UAV-Based Analysis of Armour Rock

Granulometry and Hydraulic Stability

Alisson Villca

, Muhammad Ali Sammuneh

, Poupardin Adrien

, Jena Jeong

Rani El Meouche

and Georges Chapalain

Institut de Recherche (IR), ESTP Paris, 94230 Cachan, France

Cerema Risques Eau Mer, Margny-Les-Compiègne, 60280, France

Keywords: Granulometry, UAV, Photogrammetry, Image Processing, Stability, Monitoring.

Abstract: Dikes worldwide play a crucial role in mitigating flooding risks. Often, armour rocks are placed within the

wave impact zone to protect the dike from wave loading. To ensure a dike is in optimal condition the

assessment of the hydraulic stability of armour rocks is necessary. This study presents a Granulometric

analysis technique, which is based on UAV photogrammetry and optical digital granulometry to evaluate

the spatial distribution and possible variations in time of armour rocks granulometry and hydraulic stability.

This is a new non-invasive technique with which spatial, temporal changes can be studied. Our study area is

located in Camargue, south of France. This low-laying region, exposed to multiple storms, is among the

most endangered zones by sea level rise. We concluded that monitoring of the dike is possible using this

technology the optical granulometric analysis could be performed on UAV images. We conducted

granulometry distribution calculations for armour rocks, even when they were covered with moss. Our

findings show the spatial variation of granulometry along the dike. In specific areas of interest where

hydraulic stability was assessed, based on the granulometry results, we have found areas with low hydraulic

stability that need to be investigated more thoroughly.

https://orcid.org/0009-0005-8274-9940

https://orcid.org/0009-0008-4977-7386

https://orcid.org/0000-0001-5063-6638

1 INTRODUCTION

The grading curve analysis along with characteristic

sizes of non-cohesive materials is a traditional and

important method that can provide important

parameters for hydraulic modelling. To obtain such

information laboratory and in situ tests exist.

Laboratory sieving involves time-consuming and

effort-intensive activities just the same in situ

techniques, which uses a grid system to measure

single pebbles (Wolman, 1954) or uses instead a

sampling line (Fehr, 1987), can be just as time

consuming.

To overcome these limitations tools for

automatic optical granulometry were developed.

These non-intrusive, low cost methods can obtain

grain size distributions for non-cohesive materials

(Graham et al., s. d.) (Detert & Weitbrecht, 2012)

(Buscombe, s. d.). For the granulometric, analysis of

riverbed depositions a matlab-based tool was

developed, BASEGRAIN, that can recognize,

classify and analyse grain images (Detert &

Weitbrecht, 2012).

Basegrain analysis results can still be meaningful

despite being used with photos taken in suboptimal

conditions. (Detert & Weitbrecht, s. d.)

The potential of the combination of Unmanned

Aerial Systems (UAS) technology and optical

granulometry was studied in previous works. A peak

discharge estimation in the town of Mandra, Greece

(Andreadakis et al., 2020). The median particle size

derived from the granulometric curves were used to

estimate run-off. They compared the results of run-

off estimations calculated with data from UAV and

GNSS surveys which showed minimal difference.

(Lagogiannis & Dimitriou, 2021) combined UAV-

sensed data with empirical hydraulic equations to

produce accurate discharge estimations in 10 out of

17 sites. The estimation of the manning coefficient

was based on percentile of particles (d

, d

and d

)

Villca, A., Sammuneh, M., Adrien, P., Jeong, J., El Meouche, R. and Chapalain, G.

UAV-Based Analysis of Armour Rock Granulometr y and Hydraulic Stability.

DOI: 10.5220/0012703700003696

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 10th Inter national Conference on Geographical Information Systems Theory, Applications and Management (GISTAM 2024), pages 62-70

ISBN: 978-989-758-694-1; ISSN: 2184-500X

derived from optical granulometric curves. Optical

granulometry was applied to riverbeds in Japan, the

accuracy of the optical method was compared to

field measurements showing great coincidence

except for the finer grains where the resolution of

the image plays a key role (Kadota et al., s. d.).

The aim of this study is to combine these two

technologies: UAV photogrammetry and digital

optical granulometry to carry out the monitoring of

an earthen dike’s armour rocks and establish the

procedure for the treatment of UAV imagery to

execute the granulometry analysis and ultimately

evaluate the hydraulic stability of the armour rock

layers. In section 2 we will see the study-site.

Section 3 is divided into three main sections UAV

photogrammetry, optical granulometry and hydraulic

stability analysis. In section 4 the results are

discussed and finally in section 5 conclusions are

drawn.

2 CASE STUDY

Located in the south of France, Camargue is a low-

lying region, exposed to multiple storms, see figure

1. It is among the most endangered zones by rise in

sea level (Pörtner & Roberts, s. d.).

Figure 1: Location of study area in south of France. In the

top-left view France is highlighted. The red area

represents sentinel 2 tile 31TFJ projected on the UTM –

zone 31N (Universal Transverse Mercator). The top-right

view shows the tiled sentinel-2 image (date: 28 February

2023) and the red box highlights the zoomed area of the

bottom view. The dike “Quenin” length is shown in red.

The two-kilometer long earthen dike is located to

the west of the Pharaman lighthouse. During storm

surges it has to protect the salt marshes from

inundation.

3 MATERIALS AND METHODS

3.1 UAV Photogrammetry

3.1.1 Drone Specifications

In this study a DJI Phantom 4 Advanced was

deployed with a 1-inch 20-megapixel CMOS sensor.

The maximum flight time being 30 minutes limited

the survey area, thus the total area was divided into

three survey zones. The mission took place on 12

April. Table 1 summarizes the principal

characteristic of the sensor.

Table 1: Sensor specifications.

Sensor 1” CMOS, Pixels: 20M, Size:

12.83mmx8.55mm

Lens FOV 84° 8.8 mm/24 m

Ima

e size 3:2 As

ect Ratio: 5472 × 3648

Filet

eJPEG

Ground sampling distance (GSD) can be

approximated by the following formula:

GSD =













)

Where 𝑆



and 𝐼𝑀



is the sensor and image

width respectively. The aircraft height is 𝐻 and 𝑓 is

the camera focal length. In our case, GSD is

approximately 31.1mm for a 120m flight height. The

selected GSD should allow detection of the armour

rocks with mean D

values from 250 to 850mm.

Following a rule of thumb so that every grain is

represented by a minimum number of pixels b>10px

(Detert & Weitbrecht, 2012) we note that the for the

lower D

value the number of pixels is lower than

10px, but for the larger diameter the condition is

met.

3.1.2 Flight Planification

During the data acquisition phase, Figure 2 two

drone surveys were done for the April mission. The

first survey took nadir images at 70% overlap along

the dike (149). The second mission took nadir

images with 30% overlap in an area 900m from the

coast to estimate shallow bathymetry (238).

It is necessary to verify in situ if the required

overlap is achieved. The wind may affect the drone

speed and reduce the number of overlapped images

in certain zones. To overcome this inconvenience a

combination of the images from both missions was

used. From the overlap map, it can be seen that an

offset distance from the flight area and study area

UAV-Based Analysis of Armour Rock Granulometry and Hydraulic Stability

must be foreseen to ensure more than five

overlapped images in the study area.

3.1.3 Ground Control Points

Ground Control Points (GCP’s) and Checkpoints

(CP’s) were marked with a red spray following a

zigzag pattern on the dike crest see figure 3

coordinate measurement were made on the same day

with a GNSS multiband antenna IP 67 and a ZED-

F9P RTK receiver, in total 13 points were measured.

Figure 2: a) GCP’s and CP’s distribution along the dike,

b) Combined missions overlap map.

The centipede GNSS network, an open source

collaborative network of more than 300 homemade

RTK bases across France, was used. It allowed us to

benefit from RTK centimetre positioning for free (Le

Reseau Centipede RTK, s. d.). The smartphone SW

maps application was used to collect, store and

visualize the coordinates (SW Maps - Mobile GIS,

s. d.). The consistency and reliability of this

equipment, software and network has been validated

in previous studies. (Sammuneh et al., 2023)

Figure 3: GCP’s following a zigzag pattern on the crest of

the dike.

3.1.4 Photogrammetric Processing

The data processing is done using Pix4D (version

4.5.6), were all GCP’s and CP’s targets were

manually marked at their centers to ensure accurate

geo-referencing. For accuracy assessment, Root

Mean Square Error (RMSE) is used, see table 2, for

a combination of both surveys with 178 images with

70% and 30% overlap. Previous studies have

evaluated the level of accuracy that can be achieved

with UAV equipment (El Meouche et al., 2016)

(Jiménez-Jiménez & Ojeda-Bustamante, 2021). Just

the same vertical error in our case (Z) is significantly

larger than planar error (X, Y).

Table 2: Results of the photogrammetric block adjustment.

Survey April 2023.

RMSE X(m) Y(m) Z(m)

GCP’s 0.005 0.011 0.017

CP’s 0.070 0.030 0.347

The workflow diagram in figure 4 shows the

different steps from data acquisition, data processing

and data analysis. The pix4D software proposes a

photogrammetric technique based on structure from

motion algorithm based on three stages. First, the

position of the images is extracted from the EXIF

metadata and specific features called key-points are

found on every image in order to find matching

points in overlapped images. This makes it possible

to know the camera position and orientation to carry

out an internal and external camera calibration

during the bundle block adjustment. Then automatic

tie points are detected in the image to compute its

3D position and point cloud densification allows

more tie points to be created. Finally, the products

obtained include the digital surface models, ortho-

image and reflectance maps.

Rematch and optimize was used to compute

more matching points between images and therefore

more automatic tie points through re-optimization of

internal and external camera parameters. The

internal camera parameters from the EXIF metadata

are focal length, two principal point offsets (𝑥,𝑦),

three radial (𝑅



,𝑅



,𝑅



) and two tangential (𝑇



,𝑇



)

distortion coefficients. The external camera

parameters are defined by the position of the camera

projection center (𝑇



,𝑇



,𝑇



) and the rotation matrix

defined by the camera orientation

( 𝑅



(

𝜔

)

,𝑅



(𝜙),𝑅



(𝜅) ). (Professional

Photogrammetry and Drone Mapping Software, s.

d.)

3.2 Digital Optical Granulometry

Digital optical granulometry is a non-invasive

technique based on the analysis of digital images

after a scaling factor is applied. In this study the

Basegrain software is used to carry out the

granulometry analysis (Detert & Weitbrecht, 2012).

Areas of interests were chosen along the dike’s

longitudinal axis every 100m approximately. At the

same time 3 zones of interest were defined along the

transversal axis of the dike, see figure 5. Zone A is

the area in the inner slope with small granular

)

GISTAM 2024 - 10th International Conference on Geographical Information Systems Theory, Applications and Management

Figure 4: Workflow of photogrammetric and optical

granulometry analysis processing.

material compared to the armour rocks. Zone B is

the outer slope closest to the crest-dike. Finally, the

C zone has armour rocks, covered with moss, in the

wave impact area.

Figure 5: 20 Areas of interest along the dike and three

zones of interest, zone A (inner slope), zone B (outer slope

near the crest), zone C (outer slope near the wave impact

zone). Universal Transverse Mercator, Zone 31N

coordinates.

3.2.1 Basegrain Software

Basegrain has multiple steps to obtain the

granulometric curves, see figure 6, first a procedure

to detect the interstices between grains by double

grayscale threshold (1), to determine further

interstices a morphological bottom hat transform is

applied (2). Then the canny and sobel method is

used to find grain edges (3) then the separation of

single grains is made by watershed transform (4).

Next, grain areas are measured and replaced with

ellipsis of the same normalized second central

moments (5). Finally, the analysis of results follows

the Fehr’s approach. (Detert & Weitbrecht, 2012)

Figure 6: The result of the different Basegrain steps (1 to

5) for zone of interest A9.

3.2.2 Basegrain Parameters

In this study for zones A and B the default

parameters were sufficient to detect a great majority

of the armour rocks, however for zones C some

changes had to be made as these armour rocks were

covered with moss and the automatic detection with

default parameters worked poorly as seen in figure

7. This is because in locations with partly wetted

stones the detection algorithm separates the armour

rocks in various parts.

Figure 7: Automatic armour rock size automatic detection

after steps 1 to 5 for zone C9, a) Automatic grain detection

with default parameters, b) with modified parameters.

The modified parameters include for step (1) the

blocSizG (block size gray threshold) and the

facgraythr1 (factor gray threshold 1). The first is

the size of the block in which Otsu’s thresh value is

determined and was modified from 32 to 4. The

second is the multiplier that determines definite

)

UAV-Based Analysis of Armour Rock Granulometry and Hydraulic Stability

interstices; it was changed from 0.8 to 0.5. For step

(2) the puxCutoff (bottom-hat interstices) was

modified from 1 to 8. At least 5 minutes manual

splitting, merging or removal was necessary for all

images.

Once the granulometry curves have been

obtained the median sieve size D

, can be calculated

for every area of interest in zone C. The median

sieve size D

and the median nominal diameter D

n50

are proportional, 𝐷



= 0.84𝐷 experimentally

determined for different rock types and grading.

The median mass M

and median size sieve D

are

related using the conversion factor F

= 0.60 (Rock

Manual 2007)

= 𝜌 ×0.6 𝐷





(2)

3.3 Stability Criteria for Armour Units

After various model tests with two-diameter thick

layer of armour rocks a formula was developed by

(Van der Meer 1998) to assess the stability of rock

protection under wave attack. Because 𝜉



<𝜉



we will use the formulae for plunging waves (5)

𝜉





6.2𝑃

.

√

tan ∝





.

(3)

𝜉



=𝑡𝑎𝑛𝛼/(











))

.

(4)









=6.2 𝑃

.















.

𝜉



.

(5)

For the breaker parameter 𝜉



, 𝛼 is the dike slope

angle, 𝑇



is the mean period and 𝑔 is the

gravitational constant. Where𝐻



, is the significant

wave height at the toe of the structure, Δ =

𝜌



/ 𝜌



− 1 is the dimensionless relative buoyant

density of the armour rocks. Where 𝜌



and 𝜌



are

the densities of rock and seawater Δ is around 1.58

for granite in seawater.

𝑃, is the notional permeability factor, 𝑆



the

damage level, 𝑁



the wave number and 𝜉



called

the breaker parameter. P represents the influence of

the permeability of the structure on the stability of

the armour layer, in this case we will assume P =

0.5. This means the armour rock layer has a

thickness of at least 2𝐷



on top of the core

material (see figure 8). The number of waves 𝑁



depends on the storm duration, in our case 6 hours.

Figure 8: Notional permeability factor for various

structures. (Van Der Meer, 1988).

The damage level 𝑆



depends on the slope angle

(see figure 9) of the structure and takes into account

settlement and displacement a physical description

would be the number of cubic stones with side 𝐷



eroded within a 𝐷



−wide strip of the structure.

(Van Der Meer, 1988). The slope angle in our case

is between 20° and 50°, a slope of 1:1.5 is the

closest value corresponding to a 34° slope angle.

Figure 9: Limits of Sd for a two-diameter thick armour

layer. (Van Der Meer, 1988).

For the local significant wave height at toe of the

structure 𝐻



we will use the results of a previous

study where one dimensional waves were

propagated from deep water taking into account the

actual bathymetry with Tomawac. A set of

simulations were run with varying boundary

significant wave height 𝐻



= 3m, 5m, and 8m,

initial mean water level η = 0.2m, 0.4m, 0.5m, 0.8m,

1.1m, 1.5m and frequencies 𝑓 = 0.187 Hz, 0.147 Hz,

0.117 Hz. (Paul et al., 2020)

GISTAM 2024 - 10th International Conference on Geographical Information Systems Theory, Applications and Management

4 RESULTS AND DISCUSSION

4.1 Granulometry Results

The final graph analysis chosen to represent the

results were the quasi-sieve throughput (qi[-]), based

on the top view b axis by number on a logarithmic

scale. The Basegrain derived grain distribution for

the different zones of interest are shown in figures

10, 11 and 12. For zone A, the inner slope area, the

regularity of the grain size distribution is evident

along the dike.

Figure 10: Granulometric curves for zone A, located in the

inner slope of the dike.

For zone B, the outer slope zone near the crest,

there is a strong dominance of coarser material, at

the end and the beginning however, the finer grains

percentage increases. Finally, for zone C, the outer

slope near the wave impact area, there is dominance

of coarser material along the dike except for some

areas in the middle 1, 2, 4, 7, 8, 11 and 13.

For zone C, granulometry is very heterogeneous

already some areas of interest stand out at the

beginning and end of the dike, areas of interest

number 4, 7, 11 and 13 are notable for its low

median diameter value 𝐷



and percentiles values in

general. The areas of interest stop at 18 because in

this location the wave impact area has no armour

rocks but sand.

Figure 11: Granulometric curves for zone B, located in the

outer slope of the dike near the crest.

Figure 12: Granulometric curves for zone C, located in the

outer slope of the dike near the wave impact area.

In figure 13 the estimated diameters of the coarse

fraction grain percentiles

, D

, and D

and indicate the armour rock size for

a particular percent finer value. For zone A, the

homogeneity of detected granulometry is again

confirmed while for zone B this homogeneity is

present only in the central part of the dike both the

beginning and end show much greater quantities of

fine material.

UAV-Based Analysis of Armour Rock Granulometry and Hydraulic Stability

Figure 13: Grain size percentiles values D

, D

, and D

, estimated using Basegrain for

zones A, B and C versus the areas of interest in the x axis.

A limitation of optical granulometry technique in

our case for zone A is that spatial resolution is not

sufficient to represent every grain main axis b with

at least 10 pixels. For zones B and C this condition

is met. Another limitation is that wetted or moss-

covered rocks will require the tuning of Basegrain

parameters as we did and some manual editing.

However, a huge advantage is that both spatial and

temporal monitoring of granulometry can be done

with relatively low cost and time.

4.2 Hydraulic Stability Results

To study the stability of the armour rock layer in

zone C the significant wave height for the following

areas is available 4, 5, 6, 7, 8, 10, 11, 12, 13 and 14

from a previous study. The hydraulic stability results

are shown in figure 14.

The significant wave heights at the toe of the

dike 𝐻



depend on the boundary wave height 𝐻



initial mean water level 𝜂 and position along the

dike (areas). From figure 14a) where 𝐻



3𝑚,𝑇 = 5.34𝑠 damage levels for all areas are below

initial damage limit except for areas 11 and 13 for

all initial still water levels 𝜂 = 0.2−0.8𝑚. Let’s

remember that areas 4, 7, 11 and 13 have low

median diameter values. As boundary wave heights

increases in figure 14b) 𝐻



= 5𝑚,𝑇 = 6.80𝑠

other areas such as 4, 7, 8 and 14 now surpass the

initial damage limit value (𝑆



<2) for initial still

water levels 𝜂 = 0.8 − 1.1𝑚. But only areas 11, 13

and 14 go beyond failure damage level (𝑆



8).

Figure 14: Significant wave height at toe of the dike 𝐻



(left vertical axis), damage levels 𝑆



along the dike (right

vertical axis), for areas of interest 4, 5, 6, 7, 8, 9, 10, 11,

12, 13 and 14 (upper horizontal axis) for varying initial

still water levels 𝜂 (see legend). The coordinate of the area

of study (UTM Zone 31N) are in the lower horizontal axis.

Finally in figure 14c) where significant wave height

at boundary increases to 𝐻



= 8𝑚,𝑇 = 8.54𝑠 the

areas going beyond failure damage level are adding

up 4, 7, 8 and 12 for high initial still water levels

between 𝜂 = 0.8 − 1.1𝑚 . The areas remaining

under initial damage level under all circumstances

are 5, 6 and 10. Notably area 14 reaches

)

GISTAM 2024 - 10th International Conference on Geographical Information Systems Theory, Applications and Management

intermediate damage levels for low 𝜂 = 0.4 − 0.5𝑚

values. Damage levels exceeding failure levels

reaching 15-20 values mean an S shaped profile is

developing this is the case of areas 11 and 13

for 𝐻



=3−5𝑚. The number of areas increases

to 6 (areas 4, 7, 8, 11, 13 and 14) when 𝐻



=8𝑚.

5 CONCLUSIONS AND WAY

FORWARD

This study presents a non-invasive technique of

granulometric analysis based on UAV-based

photogrammetry and optical digital granulometry in

order to evaluate the stability of its armour rock

layer on the wave impact zone. This technique is

tested on an earthen dike located in the south of

France were monitoring of an earthen dike exposed

to storms is necessary.

Measurements were obtained in April 2023, the

images were processed in pix4D to obtain a

georeferenced ortho-image then the Basegrain tool is

used to analyse armour rock size properties in 20

areas of interest along the dike, for three zones of

interest A (inner slope), B (outer slope near the

crest) and C (outer slope in the wave impact zone).

The adjustment of some software parameters

allowed automated detection of armour rocks

covered with moss.

Once the grain size distribution is obtained the

spatial variation along the dike was evaluated. The

median sieve size by a conversion factor let us

approximate the median nominal diameter to

evaluate the stability of the armour rock layer for

different sea-states.

The short time application and flexibility of the

UAV and optical granulometry, in comparison with

traditional methods, makes this approach an

effective tool for approximation of granulometry and

stability of armour rocks. The ability of multiple

data collection offers the potential of spatial and

temporal monitoring of the dike.

Several assumptions were made to apply the van

der Meer formulas, parameters like, notional

permeability factor, Number of waves, storm

duration, and dike slope. Looking ahead, this study

aims examining the variability of the different

parameters assumed to be constant along the dike.

REFERENCES

Andreadakis, E., Diakakis, M., Vassilakis, E.,

Deligiannakis, G., Antoniadis, A., Andriopoulos, P.,

Spyrou, N. I., & Nikolopoulos, E. I. (2020).

Unmanned Aerial Systems-Aided Post-Flood Peak

Discharge Estimation in Ephemeral Streams.

Buscombe, D. (s. d.). SediNet : A configurable deep

learning model for mixed qualitative and quantitative

optical granulometry.

Detert, M., & Weitbrecht, V. (s. d.). Determining image-

based grain size distribution with suboptimal

conditioned photos.

Detert, M., & Weitbrecht, V. (2012). Automatic object

detection to analyze the geometry of gravel grains – a

free stand-alone tool.

Fehr, R. (1987). Einfache Bestimmung der

Korngrössenverteilung von Geschiebematerial mit

Hilfe der Linienzahlanalyse. Schweizer Ingenieur und

Architekt, 105(38), 1104‑1109. https://doi.org/10.5169

/seals-76710

Graham, D. J., Rice, S. P., & Reid, I. (s. d.). A transferable

method for the automated grain sizing of river gravels.

Jiménez-Jiménez, S. I., & Ojeda-Bustamante, W. (2021).

Digital Terrain Models Generated with Low-Cost

UAV Photogrammetry : Methodology and Accuracy.

Kadota, A., Asayama, C., & Ndwambi, I. D. (s. d.). Image

analysis of grain size distribution around area of sand

deposition.

Lagogiannis, S., & Dimitriou, E. (2021). Discharge

Estimation with the Use of Unmanned Aerial Vehicles

(UAVs) and Hydraulic Methods in Shallow Rivers.

Le Reseau Centipede RTK. (s. d.). Centipede RTK.

Consulté 17 mai 2023, à l’adresse https://

docs.centipede.fr/

El Meouche, R. E., Hijazi, I., Poncet, P., Abunemeh, M.,

& Rezoug, M. (2016). UAV PHOTOGRAMMETRY

IMPLEMENTATION TO ENHANCE LAND

SURVEYING, COMPARISONS AND POSSIBILITIES.

Paul, T., Lutringer, C., Poupardin, A., Bennabi, A., Jeong,

J., & Sergent, P. (2020). Wave overtopping and

overflow hazards : Application on the Camargue sea-

dike.

Pörtner, H.-O., & Roberts, D. C. (s. d.). Climate Change

2022 : Impacts, Adaptation and Vulnerability.

Professional photogrammetry and drone mapping

software. (s. d.). Pix4D. Consulté 17 mai 2023, à

l’adresse https://www.pix4d.com/

Sammuneh, M. A., El Meouche, R., Eslahi, M., &

Farazdaghi, E. (2023). Low-Cost Global Navigation

Satellite System (Low-Cost GNSS) for Mobile

Geographic Information System (GIS). In M. Ben

Ahmed, A. A. Boudhir, D. Santos, R. Dionisio, & N.

Benaya (Éds.), Innovations in Smart Cities

Applications Volume 6 (p. 105‑117). Springer

International Publishing. https://doi.org/10.1007/978-

3-031-26852-6_10

SW Maps—Mobile GIS

. (s. d.). Consulté 1 février 2024, à

l’adresse http://swmaps.softwel.com.np/

UAV-Based Analysis of Armour Rock Granulometry and Hydraulic Stability

Van Der Meer, J. W. (1988). Deterministic and

Probabilistic Design of Breakwater Armor Layers.

Journal of Waterway, Port, Coastal, and Ocean

Engineering, 114(1), 66‑80. https://doi.org/10.1061/

(ASCE)0733-950X(1988)114:1(66)

Wolman, M. G. (1954). A method of sampling coarse

river-bed material. Eos, Transactions American

Geophysical Union, 35(6), 951‑956. https://doi.org/

10.1029/TR035i006p00951

GISTAM 2024 - 10th International Conference on Geographical Information Systems Theory, Applications and Management