draft ni surface velocity guidelines ver 1 · 2020. 10. 27. · 3djh ri 1dwlrqdo ,qgxvwu\...

$: DRAFT NI Surface Velocity Guidelines Ver 1 · 2020. 10. 27. · 3djh ri 1dwlrqdo ,qgxvwu\ *xlgholqhv iru k\gurphwulf prqlwrulqj 7klv grfxphqw lv rqh sduw ri wkh 1dwlrqdo ,qgxvwu\$
124

WATER MONITORING STANDARDISATION TECHNICAL COMMITTEE

National Industry Guidelines for hydrometric monitoring

PART 11: APPLICATION OF SURFACE VELOCITY METHODS FOR VELOCITY AND OPEN CHANNEL DISCHARGE MEASUREMENTS. NI GL 100.11–2020

October 2020

National Industry Guidelines for hydrometric monitoring NI GL 100.??–2020

Page 2 of 48

Copyright

The National Industry Guidelines for hydrometric monitoring, Part 11 is copyright of the Commonwealth, except as noted below:

COPYRIGHT MATERIAL, TO BE COMPLETED

Creative Commons licence

With the exception of logos and the material from third parties referred to above, the National Industry Guidelines for hydrometric monitoring, Part 9 is licensed under a Creative Commons Attribution 3.0 Australia licence.

The terms and conditions of the licence are at: http://creativecommons.org/licenses/by/3.0/au/

To obtain the right to use any material that is not subject to the Creative Commons Attribution Australia licence, you must contact the relevant owner of the material.

Attribution for this publication should be: © Commonwealth of Australia (Bureau of Meteorology) 2020.

Page 3 of 48

Acknowledgements

This guideline was developed by a technical reference group managed by Mark Randall, Queensland Government, for the Water Monitoring Standardisation Technical Committee (WaMSTeC). WaMSTeC subcommittees conducted a draft review process and coordinated extensive industry consultation both domestically and internationally. 2020 WaMSTeC draft development and subcommittee review members: Mark Randall, Queensland Government, Department of Natural Resources, Mines and Energy (DNRME) (Guideline Sponsor) Mic Clayton, Snowy Hydro, NSW Daniel Wagenaar, Xylem Jacqui Bellhouse, Water Corporation, WA, (WaMSTeC Chairperson) Michael Whiting, Department of Water and Environmental Regulation, WA Greg Carson, HydroTasmania. Linton Johnston, Australian Bureau of Meteorology Special thanks for those providing technical contributions and peer review:

Professor Ichiro Fujita, Ph.D, Professor Emeritus, Kobe University, Construction Engineering Research Institute

Ray Maynard, Senior Hydrographer, Queensland Government, DNRME

(Note that at the time of contribution, individuals may have been employed with different organisations and some organisations were known by other names).

Page 4 of 48

Foreword

This guideline is part of a series of eleven National Industry Guidelines for hydrometric monitoring. It has been developed in the context of the Bureau of Meteorology's role under the Water Act 2007 (Cwlth) to enhance understanding of Australia’s water resources.

The Bureau of Meteorology published the first ten of these guidelines in 2013 as part of a collaborative effort amongst hydrometric monitoring practitioners to establish standardised practice. They cover activities relating to surface water level, discharge and water quality monitoring, groundwater level and water quality monitoring and rainfall monitoring. They contain high level guidance and targets and present non-mandatory Australian industry recommended practice.

The initial versions of these guidelines were endorsed by the Water Information Standards Business Forum (the Forum), a nationally representative committee coordinating and fostering water information standardisation. In 2014, the functions and activities of the Forum transitioned to the Water Monitoring Standardisation Technical Committee (WaMSTeC).

In 2017, as part of the ongoing governance of the guidelines, WaMSTeC initiated a 5-yearly review process to ensure the guidelines remain fit-for-purpose. They now include additional guidance for groundwater monitoring, and other updates which improve the guidelines' currency and relevance. WaMSTeC endorsed these revised guidelines in December 2018.

Guideline 11 has been developed to operationalise the use of emerging technology and methodologies associated with surface velocity discharge measurements. Surface velocity measurements achieved through video image analysis are becoming increasingly popular driven by the use of drone technology. Surface radar systems again provide a popular non contact method of discharge measurements.

Guideline 11 fills a procedural void that exists both nationally and internationally.

Industry consultation has been a strong theme throughout development and review of the guidelines. The process has been sponsored by industry leaders and has featured active involvement and support from the Australian Hydrographers Association, which is considered the peak industry representative body in hydrometric monitoring.

These guidelines should be used by all organisations involved in the collection, analysis and reporting of hydrometric information. The application of these guidelines to the development and maintenance of hydrometric programs should help organisations mitigate program under-performance and reduce their exposure to risk.

Organisations that implement these guidelines will need to maintain work practices and procedures that align with guideline requirements. Within the guidelines, the term “shall” indicates a requirement that must be met, and the term “should” indicates a recommendation.

The National Industry Guidelines can be considered living documents. They will continue to be subject to periodic WaMSTeC review at intervals of no greater than five years. In the review phase, WaMSTeC will consider any issues or requests for changes raised by the industry. Ongoing reviews will ensure the guidelines remain technically sound and up to date with technological advancements.

Page 5 of 48


This document is one part of the National Industry Guidelines for hydrometric monitoring series, which can be found at http://www.bom.gov.au/water/standards/niGuidelinesHyd.shtml.

The series contains the following parts:

Part 0: Glossary

Part 1: Primary Measured Data

Part 2: Site Establishment and Operations

Part 3: Instrument and Measurement Systems Management

Part 4: Gauging (stationary velocity-area method)

Part 5: Data Editing, Estimation and Management

Part 6: Stream Discharge Relationship Development and Maintenance

Part 7: Training

Part 8: Application of Acoustic Doppler Current Profilers to Measure Discharge in Open Channels

Part 9: Application of in-situ Point Acoustic Doppler Velocity Meters for Determining Velocity in Open Channels

Part 10: Application of Point Acoustic Doppler Velocity Meters for Determining Discharge in Open Channels

Part 11: Application of Surface Velocity Methods for Velocity and Open Channel Discharge Measurements (this guideline)

Page 6 of 48

Table of Contents

1 Scope and general ........................................................................................................ 7 1.1 Purpose................................................................................................................ 7 1.2 Scope ................................................................................................................... 7 1.3 References ........................................................................................................... 8 1.4 Bibliography ......................................................................................................... 9 1.5 Terms and definitions ........................................................................................... 9

2 Surface Velocity Methodologies .................................................................................... 9 2.1 General description .............................................................................................. 9 2.2 Advantages and disadvantages of using surface velocity measurements........... 10 2.3 Surface velocity measurement methods ............................................................. 10 2.4 Important considerations for using surface velocity methods .............................. 11

3 Image Velocimetry....................................................................................................... 11 3.1 Large Scale Particle Image Velocimetry (LSPIV) ................................................ 12 3.2 Space Time Image Velocimetry (STIV) .............................................................. 15 3.3 Error sources in LSPIV and STIV processing methods ....................................... 18 3.4 Image Orthorectification ..................................................................................... 20 3.5 Surface Velocity Site Selection and installation criteria ....................................... 25 3.6 Fixed camera setup and data collection procedures ........................................... 27 3.7 Image Velocimetry Using Remotely Piloted Aircraft ............................................ 29

4 Surface Velocity Radar ................................................................................................ 30 4.1 SVR techniques ................................................................................................. 31 4.2 Site selection and installation methods ............................................................... 33 4.3 SVR instrument management ............................................................................ 37

5 Surface velocity coefficients ........................................................................................ 38 5.1 Vertical velocity profiles for calculating surface coefficients ................................ 38 5.2 Site specific velocity profile analysis ................................................................... 41 5.3 Typical surface coefficient values ....................................................................... 42

6 Uncertainties in Discharge Measurements .................................................................. 43 6.1 Description of measurement uncertainty ............................................................ 43

Appendix A Critical conditions for image velocimetry measurements ................................... 45 A.1 Angle of repose θ .................................................................................................. 45 A.2 Image resolution ................................................................................................... 45

Appendix B Training ............................................................................................................. 47

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Scope and general

Page 7 of 48


Part 11: Application of Surface Velocity Methods for Velocity and Open Channel

Discharge Measurements.

1 Scope and general

1.1 Purpose

The purpose of this document is to provide guidelines for recommended practice to ensure that the collected measured streamflow data are:

a) accurate; b) defendable; and c) consistent across water monitoring organisations operating under these

guidelines.

This is the minimum guideline that shall be followed to allow the collected data to withstand independent validation and integrity checks. Additional field procedures may vary between organisations and States.

1.2 Scope

This document deals with the application of surface velocity methods for determining velocity and discharge in open channels. It specifies the required procedures and methods for collecting data by Australian operators. It specifies procedures for the collection and processing of surface water velocity data collected by image velocimetry and radar systems.

This document does not include or rewrite instrument manufacturers’ operating instructions for their individual instruments. Nor does it detail Standard Operating Procedures (SOPs) of organisations using these instruments. However, it is expected that those SOPs are sufficiently robust to withstand independent scrutiny and alignment with this document.

This document contains images and examples sourced from instrument manufacturers or suppliers. Inclusion of these images, with reference to the source, is solely for the purpose of providing examples, additional information and context, and is not to be interpreted as endorsement of any particular proprietary products or services.

This guideline is an original attempt to operationalise new methodologies in image velocimetry. To capture advancements in these evolving techniques this guideline is considered a living document. Feedback can be supplied to [email protected].

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Scope and general

Page 8 of 48

1.3 References

This document makes reference to the following documents:

Bazin, H. (1865). Recherches Expérimentales sur l'Ecoulement de l'Eau dans les Canaux Dé-couverts. Mémoires présentés par divers savants à l'Académie des Sciences, Paris, Vol. 19, 1-494

Bradley, A. A., A. Kruger, E. A. Meselhe, and M. V. I. Muste, Flow measurement in streams using video imagery, Water Resour. Res., 38(12), 1315, doi:10.1029/2002WR001317, 2002.

Chiu, C.-L., and Tung, N.C., 2002. Velocity and regularities in open-channel flow. Journal of Hydraulic Engineering, 128 (94), 390-398.

Chiu, C.-L., Tung, N.C., Hsu, S.M., and Fulton, J.W., 2001, Comparison and assessment of methods of measuring discharge in rivers and streams, Research Report No. CEEWR-4, Dept. of Civil & Environmental Engineering, University of Pittsburgh, Pittsburgh, PA.

Chiu, C.-L., 1989, Velocity distribution in open channel flow, Journal of Hydraulic Engineering, 115 (5), 576-594

Fujita I, Muste M, Kruger A (1998) Large-scale particle image velocimetry for flow analysis in hydraulic engineering applications. J Hydraul Res 36(3):397–414

Fujita Ichiro, Principles of Surface Velocity gaugings ,Presentation, The 4th IAHR-WMO-IAHS Training Course on Stream Gaugings, 2018

Fujita I, Watanabe H, Tsubaki R (2007) Development of a non-intrusive and efficient flow monitoring technique: the space time image velocimetry (STIV). Int J River Basin Man 5(2):105–114

Fulton, J.W. and Ostrowski, J., 2008, Measuring real-time streamflow using emerging technologies: Radar, hydroacoustics, and the probability concept, Journal of Hydrology 357, 1–10.

Hauet, Alexandre & Morlot, Thomas & Daubagnan, Léa. (2018). Velocity profile and depth-averaged to surface velocity in natural streams: A review over a large sample of rivers. E3S Web of Conferences. 40. 06015. 10.1051/e3sconf/20184006015.

ISO 748:2007 Hydrometry – Measurement of liquid flow in open channels using current meters or floats.

ISO 15769:2010 Hydrometry — Guidelines for the application of acoustic velocity meters using the Doppler and echo correlation methods

Keulegan, G. H. (1938). Laws of Turbulent Flow in Open Channels. J. Research, National Bu-reau of Standards, Vol. 21, 707-741.

NI GL 100.02–2019 Part 2: Site Establishment and Operations

NI GL100.09-2019 Part 9: Application of in-situ Point Acoustic Doppler Velocity Meters for Determining Velocity in Open Channels for application of velocity indexing processes

Smart, G. & Biggs, H. (2020). Remote gauging of open channel flow: Estimation of depth averaged velocity from surface velocity and turbulence. Proceedings of River Flow 2020, Delft, Netherlands.

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Surface Velocity Methodologies

Page 9 of 48

Tauro, F., Piscopia, R., & Grimaldi, S. (2017). Streamflow observations from cameras: large-scale particle image velocimetry or particle tracking velocimetry? Water Resources Research, 53, 10,374–10,394. https://doi.org/10.1002/2017WR020848

1.4 Bibliography

Cognisance of the following was taken in the preparation of this guideline:

Fudaa-LSPIV User Manual (Ver. 1.7.3)

Hydro-STIV Operation Manual (Ver 1.0)

1.5 Terms and definitions

For the purpose of this document, the definitions given in National Industry Guidelines for hydrometric monitoring, Part 0: Glossary, NI GL 100.00–2019 apply. The following key terms relate to implementation of these guidelines:

Shall indicates a mandatory requirement

Should indicates a recommendation, and

May indicates an allowable option.

2 Surface Velocity Methodologies

The use of surface velocities to calculate discharge is not new—the method has been used for many hundreds of years. However, advances in technology now allow this method to be applied in a more precise and accurate manner.

2.1 General description

Surface velocity methods are based on measuring the water velocity at the surface, to calculate a stream discharge value. This process requires the tracking (timing) of surface structures, floats or natural tracers as they travel over a known distance. This allows a measured velocity to be calculated that is assumed to be representative of the actual water surface velocity at that measurement point. Surface velocity methods employ the following process to derive discharge:

Surface velocities are measured at a single point or multiple points across the channel width.

Measured surface velocities are corrected to a mean velocity value for the full flow depth. This is achieved through applying coefficients generally known as the ‘surface alpha’.

Mean velocities are applied to the channel cross sectional area to calculate a discharge value.

Multiple mean velocities can be applied to a cross sectional area (in the same manner as a wading measurement) and discharge is calculated by either the mid or mean section method.

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Surface Velocity Methodologies

Page 10 of 48

2.2 Advantages and disadvantages of using surface velocity measurements

The resurgence of surface velocity methods is based on the practical advantages this method can provide over other discharge measuring techniques. However there are challenges, of which the potential user needs to be aware.

2.2.1 Advantages

Improved staff safety for high flow measurements.

Non-intrusive measurement technique.

Can be used to make velocity and/or discharge measurements.

No depth restrictions for velocity analysis.

Potential for remote applications where staff are not required on site during the measurement.

Inexpensive measuring equipment.

Suited to high flow and high debris flood environments which present difficulties for ADCP measurements.

Comparable accuracy to currently accepted discharge measurement techniques, when applied correctly.

Large volume of citizen science flood videos from social media sources that can be used.

Provides a visual record of flow events.

2.2.2 Disadvantages

Surface tracers used to calculate velocity must be advected at exactly the surface velocity.

Wind can potentially bias surface tracer movement, so they are not exactly advected with the flow.

Surface alpha values can be site and stage level specific.

As a velocity area method, changes in cross sectional area need to be accounted for.

A lack of surface tracers, or the inability to track surface tracers will compromise velocity data determination.

2.3 Surface velocity measurement methods

This guideline concentrates on the two dominant methods currently being applied to calculate surface velocities and open channel discharge. These methods are:

Image velocimetry – Large Scale Particle Image Velocimetry (LSPIV), and Space Time Image Velocimetry (STIV)

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Image Velocimetry

Page 11 of 48

Surface velocity radar – Fixed and mobile solutions.

Other similar methods exist, such as float techniques and particle/optical flow tracking. These techniques are not covered by name in this guideline, although many of the principles and criteria included here apply equally to them.

2.4 Important considerations for using surface velocity methods

Surface velocity methods are not suitable for all measurement sites and it is important that practitioners applying these methodologies understand where these methods will not provide accurate velocity/discharge data.

Important site conditions to avoid include:

strong prevailing wind at the water surface

no visible surface tracers, textures, or ripples that are exactly advected with flow

tidally influenced flow, backwater effects, or any other site conditions that prevent accurate quantification of a surface alpha coefficient (relates to discharge calculations only)

very clear water with no visible tracers or turbulence.

Other important considerations are:

Sites suitable for current meters or ADCPs might not necessarily be suitable for image velocimetry measurements.

Radar systems require a secure mounting above the water surface (e.g. bridge, cableway).

Fixed camera sites need a secure camera mounting with a good view of the channel width.

Measurements undertaken using a drone require the drone to hover in a stable position during data collection.

3 Image Velocimetry

Image velocimetry is an increasingly popular surface velocity measurement technique that uses a recorded video of the water surface tracers to calculate surface velocities. These velocities can be calculated at multiple points across a channel. Image velocimetry relies on the accurate scaling of the pixels to metres, in relation to the camera sensor, channel width, and water level. This is achieved via an orthorectification process (described in detail at Section 4).

Once the pixels are calibrated, surface tracers can be tracked over a precisely known distance (via the pixels). The duration of the video provides the time reference required to calculate velocity.

There are currently two distinct methods or algorithms available for image velocimetry processing:


Page 12 of 48

Large Scale Particle Image Velocimetry (LSPIV), which utilises a cross correlation algorithm, and

Space Time Image Velocimetry (STIV), which uses a gradient tensor analysis.

Both algorithms were developed by Professor Ichiro Fujita, formerly of Kobe University, in Japan. LSPIV was developed in 1997 and STIV in 2004.

3.1 Large Scale Particle Image Velocimetry (LSPIV)

Since LSPIV was first developed (Fujita, et al. 1998) the methodology has become known by other abbreviations, depending on the software application. However the core processing algorithm to derive velocities is always a cross correlation analysis of surface tracer movement.

For the purpose of this guideline the generic term of LSPIV is used.

LSPIV is currently the only method available to process surface velocities in 2D (i.e., direction and magnitude).

The guidance provided in this document shall apply to all image velocimetry processing techniques that use a cross correlation method of surface tracer analysis to calculate surface velocities, and ultimately stream discharge.

3.1.1 LSPIV analysis

All programs using cross correlation analyses require that the video frames be separated into individual images, therefore producing a series of time-lapse images. This is due to computational demands of the cross correlation analysis.

A video recorded at 30 FPS (frames per second) can therefore produce 30 separate images for each second of video available for extraction from the original movie.

The number and time interval of images/frames extracted from the video shall allow the movement of tracers to be clearly visualised within the interrogation and search area sizes selected by the user.

Each extracted image/frame represents a time interval used to calculate the velocity.

The number of extracted images per second used can impact the accuracy of the velocity analysis. The user should aim to use as many images as possible for analysis and shall not use less than 15 images per second (15 FPS).

The number of images selected can impact on the computational time required to undertake the cross correlation analysis and discharge calculation.

3.1.1.1 Interrogation area

The Interrogation Area (IA) is a small area set in the images, where the analysis software will identify tracers and particle pattern displacement, which are cross correlated between sequential extracted images.


Page 13 of 48

Multiple interrogation areas are established across the Region Of Interest (ROI), forming a grid where the velocity analysis is required.

The IA shall be large enough to contain tracer/particle movement, but small enough to be representative of the flow.

Two correlation methods commonly used are Direct Cross-Correlation (DCC) and Fast Fourier Transform (FFT).

3.1.1.2 Search area

The Search Area (SA) is a user-defined area contained within the IA. The SA is the area where tracer displacement between the IA of sequential images is identified.

The size of the SA shall be selected in relation to the magnitude of the tracer displacement between sequential images. Therefore the velocity magnitude and tracer quality are determining factors when setting the size of the SA.

Figure 1. Interrogation and Search Area settings in relation to image frames and particle displacement tracking. The linear correlation coefficient is computed between the pixels from the Interrogation Area and the corresponding pixels in the Search Area. This is repeated across the entire Region Of Interest between all image pairs. (Source: Tauro et al 2017).

3.1.1.3 Post processing

Cross correlation analysis of tracer displacement requires post processing filters to be applied, to minimise the impacts of erroneous data. Filters may include:

Correlation – strong correlations can be a result of fixed objects such as channel banks, rocks, and trees. A maximum correlation threshold is set to remove these returns from the velocity analysis. Minimum correlation thresholds shall be set in accordance with the tracer material. Soft tracers such as foam, and turbulence will require a lower minimum threshold value than harder tracers such as logs and added artificial tracers.


Page 14 of 48

Velocity – maximum and minimum velocity thresholds can (software dependant) be set in both the x and y directions.

Filter thresholds applied shall be set in accordance with the site, velocity and tracer conditions. Different flow events may require different thresholds.

The strength of the filters used shall be documented with the measurement details.

Following the application of filters the initial results are averaged over all image frames to provide an average surface velocity and direction for each interrogation area.

LSPIV analyis is unable to differentiate between ripples or turbulence representative of surface velocities and wave noise that is not representative of surface velocities. A measurement video containing few physical tracers passing through the analysis location can cause analysis bias—usually a negative bias in velocity and discharge. These errors cannot be easily identified or corrected for in post processing procedures. LSPIV software containing algorithms or deep learning to mitigate this bias should be used at site locations where these conditions are unavoidable.

3.1.2 Discharge Calculation

Discharge is calculated via the velocity area method using either the mid or mean section method.

The number of verticals (velocity points) shall be set in accordance with National Industry Guidelines for hydrometric monitoring, Part 4: Gauging (stationary velocity-area method) NI GL 100.04-2019 (refer to section 6.2.2).

Channel width < 0.5 m n = 5 to 6

Channel width > 0.5 m and < 1 m n = 6 to 7

Channel width > 1 m and < 3 m n = 7 to 12

Channel width > 3 m and < 5 m n = 13 to 16

Channel width > 5 m n ≥ 22

Each vertical/velocity point shall contain no more than 10% of the total discharge.

Due to LSPIV’s gridded nature of the velocity distribution, surface velocities used in the discharge calculation are dependent on the positioning of the site cross section within the velocity grid of the ROI. Different placements of the cross section within the grid can therefore result in different discharge results, particularly during unsteady flow conditions.

Once surface velocities have been calculated the user shall identify the nature of the surface velocities and apply one of the following requirements when calculating discharge:

1. If surface velocity magnitudes are consistent within the ROI as flow moves downstream, only a single cross section shall be used to calculate discharge.


Page 15 of 48

2. If surface velocity magnitudes have significant variance within the ROI as flow moves downstream. A minimum of two cross sections shall be used. The cross sections shall capture the velocity variance within the ROI. Final discharge shall be an average discharge value from all cross sections (see Figure 2).

Figure 2. LSPIV analysis showing velocity magnitude variance within the Region of Interest. Three cross section locations have been used to capture that variance and provide an average discharge result. (Source: Jerome Le Coz, IRSTEA, FUDAA software)

3.2 Space Time Image Velocimetry (STIV)

STIV is an alternative method of surface image analysis which does not use the cross correlation method to quantify surface tracer movement, instead using a gradient tensor method as first described by Fujita, et al. (2007).

STIV only produces 1D velocities in the direction of flow.

STIV utilises a complete video for analysis, and there are no user specified frame extraction intervals, interrogation/search area requirements or post processing filters. STIV analysis shall still be checked for errors, and corrections applied if necessary.

3.2.1 STIV Velocity Measurement

STIV uses specified analysis Search Lines in the direction of flow, set by the user. Each line is of a known length and, together with the known time duration of the video, provides a basis for an average surface velocity to be calculated.


Page 16 of 48

Each pixel along the known length of each Search Line (also one pixel wide) is stacked for every frame of the video. As tracers/surface structures travel along each pixel of the search line they naturally create a visible line angled down from top left to bottom right of the space time image.

A coherency analysis of the Space Time Image (STI) identifies the mean angle and therefore the mean surface velocity for each Search Line.

Figure 3. STIV analysis using the gradient tensor method. All of the pixels along a Search Line are stacked below each other for every frame. Tracer movement along the Search Line produces an angled line pattern which represents the surface velocity.

(Source: https://hydrosoken.co.jp/en/service/hydrostiv.php viewed 20/06/2020)

The latest STIV algorithms contain a deep learning technique developed for detecting the most probable orientation angle appearing in the space time image (STI). This allows only the texture orientation (related to turbulence advection) to be extracted, while discarding the effects of gravity waves and environmental noise which would otherwise cause bias in the velocity analysis.

3.2.1.1 Post Processing

STIV analysis should be manually checked in post processing using the space time imagery or spectrum analysis to ensure differentiation between ripples or turbulence representative of surface velocities and wave noise that is not representative of surface velocities. A measurement video containing few physical tracers passing through the analysis location can cause analysis bias—usually a negative bias in velocity and discharge. Biasing can also occur from static objects or wave patterns that provide a stronger but false signal during the coherency analysis.

STIV post processing does allow these errors to be identified and manually corrected without reprocessing the whole data set. STIV software containing algorithms or deep learning to mitigate this bias should be used at site locations where these conditions are unavoidable.


Page 17 of 48

STIV software allows users to manually define or correct the automated processed angle from the space time image (STI) without having to reprocess the video data. Biasing can occur from static objects or wave patterns that provide a stronger but false signal during the coherency analysis.

Figure 4. Space Time Image showing bias created by gravity waves (yellow lines) and manual corrections (red lines) for turbulence tracers which represent the actual surface velocity. (Source: Mark Randall).

When post processing a STIV measurement the user shall check:

1. All Space Time Images generated in a measurement should be examined to ensure that the mean orientation angle has been correctly identified during the automated processing.

2. All identified errors shall be manually corrected. When correcting velocity data the user needs to consider the velocity distribution across the cross section. Velocity changes between Search Lines shall look reasonable and any possible anomalies explained (e.g. obstruction upstream causing reduced velocity).

3. If a visual check of the Space Time Image is not possible, then the spectrum analysis shall be used to identify peak intensity.

4. All calculated velocities shall appear reasonable, based on the video image playback.

5. Manually checking the data allows the accuracy of the measurement to be confirmed as part of a QA/QC process.

3.2.2 Discharge Calculation

Refer to section 3.1.2.

STIV analysis calculates an average velocity over the distance covered by the search line. It s not a single point measurement as produced by LSPIV.


Page 18 of 48

Only one cross section placement is therefore required within the ROI. The discharge value will be the same for any placement along the length of the velocity search line

3.2.2.1 Length of STIV velocity analysis search lines

STIV analysis calculates an average velocity over the distance covered by the search line allowing the effects of velocity pulses, and turbulence to be averaged out.

Search line length should be set to allow any visual tracers to travel the full length of the search line during the exposure time of the video.

Length and positioning should be set to avoid areas with excessive standing waves, and non moving structures which will increase the requirement of manual velocity corrections.

3.3 Error sources in LSPIV and STIV processing methods

The choice of processing method can influence the quality of the results obtained, due to the differences in the processing methods and environmental conditions. LSPIV analysis can produce varying results depending on the parameter choices of the user. It is important that users new to LSPIV are aware of the impacts of these choices.

Figure 5. Difference between LSPIV and STIV processing techniques. (Source: Professor Ichiro Fujita, Principles of Surface Velocity gaugings, Presentation, The 4th IAHR-WMO-IAHS Training Course on Stream Gaugings, 2018)

LSPIV analysis requires good correlations of the tracer particles used to calculate velocity, and can therefore have problems if tracer particles are difficult to determine, either by their absence, environmental conditions or inappropriate camera angle.


Page 19 of 48

Figure 6. LSPIV velocity bias caused by stationary objects. The top image shows the Region of Interest and cross section location where the tagline has been used to collect acoustic Doppler data. The bottom image shows how velocities have been biased low by the stationary tagline stretching across the channel . (Source: Mark Randall, RIVeR software)

LSPIV analysis requires a minimum camera angle of 15 degrees below the horizontal from the camera location to the furthest analysis point. Angles below 15 degrees will compromise the accuracy of the cross correlation analysis. STIV requires a minimum of 2 degrees.

STIV analysis is not subject to the same potential errors introduced into the analysis by the user’s choice of processing and post processing parameters and filters. These choices are not required for STIV analysis making it a simpler process to follow. STIV can still be susceptible to velocity bias caused from tracer, object, and environmental factors described for LSPIV, but generally is a more robust method when environmental conditions are not ideal. Unlike LSPIV, STIV analysis can be manually corrected by the user without repeating the analysis process, by visually identifying the correct tracer angle from the interference (Figure 7).

Figure 7. STIV velocity bias caused by sun glare on the water gives a more dominant signal than the tracers representing the actual velocities, biasing incorrectly the automated analysis to 0 m/s. STIV allows manual correction of velocity errors. In the image above the user has defined three velocities of 1.96, 2.02, and 1.9 m/s from the visible tracer patterns. Mean surface velocity for this location therefore equals 1.96 m/s. (Source: Mark Randall)


Page 20 of 48

3.4 Image Orthorectification

Image orthorectification refers to the calibration process required to determine the ground size of an image pixel in metres. Orthorectification is done using the pin-hole camera model, and is required because the actual ground size of each pixel is not equal when a camera is pointing at an oblique/tilted view point (see Figure 8).

The orthorectification process therefore determines each pixel size in relation to the angle and distance from the camera depending on the layout of the channel and water level contained in the image.

The correct pixel size is required because this represents the ‘known distance’ used in the velocity calculation.

There are three methods of image rectification used in image velocimetry:

3D orthorectification (perspective projection conversion or transformation)

2D orthorectification (homography transformation)

Scaling

The choice of orthorectification is determined by the relationship of the camera sensor angle to the water level plane (see Figure 8).

A camera sensor looking vertically (nadir) to the water plane only requires a scaling orthorectification whereas a camera mounted at an oblique angle requires a 3D or 2D orthorectification due to the distortion of the pixel shape as distance from the sensor increases.

A 3D or 2D orthorectification requires ground control points (GCP) with known coordinate data in XYZ or XY format.

Figure 8. Shows how camera angle creates variation in the size of the ground footprint of pixels. The left image shows a vertically positioned camera from a drone, all pixels have the same size footprint on a flat surface. The right image shows when the camera is tilted the pixels represent a different size footprint on the ground. Orthorectification calibrates those footprints correctly. (Source: https://www.esri.com/arcgis-blog/products/arcgis-pro/imagery/ortho-mapping-workspace/ viewed 20/05/2020)


Page 21 of 48

Figure 9. An orthorectified image (Left) based on the real world coordinates of each ground control point and the original camera image (Right). (Source: Mark Randall, HydroSTIV)

Orthorectification requires a number of GCP located within the Field Of View (FOV). Each GCP requires real world coordinates specific to its location and in relation to the position of the camera.

GCP survey data should be checked for relative accuracy before attempting orthorectification. Survey errors will complicate the orthorectification process. Large errors will cause the process to fail.

Once installed, a GCP calibration is only valid for that camera mounting position, image resolution, zoom, and focal length. A new calibration shall be developed in the processing software to rectify any changes. If not, velocity data will not be valid for discharge calculations.

GCPs shall be visible from the camera mounting position at the time of installation.

Whilst it is not a requirement that the camera be mounted perpendicular to the flow, this will allow a simpler orthorectification process.

Depending on the orthorectification process and software used it may be necessary to use known information about the camera intrinsic and extrinsic parameters to improve the accuracy of the orthorectification. This information can include:

Camera mounting position and elevation (XYZ)

Focal length

Pitch, Roll, and Yaw.

Orthorectification does not necessarily correct radial distortion produced by the camera lens. Radial distortion typically produced by wide angle and fish eye lenses will cause systematic errors to be reported during orthorectification. These errors should be corrected via a lens calibration process before orthorectification.

Lenses that cause significant radial distortion shall not be used for image velocimetry.


Page 22 of 48

3.4.1 3D Orthorectification

A 3D orthorectification uses real world coordinates in the XYZ plane with Z representing elevation.

A 3D orthorectification shall be undertaken if a camera system is mounted and used to collect videos at multiple changing water levels.

GCP points shall cover the range of water levels expected to be recorded and used for image velocimetry.

A minimum of 8 GCP shall be used and located throughout the FOV in a zigzag manner, both horizontally and vertically.

A perspective projection conversion should be used whenever possible. If a perspective projection transformation is used, unquantifiable errors may be present in the orthorectification.

3.4.1.1 Perspective projection conversion

Requires cameras intrinsic and extrinsic parameters.

Little bias in correction errors over entire image.

Most accurate method.

3.4.1.2 Perspective projection transformation

Transformation matrix obtained from GCP data only.

Camera intrinsic and extrinsic parameters not used.

Conversion error at each GCP location good but accuracy may decrease for elsewhere in the image away from GCP. These errors are unquantified.

3.4.2 2D Orthorectification

A 2D orthorectification (homography conversion) uses real world coordinates in the XY plane. The XY coordinates in the real world plane can be projected directly onto the plane of the camera sensor i.e., there is a homography between the two planes.

A 2D orthorectification can only be used to measure flow at a single water level because it does not use surveyed elevation data (Z) but assumes that Z is equal to the water plane. This means that changing water levels are not accounted for in a single 2D orthorectification, hence the method is best suited for single, opportune flow measurements rather than a more permanent installation.

If employing 2D orthorectification:

only four GCP are required, and shall not be aligned,

GCPs shall be set on the same horizontal plane as the water level, and

the technique should only be used if it is not possible to obtain eight or more GCP.


Page 23 of 48

Figure 10 A 2D orthorectification process. In this LSPIV software the distances between each yellow GCP are entered. The six coloured lines represent the six distances required. All four yellow GCP are at water level.( Source: USGS RIVeR software)

3.4.3 Scaling

Scaling can be used when a camera is mounted vertically to the water surface so that the camera sensor and water surface are parallel with each other. This is typically the case with drone collected videos. To scale the pixels directly, a known distance between two points or close to, water level shall be visible within the video.

Scaling pixels in this manner reduces potential errors introduced by 3D and 2D orthorectification.

3.4.4 Orthorectification errors

Errors in the orthorectification of the video imagery shall be quantified as these will translate into errors with the pixel scaling used in the velocity calculation, and ultimately the discharge calculation.

Figure 11. Ground control points used in an orthorectification. Orthorectification errors relate to the software’s ability to map the pixel coordinates to the real world co-ordinates. In this example the largest error is 0.298m. (Source: Mark Randall)


Page 24 of 48

Orthorectification errors ideally should be less than ± 1 metre to minimise errors in the pixel scaling and therefore calculated velocities. The orthorectification errors represent the software’s ability to map the pixel coordinates to the real world coordinates of the GCPs (reprojection errors). The errors are not an evaluation of the accuracy of the survey data. Any survey data errors would adversely affect the accuracy of the orthorectifications.

Orthorectification errors shall be quantified for each GCP. Any GCP with errors larger than ±1 metre should be removed from the orthorectification process, to prevent errors in the scaled pixel values (see Figures 11, 12 and 13).

Figure 12. GCP orthorectification results showing errors larger than ± 1m for GCP #2 and #4. (Source: Mark Randall, HydroSTIV)

Figure 13 (below) highlights the importance of identifying and eliminating GCP errors in the orthorectification process. The left image has GCP errors from figure 11 excluded from the final orthorectification pixel scaling equals 0.033 m/pixel and discharge of 38.951 cumecs. The right image has the GCP errors included and the pixel scaling is now 0.058 m/pixel. Although the velocity distribution is identical between each analysis, due to the larger pixel size the velocities are now faster, increasing the discharge result to 40.714 cumecs. The 2.5 cm difference in scaling equates to 4.5% difference in calculated discharges. The reference ADCP measurement conducted at the same time was 38.578 cumecs.


Page 25 of 48

Figure 13. The impacts of the orthorectification errors on pixel scaling, velocity and discharge. The left image has GCP errors from figure 11 excluded from the final orthorectification. The right image has included the GCP errors which are transferred to the velocity analysis. (Source: Mark Randall, HydroSTIV)

3.5 Surface Velocity Site Selection and installation criteria

The site selection criteria as stipulated in National Industry Guidelines for hydrometric monitoring, Part 2: Site Establishment and Operations, NI GL 100.02–2019 (Section 4.2.1.1 and 4.2.1.1.B: Discharge monitoring sites) shall equally apply to the selection of all surface velocity sites.

Cross section management should be conducted in the same manner as horizontal acoustic Doppler current profilers (National Industry Guidelines for hydrometric monitoring, Part 9: Application of in-situ Point Acoustic Doppler Velocity Meters for Determining Velocity in Open Channels for application of velocity indexing processes, NI GL 100.09–2019).

Inability to comply with these guidelines may result in the collection of lesser quality velocity data and subsequent stage/discharge relationships. The user shall take additional measures to validate the surface velocity discharge measurements with independent check measurements.

3.5.1 Site selection criteria for installing fixed cameras

A fixed installation of a camera at a measurement site allows videos to be recorded and stored for discharge processing, eliminating the need for staff to be onsite. This provides a remote area monitoring solution along with the ability to undertake multiple discharge measurements at multiple sites during widespread flood events.

Onsite cameras also provide a useful visual record of flow events.

Depending on the desired target flow to be measured the following considerations shall be applied when installing a fixed camera for image velocimetry. High flow measurements may eliminate the influence of some installation factors whereas low flow sites may require more care and stricter adherence to the considerations.


Page 26 of 48

1. A suitable site shall be selected such that visible surface movement that is advected with flow occurs for the range of flows to be measured.

2. Sites where excessive shadows and sun glare are present will hide the movement of surface tracers in the video and should be avoided. This is of particular importance at low flow sites that may not contain strong tracers.

3. Very clear water which allows the river bed to be visualised easily will require manual processing if tracer materials aren’t present.

4. The camera shall have a clear Field Of View (FOV) of the water’s surface. Some minor obstructions can be accommodated in the FOV, but the user shall be aware that these could impact LSPIV and any automated analysis.

5. The ROI shall be visible to the camera through the full expected stage range at the site. Points of flow disturbance in either an upstream or downstream location should also be avoided for automated analysis.

6. The camera shall be fastened to a secure structure that will eliminate camera movement by wind/vibrations. Camera shake prevents accurate determination of tracer movement.

7. Night measurements shall require a suitable infrared or far-infrared camera.

8. Camera mounting elevation shall be at a height that allows the critical angle between the camera and the water plane to remain above the minimum specified angle. This angle is dependent on the channel width and water level (Figure 15)

For LSPIV installations, the minimum angle at the furthest analysis point of the camera shall be 15 degrees.

For STIV applications, the minimum angle shall be 2 degrees.

Angles below these will cause velocity bias and/or tracer resolving issues for those analysis methods.

Figure 15. A fixed camera installation for STIV showing minimum critical angle measurement θ. The distance to the furthest analysis point is represented by L and the camera height above the water level plane equals H. As water level changes so will the value of θ. This value should not go below the minimum critical value depending on LSPIV or STIV analysis. (Source: Appendix 1, Professor Ichiro Fujita, email correspondence July 2020)


Page 27 of 48

9. Sites that can be impacted by strong prevailing winds should be avoided if the wind visibly impacts surface movement. Targeted flow regiemes sould be taken into consideration.

3.6 Fixed camera setup and data collection procedures

The following criteria shall apply to the camera setup and settings when establishing a fixed camera site.

Figure 16. A fixed camera installation for STIV showing the calculation of pixel size in cm’s at the furthest velocity analysis point from the camera. L is the distance from the camera to the furthest velocity measurement point/search line. Ø represents the angle either side of the central point of the Region of Interest or STIV search line. W is the length of the Region of Interest or search line set in the direction of flow. Χ is the equivalent measurement of W but measured in pixels.(Source: Appendix 1, Professor Ichiro Fujita, email correspondence July 2020)

1. Consider the quality of the camera and lens being used. It shall be of a suitable quality for use at the installation site.

2. Image resolution in the stream wise direction should be set in relation to the furthest analysis point from the camera location (water level) and either the size of the ROI where the velocity analysis is taking place or the length of STIV measurement line being used. The video resolution should be set to allow the pixel scaling at these measurement point to be approximately 5 cm per pixel (refer to Figure 16 and Appendix A).

3. Camera frame rate sampling should be stable. LSPIV software, due to the subsampling of frames used, are incapable of accommodating frame interval variability, resulting in significant velocity errors.

4. Video shall consist of a minimum 25 frames per second

3.6.1 Exposure time and sampling intervals

Fixed camera scheduling capabilities and data storage availability are some of the additional factors to consider when establishing a fixed site.


Page 28 of 48

3.6.1.1 Exposure time

The Exposure time is the total length of video in seconds that will be processed to calculate surface velocities used in a discharge calculation and reported against a sampling interval.

The Exposure time shall be a minimum of 30 seconds. This is the minimum exposure time stipulated in ISO 748:2007 Hydrometry – Measurement of liquid flow in open channels using current metres or floats.

Most organisations use exposure times of 40, 50, or 60 seconds for point velocity measurements regardless of the instrument type. Increasing the exposure time will reduce data anomalies caused by environmental and local site interferences.

Image velocimetry uniquely allows all measurement points across the channel to be measured simultaneously rather than consecutively. Total exposure time is currently a cumulative time reference based on the time to complete a measurement with 800+ seconds, a benchmark reference for ADCP data. An ADCP requires slow, steady movement when traversing a channel to minimise the accumulation of errors. The exposure time of 800 seconds helps to regulate the ADCP traverse speed. An ADCP does not measure velocity at any single measurement point for 800 seconds.

An image velocimetry video 40 seconds in length with 20 velocity measurement points is therefore equivalent in exposure time to a wading measurement or ADCP section by section measurement also consisting of 20 points measured for 40 seconds. The latter measurements are reported as a cumulative exposure time of 800 seconds rather than the actual 40 second exposure that is taking place at each measurement point.

It is this equivalent exposure time at each measurement point that allows image velocimetry data to provide comparative discharge results.

3.6.1.2 Sampling interval

Sampling intervals are dependent on expected stage changes during the measurement, and thus the appropriate interval depends on catchment shape, reach length, reach slope, land use, soil type, vegetation and precipitation. These factors can impact the shape of the flow hydrograph significantly and for this reason it is important to adjust the sampling interval to ensure an accurate representation of the hydrograph.

Site hydrographs shall be reviewed when determining the sampling interval.

3.6.2 Camera management

Organisations shall maintain camera management records specific to each instrument that contain:

1. Installed software/firmware and operational programmes.

2. Lens cleaning

3. Site configuration details


Page 29 of 48

3.7 Image Velocimetry Using Remotely Piloted Aircraft

Use of remotely piloted aircraft, commonly referred to as drones, provides a mobile solution for image velocimetry measurements, and only requires pixel scaling to calibrate pixel size. Drones can therefore be used to quickly and safely measure discharge in flow conditions unsuitable or too dangerous for deployment of an Acoustic Doppler Current Profiler (ADCP) on a manned boat.

3.7.1 Drone Safety

If flying a drone the pilot shall be aware and abide by the relevant regulations as outlined by the Civil Aviation Safety Authority (CASA) as well as local authority regulations and relevant Standard Operating Procedures (SOP)

https://www.casa.gov.au/drones

3.7.2 Site selection and Data Collection Procedures

Hydraulically a site should be selected in accordance with Section 3.5 when conducting a discharge measurement.

3.7.2.1 Data Collection Procedures

When using a drone to collect videos for image velocimetry analysis and discharge calculations the following collection procedures apply:

1. The camera shall be in a nadir position for pixel scaling.

2. The drone (not camera) should be facing downstream to align with the cross section. This is a recommended data collection procedure and not a software analysis requirement or restriction.

3. The entire channel should be contained within the image.

4. The drone shall be hovering and stable to reduce camera shake. Any significant movement of the drone could impact the velocity analyses results.

5. There shall be at least one easily identifiable feature within the video that has a known and reasonably large measured distance for scaling the pixels.

6. A cross section for the video location is required for discharge calculations.

7. Videos shall be recorded with a frame rate of not less than 25 frames per second.

8. A maximum scaling value of 5 cm is recommended for resolving ripples advected with flow when physical surface tracers are minimal or non-existent (Appendix A)

9. Values greater than 5 cm per pixel are still acceptable for use in image velocimetry.

10. Consider environmental elements such as sun glare, shadows, and wind affects. These may impact the drones ability to visualise surface movement accurately, especially in lower flow environments.

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Surface Velocity Radar

Page 30 of 48

Figure 14. An image from a drone flood video in Norway. Tracer particles that are advected with the flow are clearly visible across most of the channel width. This video is very good for image velocimetry analysis. (Source: Kristoffer Florag-Dybvik, Norwegian Water Resources and Energy Directorate (NVE))

4 Surface Velocity Radar

Surface Velocity Radar (SVR) methodology is a non-contact measurement process to determine the water surface flow velocity using Doppler frequency shift methods.

The methodology utilises a radar sensor system that transmits a signal with a constant frequency at a specific angle to the water surface. Moving water surface reflects the signal back to the sensor and a resulting Doppler shift is measured in the returned signal frequency.

SVRs generate a spread beam (the angle may vary by manufacturer) and the Doppler signals form a measurement area (zone of measurement) on the water surface. The size of the measured area depends on the inclination angle and the distance between the sensor and the reflecting water surface. The velocities in this area will generate a frequency shift distribution depending on the flow conditions. The shift distribution enables the water surface velocity to be determined in the zone of measurement. Depending on the complexity of the SVR technology various analytical tools may be available to analyse and evaluate spectral data to improve signal processing outputs within the SVR.

The local velocity measured by the SVR may then be utilised to establish stream section velocities using velocity indexing techniques (refer to National Industry Guidelines for hydrometric monitoring, Part 9: Application of in-situ Point Acoustic Doppler Velocity Meters for Determining Velocity in Open Channels for application of velocity indexing processes, NI GL 100.09–2019, for application of velocity indexing processes).

Coupled with level measurement data/systems and known stream cross section data at the measurement location, SVR data can be used to derive discharge data for the site.

Applications of SVR technique can include:

1. Standard stream gauging site operations with real-time stream flow data generation


Page 31 of 48

2. Stream discharge monitoring where variable downstream hydraulic conditions could affect standard stage discharge relationships (e.g. tidal influences, backwater effects).

3. Provision of velocity information at a site where stage/discharge rating changes create operational issues. The velocity data may assist in identifying when a discharge relationship change occurs. Data collected can contribute to the development of a new discharge relationship using stage/area/velocity analysis processes and techniques.

Figure 17. Principle of SVR application to determine local surface velocity and hence derive mean stream velocity and stream discharge via velocity indexing techniques. (Source: Sommer RQ 30 Discharge Measurement System User Manual, Version 2, 29/7/2014,)

4.1 SVR techniques

Use of SVR can occur via a number of deployment techniques. Selection of an SVR technique will depend on the monitoring site objectives and organisational requirements. Techniques can be described under three groups:

1. Permanent Fixed SVR installations – mounted on fixed structures in or above the stream, for example, on bridges or piers. Suspension by cableway above the stream may also be considered (see Figure 18).

Figure 18. Fixed Mount SVR (Source: https://www.sommer.at/en)


Page 32 of 48

2. Multipoint SVR gauging technique system – surface velocities measured at defined distances/points across the water surface by deploying the radar sensor via stream gauging cableways or bridge gauging techniques. Effectively this is an adaptation of surface velocity gauging techniques (see Figure 19).

Figure 19. SVR Cableway stream gauging deployment (Source: https://www.sommer.at/en)

3. Handheld SVR – commonly known as speed guns, provide potential to obtain surface velocity measurements at locations where infrastructure for multipoint SVR is unavailable, or where the available infrastructure prevents effective use of gauging trolleys, hand rail mounted SVRs and other similar deployment methods (see Figure 20).

Figure 20. Handheld SVR (Source: https://www.stalkerradar.com/)

4.1.1 Limitations

Limitations that may impact on the effective data collection by SVR systems are described further in Section 8 but a summary of potential limitations for SVRs are:

1. Lack of strong and consistent water surface movement to enable return a definable Doppler shift – this is a particular problem in slow moving deep stream reaches where ‘mirror’ conditions may occur.


Page 33 of 48

2. Some manufacturers define a minimum of 3 mm water surface disturbance for good SVR signal definition to be returned

3. Standing waves may impact the Doppler shift angle (used in some SVR computation processes) resulting in reduced accuracy

4. Wind impacts, most notably on deep slow moving streams, may generate SVR measurements that are not representative of the ‘real’ local stream surface velocity.

5. As water level rises the SVR measurement zone reduces, thus a smaller portion of the stream surface is analysed. This can lead to limitations, particularly in streams with large stage ranges.

6. Inclination of sensors past the manufacturers recommended operation parameters for SVRs will result in erroneous or no velocity data. The operation manual for the SVR will advise the recommended inclination limits.

4.2 Site selection and installation methods

Hydraulically a site should be selected in accordance with Section 3.5 when conducting a discharge measurement.

4.2.1 Fixed SVR installations

The methods of installation and use of SVR may vary between sites, or the operational requirement of the site being measured may define the installation requirements.

The installation should be appropriate for the final purpose of the measurement site, local stream features and infrastructure permissions.

For fixed installation or deployments over extended time frames for project-specific monitoring, data collection/transmission platforms shall be designed to provide sufficient power for the SVR and associated data collection/transmission systems, and comply with relevant organisational standards for the installation

1. Safe access for authorised personnel to maintain and service the SVR.

2. Sensor stability and rigidity (in the case of fixed SVR)

3. Protection from adverse weather conditions that may influence the velocity data

4. Protection from vandalism and fauna impacts

5. Positioned so that there is no interference with the activities or operation of persons or equipment on in and around the stream (shipping/boating operations, pedestrian footways on or under the installation)

6. The viewing direction of the SVR is set perpendicular to the bulk of the flow angles expected through the range of flows for the site

7. There shall be a suitable cross-section for gauging located at the installed SVR to confirm the validity of the index velocity rating

8. The installation ensures that if the SVR is removed or moved during servicing and maintenance activities, that the SVR can be accurately positioned back at


Page 34 of 48

the same location, maintaining horizontal and vertical aspects that existed prior to removal, and that the inclination angle prior to servicing is maintained

9. The SVR should sample velocities away from the stream edge and in an area of maximum velocity.

10. Expected measurement velocities shall be within the SVR operational boundaries as stated by the manufacturer.

11. Channel geometry of a regular shape assists with IVR development at the SVR location. Where highly irregular channel shapes cannot be avoided (e.g. overbank flows, or secondary channel breakout during flood events) more complex analysis of the IVR may be required

12. The surface velocity measurement target zone of the SVR shall be visible to the sensor through the full expected stage range at the site. Points of flow disturbance in either an upstream or downstream location should also be avoided.

13. Free Field Of View shall be maintained. The SVR interprets all movements such as moving trees, bushes or grass, and these shall not be present in the Field Of View.

14. For similar reasons to point 12 above, consideration should be given to locations that can protect the SVR from precipitation or falling water. Consideration should be given to installing sensors that have provision to blank out false velocity signals if considered an operational issue at the site.

15. The SVR shall be installed in a position that will measure a representative local velocity sample throughout the stage range to be measured, which can be applied to IVR processing.

16. Stationary waves within the SVR measurement zone should be avoided. Stationary waves can cause errors in angle correction when the radar impulse is reflected from the stationary wave and not the plane water surface.

17. A measurement location shall be chosen such that the water surface moves observably and some translatory non-wind induced roughness features (ripples and waves) are present. A more rippled water surface and higher flow velocity will improve the reliability of the SVR measurement. For very slow moving rivers where this requirement may not be fulfilled the velocity measurement may be unreliable.

18. For slow moving, deep rivers the velocity measurement may be adversely influenced by surface velocities generated by wind that is not representative of the real surface velocity. Sites with potential wind influence should be protected as much as possible against wind impacts. Collection of wind velocity and wind direction data at a site may be considered to enable additional quality processing in water data processing systems.

4.2.1.1 Bridge installations

Mounting on bridges often provides a simple cost-efficient installation option as existing infrastructure can be used. The SVR may be installed either on the structure itself or on the railing of the bridge.


Page 35 of 48

The following issues should be considered when designing and undertaking the installation:

1. Preferred viewing direction is upstream to provide a water surface less likely to be impacted by water surface disturbance by bridge piers.

2. Avoid drainages of water from the structure impacting the SVR measurement field

3. Preferably for a bridge measurement, the bridge should be level. Steeply-arched bridges should be avoided. This is due to the concept that as the SVR moves further away from the water surface the zone of measurement moves further away – effectively creating a curved measurement section. This could be corrected by employing inclination compensation to keep the zone of measurement in the required line (and if the equipment in use has the ability to compensate for the inclination calculation)

4.2.1.2 Extension arms

If no bridges or similar structures are available the SVR can be mounted on extension arms protruding from one bank of the stream. It is suggested to install rotatable arms to simplify the maintenance and access to the SVR

The following guidance applies when designing and undertaking the installation:

1. The extension arm should be positioned such that the SVR can measure the representative velocity zone in the section.

2. Preferred viewing direction is upstream to provide a water surface less likely to be impacted by water surface disturbance by bridge piers

3. The arm assembly shall be rigid and firm with no movement in the deployed position.

4.2.1.3 Cable ways

The radar sensor can be mounted on a cable way or ropes crossing the river. The following guidance applies when designing and undertaking the installation:

1. Swinging of the sensor should be minimised by tensioning the assembly to anchors and counterweighting the SVR to stabilise the inclination.

2. If inclination cannot be sufficiently stabilised, an inclination measurement shall be performed prior to every measurement.

3. Changes in the height position should be minimised. This issue can occur most commonly with SVR systems that integrate a non-contact downward looking level sensor where cable thermal expansion and contraction may occur, creating variations in water level measurements. SVR equipment validation/maintenance checks

4.2.1.4 Multipoint SVR (surface velocity gauging method, velocity-area adaptation)

In extremely wide stream sections consideration may be given to installing multiple fixed SVRs across the section. This is achievable where there is a bridge across the stream.


Page 36 of 48

Multiple SVR installations may also be beneficial in overbank flow and secondary channel stream sections, by segmenting the stream section into discrete velocity/area sub sections

A single SVR unit may also be used to measure multiple velocity points by moving the SVR across the channel. This SVR measurement process is the same as a traditional velocity area gauging. A cable way or bridge is commonly used to assist in the movement of the SVR across the channel to measure multiple points.

The following guidance applies to the SVR multipoint discharge measurement technique:

1. Known cross sectional information from surveys or previous gaugings provide a depth at each SVR station for the current stream level under the cableway or bridge section. If the cross section changes, for example during a flood event, a survey of the cross section shall be undertaken to define the area and depth relationships relevant to the discharge measurement.

2. Either:

Surface Alpha shall be defined for each depth station to correct the surface velocity to a mean velocity in the vertical (refer to Section 5 for more details on Surface Alpha), or

a mean Surface Alpha for the section shall be applied across all verticals.

3. Exposure time should be a minimum of 40 seconds at each station.

4. The number of verticals (velocity points) shall be set in accordance with National Industry Guidelines for hydrometric monitoring, Part 4: Gauging (stationary velocity-area method) NI GL 100.04-2019 (refer to section 6.2.2).

5. Overall time of exposure for a completed stream discharge measurement shall exceed 800 seconds (NI GL100.08–2019) Total exposure times less than this may result in downgrade of quality of the discharge measurement.

4.2.1.5 Special applications scenarios

In special applications the criteria in section 4.2 may be modified to suit the organisational requirements of the monitoring site or project. Examples include:

1. projects primarily investigating stream velocities under a variety of flow conditions

2. where data are obtained for assessment of backwater or non-standard streamflow conditions e.g., tidal impacted stream sections.

Appropriate documentation for the specific requirements and operation in these circumstances should be maintained within the organisations relevant data processes and documentation systems.

4.2.2 Handheld SVR (surface velocity gauging method, velocity-area adaptation)

Handheld SVR measurements may be used in situations where multipoint SVR techniques are impractical. For example, a Handheld SVR may be appropriate for


Page 37 of 48

obtaining surface velocity information from bridges with construction features (such as bridge supports or diagonal girders) that physically interfere with a multipoint SVR arrangement. Handheld SVR units can be used between obstructions.

All guidance for multipoint SVR measurements in sections 4.1.1 and 4.21 shall apply.

The number of measurement points and exposure time in sections 3.1.2 and 3.6.1.1 shall apply.

Surface velocity alpha coefficients shall be determined as specified in section 5.

All information specific to a Handheld SVR measurement shall be documented.

In addition:

1. The operator shall maintain the handheld SVR in a steady position while undertaking velocity sampling at each station. This can be assisted through use of tripods/monopods to support the SVR.

2. The same inclination of the handheld SVR shall be applied at each station of the measurment to maintain a consistent reflective angle. This will also assist in maintaining a straight measurement section as the operator moves across the bridge over the stream.

3. The SVR should have an inclination gauge/indicator to enable the operator to adjust inclination correctly.

4.3 SVR instrument management

Organisations shall maintain SVR management records specific to each instrument that contain:

1. Service and repair schedules

2. Service and repairs completed records

3. Validation and operational check records

4. Installed software/firmware and operational programmes.

Validation checks of the SVR and its measurement of surface velocities may be achieved by:

1. Manufacturers recommended processes or software analysis tools

2. Undertaking alternative surface velocity determination techniques. For example:

concurrent STIV analysis of the area in the zone of measurement for comparative velocity analysis

ADCP measurement in the zone of measurement followed by a review of velocity curves and analysis of surface extrapolation fits for the measurement

3. Comparison with historical discharge measurements where surface velocity measurements were undertaken.

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Surface velocity coefficients

Page 38 of 48

5 Surface velocity coefficients

Each of the measurement techniques described in this guideline measure surface velocity within the channel. The measurement extent varies between the different techniques, from the full channel width for a particular length (in the case of image velocimetry) to only a portion of the water surface (in the case of a single point surface velocity radar measurement). When calculating discharge, surface velocity coefficients (commonly referred to as surface alpha or alpha) shall be developed to translate the measured surface velocities to a mean channel velocity. There are three techniques for determining and applying the surface velocity coefficients:

1. Surface Alpha based on Log Law, Power law or direct surface to mean velocity ratios. These methods use multiple measurement points.

2. Probability Concept Method for estimating the mean-channel velocity Surface Velocity Rating. This method uses a single measurement point.

3. Index Surface Velocity Method. Uses multiple points.

Figure 21 Typical vertical profile of downstream velocity over a gravel bed demonstrating an augmented logarithmic profile. (Source: Smart et al 2020)

5.1 Vertical velocity profiles for calculating surface coefficients

Calculation of surface velocity coefficients follows established velocity profile theories. Sections 5.1.1 and 5.1.2 have been directly reproduced from Smart et al (2020).

5.1.1 Log law velocity profile

With no flow acceleration, secondary flow or surface wind drag, a logarithmic velocity profile (Keulegan, 1938) can be assumed to extend to the surface. The depth averaged velocity is then related to mean surface velocity 𝑢𝑠 and friction velocity 𝑢∗ as follows:

here 𝑢 is the time averaged streamwise velocity at elevation 𝑧 above the log profile zero plane, 𝑈 is the depth average of 𝑢, 𝜅 is the Von Kármán constant, z0 is the log law


Page 39 of 48

roughness scale, 𝐻 is flow depth above the log law origin, 𝑔 is gravitational acceleration, 𝑆 is slope, and 𝛼 is the velocity index. Equation 9 is similar to the experimental findings of Bazin (1865) who reported that 𝑢𝑠−𝑈=(20/3)√𝐻𝑆 for channel widths > 5𝐻.

5.1.2 Power law velocity profile

The velocity coefficient can also be estimated from the power law velocity profile as follows:

5.1.3 Measured surface velocity

Surface Alpha coefficients may be determined from ADCP and current meter/flowtracker measurements that measure actual or near surface velocities and the mean velocity. Velocity and depth data should be normalised. The surface alpha value is the ratio between the surface and mean values on the velocity profile as shown in Figure 22.

Figure 22 Normalised velocity profile showing surface and mean velocity profile locations (Source: Mark Randall)

5.1.4 Probability concept method for calculating surface alpha

If the velocity profile is non-standard, e.g. where the maximum velocity occurs below the water surface, then an alternative method of computing surface alpha should be used. The probability concept method is one such example.


Page 40 of 48

The following information has been reproduced from the United States Geological Survey (USGS) https://my.usgs.gov/confluence/display/SurfBoard/How+to+translate+Surface-water+Velocities+into+a+Mean-vertical+or+mean-channel+Velocity viewed15/05/2020.

The Probability Concept (Chiu 1989; Chiu and Tung, 2002; Fulton and Ostrowski, 2008) was developed for computing uavg or uvertical at a cross-section. Two parameters, ϕ , which converts maximum point velocity in a section to the average section velocity, and (umax), the maximum-instream velocity, are needed to compute uavg. Velocities are measured along a single vertical as a function of depth beginning at the channel bottom and concluding at the water surface or by collecting pairs of channel uavg and umax for a of channel flow conditions. If using a single vertical, it is referred to as the "y-axis" and all data collection efforts should focus on that single vertical. This is the profile that contains the maximum information content (minimum velocity, maximum velocity, and depth) required to derive the parameters umax, ϕ, and h/D used to compute uavg.

Research suggests (Chiu et al., 2001; Fulton and Ostrowski, 2008) the location or stationing of the y-axis is generally stable for a given transect and does not vary with changing hydraulic conditions including variations in stage, velocity, flow, channel geometry, bed form and material, slope, or alignment. However, field verification of these parameters should be conducted periodically, and a stage-area rating shall be maintained. The y-axis rarely coincides with the thalweg in open or engineered channels; umax can be computed or measured directly using image velocimetry or velocity radars.

5.1.5 Surface velocity index rating development

The development of the index surface velocity is similar process followed for acoustic Doppler velocity meters (ADVM) described in NI GL 100.09–2019 (Application of In-Situ Point Acoustic Doppler Velocity Meters for Determining Velocity in Open Channels).

The index surface velocity method comprises the following ratings to develop mean velocity and wetted area for discharge calculations.

1) Stage-area rating; and 2) Index surface velocity rating.

The variables associated with the index surface velocity rating development, comprises of the surface velocity measured using Surface


Page 41 of 48

Radar, LSPIV, STIV, etc. and measured mean velocity (measured discharge dived by the corresponding area obtained from the stage-area rating) in the channel.

5.1.5.1 Index alpha method

Develop index alpha based on surface velocity, stage and alpha calculated from discharge, stage and surface velocity measurements at the measurement site. The index alpha can be extrapolated using surface velocity and stage to determine the alpha.

5.1.5.2 Index rating extrapolation

The development of index surface velocity rating is dependent on two key components, discharge measurements using conventional hydrometric methods and surface velocity measurements. In cases such as flood events and or when the site and hydraulic conditions are not suitable for direct measurements, an alternative method needs to be used for extending the existing index surface velocity rating.

5.2 Site specific velocity profile analysis

A site specific surface alpha should be calculated whenever possible using the methods described. Analysis of old ADCP or current metre data may assist.

Figure 23 Measuring velocity profiles at multiple verticals across a channel. Verticals near the banks are removed to avoid incorporating velocity profiles that are not representative of the main flow (Source: Hauet et al).

When site velocity profile analysis is undertaken, it should be determined whether a single surface alpha is representative or whether different sections of the channel require their own surface alpha value. In particular, large floodplains may have a very different surface alpha than the main channel, shown in Figure 24.


Page 42 of 48

Figure 24 Measuring velocity profiles at three verticals across a channel. The left profile is the main channel, centre profile is at midway depth location between channel and flood plain, right profile is in the flood plain. The floodplain consisted of a large amount of small trees whose canopy vegetation has slowed down the surface velocities. Velocities below the surface, at the tree bases have not caused the same flow resistance resulting in a very different velocity profile than the main channel. (Source: Mark Randall).

Surface alpha values shall be documented with the surface velocity measurement specifying if they are theoretical or site specific values.

5.2.1 Channel characteristics impact on alpha

The channel characteristics and hydraulic conditions (section 3.5) at a monitoring site can have an impact on alpha. Changes in water depth

1. Channel aspect ratio ( )

2. Channel bed roughness

3. Unsteady flow or backwater conditions

4. Hydraulic structures

5. Vegetation

6. Wind impacts

Any suspected site characteristic impacts on the surface coefficients shall be investigated and documented by determining the velocity profile.

5.3 Typical surface coefficient values

The guidance below provides a basic indication of typical surface alpha coefficient values based on the power law exponent m.

Several observations published by Huet et al based on the surface alpha coefficient analysis of 3611 gaugings from 176 French hydrometric stations suggests that:

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Uncertainties in Discharge Measurements

Page 43 of 48

Surface alpha coefficients increase with water depth (D).

o For shallow water (0 m<D>1 m) alpha is close to 0.8

o Alpha increase up to 0.9 for D of 9 m in single channels. Compound channels may be lower. Deep water holes could also cause higher alpha compared to more planar beds.

Link between alpha and the relative roughness is not as clear.

o Mean value of 0.8 for relatively-rough natural rivers (sandy, pebbly, boulder rivers)

o Mean value of 0.9 for artificial concrete channels.

A site specific alpha should always be calculated when possible.

6 Uncertainties in Discharge Measurements

The following information regarding hydrometric and discharge measurement uncertainties has been sourced from ISO/TR 24578:2012 which makes reference to ISO 5168, ISO/TS 25377, ISO/IEC 98-1, ISO/IEC 98-3 and ISO 15769. It is the responsibility of the operator to refer to the original documents for future ISO updates on measurement uncertainties.

6.1 Description of measurement uncertainty

All measurements of a physical quantity are subject to uncertainties and therefore the result of a measurement is only an estimate of the true value and only complete when accompanied by a statement of its uncertainty. The discrepancy between the true value and the measured value is the measurement error which is a combination of component errors that arise during the performance of the various elementary operations of the measurement process.

When a measurement depends on several component qualities then the total measurement error is a combination of all the component errors. Therefore, the determination of a measurement’s uncertainty is a combination of all the identified component measurement errors, quantification of their corresponding uncertainties and then a combination of those component uncertainties.

The component uncertainties are combined in a manner that accounts for both systematic and random errors and are termed ‘standard uncertainties’ which correspond to one standard deviation of the probability distribution of measurement errors. One standard deviation equates to a confidence level of 68%. The uncertainty at two standard deviations is twice the standard uncertainty which if estimated can be multiplied by two to obtain the uncertainty at two standard deviations, or 95% confidence level. The multiplication factor is termed as the coverage factor. Therefore, if the uncertainty is expressed at three standard deviations the coverage factor would be three and represent a confidence level of 99%.

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Uncertainties in Discharge Measurements

Page 44 of 48

When stating uncertainties, it is also necessary to state the confidence level or the coverage factor i.e. the number of standard deviations.

NOTE: For example, if a discharge measurement of 50 cumecs had an uncertainty of 9% at the 95% confidence level the statement of uncertainty should be documented as follows:

Discharge = 50 m³sˉ1 with an uncertainty of 9% at the 95% confidence level based on a coverage factor of k=2.

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Appendix A Critical conditions for image velocimetry measurements

Page 45 of 48

Appendix A Critical conditions for image velocimetry measurements

The following information was kindly provided for these guidelines by Professor Ichiro Fujita, email correspondence July 2020.

A.1 Angle of repose θ

The critical value θ is assumed to be 2 degrees for STIV and 15 degrees for LSPIV.

A.2 Image resolution

The relative location of the camera and Region of Interest or Search Lines is shown in the figure below. A simple relationship between the two is as follows:

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Appendix A Critical conditions for image velocimetry measurements

Page 46 of 48

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Appendix B Training

Page 47 of 48

Appendix B Training

Who needs this competency?

This learning material covers the skills and knowledge required for a person to use and understand National Industry Guidelines for hydrometric monitoring

Learning outcomes

At the completion of this learner resource you will be competent in the following:

use the guideline document for reference

use the guideline in day to day operations

access the material referenced in the guideline document

understand procedural standards for using acoustic instruments to gather water data

use and understand related internal procedures and work instructions.

Health and safety considerations

Health and safety legislation shall always be considered when implementing National Industry Guidelines, workplace procedures and work instructions.

Employees carrying out work related to the National Industry Guidelines should be adequately trained in all relevant health and safety matters.

Environmental considerations

Compliance with this guideline may involve working in the environment. As such care should be taken to:

prevent unnecessary damage to river banks

prevent unnecessary disturbance of the river system

carefully construct any infrastructure to minimise impacts on the environment and river flow conditions

plan access roads to sites to minimise impacts during all seasonal conditions.

What resources will I need?

Workplace policies and procedures

Manufacturer manuals, requirements and specifications

Codes of practice

Workplace equipment, tools and instruments

Workplace reports

National Industry Guidelines for hydrometric monitoring NI GL 100.11–2020 Appendix B Training

Page 48 of 48

Workplace maps, plans and instructions

Permits and access to locations and worksites

Other useful resources

Relevant Health and Safety Act

Manufacturer’s Instruction manuals

Organisations procedures and work instructions

Herschy, Reginald W. (1985), Stream flow Measurement, Elsevier Applied Science Publishers, New York, NY, USA

Australian Standards

World Meteorological Organization (WMO)

World Meteorological Organization 2008, Guide to Hydrological Practices, Volume I: Hydrology – From Measurement to Hydrological Information. WMO-No. 168. Sixth edition, 2008. ISBN 978-92-63-10168-6, viewed 2 October 2018, <http://www.whycos.org/hwrp/guide/index.php>

World Meteorological Organization 2009, Guide to Hydrological Practices, Volume II: Management of Water Resources and Application of Hydrological Practices, WMO-No. 168, Sixth edition, 2009, viewed 2 October 2018, <http://www.whycos.org/hwrp/guide/index.php>

World Meteorological Organization 2010a, Manual on Stream Gauging, Volume I: Fieldwork. WMO-No. 1044, 2010. ISBN 978-92-63-11044-2, viewed 2 October 2018, <http://www.wmo.int/pages/prog/hwrp/manuals.php>

World Meteorological Organization 2010b, Manual on Stream Gauging, Volume II: Computation of Discharge. WMO-No. 1044, 2010. ISBN 978-92-63-11044-2, viewed 2 October 2018, <http://www.wmo.int/pages/prog/hwrp/manuals.php>

draft ni surface velocity guidelines ver 1 · 2020. 10. 27. · 3djh ri 1dwlrqdo ,qgxvwu\...

Documents