Laser ranging measurements of turbulent water surface
Gašper Rak1, Franc Steinman1, Marko Hočevar2*, Matevž Dular2, Matija Jezeršek2 and Urban Pavlovčič2
ISIMet 2
International
Symposium on
Image based
Metrology
Hawaii, Maui
December 16-21 2017
Abstract
Laser ranging is a measurement method, applied in wide range of applications. We use
laser ranging for measurements of water surface of turbulent flows. Measurements were
performed in three cross sections on the confluence of the main and tributary flow. Both
flows exhibited high Reynolds and Froude numbers where free - water surface profiles were
turbulent, non - stationary and non - homogeneous. A commercial LIDAR and fast camera
were used for measurements. The fast camera operated on the principle of laser
triangulation, using only illumination provided by the LIDAR laser beam. While we cannot
compare measurement results with state of the art measurements, results show a good
agreement between both methods. Average turbulent water profile measurements agree
very well, instantaneous profiles however agree less. We present an explanation for the
agreement of measurement results.
Keywords
LIDAR — laser triangulation — free surface — turbulence — non-stationary surface — fast camera
1 University of Ljubljana, Faculty of civil and geodetic engineering, Ljubljana,Slovenia 2 University of Ljubljana, Faculty of mechanical engineering, Ljubljana,Slovenia
*Corresponding author: [email protected]
1. Introduction
Water surface measurements of turbulent free surface
flows are an important part of hydraulic measurements.
Such flows are encountered in a wide range of applications
in civil, chemical, environmental, mechanical, mining and
nuclear engineering etc. In turbulent free surface flows, air
bubble entrainment is the result of the deformation of the
surface. When the turbulent shear stress is greater than the
surface tension stress that resists the interfacial breakup,
bubble entrainment is possible. Water - air interface has
been studied experimentally, numerically and theoretically
over last few decades, overview was provided by H.
Chanson in [1].
Distance measurements to various objects are often
conducted with one of laser ranging methods due to their
inherent noncontacts and high speed capabilities. Laser
ranging is based on principles of interferometry,
triangulation or time of flight [2]. We will consider here the
latter two principles. Within time of flight measurements
principle, LIDAR technology is one of the most widely used
and promising remote sensing technologies [3; 4]. LIDAR is
composed of a laser source, scanner, optics, photodetector
and receiver electronics. LIDAR robustness and versatile
use is reflected in a number of completely different fields of
science and engineering, among them agriculture,
archaeology, surveying, autonomous vehicles, robotics,
military, atmospheric remote sensing, meteorology, etc.
LIDAR method is however not often used for water surface
measurements. Blenkinsopp et al. [5] used LIDAR to
measure the time - varying free - surface profile across the
swash zone. The water surface in the swash zone’s area
exhibit bubbles on the surface, which increases the
probability of diffuse reflections and hence measurements
with LIDAR. The results show good agreement with
measurements using ultrasonic sensors. Allis and
Blenkinsopp have independently applied the method for
laboratory profile measurements of time - varying free -
surface of propagating waves [6; 7]. Some authors use
LIDAR for to measure water surface in connection with the
bathymetry. Recently the influence of ocean wave patterns
on 3D underwater point coordinate accuracy for LIDAR
bathymetry was investigated by Westfeld et al. [8].
Laser triangulation measurements of surfaces are common
[2], but work related to triangulation measurements of water
surface is limited and almost non - existing for highly
turbulent and aerated flows. Mulsow et al. [9; 10] system
used a modified triangulation method capable of
measurement of reflected laser line projection. Other optical
methods [11] such as particle image velocimetry and stereo
vision photogrammetry or acoustic methods like Doppler
velocimetry are more common, but fail to provide
measurements for highly aerated turbulent flows.
Conventional methods are still most commonly used for
measuring of water levels, such as resistance - type probes
[12], U - manometers [13], point gauges, ultrasonic sensors
[14], etc.
In this study we are presenting measurements of turbulent
water surface of highly aerated flows at high Reynolds and
Froude numbers. By simultaneous using of LIDAR and laser
triangulation fast camera we provide deeper understanding
of how the laser beam is scattered on the turbulent and
aerated open surface and how both methods interpret the
data. Comparison of both measurement results is shown
while also interpretation of measurement results is provided
in the discussion section.
1.1 Flow at the confluence
The confluences occur in streams (natural and artificial river
channels, torrents, etc.), as well as in various types of
facilities and infrastructure (fish ways, drainage of surface
water from impervious surfaces, etc.). At confluences,
especially ones with incoming supercritical flows, a
distinctively three - dimensional flow of water is formed.
Surface profiles exhibit an non - stationary and non -
homogeneous profile of turbulent free - water surface. The
non - stationary structure of the water flow forms in the
transversal and longitudinal directions. Previous studies
[12; 15; 16] have largely neglected measurements of
turbulent water surface profiles in the transversal direction
of a supercritical flow channel’s confluence. One of the
reasons for this are limitations of conventional
measurement methods (piezometers, ultrasonic sensors,
point gauges, etc.), which do not allow for dynamic
measurements with high spatial resolution.
1.2 Laser ranging measurements of turbulent water
surface
In comparison with other measurement methods, the laser
ranging of turbulent water surfaces is limited due to
occurrences of specular reflections. In such cases the laser
beam is in most cases reflected away from the sensor and
measurement fails. Contrary the diffuse surface reflects at
least some laser light back to the photodetector and
distance can be measured provided that the intensity of
reflected laser light is above the threshold.
The use of laser ranging methods for measurements of non
- stationary and non - homogeneous open surface flows
presents several challenges; among them are successive
specular reflections, non - equidistant sampling,
thresholding of returned laser light, problems at channel
walls, etc. We estimate that the specular reflection of the
LIDAR beam from the air/water interface represents the
most important issue of the laser ranging measurements.
Refraction effects of the LIDAR pulse passing the air/water
and water/air interfaces have to be taken into account. This
includes the consideration of the reduced velocity of laser
light in water, being approximately 75 % of the velocity in
air. The water surface of highly turbulent open surface
flows, similar to the confluence flow used as an example in
this study, is difficult to determine (Figure 1). It is abundant
with droplets flying above the surface (overestimating the
surface height), numerous steep waves (preventing
measurements of concave surfaces), entrapped bubbles
below the surface (underestimating the surface height),
presence of foam (increasing the measurement
uncertainty), etc. The surface is also spatially very non -
homogenous and non - stationary in time.
In our experiment, besides being essential for LIDAR
measurements, the same LIDAR beam was used for laser
triangulation measurements using fast camera. This in part
explains the very good agreement between both laser
ranging methods, as we will show later. We assume that in
clean tap water of the confluence all laser beam reflections
form on the water/air interface and that they are all
specular. This assumption is justified by a very limited
number of impurities in the clean tap water. Due to a very
high turbulent nature of the confluence open surface flow,
the laser beam may be reflected many times, before it
reaches laser scanner or fast camera. Fast camera images
show regions of such successive reflections as bright pixels.
Although individual reflections are all specular and laser light
reflected to the fast camera clearly cannot reach laser
scanner receiver (and vice versa), we propose to look into
multiple consecutive reflections as a statistical process. In
such way, a high number of consecutive specular reflections
may be regarded as if they would exhibit behaviour of near -
diffuse reflections. The near - diffuse reflection is similar to
real diffuse reflections, although on a much larger spatial
scale [17]. For real diffuse reflections, the most general
mechanism by which a surface gives diffuse reflection, most
of the scattered light is contributed by scattering centres
beneath the surface [17], involving series of consecutive
multiple partial reflections. The near - diffuse reflections are
in our experimental configuration detected by LIDAR and
laser triangulation fast camera, although both sensors are
located in separate locations.
2. Measurements
Measurements were performed on the measuring station
shown in Figure 2. The measuring station is an open channel
confluence, arranged as usual for hydraulic model
experiments [12; 15]. Main flow and tributary flow are
arranged at an angle 90 ° using sharp edges. The length of
the main flow channel is 6 m, while the lengths of main and
tributary channel upstream of the confluence are 1 m.
Channel width is 0.5 m for both main and tributary flow. All
sides are rectangular and made from glass. The bottom of
the measuring station is horizontal. All joints were carefully
manufactured to avoid local flow separation.
A clean tap water was used for the experiment. Water was
provided by a centrifugal pump, filling the constant hydraulic
head reservoir. From the constant hydraulic head reservoir,
water flowed by gravity to the measuring station shown in
Figure 2. Water volume flow rate was measured for both
channels using electromagnetic flowmeters. ABB
Watermaster flowmeters were used. The selected volume
flow rates were set using valves while flaps were used to
select height of main and tributary flows. The flaps were
mounted at the outflow from pressure vessels. After the
outflow from the measuring station, water flowed into the
lower reservoir.
Unlike other authors [e.g. 6; 7], who applied laser scanning
to time - varying free - water surface profiles of wave
propagation in a wave flume, no additives were added to
improve reflection in our study.
1.1 Measuring equipment
Measuring equipment consisted of a commercial LIDAR
device, a fast laser triangulation camera and a secondary
camera. Measuring equipment installation is shown in Figure
2.
Laser ranging measurements of turbulent water surface— 2
Figure 1. Specular reflections at the turbulent water surface.
Figure 2. Measuring station
The LIDAR was mounted perpendicularly above and in the
middle of the measuring station channel 1 m above the
channel bottom wall. The LIDAR Sick LMS400 was used. It
operated with scanning frequency 270 Hz and angular
resolution 0.2 °. The LIDAR Sick LMS400 operates in
visible red light with wavelength λ = 650 nm. According to
the data provided by the producer, the systematic
measurement uncertainty is ± 4 mm and statistical
measurement uncertainty is ± 3 mm. The beam diameter is
1 mm.
The LIDAR data were transmitted to the measuring
computer by an Ethernet connection. A dedicated driver
and communication software was written in National
Instruments software package Labview, enabling reliable
LIDAR data acquisition and storage to local disk with no
lost measurement samples. The LIDAR was calibrated
before measurements with a strip of white paper, attached
to the bottom of the channel.
Laser triangulation method was composed of a fast camera
mounted on the downstream of the channel and a laser
scanning projector provided solely by the existing LIDAR
(Figure 2). The camera’s angle was set at 35 °, pointing
downwards and upstream. We have used the fast camera
Photron SA - Z. The camera was operating at a framerate
100000 frames/s at a resolution 640 x 280 pixels. Every
recorded fast camera image corresponded approximately to
a single LIDAR distance measurement with angular
resolution 0.2 °. The lens used for the fast camera was F -
mount 28 mm f / 1.8 G Nikkor. The red filter with cut - off
wavelength at 600 nm was mounted on the lens, with the
purpose to filter out the blue light used by secondary camera.
The calibration procedure for the triangulation method was
performed according to Jezeršek and Možina [18].
The secondary camera was oriented normal to the water
Laser ranging measurements of turbulent water surface— 3
surface in order to get visual information about the flow at
the measurement location. The secondary camera used
was a black and white Fastec Hispec 4. The diffuse
illumination for the secondary camera consisted of a large
number of LEDs with total power around 100 W at
wavelength of 465 nm. The secondary camera used C -
mount lens with 35 mm focal length and recorded images
with frequency 270 Hz, which was equal to the scanning
rate of LIDAR. The image resolution was 1696 x 360 pixel
and the exposure time was 3.7 µs. The secondary camera
detected both lights, i.e. blue LED and red LIDAR laser
beam.
Measurements duration in every measurement cross
section was 2 s. All measurement equipment was
synchronised using a mechanical shutter. The mechanical
shutter was mounted above the measuring station channel,
between the water surface and LIDAR with the secondary
camera. Before the acquisition, the shutter was closed,
such that the laser beam from LIDAR was cut and the laser
triangulation fast camera images were dark. Then, LIDAR,
fast camera and secondary camera were started. Shutter
opening enabled acquisition and recording of all three
sensors simultaneously.
The entire experiment was performed in dark environment. 1.2 Selection of operating point and
measuring positions
A single operating point was selected for analysis as shown
in Table 1. Measurements were performed in three
measurement cross sections (CS1 - CS3) as shown in
Figure 3. Measurement cross sections were located 900 mm
(CS1), 1100 mm (CS2) and 1300 mm (CS3) downstream
from the confluence.
Reynolds number evaluates the ratio of inertial forces to
viscous forces within a fluid. Reynolds numbers Re were
calculated as
vhRe
where is water density, v is water velocity, h is water
height and is water viscosity. Froude number evaluates
the ratio of the flow inertia to the external gravity field, Froude numbers Fr were calculated as
gh
vFr
where g is gravitational acceleration.
Table 1: Selection of the operating point.
variable value
main flow flow rate [l/s] 35.5
tributary flow flow rate [l/s] 26.6
main flow height [m] 0.02
tributary flow height [m] 0.02
main flow Re [-] 7.1x104
tributary flow Re [-] 5.3x104
main flow Fr [-] 8
tributary flow Fr [-] 6
Consecutive sample images of the confluence flow in the
selected operating point are shown in Figure 4. Images show a
very high turbulent nature of the flow.
Figure 3. Dimensions of the confluence and locations of measuring cross sections CS1, CS2 and CS3. All measurements
are in mm.
Images in Figure 4 were recorded in a separate experiment
with secondary camera, moved to an approximate location of
laser triangulation fast camera (Figure 2). For this separate
experiment framerate 50 Hz, shutter 500 µs and additional
blue LED illumination for the entire downstream section of
the confluence were used.
Figure 4. Consecutive sample images of the confluence flow recorded in a separate experiment with secondary camera, time interval
among consecutive images in the above case was 0.02 s.
Laser ranging measurements of turbulent water surface— 4
3. Measurements data analysis procedure The LIDAR measurements data consists of a set of
angle/distance pairs in a sensor’s radial coordinate system.
Each distance measurement was converted into coordinates
of the local two - dimensional Cartesian coordinate system.
To enable comparison with results of laser triangulation
measurements, a linear interpolation between consecutive
measurement points was performed in order to create an
equidistant grid of data in Cartesian coordinate system.
The laser triangulation fast camera images were analysed
using algorithms, developed in LabView and C++
programming languages. The images were first filtered by
Gaussian filter (Figure 5).
Figure 5. Image analysis for laser triangulation method.
Above: original image, below: processed image. Red square
marks the target detected by the laser triangulation method.
Dimensions of the filter kernel were 4.5 % of the image size
in horizontal and 1.7 % in the vertical direction. That is how the
intensities of specular reflections from multiple neighbouring
locations were merged together. Then the location of reflected
laser light was determined as the location of the maximal
intensity on the image. Finally, the location of the point in the
3D space was reconstructed by triangulation principle [18].
Measurements by LIDAR and laser triangulation were
synchronised in time by detecting and deletion of that signals,
where mechanical shutter was visible.
4. Results and discussion Results of measurements of average turbulent water surface
heights and corresponding standard deviations are shown in
Figure 6. Measurement results in CS2 and CS3 show
remarkably good agreement between both laser ranging
measurements. Both measurements agree to around ± 10 mm
in the channel cross section, except for the regions near the
walls. In average the LIDAR measures water level at slightly
deeper locations (see blue line on Figure 6). Such agreement
was achieved for every average point of the cross section.
Agreement between both methods is slightly worse for CS1,
where deviation is within ± 25 mm. This is emphasised on the
left side of the channel, approximately up to 300 mm from left
wall. Sample image of the secondary camera on the right of
Figure 6 for CS1 reveal high flow dynamics and turbulence in
this region.
Measurement results near the channel walls exhibit higher
measurement uncertainty in CS1 in comparison with central
region. The wall region, affected by such behaviour, extends
to around 50 mm from both walls. The localisation near walls
was better for cross sections CS2 and CS3, featuring less flow
dynamics than CS1. Due to poorer spatial localisation of
triangulation method in comparison with LIDAR method, we
have in Figure 6 omitted results near both walls for 50 mm
intervals for triangulation method and for 25 mm intervals for
LIDAR method. We will discuss localisation issues of both
methods later. Here, additional reflections from glass walls
were present, influencing performance of laser triangulation
fast camera detection, image analysis algorithm and LIDAR
internal algorithm.
Figure 6. Results of measurements of average turbulent water surface heights (left) and corresponding sample images of the
recorded sequence with secondary camera (right). Results are shown for LIDAR (blue) and triangulation method (red). Above: CS1, middle: CS2 and below: CS3.
Laser ranging measurements of turbulent water surface— 5
Figure 7. Samples of instantaneous turbulent water surface heights. Blue: LIDAR and red: triangulation method. Above: CS1,
middle: CS2 and below: CS3.
Samples of instantaneous turbulent water surface
profiles are shown in Figure 7. We can notice, that both
laser ranging methods failed to provide valid
measurements for all measurement positions as
described in section 3 "Analysis". Amount of rejected
measurements was 14.6 % for CS1, 14.9 % for CS2 and
13.2 % for CS3 for LIDAR and 12.2 % for CS1, 7.2 % for
CS2 and 9.2 % for CS3 for laser triangulation method.
The amount of rejected measurements provides us with
estimation, how the laser light originating from
successive reflections at different air / water boundaries
such as droplets, bubbles and waves is detected by
LIDAR and laser triangulation fast camera. In the
following we will provide explanation why these values
do not agree for both laser ranging methods, as one
may guess due to the fact, that both operate using the
same laser beam from the LIDAR.
The cross sections differ among each other on the
amount of the turbulence of the surface. The CS1 has
more turbulent surface, because it is located nearer to
the confluence as CS2 and CS3. The cross section CS3
is located the farthest downstream from the confluence
and features the least turbulent surface. We did not
notice a significant influence of the surface turbulence
on the amount of rejected measurements for both
methods. We have observed that LIDAR method
expressed more rejected measurements than laser
triangulation method. However, the number of rejected
measurements is heavily influenced by the settings of
the fast camera image processing algorithm, while we
were unable to influence the LIDAR algorithm.
The laser beam of LIDAR is sent and received at the
same angle of mirror rotation. Because of very short
time required for the laser pulse to travel from, reflect
and return back to the LIDAR, the turbulent water
surface is measured as perfectly stationary. Only the
laser light returning from the point of direct impact and
narrow spatial angular section on the radius vector from
the LIDAR to the impact point is detected by the
LIDAR’s photodetector (in Figure 1 shown as dark yellow
part of the LIDAR measurement cross section). The
reflected beams originating from other positions are
therefore rejected.
Reflections of the laser beam to the fast camera are
recorded in a different way. In a single recorded image,
fast camera lens collects all reflected beams, heading in
the direction of the lens. These reflected beams originate
from a larger volume, being illuminated through
successive specular reflections of the water surface.
Therefore the laser triangulation method offers slightly
poorer localisation in comparison with LIDAR ranging.
Fast camera ability to record weak reflected beams is
compromised by the need of high image acquisition
frequency and very short image integration time, leading
to high noise and low image quality.
Direct comparison of single measured points between
LIDAR and laser triangulation shows Figure 8. Three
manually selected typical images recorded by the fast camera
are shown. Red circles show points, detected by LIDAR and
green circles show points measured by laser triangulation
method. Successive specular reflections are clearly visible
on those images as a cloud of multiple dark spots. We can see
that laser triangulation detects approximate centre of brightest
point cloud. Notably, measurements provided by LIDAR are not
in exactly the same locations as the centres of intensities of
reflections (see also Figure 1). The difference is however not
large. We estimate that mismatch is caused by the
difference of observation positions. The reflected beams,
returning to the LIDAR are reflected from different
position than those reflected to the laser triangulation fast
camera. This observation confirms our assumption, that
specular reflections are the most important issue of the
laser ranging measurements.
Current algorithm of the laser triangulation turbulent water
surface detection is based on finding brightest area which
originates from laser beam scattering (see section 4
“Analysis”). Since the image smoothing (Gaussian filter)
is performed before maximum search, it’s filter width has
Laser ranging measurements of turbulent water surface— 6
a major impact on result of detected point. Wider the
filter is, more robust the detection is, but also the spatial
resolution is smaller. So, the selected parameters of the
laser triangulation method represent a compromise
between resolution and spatial noise reduction.
Average intensities of recorded laser triangulation fast
camera images are shown in Figure 9 for all three
measured cross sections. Purely for the sake of clear
demonstration, the averaging was done over 700
acquired fast camera images, which correspond to two
LIDAR scans. Therefore results differ slightly from
results presented in Figure 6, due to much shorter
integration time. For instance, a local increase of water
level height for CS3 on the right is a consequence of
non - stationary local flow turbulence. Second important
information on Figure 9 is depiction of average laser
beam scattering through the droplets (see the middle
image), steep waves (left and right sides of middle and
below images) and entrapped bubbles below the
surface (centre top and middle image). All these
features may be regarded as a near - diffuse reflections
as discussed in Section 1.2., where the width of contour
on image corresponds to the average light penetration
depth into the measured turbulent water surface. Thus,
the wider contour corresponds to more dynamic water
behaviour.
Figure 8. Comparison of LIDAR and image triangulation
methods. Shown are three selected samples of instantaneous
fast camera images for the cross section CS2. Red squares
show LIDAR measurements. Blue squares show
measurement results of image triangulation method.
In contrast to triangulation method image analysis
algorithm we have no access to the algorithm of the
LIDAR signal processing, because we have used a
commercial LIDAR device. It is optimised for
measurements of solid surfaces with diffuse surfaces
where single high contrast laser pulse is usually
detected. For the understanding of LIDAR performance,
it would be necessary to evaluate the time series of
returned LIDAR beam intensity. Comparison with
recorded fast camera images would allow optimizing
LIDAR performance for the purpose of water surface
measurements of turbulent free surface flows.
The experimental setup allows comparison of every
single LIDAR measurement (for each scan and angular
position) with a corresponding laser triangulation fast
camera image, recorded at approximately the same time.
However phenomena comparison is limited because laser
triangulation fast camera images do not show the
turbulent water surface from the same location. The
secondary camera provides only qualitative data. To be
able to deeper understand the series of specular
reflections another fast camera should be mounted in the
same location. Without the use of red filter to screen the
blue diffuse illumination for the secondary camera, the
second fast camera would acquire instantaneous
turbulent water surface images, illuminated with the blue
light. In this new configuration, synchronising both fast
cameras and comparison with LIDAR should provide full
understanding of the scattering process and LIDAR
operation, how and when LIDAR detects turbulent water
surface air/water boundary and the roles of flying
droplets, foam or bubbles below surface for turbulent
open flow surface measurements.
Figure 9. Average intensities of acquired laser triangulation fast
camera images sequence. Above: CS1, middle: CS2 and below: CS3.
5. Conclusions
The paper presents the first to us known comparison of
two laser ranging methods for measurements of turbulent
open flows. We have compared LIDAR and laser
triangulation methods for the case of a 90 ° confluence
flow and high Reynolds and Froude numbers. Analysis
Laser ranging measurements of turbulent water surface— 7
was performed such that LIDAR laser beam was at the
same time used also for laser triangulation with the fast
camera and that both measurements methods were
synchronised in time. We have noticed that:
- performance of both laser ranging measurement
methods is largely comparable for measurements of
turbulent open surface height,
- both laser ranging methods agree well in estimation of
average turbulent open surface height,
- methods do not provide same instantaneous height
measurements for every position on the surface profile,
- LIDAR method feature more rejected measurements
for flows with less turbulent surface, while laser
triangulation method in current implementation provides
more rejected measurements for flows with highly
turbulent surface and
- LIDAR method is localized better than laser
triangulation method.
We used slow secondary camera, located near the
LIDAR, to provide qualitative data of the flow. We
propose to use an additional synchronised fast camera
in the future to better understand the roles of flying
droplets, foam and bubbles below surface for LIDAR
and laser triangulation method performance. Such
configuration would also allow for optimizing of LIDAR
algorithm for measurements of water surface
measurements for turbulent free surface flows.
Acknowledgments: The authors acknowledge the
financial support from the state budget by the Slovenian
Research Agency (Programmes No. P2 - 0392, P2 -
0270, P2 - 0167 and P2 - 0401). REFERENCES
[1] H. Chanson. Hydraulics of Aerated Flows: Qui Pro
Quo?, Journal of Hydraulic Research, 51 (3): 223-243,
2013
[2] M. C. Amann, T. Bosch, M. Lescure, R. Myllylä, M.
Rioux. Laser ranging: a critical review of usual
techniques for distance measurement, Optical
Engineering, 40 (1): 10-19, 2001
[3] J. Shan, C. K. Toth. Topographic Laser Ranging and
Scanning: Principles and Processing, CRC Press, 2008
[4] R. S. Wang, Y. Hu, H. Y. Wu, J. Wang. Automatic
extraction of building boundaries using aerial LiDAR
data. Journal of Applied Remote Sensing, 10 (1): 16-
22, 2016
[5] C. E. Blenkingsopp, M. A. Mole, I. L. Turner, W. L.
Peirson. Measurements of the time-varying free-
surface profile across the swash zone obtained using
an industrial LIDAR. Coastal Engineering, 57: 1059-
1065, 2010
[6] M. J. Allis, W. L. Peirson, M. L. Banner. Application of
LIDAR as a measurement tool for waves. Proceedings
of the Twenty-first International Offshore and Polar
Engineering Conference. USA, Maui, Hawaii,
International Society of Offshore and Polar Engineers:
19-24, 2011
[7] C. E. Blenkingsopp, I. L. Turner, M. J. Allis, W. L.
Peirson, L.E. Garden. Application of LiDAR technology
for measurement of time-varying free-surface profiles in
a laboratory wave flume. Coastal Engineering, 68: 1-5,
2012
[8] P. Westfeld, H. G. Maasa, K. Richter, R. Weiß. Analysis
and correction of ocean wave pattern induced systematic
coordinate errors in airborne LiDAR bathymetry, ISPRS
Journal of Photogrammetry and Remote Sensing, 128:
314-325, 2017
[9] C. Mulsow, M. Schulze, P. Westfeld. An optical
triangulation method for height measurements on non-
stationary water surfaces, International Archives of the
Photogrammetry, Remote Sensing and Spatial
Information Sciences: 213-217, 2006
[10] C. Mulsow, H. G. Maas, P. Westfeld, M. Schulze.
Triangulation methods for height profile measurements
on instationary water surfaces, Journal of Applied
Geodesy, 2 (1): 21-29, 2008.
[11] M. Y. Brazhnikov, G. V. Kolmakov, A. A. Levchenko and
L. P. Mezhov-Deglin. Observation of capillary turbulence
on the water surface in a wide range of frequencies, EPL
(Europhysics Letters), 5 (4): 510-516, 2002
[12] E. Mignot, N. Riviere, R. Perkins, A. Paquier. Flow
patterns in a four-branch junction with supercritical flow.
Journal of Hydraulic engineering, 134 (6): 701-713, 2008
[13] M. M. Pinto Coelho. Exprimental determination of free
surface levels at open-channel junction. Journal of
Hydraulic Research 53, 3: 394-399, 2015
[14] J. D. Nobrega, H. E. Schulz, D.Z. Zhu, Free Surface
Detection in Hydraulic Jumps through Image Analysis
and Ultrasonic Sensor Measurements. 5th IAHR
International Symposium on Hydraulic Structures.
Brisbane, Australia: 1-8, 2014
[15] M. Schwalt, W. H. Hager. Experiments to supercritical
junction flow. Experiments in Fluids, 18: 429-437, 1995
[16] L. S. Nanía, R. Gonzalo, M. Gómez. Influence of
Channel Width on Flow Distribution in Four-Branch
Junctions with Supercritical Flow: Experimental
Approach. Journal of Hydraulic Engineering, 140: 77-88,
2014
[17] H. W. Jensen, S. R. Marschner, M. Levoy, P. Hanrahan.
A practical model for subsurface light transport.
Proceedings of ACM SIGGRAPH 2001: 511–518, 2001
[18] M. Jezeršek, J. Možina. High-speed measurements of
foot shape based on multiple-laser-plane triangulation.
Optical Engineering, 48(11): 113604-1 - 113604-8, 2009