θ − 0.0005005θ2 (4)

b = 1.675 + 0.001111θ + 0.000008869θ2. (5)

After obtaining the calibration formula for the downstreamweir, the discharge is calculated as a function of the upstreamwater level hopw, the flap crest level hkr and the weir angleaccording to Eqs. (2) and (3).

For this study, the discharge calculated using the calibrationformula is used as a value of comparison for the dischargemeasured downstream making use of the methods discussed inthe previous paragraph.

3.4. Tracer methods

Estimations of the discharge are carried out based onconductivity measurements. An advection–dispersion reactionmodel of the river will be developed to determine the discharge


Fig. 4. Weir downstream (view downstream (left) and upstream (right) direction).

in the river. Therefore, input and output signals of the tracersin the river are used. If the water volume is known and thetime the signal needs to move from upstream to downstreamis registered, the discharge can be estimated.

Further research is beyond the scope of this paper.

3.5. Gauge data

Registration of the water level is executed in eachmeasurement section, upstream and downstream of the stretch.Each weir is equipped with a gauge, so water level measurementis rather easy. Variation of water level during the measurementis registered by reading the gauge during the measurementof the discharge. The presence of 3 limnimeters of FlandersHydraulics Lab (HIC) in the stretch between weir 3 and weir 4allows us to study the evolution of the slope of the water surfaceas a function of time.

4. Accuracy of measurement methods

Hydraulic measurements are uncertain because the obtainedvalues are only an estimation of the exact value, as explainedin e.g. [4]. So, the measurement error has to be included in anydata analysis. To learn about the accuracy and the reliabilityof the results, all various sources of errors in the measurementhave to be examined. Random errors and systematic errorsare distinguished. Some of the random errors (also calledexperimental errors) can be reduced by increasing the numberof observations. These errors are the most important to beconsidered in river gauging. Systematic errors are present inthe water level registration, the current meter accuracy or in theposition of the flap crest. By calibration of the instruments andserious control of all measurements, the errors can be reducedto a minimum.

The discharge is calculated from discrete velocity measure-ments (using the velocity-area method); with F, a factor thatrelates the discrete sum over the finite number of verticals tothe integral of the continuous function over the cross-section:

Q = Fn∑

i=1

m∑j=1

Ui j∆Ai j . (6)

To generate a reliable measurement, a sufficient number ofverticals needs to be gauged. In normal circumstances a total of13 verticals (m = 13) is considered to deliver a value 1 of thecoefficient F . If too little verticals are measured, the dischargeis underestimated and a value greater than 1 for F is applied.For a single determination of the discharge, the uncertainty atthe 95% confidence limit is up to 7%. The choice of a sufficientnumber of verticals can reduce this value [4].

The uncertainty on the discharge is given as:

u(Q)2= u2

m + u2s +

∑((bi divi )

2(u2bi + u2

di + u2vi ))

(∑

(bi divi ))2 (7)

u2vi = u2

pi + u2ci + u2

ei (8)

u2s = u ∗

2bi +u ∗

2di +u ∗

2ci (9)

with: um = uncertainty on the limited number of verticals; ubi ,udi , uvi = relative (percentage) standard uncertainties in thewidth, depth and mean velocity at vertical d, us = systematicuncertainty due to calibration errors; u pi = uncertainty due tothe limited number of depths at which velocity measurementsare made; uci = uncertainty on the characteristics of themeasurement instrument, uei = uncertainty due to fluctuationof the velocity during the measurement, u ∗bi = percentagesystematic uncertainty in the instrument measuring width,u ∗di = percentage systematic uncertainty in the instrumentmeasuring depth, u ∗ci = percentage systematic uncertainty inthe registration of the current meter [8,5].

For the values of the uncertainties, reference is made to ISO748, ISO 2537, ISO 6416 and ISO 15768 [8].

For us , an estimated practical value of 1% is taken.If the segment discharges (bi divi ) are nearly equal and the

random uncertainties X i are nearly equal and are of value Xthen:

u(Q)2= u2

m + u2s +

1m

(u2b + u2

d + u2e + u2

p + u2c) (10)

which is the simplified error equation with m being thenumber of verticals. Calculations in [5] show that there is onlylittle difference between the original and the simplified errorequation.


Measurements carried out with a propeller are the mostaccurate if they are performed standing in the river, as thismakes it possible to have an accurate measurement of the waterdepth. Determination of the water depth from a bridge is moredifficult with deviations of 1–2 cm. This leads to an error of3.5%–7% on a water depth of 30 cm. Measurements performedin the river (standing in the river or from a boat) allow us tochoose a section which is smaller, with better flow conditionsand to take into account the condition of the bottom. By thismore accurate results can be obtained [7,12].

Accuracy of electromagnetic devices is up to 0.5% while forpropellers the accuracy is 1% of the actual value. Deviations arelarger for lower ranges of velocities. Also flow characteristicsand the precision of the measurement depth influence theaccuracy of the result. This leads to an overall accuracy of2%–5% for the determination of the discharge. Calculations ofthe discharge, using the formulas of a calibrated weir lead to anaccuracy of 5% (free overflow) to 10% (submerged flow) [7].

In the analysis of the laboratory measurements, absolute andrelative deviations are calculated to check the accuracy of themeasurements.

Calculating the accuracy according to Herschy [5] resultsin the following values. For thin plate weirs, an accuracy of1% is obtained. Using the measurement techniques and valuesof the river Aa, the following calculation is made for one ofthe field measurements. The uncertainty values mentioned arepercentage standard deviations at the 95% confidence limits.These values are used because the true value of the dischargeis unknown. An estimate of the true value has therefore to bemade by calculating the uncertainty in the measurement, theuncertainty being defined as the range in which the true valueis expected to lie expressed at the 95% confidence level. Fordischarge calculation, it is not possible to calculate the standarddeviation, because no repeated measurements can be performedunder the same conditions. Therefore, an estimate of the truevalue has to be made by examining all the various sources oferrors in the measurement.

On December 6th 2006, a discharge of 3.08 m3/s was mea-sured with an electromagnetic current meter. Measurementswere carried out on 20 verticals and the average velocity inthe cross-section was 0.198 m/s. Furthermore; um = 5%,ue = 15%, ub = 0.1%, ud = 1%, u p = 5% (for 5 mea-surement points), uc = 1.5%, u ∗b = 0.5%, ud = 0.5%,u ∗c = 1.0%. For the value of the systematic uncertainty, 1% isused as practical value, this value is almost independent of theinstrument type (propeller or electromagnetic device). The ran-dom accuracy is much more stringent. The high value of ue isdue to a rather low velocity in the river and a short measurementtime. For velocities over 0.3 m/s, an exposure time of 30 s issufficient, for lower velocities, the measurement time is prefer-able up to 3 min. This is necessary to eliminate the fluctuationof the velocity. The measurements on the river Aa are carriedout in dry periods and with a fixed position of the weir, whichcan result in lower velocity fluctuations and can allow a lowervalue of ue. With the above mentioned values, the random un-certainty on the discharge is 6.1%, to obtain a value within the95% confidence interval.

5. Laboratory measurements

5.1. Introduction

Laboratory measurements are carried out to calibrate instru-ments, to compare the results of the different measurement in-struments and to check certain measurement influences underwell-known and -controlled conditions.

5.2. Calibration

The electromagnetic devices and the hydrometric propellersare calibrated in the lab. Next to the hydrometric propellers(OTT C31 Universal Current Meter) and the electromagneticinstrument (OTT, Nautilus C2000/SENSA Z300), a thirdinstrument was tested (Valeport, Single Axis ElectromagneticFlow Meter, Model 801).

Calibrations are executed in a rectangular channel (length of40 m, width of 1 m and depth of 1 m) in quiescent water. Thepropeller is installed on a trolley in the middle of the channeland is pulled through this channel with a constant velocity.Distance, time and revolutions of the propeller or the readingof the electromagnetic instrument are acquired by a computer.The same measurements are carried out for different velocitiesallowing calibration formulas to be developed (11) with anaccuracy which is better than 1% of the actual velocity. For theused propeller (C31-87200, A-67619):

V = 0.129n + 0.019 0 < n < 4.374 (11)

V = 0.128n + 0.022 4.374 < n < 10 (12)

with V = velocity (m/s); n = N/t ; N = number of revolutions(1/s); t = measurement duration (s).

Also, further measurements are carried out with twodifferent propellers (22 558 and 23 088), also calibrated. Themeasurement range (V min and V max) of each propeller isdetermined. These values indicate the importance of usingthe instrument with specific measurement range under specificconditions. For propeller 22 558, V min = 0.025 m/s andV max = 0.583 m/s, while propeller 23 088 is more suitablefor higher velocities (V min = 0.008 m/s and V max =

1.051 m/s).Both electromagnetic devices are also calibrated in the lab

where the following formulas (V(m/s)) are obtained.For the Valeport, 801:

0 < V (E M) < 0.22 : V = 0.9514 ∗ V (E M) + 0.0129 (13)

0.22 < V (E M) : V = V (E M). (14)

For the OTT, Nautilius:

0.049 < V (E M) < 0.276 : V

= 0.8582 ∗ V (E M) + 0.0075 (15)

0.276 < V (E M) < 0.883 : V

= 1.0227 ∗ V (E M) − 0.0401 (16)

0.883 < V (E M) < 2.205 : V

= 1.014 ∗ V (E M) − 0.0305. (17)


Table 2Velocities (m/s) measured in the lab by the different measuring instrumentsand corrected with the calibration formulas

Q (l/s) 12.22 69.93 100.52U Gcanal 0.058 0.250 0.326E MValeport 0.075 0.294 0.393E MOTT 0.069 0.273 0.368Propeller 0.069 0.299 0.396/0.398/0.382

Three values are indicated for the propeller measurement due to the threedifferent propellers (OTT: 2-23092 1-22566, A-67619) used in the test.

Table 3Measurements in the lab with a hydrometric propeller and different numbers ofrotations and measurement times

Propeller 22 558 23 08810 rot. 20 rot. 10 s 10 rot. 20 rot.

Qcal (l/s) 59.42 59.42 59.42 59.42 59.42Qmin (l/s) 59.33 59.15 59.31 59.32 59.36Qmax (l/s) 60.08 60.07 59.85 59.91 59.98abs. dev. (l/s) 1.34 1.34 1.34 1.81 1.81rel. dev. (%) 2.25 2.24 2.25 3.04 3.04stand. dev. (l/s) 0.224 0.303 0.199 0.199 0.229

5.3. Comparison

Comparative tests with the 3 devices were performed in aflume of the Hydraulics Laboratory, with a length of 12 mand a width of 70 cm. Three different discharges werechosen and the corresponding velocities in the middle of thechannel were measured with the hydrometric propeller andthe electromagnetic instruments. The results are presented inTable 2 and show only little difference. The value U Gcanal isobtained by using a calibrated weir at the upstream end of thetest flume, it is lower because this is an average uniform valueover the rectangular section.

5.4. Influence of the measurement time

In order to evaluate the influence of unsteadiness of steadyflow, for a chosen discharge, more measurements are carried outwith the hydrometric propellers and the electromagnetic device.Therefore, two different propellers (22 558 and 23 088) and theelectromagnetic instrument (Valeport) are used for a varyingmeasurement time or varying number of rotations.

For each of the situations (different propellers and differentnumber of rotations), the measurements were repeated 10times. Each of the measured discharges is compared tothe calibration value (discharge measured with a calibratedweir) and the minimum and maximum discharge of the 10measured values (cf. infra) is mentioned in Table 3. Alsothe accuracy of the results is checked by calculation of theabsolute and relative deviations, next to the standard deviation(Table 3). The absolute deviation is determined for each ofthe 10 measurements, calculating the absolute deviation on thevelocity and on the geometric dimensions. For each of the10 measurements, the absolute deviation is more or less thesame, and the average value is mentioned in Table 3. The smallvariation in the absolute deviation is confirmed by the small

Table 4Measurements in the lab with an electromagnetic device and differentmeasurement times

Valeport 10 s 20 s

Qcal (l/s) 59.06 59.06Qmin (l/s) 59.80 59.71Qmax (l/s) 61.27 60.75abs. dev. (l/s) 1.41 1.41rel. dev. [%] 2.33 2.33stand. dev. (l/s) 0.363 0.352

variation on the discharge under these laboratorial conditions.Measurements are performed at 3 verticals in the cross-section(10, 35 and 50 cm) at different heights (4, 8, 12, 16, 20 and24 cm) while the water depth in the flume was 27 cm.

The average velocity for the measurements is 0.317 m/s.Looking at the measurement range of the propellers, it isexpected (and confirmed in Table 3) that propeller 23 088 ismore suited for these measurements than propeller 22 558,which is rather meant to be used for higher velocities.

Hydrometric propellers used in good flow conditions andin an accurate way can measure with an accuracy to 1%of the actual value, which is confirmed by the values inTable 3. The measurement of the time with a fixed numberof rotations or the measurement of the number of rotationsduring a certain time period to calculate the velocity does notshow significant differences, which indicates that measurementtime and number of rotations are of minor importance underlab conditions. Higher deviations are observed for the resultsof the second propeller, which is attributed to differences incalibration formulas and/or application range. It should bementioned that, the measured average velocity is situated in themiddle of the range for propeller 22 558, while propeller 23 088is more suitable for higher values of the velocity.

The same measurements are performed with an electromag-netic device (Table 4) at 3 verticals (10, 35 and 55 cm) at differ-ent heights (5, 15, 25 cm). These heights are chosen taking intoaccount the spherical measurement volume (diameter of 12 cm)of the electromagnetic device. At each point, 10 measurementsare carried out and the constant value of the discharge (compar-ing Qmin and Qmax of the 10 measurements) indicates stableconditions and accurate measurements. Velocities up to 0.3 m/slead to a relative deviation of 1.5% which corresponds withthe technical information (deviation up to 3% for velocities of0.1 m/s). For higher velocities, the deviation is smaller.

As a conclusion, all used instruments, hydrometricpropellers and electromagnetic devices, lead to accurate resultsin well-controlled lab conditions. Variations in measurementtime are of less importance. It also becomes clear thatapplication of a device (propeller) which is appropriate for thevelocities under concern leads to better quality of the results.

6. Field measurements

6.1. Importance of measurements

Besides the laboratory tests, a large number of fieldmeasurements were carried out. In the stretch of the river Aa,


discharges were measured during the period 2004–2006 once amonth and in 2005 (April, August) and 2006 (February, May,July, September) extended campaigns were organised.

Velocity measurements were performed using the hydromet-ric propellers and the electromagnetic instruments. The dis-charge was calculated using the method of integration of thevelocity field over the cross-section.

To check the sensitivity of the results to measurement errors,the discharge measurements of the different methods are com-pared. The results are also related to the values obtained by HICand to the values calculated from the calibration formula of thedownstream weir. For the comparison, river characteristics (asthe wetted section and the river width) are used as measured.

The Hydrological Information Centre (HIC) of FlandersHydraulics Research has registered the water levels upstreamand downstream along the studied stretch during 2004, 2005and 2006. Also the discharge is determined in this area. Itshould be mentioned that the values of HIC are estimatedstarting from the measurements in two neighbouring stationsand taking into account the surface of the correspondingcatchment areas. Due to this approximate determination ofthese discharges, differences with the values obtained from themeasurements can be remarked. The calibration parameters ofthe weir allow a more accurate calculation when the positionof the weir and the water height over the weir are registeredcontinuously, which is not possible during the reported period.Also, it is important that for a reliable measurement the weirshould be kept free from aquatic weeds.

Table 5 presents the results of the discharges measured(Q upstream and Q downstream), their average values, theestimations of HIC and the discharge obtained using thecalibration formula of the weir. Individual deviations are due todifferences in determination methods. Largest differences canbe remarked to have occurred in Spring 2006, certainly for thevalues of HIC compared to the others.

There is measured only a small difference between theupstream and downstream discharge values, due to groundwaterinflow/outflow, to little tributary inflow between the sectionsand to inaccuracy of the measurements. In general, the differentvalues follow the same trend. Deviations that sometimes occurcan be explained by vegetation patches disturbing the free flowover the weir.

Next to the comparison of the global discharge once amonth, during the measurement campaigns, discharges weremeasured repetitively in the same sections with differentinstruments and on the same day (Table 6). As can be seen, onlysmall deviations occur while measuring in the same section.Also when different measurement instruments are used, theresults are comparable. Exception is made for the use ofpropellers in periods with a wealthy plant growth (cf. infra).Remarkable also is that the results for the discharge determinedby HIC are higher in 2005 and lower in 2006, with an analogueoffset value.

6.2. Continuous discharge measurement

During the measurement campaign in April (2005), ahydrometric propeller has been installed on a fixed location

Table 5

Comparison of the results of the discharge measurements (m3/s)

Date Qupstream

Qdownstream

Qaverage

QHIC

Qcalibration-weir

29/09/2004 0.98 1.02 1.00 1.3728/10/2004 1.47 1.27 1.37 2.1323/11/2004 2.81 3.00 2.91 3.25 3.7915/12/2004 1.33 1.28 1.31 1.73 1.25

25/01/2005 2.98 3.21 3.10 3.01 2.6723/02/2005 2.86 3.20 3.03 2.96 3.3117/03/2005 1.64 1.84 1.74 1.92 1.8412/04/2005 1.81 1.62 1.72 1.70 1.7323/05/2005 1.42 1.18 1.30 1.06 1.2523/06/2005 0.69 0.60 0.65 0.52 0.5214/07/2005 1.06 0.94 1.00 1.13 1.3622/08/2005 0.72 0.73 0.73 1.0028/09/2005 1.05 0.77 0.91 1.37 0.9426/10/2005 1.20 1.26 1.23 1.32 1.6030/11/2005 2.18 1.99 2.09 2.02 2.1121/12/2005 1.80 1.86 1.83 1.81 2.53

11/01/2006 1.28 1.44 1.36 1.18 1.5907/02/2006 1.24 1.35 1.30 0.89 2.3915/03/2006 2.56 2.63 2.60 1.60 2.6719/04/2006 1.97 1.95 1.96 1.25 1.9815/05/2006 0.85 0.99 0.92 0.70 0.9405/07/2006 0.64 0.66 0.65 0.64 0.5226/06/2006 0.90 0.95 0.93 0.79 0.7618/10/2006 0.89 0.92 0.91 0.81 1.1509/11/2006 0.96 0.97 0.97 0.83 1.3606/12/2006 3.08 2.91 3.00 2.74 3.96

in the downstream end of the river stretch for three days. So,continuous velocity measurements (‘velocity measurements’)were carried out. With this velocity and compared to thevelocities at the same place during the discharge measurements,a value for the continuous discharge can be determined. Inthe same period, discontinuous discharge measurements areperformed in the same section for calibration. Fig. 5 showsthe result of the continuous propeller measurements (velocities)and the continuous discharges (‘continuous discharges’) basedon the discontinuous discharge values. The discontinuousdischarge measurements (‘triangles’) were carried out in theneighbourhood of the downstream weir (weir 4) on April 12th(1.62 m3/s), April 13th (1.46 m3/s), April 14th (2.79 m3/s)and April 14th (2.89 m3/s). Also some ADCP results (‘dots’)and the values presented by HIC (‘discharges HIC’) are added.As can be seen, the increase in discharge on April 14th is alsoregistered, due to heavy rainfall during the night.

It is checked whether it is possible to deliver accuratecontinuous discharge measurements with the fixed propeller.The propeller measures velocities in 1 point from whichthe discharge over the cross-section has to be determined.Therefore, the measured values for the velocity are correctedusing the discrete discharge determinations in the downstreamsection.

The position of the propeller was based on the discontinuousdischarge measurements in that section. Out of all the results,the vertical and the water depth were selected where the averagevelocity was found. With that velocity and knowledge of the


Table 6

Comparison of the results of the discharge measurements (m3/s), at the same place and on the same day

Date Time Instrument Q meas Q HIC Q calibration-weir

12/04/2005 downstream 10:30–11:21 A106202 1.70 1.77 –13:35–15:30 A67619 1.62 1.70 1.7315:45–17:00 Valeport 1.91 1.71 –

14/04/2005 downstream 11:20–12:45 Valeport 2.79 2.88 –13:15–14:20 Valeport 2.89 2.88 –

22/08/2005 downstream 11:45–13:10 Valeport 0.73 1.00 –15:40–17:00 Valeport 0.77 0.98 –

23/08/2005 upstream 10:15–12:10 A106202 0.68 1.00 –13:35–15:10 Valeport 0.80 1.01 –

23/08/2005 downstream 09:45–11:50 Valeport 0.77 1.01 0.6813:32–15:20 A106202 0.79 1.01 –

24/08/2005 upstream 09:25–11:20 A106202 0.69 1.10 –12:15–13:40 Valeport 0.78 1.09 –

07/02/2006 upstream 11:10–12:05 A106202 1.24 0.89 –12:20–13:25 Valeport 1.32 0.89 –09:50–11:30 Valeport 1.47 0.88 –

08/02/2006 upstream 09:10–10:15 Valeport 2.17 1.29 –11:20–12:48 A106202 2.84 1.89 –09:50–11:30 Valeport 2.59 1.53 –

08/02/2006 downstream 09:30–11:00 A106202 2.13 1.40 2.8212:00–13:15 Valeport 2.99 2.03 –

09/02/2006 downstream 10:40–11:45 A106202 2.64 1.63 1.7310:45–11:45 Valeport 2.37 1.63 –

16/05/2006 upstream 10:15–11:35 Valeport 0.93 0.73 –12:36–13:30 OTT 0.82 0.72 –

16/05/2006 downstream 11:41–13:30 Valeport 1.02 0.72 0.9410:16–11:50 OTT 1.01 0.73 –

17/05/2006 upstream 09:30–10:30 Valeport 1.16 0.83 –10:40–11:35 Valeport 1.03 0.85 –

17/05/2006 downstream 09:30–10:55 OTT 1.00 0.83 1.0411:20–12:25 OTT 1.14 0.89 –

18/05/2006 upstream 09:21–10:25 OTT 1.07 0.90 –10:30–11:15 OTT 1.08 0.93 –

26/09/2006 downstream 09:05–10:29 Valeport 0.88 0.79 –9:10–10:04 OTT 0.95 0.79 0.76

28/09/2006 downstream 09:00–10:12 Valeport 0.89 0.76 1.0511:15–12:06 Valeport 0.95 0.765 –

wetted cross-section, the discharge can be calculated. Thedifficulty is to find the right combination between the velocityin that fixed point and the wetted cross-section, which dependon the geometry and the bottom. Discontinuous dischargemeasurements deliver some calibration points (A = 9.5 m2

for Q = 1.70 m3/s and A = 10.9 m2 for Q = 1.91 m3/s),but a correct determination of the location where to install thefixed propeller meter is difficult due to presence of vegetation,sedimentation and erosion, geometry of the section, etc. Afterall, the wetted cross-section increases if the discharge increases.An option is to set up a table or graph with the combinationvelocity-wetted cross-section derived from the discontinuousdischarge calculation. Therefore, a lot of measurements have tobe carried out for different discharge situations. Furthermore,this relation will be different for summer and winter situationsin vegetated rivers.

For the determination of the location for the continuousmeasurement use is made of the theoretical, logarithmic PrandtlVon Karman profile, where the average velocity is found at 0.37times the water depth. This is about the position of the fixedpropeller at the beginning of the measurements (April 12th). On

April 14th, a peak flow is measured together with a water levelrise of 11 cm. In this situation the velocities are registered at adepth which is lower than the depth corresponding with averagevelocities. So, the water level rise of about 10% includes acorrection on the calculation of the discharge, by correcting theregistered velocity.

From Fig. 5, it can be seen that using a fixed velocitymeasurement instrument leads to a good indication for thedischarge if discontinuous discharge measurements are used tocorrect the velocity measurements. The curve obtained fromHIC gives higher discharge values. This can be due to tworeasons; first of all, the estimations as done by HIC introducedeviations. Furthermore, to calculate the discharge, next to themeasurement of the velocity, an accurate knowledge of thewetted cross-section is necessary.

6.3. Influence from vegetation on discharge measurements

In this paragraph, it is checked whether the presence ofvegetation influences the quality and the accuracy of thedischarge measurements.


Fig. 5. Continuous downstream discharge measurement during the measurement campaign. Measured values with the propeller and corrected values based on theparticular measurements.

Table 7Overview of discharge Q, Wetted section A, Hydraulic radius R, Wettedperimeter P and Average velocity V for the different sections

Time Q A R P V(m3/s) (m2) (m) (m) (m/s)

AprilSection 1a 10:00–11:10 2.72 12.858 0.866 14.844 0.211Section 2a 11:20–12:45 2.77 13.108 0.864 15.163 0.211Section 3a 13:15–14:25 2.89 12.813 0.881 15.549 0.225AugustSection 1b 10:00–12:05 1.47 17.119 0.996 17.179 0.086Section 2b 09:40–11:40 1.46 13.626 0.815 16.725 0.107Section 3b 12:20–14:05 1.44 14.479 0.962 15.053 0.099

In the month of April, discharge measurements werecarried out in three sections (S1: 50 m downstream weir3, S2: 50 m upstream weir 4, S3: 30 m upstream weir3). In August, measurements were performed in the samedownstream and upstream sections. The measurements wereperformed with an electromagnetic device (Valeport). Resultsare shown in Table 7. Macrophytes were only removed after themeasurement. In April, the average dry weight of the biomass

was 63.04 g/m2, while in August, values of 306.11 g/m2 weremeasured.

Table 7 presents the hydraulic characteristics of the sections.As can be seen, not withstanding the fact that the wettedsection A is larger in August then in April, a smaller dischargeis calculated which is due to the amount of vegetation.Macrophytes obstruct the flow resulting in lower velocitiesand higher water levels. The first series of measurements wascarried out on April 14th, at the time indicated in the table.

During the campaign of August, field measurements werecarried out with different devices in the same cross-sections.Next to the propeller, also two different electromagnetic devices(type Valeport) are used. Results are mentioned in Table 8.Values between brackets are propeller measurements in non-cleaned sections, which include the influence of the vegetation.

Lower values are measured with propellers, for theelectromagnetic devices no differences were seen. So, thequality of the discharge measurements is not influenced by thepresence of vegetation on condition that the measurements arecarried out with an electromagnetic device.

This conclusion is confirmed by measurements performed inJune 2006 in the Biebrza river (Poland). During a measurement


Table 8Results of the discharge measurement (V’port = Valeport)

Q (m3/s) 22/08/05 23/08/05 24/08/05 25/08/05

Upstr. section propeller 0.72 (0.68) (0.69) –V’portLvH

– 0.80 0.78 0.84

V’port UA – – – 0.91

Downstr.section

propeller – 0.79 – –

V’portLvH

0.73/0.77 0.77 0.80 0.87

V’port UA – – 0.84 0.82

Table 9Results of the discharge measurement in Poland, the same cross-sections weresampled with and without vegetation

Q (m3/s) With vegetation Without vegetation

Section 1: Novy Rogozyn 0.179 0.181Section 2: Rogozynek 0.215 0.214Section 3: Rogozyn 0.225 0.208

campaign, three sections were selected and the discharge wasdetermined before and after removing vegetation. As can beseen (Table 9), the discharges measured with and withoutvegetation correspond very well.

7. Conclusions

According to the scope of the research, one instrumentis more useful than another. Measurements in the lab differfrom those in the field, due to the controlled conditions. Alsothe accuracy will be varying. Continuous registrations cangive information over a longer time period in contrast todiscontinuous ones. Next to surface water measurements, alsogeometrical characteristics and the presence of vegetation haveto be taken into account. HIC registers, in a rather continuousway, water levels and calculates discharges, however, thevelocity measurements performed by Ghent University areuseful in the integrated view which consider groundwaterinfluence and vegetation interaction.

These traditional techniques are not all easy to useto measure flow velocities through vegetation. Hydrometricpropellers can be fouled in vegetation. For acoustic devices,the vegetation interferes with the backscatter signal and anelectromagnetic velocity meter may have an impact on theenvironment due to its size.

The ADCP, based on the Doppler technology, allows afaster sampling. Furthermore, next to the measurement ofcharacteristics of the section and the velocity, an estimationof the discharge is carried out. Although this instrument is aninteresting tool, it is rather expensive compared to the propellersand electromagnetic devices.

Out of the continuous discharge measurements, it canbe seen that all devices are useful for measurements. Dueto the presence of vegetation, HIC prefers an estimateddischarge which can slightly differ from other values. Forlab measurements, all mentioned devices are useful due

to the rather ‘perfect’ conditions. Instruments to performmeasurements on the field are determined by the environmentalconditions such as the presence of vegetation (the propellers),the velocity in the river (range of the instrument), etc.

The quality and accuracy of a numerical model stronglydepend on the data available to build and to calibrate the model.Therefore, regular measurement campaigns are an importanttool.

According to the location and the circumstances, one mea-surement technique is more advisable than another. The use of acalibrated weir is possible when the circumstances allow a linkbetween the discharge and the water height over the weir, e.g.in lab conditions. In rivers, single point measurements are oftenused. The technique is rather time-consuming, but cheap com-pared to other advanced instruments. The use of an electromag-netic device for example, can solve the problems encounteredby the use of hydrometric propellers in vegetated rivers.

Acknowledgements

This research is funded by the FWO (Fund for ScientificResearch) - Flanders (G.0306.04). It is part of the multidis-ciplinary research project ‘A fundamental study on exchangeprocesses in river ecosystems’ (University of Antwerp, VrijeUniversiteit Brussel, Ghent University, 2004–2007). The over-all objective is to study the physical and biological exchangeprocesses in margins and inundation areas of water courses andhow their interactions determine the exchange of water, dis-solved compounds and particulate matter.

The authors thank Mr. Martin Van Daele and Mr. StefaanBliki for their assistance with the discharge measurements andSarah Seghers, for making available her research conducted atthe Hydraulics Lab. The authors also express their gratitudeto the Hydrological Information Centre of Flanders HydraulicsResearch (HIC, Mr. E. Cornet) for providing the discharge andwater level data of the river Aa.

References

[1] Bass J, Wharton G, Cotton J. Flow disturbance by instream macrophytes,a hot wire anemometry and scaffold system for measuring finescalevelocity patterns and mapping sedimentdeposits and habitat structures.Poster of the University of London. 2005.

[2] Chow VT, Maidment DR, Mays LW. Applied Hydrology. 1988.[3] Fenton J. The application of numerical methods and mathematics to

hydrography. In: Proc. 11th Australasian hydrographic conference. 2002.p. 16.

[4] Herschy R. The uncertainty in a current meter measurement. FlowMeasurement and Instrumentation 2002;13:281–4.

[5] Herschy R, et al. Hydrometry: Principles and practices. Wiley-Interscience; 1978.

[6] Herschy R, et al. Streamflow measurement. Rootledge; 1995.[7] HIC. Methodology for the discharge measurements. Methodologie voor

het uitvoeren van debietmetingen. [in Dutch] Technical report. FlandersHydraulics Research - Hydrological Information Centre. 2000.

[8] ISO. International organization for standardization. http://www.iso.ch.2006.

[9] Kouwen N. Using vegetation properties to predict flow resistance anderosion rates. In: Vegetated channels: Hydraulic, morphological andecological aspects. (International workshop on riparian forest). 2003.

[10] Thomas F. Open channel flow measurement using international standards:Introducing a standards programme and selecting a standard. FlowMeasurement and Instrumentation 2002;13:303–7.

http://www.iso.ch


[11] UGent. Ijking van de stuw op de aa te poederlee. Technical report.Ghent University. Report of the Hydraulics Lab (Department of CivilEngineering), (Van Poucke L.) in order of the VLM. 1995.

[12] UGent. Doorlichting methodologie voor het uitvoeren van debietmetin-gen. Technical report. Ghent University. Report of the Hydraulics Lab(Department of Civil Engineering) in order of the WLB. 2002.

[13] Voet M, Cabus P, Van Hoestenberghe T. Ultrasone snelheidssensoren bijhet meten van debieten in onbevaarbare waterlopen. Water, Nieuwsbrief.2004.

[14] Ward C, Tepper M. The use of doppler instruments for measuringvelocity and discharge in streams less than 1.5 m depth. Technical report.SonTek/YSE Inc.; 2004.

Flow Measurement and Instrumentation - vliz.be · Flow Measurement and Instrumentation 19 (2008) 29–40 Flow Measurement and Instrumentation ¬‚owmeasinst Accuracy of discharge

Documents