Olivier Blanpain 17/09/2007 - 1 -

High resolution in situ study of multimodal sediment transport processes using

Dynamic Sediment Profile Imagery (DySPI).

Journées des thèses IRSN, 01-04 octobre 2007

___________________________________________________________________________________________

Intitulé du sujet de thèse : Modélisation numérique de la genèse des structures sédimentaires superficielles de la Manche ; application à la modélisation intégrée de l’écosystème.

Olivier BLANPAIN, deuxième année de thèse cofinancée IRSN/IFREMER (depuis le 2 novembre 2005)

Ecole Doctorale Normande Chimie Biologie – Université de Rouen

Robert Lafite (UMR 6143 M2C Rouen) : Directeur de thèse

Philippe Cugier (IFREMER/DYNECO/EB) : Encadrant IFREMER.

Pascal Bailly du Bois (IRSN/DEI/SECRE/LRC) : Encadrant IRSN.

Localisation : première année IFREMER Brest, deuxième année IRSN Cherbourg – Octeville.

___________________________________________________________________________________________

1 PhD context

Bedload sediments and particulate suspended matters play an important part in abilities and organization

of coastal ecosystems in numerous ways:

• They affect trophic web by controlling turbidity which acts on primary production.

• They can fix and carry chemical species (nutrients, pollutants) and radionuclides. Thus, bed

sediments can represent alternatively a sink and a source for dissolved and particulate elements

by exchange with the water column.

• Bed sediments form a wide range of habitat or spawning habitat for fish and invertebrates.

English Channel seabed is characterized by a wide range of sediment mixture composed from mud to

cobbles (Figure 1). Dynamic behaviour of such mixtures depends on the relative ratio of each size class. In

order to improve sediment-transport model accuracy, it is essential to take into account particle size

distributions.

First developments of a non-uniform non-cohesive sediment model have been realised at IRSN by Cugier

(2000). The superficial sediment map of the English Channel (Vaslet et al., 1979) has been digitized

(Struski, 1999) and a granulometric curve has been allocated to each sedimentary facies (Bailly du Bois,

2000; Nozière, 2001). Flume studies with sediment mixture (bimodal sand) have been performed in order

to test transport formulations and to compare model simulations with observations (Olivier, 2004).

The aim of the PhD is to set up a 2DH sediment transport model (suspended and bed load) in the English

Channel, taking into account the sediment sizes heterogeneity. It will be able to simulate heterogeneous

sediment covers. This model will allow to predict pathways and sinking zones of radionuclides fixed on

particles on the one hand, and to simulate their resuspension events on the other hand. In association with

transfer models existing at IRSN, the fate of anthropogenic radionuclides will be assessed in most

compartments of the English Channel ecosystem (water, sediments, and aquatic species).

Olivier Blanpain 17/09/2007 - 2 -

This action will contribute to IRSN mission in order to assess radionuclides dispersion in the environment

following chronic or accidental discharges.

Figure 1: Granulometric distribution map of superficial sediment in the English Channel (Vaslet et al.,

1979; Struski, 1999; Bailly du Bois, 2000; Nozière, 2001)

2 Introduction

English Channel seabed implies to model mechanical behaviour of mixed particles and especially

fine grains in a coarse environment. To deal with this specificity, sediment transport models usually

include two numerical methods able to translate dynamical features:

• the active layer concept makes possible to simulate bed armouring phenomenon;

• the hiding / exposure coefficient allows to tune thresholds of movement according to

heterogeneity degree of the mixed bed.

The need for in-situ data for the adjustment of these numerical computations of physical processes is

essential. White (1998) highlighted that despite the numerous available methods, it is still not possible to

make detailed or accurate field measurements of bedload, suspension of mixed sizes, or suspension very

close to the seabed. The device presented herein, with the addition of appropriate image processing

(Keshavarzy and Ball, 1999), provides a way to investigate in details these transport processes at the

sediment – water interface.

3 The lack for high resolution bedload process studies

Bedload transport is difficult to measure since sediment movement takes place within a few

centimetres of the sea bed. As a result, field studies usually focus on grain and flow parameters governing

average transport rates in order to derive and test expressions for bedload transport. Few in-situ

techniques have been adjusted to estimate average transport processes: 1) Particle tracking method which

offers a solution to investigate transport pathways of a variety of sediment over wide temporal and spatial

Olivier Blanpain 17/09/2007 - 3 -

scales. Black et al. (2007) provides a comprehensive review of the method and highlights its technical

limitations, 2) Ripple progression monitoring using a camera based technique (Kachel and Sternberg,

1971), or an echo-sounder based technique (Bell and Thorne, 1997). It can only be applied where hydraulic

and sedimentary conditions allow ripple formation, and 3) Bedload traps, which according to Dyer (1986),

are of very variable efficiency due to the difficulties of restricting the sampling to the movable layer.

The above methods yield averaged bedload transport hence providing poor resolution of individual or

collective grain motion in time and space. On the other hand, two techniques establish a detailed link

between boundary layer turbulence and sediment mixture dynamics. These are: 1) Self generating noise

measurements due to particles hitting against each other as they move (Thorne et al., 1983/1984).

Threshold of movement, size of the moving grain and instantaneous transport rate can be determined.

This method is most suitable for coarse grains but presents problems in calibrating the acoustic signal, and

2) Video observations. Williams (1990) used this technique to observe gravel transport. He determined

individual transport velocities and distances for 1680 particles whilst he managed to investigate bedload

response to momentarily high bed shear stresses.

4 Dynamic Sediment Profile Imagery (DySPI)

DySPI is a new field device which intends to investigate multimodal sediment transport processes

with a high resolution. It allows characterization of sediment response to turbulent fluctuations in terms of

mode of transport, instantaneous transport rate, threshold of movement, individual grain velocity,

transport thickness, sorting processes and armouring.

4.1 Apparatus and deployment

DySPI is an advancement of an about thirty year’s old device called Sediment Profile Imagery (SPI).

SPI is a remote sensing technique for mapping superficial sediment properties along with observing and

quantifying animal sediment interactions in aquatic systems (Rhoads and Young, 1970). A remotely

operated camera is used to obtain profile photographs of the sediment-water interface. SPI is an effective

technique for the assessment of benthic habitat quality. A multi-parameter organism-sediment index (OSI)

is calculated on the basis of physical, chemical and biological parameters derive from sediment profile

images (Rhoads and Germano, 1986; Nilsson and Rosenberg, 1997). OSI has been defined to interpret

images, and to reduce them to a univariate factor.

DySPI enlarge the scope of SPI by allowing bedload processes to be studied with video imagery.

Indeed, thanks to its streamlined shape, the sediment-water interface remains undisturbed during the

penetration and the main flow is not modified during video acquisition.

An inverted periscope is placed at the middle of a half hull-shaped walking beam (Figure 2). The

periscope consists of an optical mirror mounted at a 45°angle into a box with 2 Perspex face plates. The

Perspex box is filled with clear water and sugar to prevent corrosion of the mirror and to obtain the same

light diffraction as the sea water. A high definition digital video camera (resolution of 1080x1920, 50 half

frames per second) is housed on the rotation axis on top of the mirror. The field of view is centred in order

to see both the sediment vertical section reflected by the mirror and the sea floor directly. It is noted that

the finest grain size to consider is determined by the camcorder resolution. Because video frames are

taken through the periscope box, turbidity of the ambient water does not affect image quality. Light is

provided alternatively by a spotlight to illuminate the entire area of interest and by a light pencil to see a

specific volume of water. The periscope is horizontally sliced through the sediment thanks to a motorised

Olivier Blanpain 17/09/2007 - 4 -

winch that ensures the penetration is slow enough to minimise disturbance of the sediment-water

interface. The penetration depth can be adjusted to a maximum of 10 cm inside the sedimentary layers. A

drag anchor is mounted on the DySPI frame to ensure it is trimmed right in the current direction when it

descends through the water column. Thus, the periscope vertical face plate is parallel to the current

without any disturbance from obstacles upstream.

Figure 2: DySPI and autonomous sensors on the benthic frame

4.2 Video measurements

An appropriate image processing allowed the determination of moving area ratio, size of the

largest particle in motion, instantaneous transport rate and interface evolution. A spatial calibration of

the video has been done at each deployment by diving operators: a millimetric grid was placed

horizontally over the bed of sediment and vertically in the area illuminated by the light pencil. For the

entire set of images, a typical value for pixel resolution is about 125µm/pixel. The software ImageJ

(developed at the National Institutes of Health, USA) is used in combination with Fortran routines to

handle the digital images. Captured pictures are converted in black and white. Pixels are represented by 8

bit integers, ranging in intensity value from 0 to 255. Sediment vertical section view and direct view are

processed in a different way:

• Parameters obtained from the direct view are: 1) proportion of the moving grain surface and, 2)

size of the largest particles to move. A region of interest (ROI) is set based on an analysis of the

brightness distribution. In order to remove non-moving background particles, a difference image is

obtained by the subtraction of two images. Thresholding is applied to the resulting image. After

analysis of the histogram for the ROI, only pixels with a brightness variation over 20% are

considered as moving pixels. Resulting image shows both grains that appear on the last image and

grain that disappear from the previous one (Figure 3). The “AnalyseParticles” function of ImageJ is

used to circumscribe and calculate the area of each particle (in pixel). Thus, with the video

calibration, proportion of moving surface and size of the largest moving particle can be evaluated.

Particle size analyser Cilas 1/5Hz

Periscope box filled with clear water

HD video camera Sony 50Hz

ADV Nortek Vector 16Hz

ADCP Sontek 1/15Hz

Light pencil

Fluorometre 1/5Hz

OBS 1/5Hz

Motorised winch 1000W

Olivier Blanpain 17/09/2007 - 5 -

Time lapse between the 2 original images: 1/25 s

Ratio of moving area: 0.2 %

Size of the largest moving particle: 1.43 mm²

Time lapse between the 2 original images: 24/25 s

Ratio of moving area: 1.4 %

Size of the largest moving particle: 6.86 mm²

Figure 3: Difference images obtained after processing of the direct view part of video

• Parameter obtained from the vertical section is the deposition / erosion rate. Brightness and

contrast are adjusted to enhance the sediment water interface. Then, this one is hand labelled

with ImageJ. Interface position in pixel is obtained by selecting label coordinates with a Fortran

routine. Then, the interface evolution can be plotted (Figure 4) and deposition/ erosion rate

measured.

Figure 4: Water-sediment interface evolution

4.3 Application during SEDHETE field campaign

DySPI has been launched at several stations during spring tide in the Normand-Breton Gulf.

Deployment duration did not exceed a tide cycle. Mooring locations were chosen with bidirectional

currents (according to MARS hydrodynamic model) to make study of sediment structure reorganization

following turn of tide accessible. Before deployment, Shipeck garb samples, video observations and side

scan sonar investigation ensured that seabed was homogeneous around the landing point: flat bed

geometry with sediment characterised by a mixture of size with coarse grains being dominant. DySPI has

been associated with several autonomous optical (OBS, particle size analyser, fluorometre) and acoustic

sensors (ADV, ADCP) on a benthic tetrapod (Figure 2) in order to monitor boundary layer characteristics

simultaneously with video observations. Thus, observed grains dynamics have been linked with high

resolution time series of velocity, pressure, SPM, fluorescence and particle size.

2 cm

Olivier Blanpain 17/09/2007 - 6 -

5 Discussion and conclusions

The device presented herein, with the addition of appropriate image processing, provides a way to

investigate in details transport processes at the sediment – water interface. Threshold of movement, size

of the moving grain, moving area ratio, transport thickness and sorting processes can be studied in

addition to usual SPI parameters (Kennedy, 2006). Furthermore, if DySPI is moored at a bidirectional

current location, study of sediment structure reorganization following turn of tide is made accessible. This

new apparatus enables in-situ sediment processes investigation in a large range of hydro-sedimentary

conditions which are monitored as accurately as during flume experiments.

Data collected from SEDHETE field campaign will allow to characterise mechanical behaviour of mixed

particles. Sediment transport formulation, active layer thickness and hiding / exposure coefficient will be

tested in the model and chosen to fit the observations. Then, movement of the sedimentary cover in the

English Channel will be simulated taking into account specificities of mixed size facieses.

References Bailly du Bois P., 2000. Représentation continue des classes granulométriques des sédiments superficiels de la Manche à partir des travaux de Larsonneur (1971). Rapport IRSN/DPRE/SERNAT/ 2000-23, 14 p. Bell P.S. and Thorne P.D., 1997. Measurements of sea bed ripple evolution in an estuarine environment using a high resolution acoustic sand ripple profiling system. Proceedings of Oceans ’97, Halifax, Nova Scotia, MTS/IEEE. IEEE Ocean Engineering, Piscataway, N.J., 339 - 343. Black K.S., Athey S., Wilson P. and Evans D., 2007. The use of particle tracking in sediment transport studies: a review. Balson, P. et Collins, M.B., (Eds.). The Geological Society (Special Issue). Coastal and Shelf Sediment Transport. Cugier P., 2000. Développement d’un modèle numérique multicouche hétérométrique pour la simulation du transport sédimentaire en Manche ; Caractérisation des principaux processus. Rapport scientifique, IPSN/DPRE/SERNAT/2000-24, 97 p. Dyer K.R., 1986. Coastal and estuarine sediment dynamics. John Wiley and Sons, Chichster, 342pp Kachel N.V. and Sternberg R.W., 1971. Transport of bedload on ripples during an ebb current. Marine Geology, 10, 229 – 244. Kennedy R., 2006. Introduction. Journal of Marine Systems, 62, 121 – 123. Keshavarzy A. and Ball J.E., 1999. An application of image processing in the study of sediment motion. Journal of Hydraulic Research, 37, 559 - 576. Nilsson H.C. and Rosenberg R., 1997. Benthic habitat quality assessment of an oxygen stressed fjord by surface and sediment profile images. J. Mar. Syst. 11, 249–264. Nozière F., 2001. Caractérisation granulométrique des faciès sédimentaires de la Manche dans le but de contribuer à l'élaboration d'un inventaire quantitatif des radionucléides dans les sédiments de la Manche. Rapport de stage IPSN-LERFA - INTECHMER 64p. Olivier C., 2004. Développement d’un modèle numérique multicouche hétérométrique pour la simulation du transport sédimentaire en Manche ; détermination de l’équation de transport par des mesures de canal en laboratoire. Rapport scientifique, IRSN/DEI/SECRE/2004-2, 98 p. Rhoads D.C. and Young D.K., 1970. Influence of deposit-feeding organisms on sediment stability and community trophic structure. Journal of Marine Research, 28, 150 - 178. Rhoads D.C. and Germano J.D., 1986. Interpreting long-term changes in benthic community structure— a new protocol. Hydrobiologia 142, 291–308. Struski C., 1999. Création d'une base de données cartographiques de la répartition des espèces vivantes et de l'exploitation du milieu marin de la manche. Rapport de stage IPSN-LERFA 66p. Thorne P.D., Heathershaw A.D. and Troiano L., 1983/1984. Acoustic detection of seabed gravel movement in turbulent tidal currents. Marine Geology, 54, 43-48. Vaslet D., Larsonneur C., Auffret J.P., 1979. Les sédiments superficiels de la Manche 1/500000. Carte géologique de la marge continentale Française. Bureau de Recherches Géologiques et Minières, Centre National pour l’Exploitation des Océans. White T.E., 1998. Status of measurement techniques for coastal sediment transport. Coastal Engineering, 35, 17 - 45. Williams J.J., 1990. Video observations of marine gravel transport. Geo-Marine Letters, 10, 157 - 164.