snift ei al, 199

FISH CAGE PHYSICAL MODELING FOR SOFTWARE

DEVELOPMENT AND DESlGN APPLICATIONS

M. Robinson Swift

M ichael PalczynskiKenneth Kestler

Derek Miche in

Rarbaros Celikkol

University of New HampshireDurham, NH 03824

c-maih rnrswiftOchrista.unh.edu

and

Michael Gosz

Illinois Institute of TechnologyChicago, IL 606 1 6

ABSTRACT

Fish cage response to waves was investigated using a physical model in the Umversity of Vew Hampshirewave tank. The tank was built with below-waterline windows placed for convenient observation of moored cagemodels. Cage motion was measured using an opncal system comprised of a high resolution video camera, a framegrabber, and a computer with expanded random access metnory RAlVl!, Targets on the cage, consisting of twoblack painted dots on a white background, were tracked usmg image processing software. The system was,therefore, noninvasive to the fluid environment and did not aher the cage inertial characteristics. Specificexperiments werc done in support of a coinputer modeling effort which has resulted in a finite element programfor fish cage dynamics. Experimental data was obtained using a physical model reduced in complexity in orderto focus on basic parameters. Comparison of rank data with coinputcr predictions indicated that the computersimulation reproduced the fundamental features of the observed cage motion.

IN TROD U CT ION

Physical and computer models are essentialtools for the design of offshore net pen systems.To avoid failure, net pens and their inoorings mustbc engineered to withstand both severe storm eventsand the cutnuiative effects of long-term wave andcurrent loading. We developed methods for testingphysical models in the new University of NewHampshire UXH! wave tank. The experimentalmethodology was then used to generate data forcomparison with recently developed finite elementcomputer models of fish cage response to wavesand current,

The new UNH wave tank was designedand built with offshore aquaculture applications inmind. The tank itself and the building housing thisand other facHities were constructed in l 994, while

the wavernaking system was added in 1996 asdescribed by Washburn �996!. The next. step wasto incorporate a measurement system for physicalmodel motion response, In the study described herc,this need was realized using an optical systein. Thisstrategy was chosen because it offered precisionmeasurements without altering the fish cagedynarnlca.

The physical modeling approachcomplemented the UlieH finite element computerprogramming effort which resulted in a net pendynamics program. As demonstrated by Gosz etal. �996!, the program can be used to predict cagemovement and structural loads for user-specifiedwave and current environments To increase

confidence in its predictions, however, it wasdetermined that an experimental program shouldbe set up to generate specialized, empirical cage

Test Object Computer

Figare t, Schematic of optical measttremcnt systt.'m.

motion data for comparison with the finite elementfnode] predictions, The objectives of this work may.therefore, bc summarized as:

~ Deve]opment of test tank methods forfish cage experiments in the UNHwave tank;

e lrnplernentation of an optical systemfor cage motion measurement;

~ Obtaining specialized data in supportof nct pen coinputer modeling;

~ Comparison with predictions frotn thcexi~ting UNH finite element program.

These objectives were addressed makingUse of unique features of the tank. The opticalmeasurement system was positioned opposite bui]t-in observation windows located halfway along thelength of the tank. Fish cage physical tnodels wercthen conveniently tnoored for clear viewing. Forthe software developtnent application, the cagephysical model was simp]ifted to focus on inajorcomponents and basic dynamic processes. Thecotrtputer progratn was applied directly to thephysical cage tnodel at its actual size, with nopotentia]ly error-producing changes in scale. Theeva]nation was done by comparing displacement

as a function of time for key points on the cage,

L'NH WAVE TANK

The wave tank is 36.6 m long, 3,66 tn wide,and 3.05 m deep. l t is usual! y filled with 2.44 tn ofwater. A tow carriage is supported and cable-driven along a single main rail on one side g!amel],l 996k A lightweight, protected outrigger supportsthe carriage on the opposite s ide wh ich is reservedfor observers. A hydraulically driven, computercontrolled, f]ap-type wavetnaker is at one end Washburn, ]996!. Software allows thc user torun a regular wave ot' specified height andfrequency or a random sea of specified spectrum.Waves are dissipated at the opposite end using avertical "beach' consisting of vertical layers ofgeotechnica] cloth suspended from an angledfiberglass frame,

Midway down the observer side, a pita]lows access to two side windows in the wall-one covering the waterlirie and upper water co]umnand the second placed just above the f]oor of thetank, A mid-width floor window can also be usedfrom a tunnel beneath the tank which is entered

Swift et aL 201

!Camera captures image. Image is transferred

to computer and st ored.

Figure 2. Information flow in the ofnicaI measurement sys em.

Position information generated.

from the base of thc ptt. These windows are veryconvenient for viewing the net pen systemexperiments and allow optical monitoring of motionvariables.

Optical measurement systemAn optical rneasurernent system for

determining test object motions was developed totake advantage of the observation windowopportunities. By using a noninvasive opticalsystem, no dynamic altering sensors are attachedto the test object. Figure 1 shows how images ofthe test object are captured by a camera and fedinto a computer for analysis. The essentialcomponents of the present UNH systcrn seeMichelin and Stott, 1996! include:

A high resolution, black and whitePulnix video camera which can operateat 30 frames/second;

~ A frame grabber to tran sf'er the imagesto a computer;

~ A personal computer with expand&random access memory RAM!,

~ Software to analyze the storedsequence of images,

The steps involved in motion rneasurernentbegin, as indicated in Figure 2, with establishing atarget on the test object, Two small black dots on

Software processes images.

a white background are painted on thc test object,Horizontal, vertical, and angular changes planarmotion! can be inferred from the movement of thespots. The UNH system has been successful inresolving the gray-scale contrast between black dotand light background, so the use of potentially error-producing light sources on the test object isunnecessary. The camera captures a sequence ofimages and transfers each fraroe, via the framegrabber, to the cotnputer for temporary storage.Later, specially written software is used to searcheach frame for the gray-scale differenc indicatingthe presence of the target dots, Dot position as afunction of frame number converted to time! isthen used to calculate test object linear and angulardisplacernent components as a function of time.

Model testingWhile the optical position measurcmcnt

system is adaptable to any fish cage model, thepresent study made use of a special model to obtaindata for comparison with finite element computerpredictions. This skeleton model consisted only ofa rectangular parallel-piped structural frame, abridle, and a single mooring linc see Fig, 3!. Thus,the comparison between empirical data andcomputer predictions represented a focusedevaluation of basic fluid mechanic processes

STILL vAT ER LEVEL BABA V*VE TA»K Vi»BPV099Am

O ASSi

I>RD699

BLACK DDI, e ss RIA,CE>>TEAE» m BcvEL BDT>> SIRES>

IT>>>Ei TB~ BI*

>RSITRB>C LINE

202 t'JTAIR Technical Report 1Vo. 2t>

Figure 3. Ske!etoit fish Cage physical mt>de1.

invulving the main net peri COinpoitents. Uponsati sfactory validation, further complexities in theform of netting and small appendages will be addedin future work,

Before wave tank testing, preliminaryexperiments on the physical model were carried outin the UNH recirculating flurne, This 12.19 rn long,1.22 m wide, and 1.22 m high facility provided asteady current environment enabling the static dragcharacteristics of the tnodel to be measured. Afterapplying the finite element model to this case andobtaining a satisfactory comparison, the physicalmodel was deployed for dynamic testing in the wavetank.

The model was moored, using the setupshown in Figure 4, so that the f'u1ly submerged cagewas directly in front of the upper sidewall window.The high resolution camera was positioned to viewthe cage through the window over its full range ofmotion.

The model was allowed to come to verticalequilibrium and was then excited by regular,

&gare 4 Expertmeitta! setup Oppoute upper Wittdow

sinusoidal waves. Frequencies used ranged from0.5 to 1.2 Hertz, and wave slopes were on the orderof I/15, After the leading edge of the wave trainpassed the model and thc model appeared to beoscillating with the waves, position measurementswerc recorded over three wave cycles. The opticalsystem software was then used to calculate timeseries of horizontal and vertical position of the twotarget points shown in Figure 4. Using relativeheight difference and the distance between targetpoints, time series for the cage pitch angle wercalso calculated.

The finite element computer program wasrun for identical conditions. The finite element cagemodel, shown in Figure 5, used the exactdimensions and weights as the skeleton physicalmodel. The excitation consisted of the same casesof regular wave forcing. lt should be noted thatthe computer program input corresponded directlyto the actual model dimensions. Thus, there wasno need for either Froude or ReynoMs numberscale-up of results.

Swift et aL 203

RF.SULTS

The wave tank, physical model responseto regular waves is summarized in Table l. Foreach separate test, wave period T, wave height H,and average amplitudes for pitch angle andhorizontal X! and vertical Y! displacement of thctwo target points see Fig, 4! are provided. Averageamplitude is one-half the peak to trough differenceaveraged over the three waves measured.

Representative time series of horizontaland vertical displacement of the target poinb areplotted in Figures 6 and 7, Thc regular waveresponse is gcncr@ly sinusoidal with the horizontalmotion of the target points nearly equal and greaterthan the vertical motion, A drift can be seen which

is due to a persistent transient initiated when theregular wave train encountered the upright cage/mooring system. Close examination of the timeseries reveals that thc angular motion is oppositeto that of an inverted pendulum, At the extremesof the horizontal displacement, the side of the cage

Figure 5 Fmite eletnent cage model.

T = wave period; H = wave he'ght; angle - pitch angle with

respect to the horizontal; X., X2 and Yl, Yz are

horizontal and vertical displacement cotnponents of the two

target points 3, 2 showrt in Figttre 4.

Table L Measured cage response to regular waves.

t;JNR Technical Report Vo. 26

2,0 3.0 60

Time seconds!6.0 60

Figure * Measttred cage response to a regu/ar wave. Pertod = 1.67 seconds aud wave height = 0.28 neuter.

2,0

-2.0

-3,00.0 0.5 1D 1.5 2.0

Time seconds!6.0

6

S.00Lcs

O1st

10.00COlO

1,0cs

0.08CL

Cl

easured cage response to a regular wave. Period = 1.25 seconds and wave height = 0.16 tneters.

gwifit ct al, X5

12n 05 1 1.5 2 2.5 3 35 4 4S 5

mme seconds!

Figure L Finite clement model predications of cage response. Period = 1.67 seconds and wave height = 0.2R meters.

2.5

1.5

Es3

OhCI

oIQCtt

ELSCLCOCl

D nk 1 1.5 2 25 3 3S 4 45 5

Tame secands!

Figure 9. Finite element inode1 predictions of cage response Period = 1.25 seconds and wave height = 0,16 meters.

LITERATURE CITED

CONCLUSIONS

ttJNR 'fggbatcat RePort Wo. 26

towards thc mooring dips down. This is apparentlydue to wave action on the mooring/bridle system having negligible inertia! kicking out the cagebottom,

Finite element inodel predictionscorresponding to the Figures 6 and 7 experimentare shown in Figures 8 and 9, respectively. Thesame type of horizontal and vertical sinusoidalmotion is seen and transient behavior is evident.Thc same bottom kick- out type angular inotion isalso evident, though vertical motion is inuresy mmetric in the computer model output. Overall,the finite element inodel is seen to replicate the basicprocesses and motion response recorded in theempirical data. Direct comparison of thequantitative results, however, shows that the finitee/ement model somewhat underpredicts the motionamplitudes.

It should be noted that coefficients in theMorison equation fluid forcing inodel werecalculated using accepted theory sec Gosz et al�1996! and were not tuned for this physical modelapplication. Symmetry seen in the predicttxI verticalmotion but not as evident in the physical modelresponse may be due to using a linear wave theoryin the program. The wave loading and o ther codingissues are currently under review in the ongoingmodel improvement effort.

The UNH wave tank is ideally configuredfor testing offshore fish cage physical models.Conveniently placed observation windows allowprecise, noninvasive measurement of cage motionusing a passive optical technique,

The skeleton model approach reducescomplexity allowing evahuation of how wellcomputer programs simulate basic processesgoverning fish cage dynamics in waves. The Goszet al, l 996! finite element program was found toreplicate the fundamental characteristics of thephysical model motion, but work is ongoing toobtain more exact numerical agreement.ACKNOWLEDGE@ NTS

The authors are grateful for the supportprovided by the National Sea Grant ONce to the

University of New Hampshire. Critical reviewsprovided by Professors Igor Tsukrov and KennethBaldwin were very helpful and are appreciated.

Darnetl, L, 1996. A towing carriage for theUniversity of New Hampshire towing andwavemaking basin. M.S. thesis, Univ, NewHampshire, Durham, NH. 178 p.

Gosz, M., K, Kestler, R. Swift, and B. Celikkol.1996. Finiie-element inodeling ofaquaculture net-pens, pp. 523-541. In:Open Ocean Aquaculture, Proceedings ofan International Conference, Portland,Maine. New Hampshire/Maine Sea GrantCollege Program Report No. UNHMP-CP-SG-96-9.

Michelin, D, and S. Stott. 1996. Optical positioningintrumcntation and evaluation. OceanProjects Course final report, Univ NcwHampshire, Durham, NH. 85 p.

Washburn, S. 1996. A wave generator andwave absorber system for the Universityof New Hampshire wave/tow tank. M.S.thesis, Univ. New Hampshire, Durham,NH. 224 p.

Taka st 207

CREATON OF OFFSHORE AQUACULTURE GROUND BYFLOATING BREAKWATER

N, TakagiNational Research Institute of Fisheries EngineeringEbidai, Hasaki, Kasitna, Ibaraki 314-0421, JAPAN

e-mail:norimasa C@nrife.affrc.go,jp

ABSTRACT

In Japan, cage culture hasexpandedmainlyincalmwa erazeas such as the Seto inland Sea The present culturecondition can he described as overly intensive, and this has caused the detenoration of water quality. hindering theexpansion of fish culture. ln order to expand lish culture, n is necessary to enhance otfshore culture. Animponaot task is to develop a nursery system which can withstand the rough wave conditions of the oceanaround Japan. Here, we introduce some offshore types of floadng breakwater now in use and a floating hreakwaterequipped with aquacuhure net cages which is under development. in order to realize offshore culture

INTRODUCTION

The coastal fisheries production is on astable level. Coastal fisheries resources have been

enlarged by the coastal fishing ground development,such as construction of artificial reefs and fisherynursery grounds for propagation. Only aquacu lturehas shown an upward tendency in area andpmduction quantity,

Aquaculture was developed in the inlandseas and calm bays because the wave conditionsin the open sea were severe, thus detrimental toaquaculture facilities. The floating breakwater wasdeveloped in order to enlarge the aquaculturegrounds, and at many locations in Japan has beenconstructed to create more suitable grounds foraquaculture.

It is necessary to maintain calm seas atthe aquaculture grounds for safety and workability,and for an optimal environment for breeding fish.Therefore, the floating breakwater has been usmito create aquaculture grounds because it has thefollowing characteristics:

I! Transmitted waves canbe controlled by the scaleand wave absorptionprinciple of the floatingbreakwater;

�! The floating breakwaterdoes not obstruct the

seawater exchange,mixing, and diffusion, sowater quality istnaintained;

�! In the deep sea >20m!,the floating breakwater istnore economical than the

gravity type;{4! The floating breakwater

is convenient for planningand tnaintenance.

Reliability for safety of the floatingbreakwater has been established with actual results.

In addition, aquaculture grounds have expandedfrom the bay and inland sea areas to the open seabecause of environtnental change andovercrowding. However, it is clear that the nortnaltype of floating breakwater costs too much in orderto achieve the required perfortnance, and in thecase of aquaculture grounds in the open sea, isvery difficult to construct. A new type of floatingbreakwater which can absorb big and long waveseffectively is warranted, Moreover, its the case ofaquaculture grounds in the open sea far from thefishing port or fishing village, the floatingbreakwater must have additional functions, suchas the cultivation of bmodstock and nursery culture.

I would like to introduce two examples ofthe floaung breakwater constructed in the opensea, and to describe the direction of its research in

Figure 1. Project site and layout platt.

the future.12 l Wave condition

Design wavefunction

Design wave forstructural stabili vFloating breakwater constructed in Takahama

District, Fukui PrefectureThis floating breakwater was planned as

part of the creation of ncw aquaculture grounds inTakahama District, Fukui Prefecture, located in thertudd!e part of Japan along the Sea of Japan, Thiswas the first one constructed along the Sea ofJapan, which protects the aquaculture ground I 3ha! from big waves, Figures 1 and 2 show itslayout and the aquaculture ground, and Figure 3shows its structure.

srgruficant wave heightsignificant wave periodWave lengthWave directionTransmission coeffiment

1.3 m 1.4 m7.5 sec 6.0 see87.tt m 56.2 rn'sfNW 7C0.6 0.5

3.9 rn11 7 scc171,4 rnN

Design Condition

Ya ural rondino nAverage water depthTidal rangeTidal currentWind ve!ocnysediment

:28m. I m. 0.5 rrv'scc

28.0 rttrsec. Silt nosed with fine sand

U3hiR Technical Rcport Yo. 26

The principle of floating breakwaterThe floating breakwater shown in Figure

3 absorbs waves by imeraction between thc airflow and the internal water movement, It haschambers on both sides and air ducts connected toeach chamber, Its scale is as foliows: l unit length68.0 m; width 14.5 m; height 8.7 m; total length228 rn � units!.

Characteristics of floating breakwaterWave function design requires a relatively

long period, so this floating breakwater iscategorized as the open sca type, It is moored by

Ta k a gi 209

Figure 2, An airplane view of the floating breakwater.

Air ducts

Figure 3, Structure of the floating bteakwa er.

six cross lines connected with anchors because

wave direction is not normal to it. Moreover, a

construction craft cannot be used because there is

not one large enough in this region, and to move itfrom another place to the project site would bevery expensive. The anchor is divided into threeparts.

Figurc 4 shows the three.-part anchor. Thisanchor was used for the first time, and is appliedto the open sea because of its size. In this case, itsapplicability to the oblique waves and problems inconstructing in the open sea were made clear. Thefloating breakwater has performed very well 3 yrafter installation, though high waves are frequentin the winter.

Fishing ground constructed in Aba District,IVagasaki Prefecture

This fishing ground was planned as partof the creation for a multipurpose calm area, whichenhances developing coastal fishing grounds. The

~ I

plan included the gravity-type breakwater and thcnew floating type that applies the results of theTakahama case described earlier. The project siteis located in the northern part of Kyushu Islandalong the east China Sea, Figure 5 shows theproject site and the layout of facilities, and Figure6 shows thc new type of floating breakwater.

�! Wave conditionDesign wavefunction

Design wave forstructural stability

Significant wave heightSignificant wave periodWave lengthWave direction

2.0 m6.7 sec70.0 mSSW

3.9 ml 3,3 sec276.0 mSSW

Transmission coefficient 0.5

Design Condition l l Natural condition

Water depthTidal rangeTidal currentWind velocitySediment

The prinFi

of floatin

chambers

principles

ciple of flgure 5 shog breakwis asymare consi

oating breakwws the cross se

ater. The sha

cal bec

dto wo

: 17-28 m38m

: 0.761 m/sec: 40.0 m/sec. clay

metrl

dere

au

rk

210 UJNR Technical Report Ve. 26

Figure 4. Parts of the new type of attchor

'g!

ater

ction of the new typepe of its absorbing ';::,.''~>jjise three absorption '';::,'!~~effectively: first is:,"kg~>!

Figure 5. Project site and layout plan.

Gf8Ylbp t,pp8 bK88k%'RIM' PM<i»g bTeakms.ter

Takagi 211

terr!p ~~ B"!

Figure 6, The floating breakwatn

absorption by scattering waves due to the motionof the floating body; second is by interactionbetween the air motion and the internal water

surface motion in the chambers; and third is bydissipation due to the water currents in thechambers. The floating body could be made smallerthan the normal type by changing the length of theinternal current in the chambers to achieve thc

required performance. Because its performancewas also estimated by numerical simulation andthe hydraulic model test, this floating breakwaterwas applied to the Aba aquaculture ground for thefirst time in Japan. Its scale is as follows: 1 unitlength 57.0 m; width 11.0 m; height 8.3 rn; totallength 200 m � units!.

212 1;3VR Technical Report Vo, 26

Characteristics of floating breakwaterThe bottom at the project site is very soft

and large waves are often generated, so it isdifficult to moor the floating breakwater. Thisfloating breakwater was very unique, whichworked effectively in thc open sca with its mooringsystem designed for ground stability, Itsconstruction began in 1993 and was completed in1995. It now works effectively although typhoonsoften attack this area. Its performance was alsochecked by the field survey.

Development of new type of floatingbreakwater

Up to now, various types of floatingbreakwater have been proposed and applied in thecreation of aquaculture grounds. Moreover, a new

Tatragi Z13

Cross section of floating breakwater

0.0 0 2 6 8 l0The ratio of wave Length to the widtlr of flaating breakwater

Figure 7. Cross section of the floating breakwater for the open sea and its performance.

Orifice

ent wave

l,o

6

o OS

&

Figure L Cross section of the floating breakwater attachedwith culture cages.

type, one applicable to the open sea, was developedand constructed. Recently, for utilizing the floatingbreakwater in rnultiples, the one with culture cageshas been proposed. By constructing this new typewhich has culture cages and the facilities formanagement, offshore aquaculture will becoinesafer and more efficient.

Current state of developmentFigure 8 shows the cross section of the

multipurpose breakwater. The front part absorbswaves by controlling air and water currents throughoriflices attached to the ceiling of each chamber.The back part is a culture cage. The main problemis reducing the wave motion which damages fishin the cage. The width and height of the absorptionpart must be considered for the safety ofaquaculture and stability of the cage. Its

UJNtt Teehnteal rteport No. 26

sorbing section

Ftttttre 9. A new type of tnn/tipnrpose breakwater.

performance and the water environment in the cagebased on the image shown in Figure 9 are nowbeing studied.

EPILOGUE AND ACKNOWLEDGMENTS

These studies will serve as the basictechnology for creating offshore aquaculturegrounds and the offshore fisheries base. So wecontinue to study steadily. Finally, I express specialthanks to the people who offered importantphotographs «nd data. In addition, development ofthe new type of floating breakwater was carriedout in cooperation with the National ResearchInstitute of Fisheries Engineering, Mihubishi Heandustry Co., Ltd.. Ishikawajima-Harirna HeavyIndustry Co., Ltd., and Hitachi EngineeringShipbuilding Co., Ltd.

Takagi, N. and S, Akeda. 1992,. Hydraul ic modeltest on formation of pisciculture ground atTakahama, Fukui Pref. Tech. Rep, Natl.Res. Inst, Fish. Eng., Aquacult Fish, PortEng. No.14, pp. 37-75,

LITERATURE CITED

Takagi, N., Y.Ohtnur, M, Ozaki, Y. Isozaki, A.Arami, A. Kadono, and A. Nagano. 1995.Experirrtental study on application of newtype floating breakwater for open sea area.Bull. Natl. Res. Inst. Fish. Eng. No,16, pp.29-57.

Kagiyama, H. 1995. Development of ne w floatingbreakwater. Sci. Tech, J. Jpn. Int. Mar. SciTechnol. Fed. 8�!: 31-40.

Ohkusu, M., M. Kashi wagi, K. Ikegarni, M. Ozaki,and Y. Isozaki. 1991. Study on perfonnanceof floating breakwater utilizing the relativemotion of inside water. J, Soc. Nav. Archit.Jpn. 169: 215-222,

Dusnry et ai. ius

AWATS: A NET-PEN AQUACULTURE WASTE TRANSPORT SHVIULA-TOR FOR MANAGEMENT PURPOSES

Robert W. DudleyUniversity of Maine

Department of Civil EngineeringOrono, ME 04469

e-mail:rwdudley!st voyager,umeres.maine.eduVijay G. Panchang

National Sea Grant Office, NOAA

Silver Spring, MD 20910and

Carter R. Newel!

The Great Eastern Mussel Farms, inc,P,O, Box 141

Tenants Harbor, ME 04860

ABSTRACT

An efficient mathematica! modeling package called Aquacu! ture Waste Transport Siinu!ator A WATS!provides first-order estimates of the physical dispersion of finfish aquacuhure wastes forregulatory purpose~.The mode!ing strategy entails the utilization of a verttca!!y averaged, iwo-dimensional !!ow model to producet!ow-fie! d information; this information is input to a particle tracking waste transport model to simulate theresulting transport of wastes. Since earlier studies have shown that the transport modeling resuhs are sensi-tive to the thresho!d shear stress at which setoed fish-pcn wastes are resuspended, fieldwork was conducted toimprove ihe parameterization of erodibi!ity in the transport model. Application of AWATS to several aquac-ulture sites in coasta! Maine se!ected by the !vtaine Department of Environsnenta! Protection! shows thai it isa convenient tool in the regulatory process.

INTRODUCTION

Due to high stocking densities and feedrates, net-pen aquaculture operations are regardedas potential polluters of the tnarine environment.Net-pen wastes, consisting primarily of fish feedand fecal pellets, can adversely impact the coastalenvironment through increased concentrations ofammonia, decreased dissolved oxygen, and theformation of bacterial mats at particularlyprob!einatic sites. While rates of' deposition andaccumulation of these wastes in the vicinity ofnet-pen opera.tions depends on many factors including stocking density, feeding rates and theamount of excess feed waste, settling rates ofwaste tnaterial, fish metabolism, grazing, bacterialdecomposition, etc.!, the degree of environmentaldeterioration depends ultitnately on the

hydrodynamic environment.A considerable effort is put forth by

regulators to tnonitor hydrodynatnic, waterquality, and benthic conditions and to evaluateenvironmental impacts of net-pen aquacultureoperations. The efficiency of this work may besignificatitly enhanced through the usc ofmathematical models that give more completeinfortnation regarding the physical conditions inthe domain. For example, Panchang et al. �997!have shown that the use of blanket guidelines forminimum current speed and water depth do notautomatically ensure favorable hydrodynamicconditions for a net-peri operation. The flow-fieldsseen in coastal Maine are complex and it is oftendifficult to discern prevailing current direction andoverall flow-fields froin discrete, site-specificmeasurements over liinited time periods. Such

data fail to ascertain the spatial and tempaialvariations of the hydrodynainic environment suchas vorticity, wind, seasonal effects, ctc.! withinlease sites or the cumulative effects of severaloperations within a coastal ernbayrnent. Thecomplex and restrictive regulatory environmentis viewed as a limiting factor in thc growth of theaquaculture industry in the United States Schneider and Fridley 1993!,

To resolve some of the above limitations,Panchang et al. �997! developed a comprehensivemodeling strategy involving an investigation oftidal and storm-induced currents, wave effects,and net-pen ~aste transport mechanisms such assettling, resuspension, and decay. This approachwas shown to be successful in assessing the impactof aquaculture operations in Cobscook Bay andToothacher Bay. First, a vertically averaged, two-diinensional flow model is constructed usingappropriate field measurements, to simulate thecurrents induced by the tides and by storzn winds.The resulting flow-fields were used as input to aparticle tracking waste transport model. Theresults showed that at some sites, inferences drawnregarding the waste distribution using acombination of modeling methods and field datacould be quite different from those drawn usingisolated field measurements. The potential of themodeling methods for site selection and irideciding a priori which sites needed a greater levelof morntoring was also demonstrated.

Before the modeling techniques can beadopted in regulatory practice, however, the workof Panchang et al, �997! suggests that twoproblems need further attention. First, a morereliable description af the resuspension of settledwastes is needed. Since resuspension involvescomplex mechanisms that are not well-understood, it was modeled using a parameterU �describing a threshold or critical currentvelocity at which settled waste material would beresuspended. Panchang et al. �997! found thatthe waste dispersion and accumulation resultswere very sensitive to the threshold of shear stressat which settled fish-pen wastes are resuspended,thus limiting the usefulness of the models for siteselection, Secondly, Panchang et aL �997! weremoti vated more by a research perspecti ve and did

not offer tools readily available to regulators.We describe efforts to improve estimates

for the critical resuspension velocity of net-penwastes, and to create a modeling package thatcould be routinely used ta aid regulators with siteevaluation and decision-making. Specifically,field measurements were made to estimate in situerodibility of net-pen waste materials. Asubmarine annular flume called the Sea Carouselwas used; this device was designed by theGeological Survey of Canada to study seabedinstabilities and the rncchanisrns involved Amoset al. I992a!. The Sea Carouse! and thc fieldmeasuremcnt programs are described in sectionl. In the interest of packaging the modelingtechnology for regulators, two reasonably well-known flow models were evaluated for accuracyand ease of use: a fiiute element model calledRMA2 and a finite-difference model calledDUCHESS, R1VIA2, developed through fundingfrom the US Army Corps of Engineers andcoupled with a sophisticated graphical interface,is a public-domain, two-dimensionalhydrodynamic model. DUCHESS, which wasdeveloped at Delft University, Netherlands, iswidely used for two-dimensional tidal and stormsurge computations e.g., Booij 1989, Jin andKranenberg 1993!. The transport model developedby Panchang et al. �997! was enhanced andpackaged with an interface used to extract flowsolutions and graphically display flow andtransport results. This work led to a package calledA WATS Aquaculture Waste TransportSimulator!, described in section 2. It was appliedto three sites selected by the Maine Departmentof Environmental Protection for testing anddemonstration purposes as part of technologytransfer efforts. Application of AWATS tomodeling an aquaculture site in Maine is presentedin section 3.

I. Fieldwork to estimate erodibilityIn the initial development of the waste

transport model, Panchang et al. �997! found thatthe transport of net-pen aquaculture waste wassensitive ta the ability of the currents to resuspendmaterial once it had settled on the bottom. Withsettling rates of 4-10 crn/sec and typical depths

Dudley et aL 2l 7

beneath pens of 15-25 m, net-pcn wastes will settlein the vicinity of the pens in a matter of minutes.ln constant low-velocity environments such asfjords, local settling can have adverseenvironlnental impaCtS; in high-velOCityenvironments the material may be resuspendedand lnore effectively dispersed, Lackingapplicable information regarding the complexprocess of resuspens ion in aquacultureenvironments, Fanchang et al. ]997! used casualdiver observations which suggested that net-pcnwastes were eroded when the flow velocityexceeded approximately 30 cm/sec, In view ofthe uncertainty, however, Panchang et al. �997!modeled multiple transport scenarios by varyingthe values of U, over a range and found that theresulting waste dispersion was very sensitive toU,. For example, waste removal from the domainused to examine a commercial lease site in DeepCove, Cobscook Bay, varied between 83% and0% when U, was varied between 10 cm/sec and40 cm/sec. The area affected by the wastes alsovaried substantially.

Erosion of sediments is a function of

bottom stress which is often expressed as shearvelocity. In this sense, U, is intended to be ameasure of the threshoM stress at which net-penwastes wouM be eroded and resuspended. Toobtain more reliable information regarding thismechanism, measurements were made under thedirection of Dr. Carl Amos of the Bedford Institute

of Oceanography BIO! at the Connors BrothersInc. comtnercial lease site at Deep Cove inCobscook Bay near Eastport, Maine Fig. 1!.Figure 2 shows the locations of erodibilityexperiments in relation to the Deep Cove net-pensystems, A device called the Sea Carousel shownin Figure 3 was used to conduct the erosionexperiments. The Sea Carousel is an annularflume designed by the Geological Survey ofCanada to measure seabed erosion Uponlowering it to the benthos from the side of a boat,a current was generated inside the flume andslowly increased in magnitude in a stepwisefashion. At each step over the course of theerosion program, a video of the erosion processwas obtained in conjunction with water samplesand turbidity measurelnents. The resulting

turbidity measuretnents can be correlated withshear velocity to provide values of criticalresuspension velocity e.g., Amos et a1. 1992b!.

The Deep Cove site contains three pensystems Fig. 2! consisting of net-covered cagesarranged in rows of 10 cages, with two rowsforming an independent floating pen system, eachholding about 5,000 fish. At this site, wcattelnpted to determine the erosion threshold andits variation in time and space. It was estimatedthat the greatest amount of sedimentation wouJdbe near the center of the three pen systems andwould decrease outwards. Since i was possiblethat the erosion threshold varied with the amount

of material already acculnulated, the Sea Carouselwas deployed at nine locations: three near thecenter of the site, four locations at different pointson the seditnentation gradient, and two controllocations closer to land deemed to be unaffected

by the net-pen operation, Data were collected attwo different times, one in April 1996 and one in

Figure 1. NottheL~~ Cohscook Bay illustrating the loca-tion of the Deep Cove field sitr.. Bathymetry in meters.

atg t;Jbltt T~ aepert No' S

Agure l, Deep Cove showing tbc locations of April and September 1996 Sea Carousel andcorrcnt tncter deploytncnts in relation to Connors Brothers inc, net-pen systems 5400,5600. and 5700. Sea Carousel deployments are numbered 1-9

~ 5 fbe Sea Caronrel ahnnt to be lnwe<ed tO tbe benthasat the Connors Brothers loc, a<toacultttre site in DeepCove near Eastport, Matne.

September 1996, since there is likely to beseasonal variation in the amounts of net-penwastes present due to higher feeding rates in thesurnrner, and more frequent storm-inducederosional events in the winter!.

During the fteldwork, locations of the net-pen sites, current gauges, and Sea Carouseldeployrnents were determined via the GlobalPositioning System GPS!. The Sea CaroUselwork provided videos of the seabed erosion,samples of suspended sediments for each velocitystep Fig. 4!, seditnent core samples, watervelocities, and turbidity data, The erodibility datafrom the Deep Cove aquaculture site wereanalyzed by Drs. Terri Sutherland and Carl AtrtQ$at810. A summary of the results is given in Table1, in termS of U«<Nh the current veloCity at 100 ctrlfrom the bottom. The U««o, �values wercdetermined from plots of suspended particulate

17tadtey et al. 219

rnatter SPM! against Ut ooi where SPM wasobserved to be significant]y higher than thepreceding atttbient SPM concentrations Fig. 5!.The U�ott~, value was taken as the incan of theU,too> speed settings at that transition pointTable 1 shows that, in general, the erosionalvelocity increases along a transect tn the directionof the net-pen. Similarly, the values are higher inthe summer than in the winter. This suggests thatU -, is indeed affected by the amount of materialpresent. The modeling strategy described lateronly allows for a constant U,. Average valuesof 0.40 m/sec for the winter/spring and 0.50 m/sec for the summer/fall are used. These numbers

are close to anecdotal evidence provided by divers Dr. R. Findlay, Deltartment of Microbiology,Miami University, personal communication! thatmaterial seems to be resuspended when the flowspeeds arc greater than about 0.30 m/scc, It isitnpor4mt to note that the U, values in Table 1include values for all sediment types encounteredin the field sessions from fine gel mud to coarsematerial and includes erosion of native material;

research is currently being perfortned by Drs.Sutherland and Amos to estimate the erosionthresholds for strictly fish feed pellets.

2. Mathetnatical models

Modeling the physical transport of finfishaquaculture waste requires detailed knowledge ofthe spatial and temporal variations in tide andwind-induced currents in the particular region ofinterest. Hydrodynanuc models, driven andvalidated with field data, simulate these currentsand provide the necessary input information fortransport models to cotnpute the resu!ting wastedispersion. Previous modeling work conductedfor Cobscook Bay indicated that a two-ditnensional flow model based on the shallowwater equations that. yields depth-averagedvelocity components is adequate for this task. Thisis fortunate, since three-ditnensional schemesrequire intensive computer resources, particularlywhen large areas such as the coastal domains ofMaine are to be modeled. In addition, datacollected near aquaculture sites in Cobscook Bayindicate thar. the large tidal forcing leads to littlevertical variation in the horizontal velocities in

those areas e.g., panchang et al- 1~93!-In the interest of assembling a user-

friendly modeling software package to be usedby regulators, we evaluated the ease of operationand accuracy of two-dimensional flow models.Both finite-element and finite-difference tnodelswere investigated. Finite elements usually affordgreater flexibility in describing complex coastalboundaries and domains where aquacultureoperations are carried out. As an example, wechose the tnodel RMA2. This is a public domaintnodel which is a part of the popular "ShallowWater Modeling System" developed by the U. S,Army Corps of Engineers and is hence readilyavailable along with a sophisticated user interface.The finite-difference tnodcl DUCHESS was

Fltpsre 4, Water samples for suspended particulate maneranalysis collected dttrirtg a Sea Carouse! erosion pro-gram coadncaed ai the Connors Etrothers inc. atlttacut-tnre site in Deep Cove. Frorrt left to right. each houiecorresponds to a water sample tatten 2 min after thc on-set of eacb step-wise velocity magnitude increment gen-erated inside the attttttlar flume.

Rat! UJh!R Toebnkal Report ivo. 26

1.50.5

Utsee! ntfa!

Ft!tttre 5. Esumate of thc erosion threshold for an Apri!!996Sea Cwouse! dcp!oyment at station 4 in Deep Cove, Maine.Ambient ctmccntrations of suspcndcd ponictt!ate matter con-centration SPM! we designated by round symbo!s, whi!etbc eroded concentrations of SPM sse designated by triangu-lar symbo!s. Based on thc significant change in SPM i!!os-trated above. Uo«~ for this particular experiment is esti-tnated to be 0. 33 m/sec.

chosen since the performance of DUCHESS hadalready been well established via prior modelingefforts Panchang et al. 1997!.

a. Hydrodyttamie modelsThc graphical user interface for RMA2

consists of a software package called Surface-water Modeling Systein SMS! developed at theBrigham Young University EngineeringComputer Graphics Laboratory ECGL! incooperation with the Army Corps of Engineers Jones and Richards 1992, ECGL 1995!. Thisenables users to graphically construct finiteelement meshes required as input to RMA2 andto display hydrodynamic solutions frotn RMA2.The SMS software provides the user with varioustools and pull-down menus to facilitate digitizingscanned topography maps, constructingcomputational meshes, and displaying andanimating solution data sets with color contouringarid vectors.

A significant amount of modeling waspursued using RMA2 to assess its suitability forcoastal modeling associated with aquaculturernanagernent. The evaluation of RMA2 includedmodeling of simple test cases as well as asystematic investigation of mesh construction andrefinement, boundary conditions, time step size,and flooding and drying mechanisms for thecoastal region of Cobscook Bay, Maine. The 15.5x 13.7 km Cobscook Bay domain Fig. 6! had been

Tts!s!e t. Su~ ~' o ~m Uo«' values from Sea Carouse! data; deploytnents Apr!! and Scptembcr l996 at Connors Brothers Inc. 'aquacu!tare farm, Deep Cove, Maine.

y ngolusly modeled and validated usingUC ESS, Initial RMA2 flow model mns usin

a 225-m resoJution mesh of the Cobscook Baydomain resulted in problems with floodingdeing- If any node comprising an elementthe drying criteria, the entire element to which itbelonged becatne "dry" and was removed fromcomputation. As a consequence, entire reachesof thc bay would be shut off due to drying inshallow, narrow areas, resulting in a discontinuousd.omain and inodel failure.

Subsequent efforts, which invol vedrefining the computational mesh and adjustmentsto various model pararnetcrs such as time step,eddy viscosity, and bottom friction, met with onlymoderate success. Due largely to the size andcoinputational demands of the Cobscook Baydomain, the most successful model run using arelatively coarse mesh and 12-min time steps ranin near-real time on our 200 MHz PC, The mesh,at its finest, had a resolution of 75 tn, the majorityof which was much coarser, with a maximutn of225 m see Fig. 6!. The resulting simulations were

Dudley et ai, 221

not as satisfactory as those descri bed by Panchanget al. �997! using the finite-difference modelDUCHESS. Ivloreover, DUCHESS could resolvetwo tidal cycles for tlte same domain with aconstant 75-rn resolution and a 40-sec time step,in about 40 min on the same PC.

In sutritnary, though RMA2 has beca usedin other applications. its implementation wasextremel y time-consuming and problematic for th isparticular application. It also presented addedcoinplexity for regulators due to its sensitivity to gridsize~, requiring greater efforts in the construction andrefinement of finite element meshes. Meshconstruction and refinement is a complex problemrequiring evaluation of dotnain geometry andbathymetry, 1Vhile a.ll modeling involves a certainlevel of trial-and-en'>r before successful simulauonsare obtained, it was felt that working with finiteelement models would be too cumbersoine from the

point of view of routine managementMost finite difference models, in

comparison, require only a single resolutiortthroughout, and entail a straightforward

gurntnary inustrtttion for the TRA.'4S aquaculture net-pen waste tndevelops at the University of Maine.

222 t;JitlL Techakd Report Iso. 26

relationship between the time step and the gridsize Although the flexibility of enhancing theresolution in specific parts of the domain iscompromised, DUCHESS does allow options forsubsequent simulations in "nested" domains.Flow modeling groundwork for computingaquaculture waste transport was performed byPanchang et al, �997! for Cobscook Bay andToothacher Bay using DUCHESS. The model hasalso been successfully applied to other fisheries-related problems Newell 1991! and has beenfound to be generally robust.

One limitation of DUCHESS is that,unlike RMA2, it lacks a convenient graphical userinterface to expedite the modeling process byaiding the user in model construction and viewingand interpreting model output. For this reason,we made efforts to interface DUCHESS with SMSto avail users of the graphical advantages of SMS.We developed a utility program calledDUCHSMS which indirectly links the twoprograms, The program facilitates constructionof the model domain using SMS, and graphicalviewing of the flow model output. I enablesbathymetry digitized with SMS to be exported ina form required by DUCHESS as input and alsotransforms DUCHESS output into a form readableby SMS. This allows easy graphical display andanimation of flow solutions obtained fromDUCHESS in SMS.

b. Transport rriodel

A transport model called TRANS wasdeveloped at the Uruversity of Maine ro simulatethe advection and dispersion of finfish aquaculturewastes; it is included in the A WATS package, andmodels the mechanisms of settling, advection, andresuspension to describe the physical transport offish-pen waste materials. To accomplish this,TRANS requires spatial and temporal flow-fieldinformation, bottom topography data, andproperties describing the net-pen wastes such asresuspension threshold U,!, settling rates, andthe location and the frequency of the introductionof wastes into the water. Paraineters describingthe aquaculture farm are input by the userproviding coordinates of each net-pen in thedomain coordinate system, as well as the size ofeach pen, its stocking density, and daily feed

quantity, Other user-specified parameters in themodel include: the simulation duration, begin andend times for food and fecal matter introductioneach day, the uneaten food ratio as a percent ofthe daily food mass introduced, the daily fecalpellet production in g/kg of fish, percentage oforgariic carbon contained in the waste dependingon the feed used, and first-order decay coefficientestimates for food and fecal rnatter.

The transport model computationsinvolve breaking the daily feed and fecalintroductions down into particles and trackingtheir dispersion throughout the model domain asthey are advected by the currents computed bythe hydrodynainic model Fig. 7!, Each particlerepresents a user-specified amount of massrepresenting a part of the total mass introducedover the course of the simulation. Each particleis tracked until it leaves the transport domain atwhich point it is considered to have been flushedaway, and is not allowed to return. As the particlessink, they are advected by thc flow-field until theyreach the bottom. For modeling purposes, wechose sinking rates of 4 cm/sec for fecal particlesand 10 crn/sec for feed particles variable uponfeed type! Panchang et al. 1993!. Once on thebottom, a check is made at each time step againstthe specified U, to deterrninc whether or not theparticle is eroded froin the bottom andresuspended in the water colutnn to be furthertransported. Partides can decrease in mass overthe course of a model run to first-order exponentialdecay. Values used for the decay coefficientdepend upon the environtnent and oxygenavailability. Values in fjords have been found tovary between 0.1 yr' to 0.5 yr' Aure andStigebrandt 1990, Hansen et al. 1991!.

At the end of the simulation, TRANSoutputs waste distribution snapshots at a user-specified time interval and a simulation summary,Particles reinaining inside the model domain atthe end of the simulation contribute to organiccarbon loading to the benthos. The loadingconcentration, in g/m', is computed by dividingthe total mass in each transport model grid by thcarea of the grid. TRANS will interpolate fortransport model grid and time step sizes that aresmaller than those of the flow model. A typicaltransport scenario is run for 15 days to approach

Dudlcs el sk 223

c. The A WATS modeling package

We have constructed a package cal!edAWATS that may be suitable for regulatory use.

IICtu aaurlcse Nstp en

The di spars on at'partible" is computedby TRANS each arnear fecal particle' rsp-rsasrds 8 dtsctettzeclamount of mess rep-reaenttng 8 very Siileststb an Ot the tatelmess Intradutm cl.P erbdes" ski k ta thebatters td user -sped fiedanting rates, ttv

std sv* dl

F eccl P artitjes" ~ee ~ o

p ~

0tstfhZ F cad Pesttcfss

unetd etlj

"s:.' «~c~.~.

The eccumul ali on of net-penmsfen sf, espreSSed es themess O I atgsrco Carbon persquare meter, iS Cam puted byTRANS ask ts ~by

T ha til 8SS ateootstt teabng net~ m at estdcen decream avsrthe toutseaf Ste mtt tssb On SCCOC dng touaermpected, trSI-Ordere~ al daaay Caefric ere s

!'sgttre 7. Sontmary illustration fOr the TRANS St!uaou!ture ttet-patt watte trans!xtn simulator atttnrtt!tnt tfeveta~ttt atthe Uttiversity of Maine.

a steady-state loading pattern, The outputsnapshots represent the estimated concentrationt>f net-pen wastes as a measure of organic carbonas it is distributed over time throughout thedomain. These snapshohs are output in a forntreadable by SMS for easy graphical display andanimation. TRANS reports al l model parametersas wc!! a» the amount oi' material f!ushed out ofthe domain, the residence time for materialintroduced on the first day of the simulation. andthe rnaximiurn !oad rate and its location in th.e

model. domain in the surnrnary file.

P ardctes' me resuspe nded tmctsding to 8 cand-kioml atgottthm. Ftx cacti bme step tnthe sssuc-eti On, T R AN S chettt s vetether Or nat e ~ iSresbng cASe bosom, if so, TRANS perssnns 8check on the current velocity at the ~" loc-ation. If it esessdsthe Criticelresuspen San vel-ccky, C, the psrbcle" is rest spsnsied arXI Stdveo-ted in the harizcrttd drectton;S On the btttama

I f no, the ~ corti nues to sink.I f yes, is the ststremtIf yss the partes isedvectedin x and yfor csts time step end settles back to the bottom.

This package con venientls applies thehydrodynamic and transport mode!s along withassociated information regardmg thc net-pens andto obtain appropriate graphical displays. A WA ISinc!udes the waste transport program TRANS, thcgraphical interface SMS, and thc flow modelDUCHESS. In the event the user does noi have

DUCHESS, output from another flotx model maybe used.! lt also includes the uti!ity programDUCHSMS which will extract t!ow and

bathymetry data for the subdomain of interest i.c ..thc general vicinity of the net-pen, specified bythc user in the form of a rectangle[ from the outputfiles of the flow rrtode! DUCHESS or a!ternati ve!

and use this information io run TRANS.

It is helpfu! to describe the AWATS

The Sav4etd soiurtonfran the hyCkcdyn amiamadel is input to TRANSshet de scribes the verbc-ally eve.aged IXinentve Ktctt es in the x end ydirections over tim e

u tsarp>tr tsaiA!

The 'parboil ere edve Cted h tmsont sly in thes tstd y direttt one ssthey ink to the batfotn:

d*=u ady- V'fR

modeling package in terms of its data fileconlponeilis:

I ! DOMAIN.DOM, the overall domain descriptorfile containing the size of the overall domain, gridspacing, number of tiinc steps, and time step sizein the flow solution file;2! DOMAIN. TOP, the topography file, containingthe depths at different grid points of the overaltdomain for which the flow model is run;3! DOMAIN.FLW, the flow soIution file,containing the x-directed and y-directed velocityfields at each time step;4! DOMAIN,OZ, a one-zero file used byDUCHESS and A WATS to differentiate between"dry" �! land points and "wet" I! computationalpoints in the overall domain;S! SUBDQMAIN.BTH, containing depths of asubdomain in thc vicinity of the net-pen thesubdornan in which waste transport simulationsare to be made!;6! SUBDOMAIN.XYZ, a bathymetry filereadable by SMS for use in constructing thedomain geometry to graphically view flow andtransport solutions;

7! SUBDOMAIN.UV, the flow solution filecorresponding to the subdomain to be used byTRANS;

8! SUBDOMAIN,DAT, a second flow solutionfile readable by SMS that can bc used to display/animate the flow solution over the domaingeometry;

9! SUBDOMAIN.FRM, containing user-definednet-pcn parameters: coordinates of the center ofeach net-pen, and the volume, stocking density,and daily feed quantity of each individual pen;I 0! TRANSIN.DAT, which contains the transportmodel grid spacing and time step, simulationduration, daily start and end times and frequencyof food/fecal maner introduction, output requests,pcn location coordinate adjustments, criticalresuspcnsion velocity U !, settling velocities forfood/ fecal material, the fraction of the introducedfood and fecal maner that is organic carbon, thefraction of thc daily feed quantity that is wasted,mass of fecal pellet production per unit mass offish, and fiirst-order exponential decay coefficientsfor food and fecal rnatter.

The first step in the modeling procedure Fig. 8! consists of obtaining tidal and/or wind-

Ftgare g. Operatiouai chart for the A WATS package.

induced velocities using a flow model solution,This involves the DOMAIN.* files. Since theoverall model is often larger than the area ofinterest near the net-pen, a subdomain defined bythe four corner points may be selected for furthermodeling. DUCHSMS uses the DOMAIN.* filesas input to provide the necessary SUBDOMAIN,BTH and SUBDOMAIN. UV files, which are usedalong with the additional data contained in theSUBDOMAIN. FRM and TRANSIN,DAT filesrequired to construct a transport simulation. Thecreation of the following output fiIes usingDUCHSMS functions requires the input of theDOMAIN.DOM file to coordinate the use of theother input files *.TOP, ~.FLW, *.OZ!:SUBDOMAIN.BTH, the bathymetry file to beused as part of the farm description file;SUBDOMAJN,XYZ, a second bathymetry filereadable by SMS for use in constructing thedomain geoinetry to graphically view flow and

Dttdtey et ai. 2?5

transport solutions; SUBDQMAIN,UV, the f!owsolution file corresponding to the subdomain tobe u«d by TRANS; and SUBDOMAIN.DAT, asecond flow solution file readable by SMS thatcan be used to display/animate the flow solutionover the domain geometry.

Executing TRANS produces two formsof output. First, the TRANSPORT,SUM file is asimulation summary describing all user-definedparameters and also reports flushing efficiency ofintroduced particles from the domain, residencetime, aud the sedimentation rate and location ofthe point with greatest accumulation in thesubdomain. The other file TRANSPORT.PLT isa data file that contains snapshots of the dispersionof net-pen wastes over the simulation, suirab! e forplotting in SMS for viewing/animation.

3. Sitnulation of net-pen waste distribution inMachias Bay, Maine

The AWATS modeling package wasapplied to six aquaculture sites in Maine: three inCobscook Bay, and one each in Blue Hill Bay,Machias Bay, and Cutler Harbor. Here, spaceperruits the description of our simulations inMachias Bay, which is located in WashingtonCounty Fig, 9! in the Gulf of Maine. The typicaltida1 range for Machias Bay is about 4 m. Theaquaculture site of interest for this domain isoperated by Atlantic Salmon of Maine, Inc. ASMI! located in Northwest Harbor off CrossIsland. The island is situated in the mouth ofMachias Bay close to thc mainland where it formsthe Cross Island Narrows to its northeast Fig. 10!.

Figure 10 shows the 13 x 14 km domaingeometry representing the entire Machias Bayarea. Thc domain cotLst!ine and bathymetry weredigitized to 75 m reso!ution in SMS using acomputer-scanned image of nautical chart ! 3326of the National Oceanic and AtmosphericAdnunistration NOAA!. These bathymetry dataare stored in the topography file MACHIAS TOP.The flow mode! DUCHESS was used with thisbathymetry to simulate tidal currents The modelwas forced with specified tidal amplitudes at theGulf of Maine/Machias Bay boundary and theCross Island Narrows boundary. Initial efforts intuning the model y ielded reasonable simulationswhich matched current data provided by the Maine

Figure 9. Locatiou of Machias Bay itt Washington Court ty,Maitre, The aquaculture operatiott at Cross islaitd ittthe mouth of Machias Bay is one of six sttes rttodetcdwith the AWATS niodeiiag package.

Department of Marine Resources DMR! in thevicinity of the aquaculture lease area: however,the flow patterns in other areas of the mode! didnot appear to be entirely realistic. For example,while the model produced high currents in theCross Island Narrows as related by anecdotalevidence! and varied over time, thc direction ofthe current never reversed over the course of anentire tida! cycle. Additional current data weretherefore collected in the Cross Island Narrowsusing an S4 current meter on 15 August !997,These data allowed the adjustment of tidalamp!itudes and phases at each open boundary,yielding greatly improved resu!ts not only nearthe Cross Island Narrows, but for the overalldomain by providing a more complete picture ofthe tidal forcing at the boundaries of the model.A snapshot of current velocities just after high tidenear Cross Is!and, take.n from the model resu!ts

LJWR t'eehedcai Report Xo,26

Figure 10. Machias Bay domain, baihyine ry shown in gray srale. Locaoon of the AS%i aquaculture site is denoted by anasterisk near Cross island. The subdomain chosen for transport mot}cling is enclosed by a box. Grid squares are l km'-,depths are given in nteterv.

stored in MACHIAS.FLW, is shown in Figure 11.For modeling nct-pen waste transport at

the ASMI aquaculture site, the subdotnain outlinedby thc box in I-igure ! ! was chosen; it is definedby specifying the coordinates of the corners.DUCHSMS was used to extract the hydrodynatnicsolution and depths for this area of interest fromthe overall domain information contained inMACHIAS.FLW and MACHIAS.TOP. Thcresulting subdotnain information is contained inASMI.BTH and ASMI.UV. Another file calledASMI.XYZ is also obtained from DUCHSMS,which is read into SMS in order to construct thedomain geometry for plotting and animating flow-field solutions and transport inodel output.

In addition to the hydrodynamic solutionfile, TRAIslS requires a farm description filedefining the locations, voluines, stockingdensities, and daily feed quantities for each pen,Eighty-six ASMI net-pens of various sizes and

configurations were located with the aid of aerialphotos from March 1996 provided by T. Riggensof the Maine DMR. Exact stocking and husbandryinformation for this site is confidential and so, formodeling purposes, general aquaculturehusbandry data obtained for ihe previous modelingstudy in Cobscook Bay courtesy of ConnorsBrothers Litnited, Aquaculture Division! wereused in conjunction with literature data Laird andVeedham 1988! to estimate pen stocking density,daily feed quantitics pcr pcn, and fecal productionper unit mass of fish for the ASMI site, It isimportant to note that this nominal aquaculturehusbandry information was used to simulate thedispersion and rates of sedimentation of wasteeffluent from this site in order to illustrate the

application of A WATS.Running a 15-day transport s.cenario for

the Cross Island site produced the sutnmary fileASMITRANS.SUM and the organic carbon

i!udfri et a1 7'i7

CONCLUSION

Figure 11. Tidal fiots-field solution ai the ASM l Cross is-land aquaculture lease site immediately after high tide.Vectors de~ote magnitude and direction of current ve-locity. Gray-scale contours also represent velocity mag-nitude. The circle indicates the approximate locationof the lease area.

concentrations file ASMITRANS.PLT which isread into SMS and plotted. Figure L2 shows asnapshot illustrating the loading pattern of finfishaquaculture waste deposition as g/tn' -organiccarbon at the Cross island aquaculture lease siteai the end of the 15- day model run. For thissimulation, the U, value was set at 40 cm/sec.Since no waste material introduced on day l ofthe simulation was transported beyond the CrossIsland domain bounds, average residence time wasnot computed for the summary. The eastern andsoutheastern portions of the Lease area receivedthc. highest loading with one point receiving ainaximum organic carbon loading rate |averagedover 15 days! of 38.9 g/ma/per day. The mean andmaximuin velocities computed by the model forthis particular area were 6.3 crn/sec and 10.0 cm/sec, respectively. Though not high enough toexceed the U,, criterion for resuspension, thecurrents in this area could supply suff icient oxygento the benthos for adequate rates of decay of theeffluent as well as high rates of water exchangein the embayment to prevent adverse impacts on

Figure 12 Contour plot of l 5-day simulated net-pcn aqua-ulture waste deposition ai the Cro~s island aqua 'ulturesite in Northwest Harhor Coniour interi a! s are lo ! g/m- organic carbon.

thc macrobenthos Drake and Arias 1997!. Findlayand Watling �994! estimated that a constant 6cm/sec current can deliver enough dissolvedoxygen to sediments to support the theoreticalmaximum aerobic oxidation of nearly 50 g/m-/day of organic carbon. The results demonstratehow AWATS can provide noi only a picture ofwaste distribution, but information regardingspatial and teinporal variations in current velocitythat can be used in conjunction with benthicoxygen demand data to determine if organicenrichment in high-load regions has thc potentialto exceed the assimilative capacity of iheenvironment.

In situ measurements near the Deep Coveaquaculture site suggested that bonom sedimentsnear nei-pen aquaculture sites are eroded ai I.'urvelocitic.s greater than about 40 cm/scc in thewinier and about 50 cm/sec in the suinrncr. Thesevalues are used in the development of the

228 L'JNR Technical Re pert Nn. 26

modeh ng package A WATS which can be used forestimating the dispersal of net-pen wastes in acoastal en v i r on ment with varying currents.Although not described here, the package may beused for storm-driven currents and wave-inducedvelocities as well. Application to the aquaculturesite in Machias Bay and others in Maine suggestthat AWATS is a convenient tool that can be usedto aid with site evaluation and direction of fieldmonitoring programs for areas of coastal Maine.

ACKNOWLEDGMENTS

This work was supported in part by theNational Marine Fisheries Service NMFS!/NOAA under the Saltonstall-Kennedy Progratnand by the University of Maine School of MarineSciences. We are grateful to Dr. James Manningof NMFS, John Sowles of the Maine Departmentof Environmental Protection, Laurice Churchilland Tracey Riggens of the Maine Department ofMarine Resources, and to Stephen Dickson of theMaine Geological Survey for their helpfulassistance and advice over the course of thispfoJec L

LITERATURE CITED

Amos, C. L., J. Grant, G. R, Dabom, and K. Black.1992a. Sea Carousel � a benthic, annularflume. Estuarine Coastal Shelf Sci. 34: 557-577.

Amos, C. L., G, R. Dabom, H. A. Christian, A.Atkinson, and A, Robertson. 1992b. In situerosion measurements on fine-grainedsedinsmts from the Bay of Fundy Mar Geol.108: 175-196.

Aure, J. and A. Stigebrandt. 1990. Quantitativeestimates of the eutrophication effects of fishfartning, on fjords. Aquaculture 90: 135-156.

Booij, N. 1989. User Manual for the programDUCHESS, Delft University coinputerprogram for 2D horizontal estuary and seasurges, Department of Civil Engineering,Delft University of Technology, Delft, TheNetherlands,

Drake, P. and A, M. Arias. 1997. The Effect ofAquaculture Practices on the Benthic

Macroinvertebrate Community of a LagoonSystem in the Bay of Cadiz SouthwesternSpain!, Estuaries 20�!: 677-688

Engineering Computer Graphics Laboratory.1995, SMS Surface Water Modeling SystemReference Manual. Brigharn Young Univ.,Provo, Utah.

Findlay, R. H. and L. Watling. 1994. Toward aprocess level model to predict thc effects ofsalmon net-pen aquaculture on the benthos,pp. 47-78. IIi: B.T. Hargrave [ed.], ModehngBenthic Impacts of Organic Enrichment froinMarine Aquaculture, Can. Tech. Rep, Fish.Aquat. Sci, 1949: xi+ 125 p,

Hansen, P. K�K. Pittman, and A, Ervik. 1991.Organic waste from marine fish farms-effects on the sea bed, pp. 105-119. In; T.Makinen ed.!, Marine Aquaculture andEnvironment. Nord 1992: 22, NordicCouncil of Ministers, Copenhagen, Deninark,

Jin, X. and C. Kranenberg. 1993. Quasi-3dnumerical modeling of shallow-watercirculation. J. Hydraulic Eng. 119: 458-472.

Jones, N. L, and D. R. Richards. 1992. Meshgeneration for estuarine flow models. J.Waterway Port Coastal Ocean Eng. ASCE!118�!: 599-614

Laird, L. and T. Needham, Editors. 1988. Salmonand Trout Farming. Halsted Press, NewYork, 271 p.

NcweH, C. R, 1991. Development of a model toseed mussel bottom leases to their carryingcapacity. Phase 2 report, ISI8809760,National Science Foundation SBIR.

Panchang, V.G., G. Cheng, and C. Ne well, 1993,Application of rnathernatical models in theenvironmental regulation of aquaculture.Final report to the National Marine FisheriesService4tNOAA.

Panchang, V G.,G,Cheng,andC,Newell. 1997,Modeling hydrodynamics and aquaculturewaste transport in coastal Maine. Estuaries20�!: 14-41.

Schneider, C, and R. Fridley 1993. Aquacultureand the marine environment: the shaping ofpublic policy Working Paper for theWorkshop, Marine Biological Laboratory,Woods Hole, MA. 10 p.

ENGINEERING TECHNIQUES FOR ENHANCEMENT OFNEARSHORE ROCKY HABITATS FOR SEA URCHIN AND ABALONE

AQUACULTURE

S, Kawarnata

National Research Institute of Fisheries EngineeringEbidai, Hasaki, Kashiina, Ibaraki 314-0421, Japan

e-mail matasan@nrife.affrc.go.jp

ABSTRACT

Commercial production of abalone and sea urchins has been markedly reduced by low availability of algalfood they consume, There are two mech anisins responsible for the liinited food: overgrazing by sca urchins,and a great loss of drift algae produced as they becoine dissipated by water inovement. The wave-inducedwater motion msy inhibit sea urchin grazing, and as such tbe importance of the wave action to protect kelpabundance from the destructive grazing leads us to an engineering possibility of developing kelp beds byincreasing water velocity. ln addition, a new device was developed to trap drift algae. The device is a bait cagewith a pendulum-like door and stoppers. The door is designed to be opened inwards by wave-induced oscilia-tory flow but the stoppers prevent its outward opening. Laboratory scale-madel cxperhnents on the trappingmechanism, effectiveness, and engineering design were carried out. Further field experiments demoisstratedthat the device could trap drift kelp and sever lose thetn until they were consomcd by aggregated sea urchins.

INTRODUCTION

The comtnercial production of abaloneand sea urchins is frequently limited by theavailability of food, Two mechanisms areresponsible for limited food in the habitats. First,kelp, Lanunarian algae, are primary food resourcesnecessary for growth of the animals but arefrequently overgrazed, especially by sea urchins.In addition, most of the kelp production inay beswept out of the shallow habitats as drift algae bycoastal water motion. These benthic herbivoresare of great itnpottance to the nearshore fisheriesin Japan. Artificial structures built to establishkelp beds and to trap drift algae have beenincorporated in a long-term govemrnent subsidyprogram. The program was the Coastal FishingGround Improvement and Development Project,called the Ensci Project, initiated in 1976 topromote the enhancement of artificial habitats inJapanese coastat waters {Stone et al. 1991!,

It is well known that new substrata tnaylead to an increased abundance of algae. Thisempirical knowledge has encouraged the

operations of constructing artificial substrata,mainly with concrete blocks and quarry rocks tocreate kelp beds. However, the operationsfrequently failed to establish kelp forests asexpected. Many of such failures seemedattributable to sea urchin grazing. Thus, a numberof attempts were made to protect kelp plants fromanimal grazing with physical barriers, such as gridfences, plastic nettings and plastic-seaweed frills,as well as with chetnical repellents Kawamata1994!. Nevertheless, no technique is available forthe artificial development project in sea urchin-dorninated barren grounds. A recent study showedthat wave-induced distu.rbance may restrict seaurchin feeding, thereby maintaining kelp forestsadjacent to sea urchin-dominated areas Kawarnatain press!, This view would be of great sa!ue toengineering practices.

On the other hand, the necessity ofpreserving drift algae in nearshore rocky habitatshas been recognized by field researchers andfishermen involved in abalone and sea urchinaquaculture, but little has been known about theirphysical behavior in the field. A series of

engineering studies revealed the effectiveness of'various types of structures of water motion in thelaboratory and in the field Kawamata 1987, 1988,199 I, Kawamata et al. 1993!. The results werereflected in the publication of a guide book issuedby the Japan Coastal Fisheries PromotionAssociation l993! for design of artificialstructures under the Ensei Preject. The previousstructures were "stable," or consisted of fixed

materials. It is predicted from the guide that anyof the stable structures could hardly trap andpreserve drift algae in many of the shallow rockyhabitats for a long enough period. Ta cope withthe difficulty, recent studies e.g., Kawamata andSuzuki 1995! developed a new trap with apendulum-like door that is inoved by wave-induced oscillatory flow.

This paper describes the itnportance of thewave-induced water motion to ecological balan':between plants and herbivores, thereby showingthe potential for establishment of kelp beds inengineering modifications. State-of-the-arttechnology of trapping drift algae is also described

ARTIFICIAL DEVELOPMENT OF KELPBEDS

It is welt known that feeding by seaurchins may cause devastating effects an benthicmarine plants Lawrence 1975, Lubchenco andGaines 1981!. Unlike other benthic herbivoressuch as abalone, sea urchins have hard teeth andthereby easily feed on the stiff stipe and holdfastof macroalgae Sea urchins may aggregate anddenude the substratum of foliose algae, farmingbarren grounds that are covered solely byencrusting coralline algae. Sea urchin-dominatedbarren grounds are widely observed a1ong thecoasts of the Japanese archipelago. Suchcommunity types show long-term persistencebecause sea urchins can survi ve and reproduce infaod-limited environments, However, preferredfoliose algae are frequently abundant in shallowwaters next to the sea urchin-dominated barrenzones. The wave-induced benthic oscillating flowincreases with decreasing depth more precisely,up to a wave breaking point!, so that the watermotion in the shallow depth constantly preventssea urchins from feeding on algae, even duringcalm sea periods, There is much evidence

supporting the hypothesis that the absence of algalplants from deeper or sheltered sites results fromherbivorous grazing but nat fram the shortage oflight intensity or nutrients, First, experimentalremoval of sca urchins led to re-establishment ofinacraalgae Iwate Prefectural FisheriesExperimental Station 1988, Agatsuma et al. 1997!,In addition, it is frequently found that underwaterfloating objects such as mooring rapes, whichbenthic herbivores can scarcely climb, areovergrown by kelp, even immediately abovebarren beds.

EFFECT OF WATER MOTION ON SEA

URCHIN GRAZING

A previous study Kawamata in press!evaluated the restrictive effect of the wave-inducedoscillating flow an feeding by the sea urchinStrorrgyloeentrotus rtudus. Thc sea urchin iscommercially important but is frequently a causalagent in clearance of macraalgae along the coastof northern Japan, from Hokkaido to centralHonshu. The method of the study was briefly asfollows. A kelp Drrrirtaria spp.! food with givendimensions was anchored to the bottom in anoscillating flow tank, where starved sea urchinswere contained. A feeding experiment was thenconducted for one or two days to examine thefeeding rate under a periodic oscillating flow. Theexperiments showed that the restrictive effect ofthe oscillating flow on feeding rates somewhatvaried with the animal size and food morphology,but indicated a mechanical constraint that strictlyinhibited urchin feeding at a moderate watervelocity, approximating 30-40 crnisec. The seaurchin's mouth is at the center of its attachmentbase, so that it must mount a thallus by detachingmore than half the number of tube feet used tocling to the substratum. Sea urchins wer edislodged when they would try to eat at suchmoderately high velocities. These findings led toa conclusion that the urchin feeding on foliosealgae is nearly inipossible beyond 40 cm/sec.

Other sea urchins seem to show similarvelocity limits for feeding. Kawamata unpublished data! examined feeding rates of seaurchin Herrricentrorus pulcherrimus in theoscillating flow in the same method as describedabove, The sea urchins of 45-rnrn test diamete~

approximating the maximuin size! showed higherfeeding rates at the higher temperature over thepeak velocity. However, the feeding rate underboth temperatures began to cease at approximately40 cm/sec.

The finding that sea urchins cannot feedon kelp at the peak velocity higher than 40 c m/secmight give quantitative estimates forunderstanding the spatial distributions of seaurchins and kelp. In shallow subtidal areas wherewaves are constantly broken, kelp are usuallyabundant, The wave-induced peak velocity uis estimated from the equation Denny 1988!:

rr = 0.3[g h + H! f '~, �!

where g is the gravitational acceleration = 9.8 rn/s'!, h the water depth, and H the local wave height.When the bottom of the area is horizontal, the wave

height is solely related to the depth, approximatingDenny �988!

H= 078 h

The wave height within the surf zone somewhatincreases as the bottom slope is steeper. Hencethe wave-induced benthic peak velocity in the surfzone is

i4 �! 0.4 gh!'~ �!

Eq, 3 indicates that the velocity almost everywherein the surf zone exceeds the limit for sea urchinfeeding. Two other typical examples might beexplained by the spatial variation in wave-inducedwater velocity. First, the lower limit of kelp bedstends to be deeper with increasing degree of waveexposure. Second, kelp occur solely on theuppermost part of rock outcrops and artificialstructures, where absence of kelp from the lowerpart is unlikely to be attributable to light intensityor drift sand Terawaki et al. 199S!.

In general, the peak velocity on thesubstratum produced by surface waves is estimatedfrom wave data through numerical computation e.g., Kawamata in press!. Several problemsremain in accurately predicting kelp abundancein the field, including engineering problems onpredicting local water flow in the vicinity ofrnicrohabitat and biological ones on algal growth

in nature. However, recent studies Kuwahara et

al. 1997, Kawamata in press! indicated that thevelocity limit for feeding may give a reasonableestimate for the area with kelp plants exposed tointensive animal grazing,

ENGINEERING TECHNIqUES FORARTIFICIAL KELP BEDS

Several observations indicate an

engineering possibility of establishing kelp bedsby increasing the water velocity: kelp overgro~the uppermost parts of concrete blocksimmediately below low water level while kelp areabsent from the lower parts Terawaki et al. 1995!;kelp are abundant on the onshore side of permeablebreakwaters but absent from the onshore side of

less perineable ones.Although it is easy to increase the water

velocity with conventional engineering structures,attention should be paid to other aspects of thewave effect on algal populations, such as breakageand dislodgment by waves. When an object isp!aced under waves, the water velocity is higheron the op of the object. The increased watervelocity inay lead to the higher maximum watervelocity at severe waves, thereby increasing therisk of kelp breakage and dislodgment. Noquantitauve information is available for estimatingthe breakage and dislodgment of kelp. In additionto this biological problem, no practical method isavailable for estimating the local water velocityon the surface of structures under waves. Despitethese problems, the velocity liinit for sea urchinfeeding will be undoubtedly an important criterionfor deciding how to design or allocate artificialstructures for kelp,

DEVELOPMENT OF DRIFT-ALGAL TRAP

BACKGROUND

Aimed at increasing food availability bytrapping drift algae, various types of artificialstructures have been constructed in nearshore

rocky fishing grounds of Japan. In general, thesestructures may be divided into two types: blockor grid. The block type settles drift algae on theupstream and downstream sides by controlling thesurrounding fluid motion while the grid typeobstructs drift algae by nettings with litt!e variationof the flow. In practice, however, previous studies

INTRODUCTION

st

Z3Z Ujlt R Tedttdetu Report Na. Za

suggested that the past attempts were too optinusticartd that any of the conventional fixed structures<an hardly control drift algae in wave-exposedsha]]ow areas for a long enough period. Drift algae«re considerably lightweight in seawater. Forexamp]e, the ratio of weight in seawater to that inair approximates 0.025-0.04 for Laminaria spp.and 0,04-0.1 for Eisenia bicyclis Kawainata]99]!. In addition, the fall velocity, which is aItteasure of the difficulty in bemg raised or movedby turbulent water flow is approximate]y 2 to 4cm'sec for Wrrtinaria and 8 em sec for F. bicyc]is Kawamata et «l. 1993!. Therefore, drift algae areeasily transported and raised up from the bottomby turbulent water flow, Occurrence of turbu]enteddies is associated not on]y with rugged bottomand wave breaking but a]so with blockage effectof traps themselves. For example, let us considerthat a very long impermeable concrete block witha height of 2 m is deployed parallel to the shoreon aflat bottotn at 6 m depth. It is predicted fromKawamata et al. �993! that drift lamirtaria maybe raised up and spread out of it by wave-inducededdies at wave heights of only 0.4 rn for a waveperiod of 10 sec. Further, a field study on 1.5-mho]low concrete cubes fitted with 10-crn meshgrids, specif ically designed to trap drift algae, hasdemonstrated that E. bicyclis plants were raisedout of the cubes by w ave-induced turbulence, evenat the peak bottom velocity of as large as 20 cm/sec Kawamata et al. 1993!. In addition, coasta]water flow may transport drift algae in al]directions. Consequent]y, drift algae may soonbe driven away from a barrier which does notenclose trapped drift algae Kawamata et al. 1993!.

The newly developed drift-algal trapconsists of a rectangu]ar cage with a door andstopper Fig. 1!. The door is a p]ate as a whole, ora flow-shield and grating at the lower and upperParts respectively, which is suspended by hingesattached to the cage. The stopper consists of elasticbodies such as rubber and springs installed to stophe door frotn swinging outwards and lesseningtlte shock of collision. By p]acing the trap on thee-abed as the door confronts prevailing waves, the

door is opened inwards by drag force exerted

primarily on the f]ow-shield when thc wave-induced water flow crosses the door inwards.Otherwise, the door is closed hy the drag andgravitational forces

As readily imagined, when drift algae aretr«nsported straight to the door by oscillatory flow,they may enter the trap with incoming water whenthe door is opened, and then are confined in thecage because the door is closed when the directionof flow is reversed. Furthermore, the trap tnayalso capture drift algae passing by the trap. Thisprocess is slightly complicated and will bedescribed later.

MODEl FOR ESTIMATING THEIMPULSIVE FORCE

Because the door and the fluid near bothsides of the flow-shield move freely and stopsuddenly at the stopper, the consequent impulsiveforce may be considerably larger than thc drag onthe trap The impulsive force F is estimatedfrom the equation Kawamata and Suzuki 1995!:

r 8 r lF =k' g + MpbIt, � > � � � +-, �!4 h, 3n g 4 r � It/2'

where k is the modulus of the stopper, I the momentof inertia of the door, C the coefficient, p thedensity of the fluid, b the width of the flow-shield,

Ftgara 1. Drift-algal trap with a tnoveable door.

r thc height of the door, h, the height of the flow-shield, and u, the instantaneous water velocity atcollision. The coefficient C is empiricallydetermined as 1 Kawamata and Suzuki 1995!.

TRAPPING MECHANISMS

A laboratory scale-model experiment Kawamata and Suzuki 1995! clarified the trappingmechanisms and efficiencies. Under progressivewaves, drift algae are transported in the wavedirection with oscillatory motion When a trap isplaced and oriented with the door facing the wave-corning direction, drift algae are trapped in threeways, First, drift algae straightly approaching thedoor arc trapped as described earlier. Second, driftalgae moving in the course outside of the trap aregradually drawn toward the side of the door, andthen are moved back to the front of the door and

enter the trap. Third, drift algae approaching thedoor from the front are once transported to theside of thc door, and then are trapped in the sameway as in the second process. The fact that driftalgae are carried obliquely to the door and thatdrift algae once pass by the door and then aredrawn back to the front of the door are explainedby the crosswise flow produced as follows, Whenthe direction of oscillatory flow turns fromoutwards to inwards of the door, the door turnsback to the upright position and stops at thestopper. However, the water iminediately outsideof the flow-shield is entrained by the fiuid passingover the flow-shield and then moves outwards,followed by the fluid from its sides. The crosswiseflow canies drift algae fmm the sides iminediatelyahead of the door so that they are readilytransported into the trap with the subsequentinward flow. When the moment of the

hydrodynanuc force on the door is large comparedwith that of the gravitational force on the door i.e,the door readily follows the flow!, drift algaecoining from the front of the door are mostlytrapped in the first process, Otherwise, they aretrapped in the third process, because the slowlytnoving door produces a high pressure region ora separated flow region! in front of the flow-shieldwhen the water begins io move inwards of the door.

TRAPPING EFFICIENCIES

Let us consider that drift algae are

transported with progressive waves near the trapfrom the front to the back of the door, and denotethe crosswise distance between the course of

oncoming drift algae and the axis of the trap as y.A laboratory scale model experiment Kawamataand Suzuki 1995! showed that a trap with the doorproperly designed has the following ability to trapdrift algae. The probability that drift algae comingin the course of y wifl be trapped can be high morethan approximately 90%! when the course iswithin the door i.e., y < b/2! and then decreasesas the course is more distant from the door, The

approaching course at which the trappingprobability begins to be zero may reach the doorwidth away from the side of' the door i.e., y =1.5b!, The relationship between the. trappingprobability and the ratio y/b seemed independentof the absolute value of the door width. Hence,the trapping efficiency defined as the integrationof the trapping probability overy divided by h mayreach almost 2. This suggests that the trap mayhave the equivalent of completely capturing driftalgae passing thmugh twice the width of the door.The trapping efficiency may remain at such a highlevel when the height of the flow-shield is greaterthan a limit, which is never lower than 0.25 m infull scale. The higher the door or the flow-shield!,the greater the impulsive force on the stopper.Thus, the optimal height of the flow-shield is thelimit, probably approximating 0.4 to 0.5 rn.

FIELD TESTS

To verify the effectiveness of the trap.field experiments were conducted with a simpletest device from August to December 1994 Kawatnata and Suzuki 1995! and werc redonewith a revised test model from August 1995. Thedevices were placed at 9-m depths on a relati velyflat boulder area on the northeastern Pacific coast

of Honshu, Japan �8'22'N,141 26'W!. The sitewas immediately offshore of a steeply sloping bedthat reached the shore. The shore was partlyprotected but constantly washed by waves. KelpK bicy elis were abundant immediately below lowwater level while the deeper area was barren witha high density of sea urchin Strorrgyfoceritrorusnudus. Abalone Haliotis discus hanriai occurred

mostly in the kelp bed but wit.h lower density.Wave action seemed to prevent sea urchins from

234 UJMt Taettateai Report No. 26

Figtsre 2. The first test device placed on the eaperitnenta!site

invading the shallow kelp bed, as described byKawamata in press!.

The first device was a stainless steel cage�.4 x 2 x 0.6 m! with a door of 53 cm height,whose lower portion was covered with a 5-mm-thick fiber remforced plastic board to create a 286-mm-high flow-shield. The mass and the momentof inertia of the door were 11.1 kg and 1.54 kg/in', res pecti vely. The trap was firmly anchored tothe bottom using underwater drilling equipmentand oriented with the door facing the shore andperpendicular to the direction of oscillatory flow Fig. 2!. Three iron springs with 300-kg capacitymounted on iron plates �5 x 30 x 1 5 cm! wereseparately embedded in front of the door as thestopper. The device successfully trapped driftFisenita nearly to the full, In addition,investigations suggested that the trap could holddrift algae until they were consumed by aggregatedsea urchins. Despite such high effectiveness,engineering problems for practical applicationsurfaced during the first test. First, stainless steelmay be corroded in seawater, so that making adoor of stainless steel might result in a heavy doorwith a lower trapping efficiency. Thus, it mightbe better to use more cormsion-resistant metal with

a smaller specific inass density. Second, tofacilitate the installation of traps, the stoppershould be combined with the main c-age and thetrap should be simply fixed by weights in a general

Htttsra 3. The titaniutn drift-algal trap test device.

way. Final.ly, drift ke]p caught in the cageaccutnulated from the innermost part of' it but didnot pile up approximately 0.4. tn above the bottom,suggesting that the height of the cage could belower without decreasing the effective capacity fordrift algae.

Improved in these respects, the seconddevice was made as a more practical modeL Thedevice was a 2.5 m !ong x 2,0 in wide x 43 cin-high titanium cage incorporated with a titaniumdoor and rubber stoppers Fig. 3!, The door was a196 cin wide x 40 cm -high titanium platereinforced with thicker plates at the margin. Thcmass and the moment of inertia of the door was

6,37 kg and 0,361 kg/m', respectively. Iron weightamounting to 1920 kg was placed in the innermostpart of the cage. The weight necessary to stabilizethe trap was estimated from Eq. 4 with a designwater velocity of 1.2 mtsec, which was determinedfmm the velocity measurement conducted in 1994.The improved device also succeeded in trappingdrift kelp up to the full approximately 80 kg wetweight! Fig. 4!, and in holding drift algae untilthey are consumed by aggregated sea urchins Fig.5!. Drift algae were almost absent around thedevice throughout the year except in early autumn,whereas drift algae frequently reinained in the trapwith intensive grazing by sea urchins. The doorwas amended because of slight damage due to itsinsuflicient stiffness. With this amendment, the

Kawasstata 235

DISCUSSION

Flgtsre 4. The improved test device trapping drift kelp tothe full.

device has been functioning over 2 yr, showingproini sing results,

Observations supported the hypothesisthat most drift algae were swept out of shallowhabitats before being consumed by animals undernatural conditions, Abalone and sea urchinsoccasionally captured small pieces of drifting kelp,but never fed on entire detached kelp plants. Itwas observed that 1 arge amounts of drift kelp wereaccumulated in a crevice near the trap in earlyautumn. However, the drift kelp soon disappearedwithout aggregating animals. The drift kelp in thecrevice occasionally oscillated itt a "huge" body,with the result that small aniinals hardly graspedthem. ln contrast, the test devices frequentlycaught drift algae and maintained them until theywere consutncd by congregated sea urchins.

The observed variation in trapped driftalgae suggested that drift algae sporadically occurat storms, especially during the first storm aftersummer, in which kelp biomass reaches maximum.Considering such sporadic occurrence of driftalgae, an effective drift-algal trap is a device whichcan catch a large amount of drift algae occurringat rough seas and can reserve them undersubsequently repeated severe waves. Although thecapacity of the test devices was too small

Figure 5. Sea urehins aggregated at the trapped drif kelp.

compared with a great consumptive capacity ofaggregated sea urchins, it could be expected f'romthe high trapping effectiveness that the devicetnight be an effective technique for nearshorerocky aquaculture.

Only one or two abalorie were found inthe device, probably because of a low populationdensity. However, observation made in lateAugust 1997 recorded nine adult shells in thedevice,

Since the densities of drift algae and theirherbivores also vary with the location, the trapshould be placed at the path on a nearshore, deeperbarren ground, through which plants detached fmmkelp beds may frequently pass. Like this field test,a small embayrnent with a relatively flat anddepressed bottom near the opening is a potentialappropriate site for application,

With high trapping efficiencies, the trapmay also be used as an "automatic feeding system"for underwater cage culture, e,g., by stockingstarving adult sea urchins or lean adult abalone ina cage with adequately small mesh grids placedclosely behind the trap.

Finally, the durability required to resistthe repeated collision should be exanuned by fieldtests, The present experiment will be continuedto validate its practical applicability.

zs6 Usia TcchakA Rtyorl!va 26

CONCLUSION

In nearshore rocky fishing grounds witha high density of sea urchins, kelp beds may beconfined to areas where the wave-inducedoscillatory flow cons4mtly prevents sea urchingrazing. The velocity limit for feeding by seaurchins is approximately 40 cmlsec, There is apossibility of developing artificial kelp bed.s in seaurchin-dominated areas by increasing the wave-induced water velocity at calm sea periods. Insuch areas, drift plants may be the primary foodof sea urchins and abalone, but most of thetn maybe swept out of hc shallow habitats by watermovement. The previous stable structures couldnot trap drift algae in most shallow habitatsbecause of the high mobility of drift algae in waterflow, The device developed by Kawamata in 1994could catch drift algae effectively and hold themunti! consumed by animals,

LITERATURE CITED

Agatsuma, Y., K, Matsuyama, A. Nakata, T.Kawai, and N, Nishikawa. 1997. Marinealgal succession an coralline flats afterremova! of sea urchins in Suttsu Bay on theJapan Sea Coast of Hokkaido, Japan.Nippon Suisan Gakkaishi 63: 672-680, [InJapanese�].

Denny, M.W. 1988, Biology and the Mechanicsof the Wave-swept Envimnment. PrincetonUniv. Press, Princeton, NJ. 329 p.

I wate Prefectural Fisheries Experimental Station.1988. Ganshou-kaitei ni okeru moba-zosei-

jouken ni kansuru kenkyu study on thecondition of kelp-bed deve!opment on therocky bottom!. Specific Research GrantProgram Report, 1975-1977. IwatePre fectural Fisheries Experimental Station,Kamaishi, Iwate. 15 p, [In Japanese].

Japan Coastal Fisheries Promotion Association.1993. Structui'a! design guide for theCoastal Fishing Ground Improvement andDeve!oprnent Project Engan-Gyojo Seibi-Kaihatsu-Jigyo Kozobutsu Sekkei-Shishin!.[In Japanese],

Kawamata, S, 1987. Drift alga traps to feedbenthic a!ga! feeders. Fish. Eng. 24�!: 53-

60. In Japanese!.Kawamata, S. 1988. Developmental studies on

block-type drift alga traps with slitopenings. Bul!. Natl. Res. Inst, Fish. Eng.9: 1-8. In Japanese!.

Kawatnata, S. 1991, Physical considerations forthe design of algal drift traps. U.S. DeptCommerce, NOAA Tech. Rep. NMFS 102171-1 80.

Kawamata, S. 1994. Importance of physicaldisturbance for artificia! development ofnearshore rocky fishing grounds, Fish. Eng.31�!: 103-110. [In Japanese].

Kawamata, S. In press. Effect of wave-inducedoscillatory flow on grazing by a subtidal seaurchin Srrongylocenrrorus nudus A.Agassiz!. J, Exp, Mar, Biol. Ecol.

Kawamata, S, and H, Suzuki. 1995. Deve!opmentof a drift-a!gal trap for nearshore rockyaquaculture, pp. 628-633 In: Proceedingsof the International Conference on

Ecological System EnhancementTechnology for Aquatic Environments S i xth In t. Con f. Aquat. HabitatEnhancement!. Japan International MarineScience and Technology Federation, Tokyo,Japan,

Kawamata, S., S, Hagino, and M. Yamamoto.1993. Modome-shisetsu no kaiso-hosoku

koka ni kansuru kenlr.yu studies on thetrapping effects of drift algal traps!. Heisei3-Nendo Engan-Gyojo Seibi-Kaihatsu-Jigyo ni Kansuru Suisancho-KenkyusyoKenkyu Hokoku reports of the researchinstitutes of Japan Fisheries Agency on theCoastal Fishing Ground Irnprovernent andDevelopment Project!, pp.! 39-164. InJapanese!.

Kuwahara, H., S. Akaike, H Hayashi, and Y,Yammhita. 1997. Mechanism of groupseaweed formation in isoyake area, Jpn.Sac. Civil Eng. 44: 1181-1185. [InJapanese].

Lawrence, J.M. 1975. On the relationshipsbetween marine p!ants and sea urchins,Oceanogr. Mar. Bio!. Ann. Rev. 13: 213-286.

Lubchenco, J. and S.D. Gaines. 1981. A unifiedapproach to marine plant-herbivore

interactions. I. Populations andcommunitics, Annu. Rev. Ecol. Syst. 12:405-437.

Stone, R.B�J.M. McGumn, L.M. Sprague, andW. Seaman, Jr. 199L Artificial habitats of

the world: synopsis and major trends, pp.31-60. In: W. Seaman, Jr. and L.M.

Spraque eds.!, Artificial Habitats forMarine and Freshwater Fisheries.

Acadetnic Press, San Diego, CA,Terawaki, T., S. Arai, and Y. Kawasaki. ]995.

Methods of subtnarine forest formation

considering local limiting factors ofdistribution. Fish. Eng, 32�!; 145-154. InJapanese!.