Sediment resuspension by wind in a shallow lake of …F...Ecological Modelling 186 (2005) 63–76 Sediment resuspension by wind in a shallow lake of Esteros del Iber´ a (Argentina):

Ecological Modelling 186 (2005) 63–76

Sediment resuspension by wind in a shallow lake ofEsterosdel Ibera (Argentina): a model based on turbidimetry

Andres Cozara, c, ∗, Jose A. Galveza, Vincent Hullb,Carlos M. Garcıaa, Steven A. Loisellec

a Area de Ecolog´ıa, Facultad de Ciencias del Mar y Ambientales, Universidad de Cádiz, Campus R´ıo San Pedro,11510-Puerto Real (Cádiz), Spain

b Laboratorio Centrale di Idrobiologia, Ministero delle Politiche Agricole e Forestali., Via del Caravaggio 107, 00147 Roma, Italyc Dipartimento di Scienze Chimiche e dei Biosistemi, Università di Siena, Via AldoMoro 2, 53100 Siena, Italy

Available online 12 February 2005

Abstract

A model based on empirical relationships is used to study frequency and magnitude of the sediment resuspension by wind-induced waves. The model has been developed for Laguna Galarza, a mesotrophic round-shaped shallow lake located inEsterosdel Iberawetland. Given the logistic and accessibility difficulties of this macrosystem, the installation of automated field stationsfacilitated continuous data acquisition. Using the wave theory, a daily spatial model of resuspension was built from simultaneoustime series of hourly measurements of infrared nephelometric turbidity, wind speed and wind direction. The model was usedto predict total suspended solids in another lake of the wetland (Laguna Ibera) showing a good agreement with observed fieldv discusst tion of thep f the recenta©

K

1

ta

f

sesoughpro-;pey,eticitalergyand

0

alues, even although Laguna Ibera has a more irregular contour and a eutrophic state. Finally, we apply the model tohe ecological impacts of resuspension on the distribution of the littoral communities and to characterize the composiarticulate suspended matter of the limnetic ecosystem. The model was useful to simulate the possible implications olterations of the wetland water level on the resuspension regime of the open water bodies.2005 Elsevier B.V. All rights reserved.

eywords:Shallow lakes; Wind resuspension; Nephelometric turbidity; Suspended particulate matter

. Introduction

Resuspension of lake sediments is a function ofhe bottom shear stress as the result of fluid motionnd the local sediment characteristics (e.g.Hakanson

∗ Corresponding author. Tel.: +39 0577 234360;ax: +39 0577 234177.E-mail address:[email protected] (A. Cozar).

and Jansson, 1983). Several hydrodynamic procescan intervene on sediment resuspension, althwind-induced waves are usually the dominantcess in shallow lakes (e.g.Leuttich et al., 1990Weyhenmeyer et al., 1997; Douglas and Rip2000). The wind surface stress induces an energwave-affected layer in which both large-scale orbmovements and the dissipated turbulent enare important. The closeness of the surface

304-3800/$ – see front matter © 2005 Elsevier B.V. All rights reserved.doi:10.1016/j.ecolmodel.2005.01.020

64 A. Cozar et al. / Ecological Modelling 186 (2005) 63–76

Nomenclature

D water depth (m)F fetch (m)Lw wavelength of the swell (m)SR resuspended solids, daily mean (mg/L)S0 background concentration of suspended

solids, daily mean (mg/L)TSS total suspended solids, daily mean

(mg/L)TT total turbidity, daily mean (NTU)TR turbidity by resuspension, daily mean

(NTU)T0 background turbidity, daily mean (NTU)W estimation of the daily wind speed (m/s)Wmax maximum daily wind speed (m/s)Wmed mean daily wind speed (m/s)W0 theoretical wind speed that starts the re-

suspension (m/s)

Greek lettersα intercept of the potential equation link-

ing TT andWβ exponent of the potential equation link-

ing TT andW

bottom boundaries in shallow lakes often generatesa completely mixed water column during the resus-pension events. In contrast with the surface or densityboundaries, the bottom boundary is rigid, and the wind-induced current speed vanishes at the water–bottominterface, through the turbulent dissipation of energyand the subsequent vertical mixing (Grant and Madsen,1986). A two-layer vertical structure of horizontalcirculation is unlike to form during these events inshallow lakes (Webster, 1990) and the bottom shearstresses associated with the horizontal currents aregenerally too small to influence suspended solidsconcentration (Leuttich et al., 1990). The main effectof the horizontal advection would be to smear ofthe horizontal distribution of solids resuspendedby the energetic wave-affected layer (Bailey andHamilton, 1997), thereby having a secondary role onthe resuspension process. Therefore, the modelling ofthe resuspension phenomenon in shallow lakes could

be simplified by focusing on the wave orbital motionand the associated turbulence.

As the air–water and water–bottom interfaces cre-ate bottlenecks for the energy exchange that involvecomplex processes, the mechanistic modelling of theresuspension process depends on multiple factors thatare difficult to estimate (e.g. wind drag coefficient, ver-tical mixing speed). Using the dependence of the depthof the energetic wave-affected layer on wavelength, apractical spatial analysis of the susceptibility to resus-pension can be performed in shallow lakes (Carperand Bachmann, 1984; Scheffer, 1998; Nagid et al.,2001). For specific sediment conditions, the resuspen-sion regime of a certain lake may be characterized usingan empirical correlation between the concentration ofsuspended solids and the wind speed if enough data areavailable (Scheffer, 1998). In this way, sediment trapsand filtering of water samples are common methods tostudy the resuspended particles flux in the water col-umn (Evans, 1994). However, while these methods canprovide valuable information on the characterizationof the suspended material, they have a limited useful-ness in the study of the sudden temporal variations ofthe resuspended solids concentration. The inaccessi-bility of the studied wetland (Esteros del Iberá) ham-pers an appropriate periodicity of sample collectionwith these methods during periods of severe storms.In this work, the building of the resuspension modelwas based on hourly time series obtained by an infrarednephelometric turbidimeter, which simplified the dataa

d bes ther ollu-t ormst talsa e oft esti-m thet llowl i-m e fi-n .g.C en-s olvedn ,1 sig-n of

cquisition.The ecological effects of the resuspension coul

ummarized at three fundamental levels. Firstly,esuspension works directly on the nutrients and pants dynamics. The sedimentation process transfhe lake bottom into a store of nutrients, toxic mend organic pollutants. In spite of the intermittenc

he resuspension process, an order-of-magnitudeate of the contribution of resuspended material to

otal flux of particulate suspended matter in the shaakes is about 80–90% (Evans, 1994). Thus, the sed

ent resuspension conditions the availability and thal fate of all biologically active particulate matter (eotner et al., 2000). On the other hand, the resuspion modifies the release rates of regenerated dissutrients from the sediment (e.g.Ogilvie and Mitchell998). Secondly, the resuspension events modifyificantly both attenuation and temporal fluctuation

A. Cozar et al. / Ecological Modelling 186 (2005) 63–76 65

light in the water column.Hellestrom (1991)calculatedthat this effect would be sufficient to reduce primaryproductivity in a shallow Swedish lake by about 85%.Thirdly, the wind is considered one of the strongestfactors influencing the zonation of aquatic vegetation.Even moderate wave heights (0.1–0.15 m) may causenegative effects on plants with morphology that maybe tolerant of the wave activity, such as long ribbon-like leaves (Doyle, 2001). Both wave energy and sed-iment remobilization have shown dramatic impacts inthe vegetal colonization of the lakes (Galinato and vander Valk, 1986; Foote and Kadlec, 1988).

It is well reported that a water level change inshallow lakes can cause large and rapid ecosystemshifts. Sediment resuspension is often involved inthe internal mechanisms causing such alterations.A drop in the lake water level usually contributesto an increase in resuspended sediments which leadto a higher trophic state and lake turbidity. This hasbeen well reported in shallow ecosystems such as inLake Chapala, Mexico (de Anda et al., 2001) or LakeNewman, Florida (Nagid et al., 2001). Likewise, therise of the water level can promote undesirable changesin the shore vegetation and the state of the lake.Engeland Nichols (1994)reported how, in the early 1970s,during a high water period of Rice Lake (WI, USA),waves damaged the littoral marsh vegetation anderoded the bottom. The lack of vegetation and theresulting softer sediment altered the dynamics of thelake causing higher turbidity. Once the water levelr uedt rbidc ion.

per-t cal,t itsm tivem tantt hisw ilityo llowl ).T icalr rk.I oft etice ts oft

2. Study site

The study has been focused inEsteros del Iberá,one of the most pristine and largest wetlands of SouthAmerica (13,000 km2). This subtropical wetland islocated between 27◦36′–28◦57′S and 58◦00′–57◦30′W(northeast Argentina). The macrosystem consists of amosaic of marshes, swamps and open water bodies.The wind-induced resuspension plays a relevant rolein the dynamics of the wetland lakes because of theirshallowness. The larger lakes of the macrosystemhave an average depth about 2.5 m. Most of the lakesremain permanently flooded with a small seasonalvariation of the water level. However the inhabitantsof the region have reported rises of the water level inthe last years, a fact which highlights the need to studythis poorly known ecosystem. Thanks to the existenceof a nearby small village (Colonia Pellegrini), a seriesof water level data from 1968 is available for LagunaIbera. The analysis of this register revealed a suddenincrease of 80 cm of the average water level cm in1989 (Ferrati and Canziani, 2005). Esteros del Iberáis mainly fed by rain. However, data analysis suggeststhat a possible cause of this water level increase isrelated to a change in ground water flux associatedto the construction of the Yacyreta hydroelectric damon the River Parana (Ferrati and Canziani, 2005). TheYacyreta reservoir is located at the north ofEsteros delIbera, separated by a thin strip of land (4–12 km wide).The wetland drains only through River Corriente in thes lopea n theb thee

inlya h1 d int malls reat t is as owl rzav y oftm or-p intot arrierr ater

eturned to normal, the littoral vegetation contino be excluded for at least a decade and the tuonditions remained due mainly to wind resuspens

Sediment resuspension as an environmentalurbation plays a significant role in geochemioxicological and biological processes. Due toultiple effects on the ecosystem, the quantitaodelling of the resuspension process is impor

o understand shallow lakes functioning. In tork, we attempted to model the spatial variabf the resuspension process of a particular sha

ake ofEsteros del Iberá wetland (Laguna Galarzahe integration of physical models and ecologesults help to build a better predictive framewon the last section, we illustrate the applicabilityhe resuspension model on the littoral and limncosystems, paying attention to the potential effec

he water level variation inEsteros del Iberáwetland.

outh. The drainage is slow due to the very flat snd the huge amount of vegetation accumulated iasin. At the present, the wetland has maintainedlevated water level that resulted from 1989.

The present modelling effort has been maddressed to Laguna Galarza (16 km2, mean dept.9 m), a mesotrophic round-shaped lake locate

he oriental border of the Ibera macrosystem. A stream (Yacare) that originates in the vast marsh ao the northeast feeds the lake, while the lake outlemall stream (Isirı) that connects with another shallake (Laguna Luna). The water level of Laguna Galaaries about 0.4 m throughout the year. The validithe model was also examined in Laguna Ibera (58 km2,ean depth 3.2 m), which has a more irregular mhology and a eutrophic status. This lake is divided

wo basins by a narrow passage that acts as a beducing the interchange of wave energy and w


masses. A small river (Rıo Miri nay) feeds the southernbasin. The bottom of both lakes is quite flat and homo-geneous. It is dominated by silt-clay, with fractions ofvery fine sand and high percentage of organic matter(Varela and Bechara, 1981). Macrophytes cover wasscant (usually below 5% lake area), and practicallyabsent in the open waters. The vegetation of the marshareas is dominated by emergent macrophytes suchasTypha latifolia, Scirpus californicus, HymenachneamplexicaulisandEchinochloa helodes. The shorelinehas vegetation that usually varies from coast to coastbut often concludes in extensive floating mats ofvegetation and organic matter called “embalsados”.

3. Methods

The bathymetry of the lakes was outlined with smallportable equipment consisting of a global positioningsystem (GPS) and an echosounder. The surveys wereperformed during June 1999 in Laguna Galarza andduring February 1999 in Laguna Ibera. Over 30 tran-sects in each lake were run to outline the structure ofthe isobaths (Fig. 1).

Two meteorological stations, one in each lake, andone hydrological station were used for autonomousmonitoring. All sensors were connected to data loggersthat received energy supply from a battery connected toa solar panel. Sensors of Vector Instruments were usedto measure wind speed (model A100R) and direction( itha ro-l wasa unaG etricd m.I tot tter( whent ncei . Ift a arer alizest ria-t timesa wasp Thet nder

the water surface. Thus, the sensor only registersresuspension events that involve mixing of the wholewater column, intervening in the light attenuation ofthe superficial layer. Since the resuspended particleshave a relatively high sinking velocity (e.g.Kristensenet al., 1992; Ogilvie and Mitchell, 1998), once windhas stopped, these particles fall out of the surface layerand are not registered by the sensor. In this way, thesensor does not register the settling particles due tothe resuspension by previous events.

Simultaneously to the monitoring period (March1999–August 2000), a general limnological study(nine samplings dates; Secchi depth, chlorophyll aconcentration, plankton composition) was carried outto obtain a biological characterization of the studiedlakes (Cozar et al., 2003). During the periods of higherphytoplankton biomass, the long-term use of theturbidimeter without continuous cleaning resulted inalgal growth on the turbidimeter window. The block-age of the window occurred even by simple depositionof filamentous branches. This fact prevented the use ofthe turbidimeter in the eutrophic Laguna Ibera, since itwas observed that intense biological growth occurredaround the sensor cage. In order to develop a consistentwind-turbidity correlation, a significant problem wasthe difficulty to discriminate between the turbiditydue to the wind-induced resuspension and that due tobiological activity or other physical process (e.g. rain).During the monitoring period in Laguna Galarza, weacquired four simultaneous turbidity-wind registerso ner orals d.T oft als

gt tifys land,w eraw redt ans e ofo theo them on.S tiont tent

model W200P). Both variables were registered wn hourly frequency. The hydrological station (Hyd

ab Datasonda 2) equipped with a turbidimeternchored in a mid-lake site (2.2 m depth) of Lagalarza. The installed turbidimeter is a nephelometector at 90◦ from an infrared light source of 880 n

nfrared nephelometric turbidity is closely relatedhe scattering coefficient due to particulate maKirk, 1994). The sensor measures the responseshe light source is on and when it is off. The differes used to eliminate the effects of ambient lighthe light environment saturates the sensor the datemoved. A second sensor measures and normhe light source output to correct its possible vaions. Each measurement was repeated threend an average value was used. The calibrationerformed using standard formazine solution.

urbidimeter was submerged few centimeters u

f approximately 2–3 weeks duration. Only oegister (1–20 September 1999) showed a temperies of turbidity driven almost exclusively by winhe remainder showed clear biological control

he temporal variability of the turbidity or abnormensor drifts by the bio-fouling problem.

Suk et al. (1998)analyzed the feasibility of usinhe infrared nephelometric turbidimeter to quanuspended solids concentration in a coastal wetith a sediment composition very similar to the Ibetland lakes. The particles size distribution cove

he range from clay to very fine sand, with meizes within the silt category and high percentagrganic matter. As in the Ibera wetland lakes,rigin of the organic matter in the sediment wasobilization of the litter from the marsh vegetatiuk and co-workers used standard formazine solu

o calibrate the sensor. They found a consis


Fig. 1. Location ofEsteros del Iberá wetland in South America and bathymetry and location of the studied lakes, Laguna Ibera and LagunaGalarza.

linear correlation (R= 0.827) analyzing 593 samplesduring a period of 6 months including periods of severestorms. The relation between turbidity and concentra-tion of suspended solids showed a slope of 1.584 mg/L

per nephelometric turbidity units (NTU). We haveinitially used the results from this study to transformthe turbidity data to concentration of suspendedsolids.


Fig. 2. Hourly temporal series of wind speed (black line) and turbidity (grey circles) from 1 to 20 of September 1999 in Laguna Galarza.

The validation of the model was carried out throughextensive spatial samplings of the total suspendedsolids (TSS) in one resuspension event in each lake.Vertical profiles of temperature (0.5 m intervals) wereperformed along the fetch line. Laguna Galarza wassampled with a grid of 24 points (9 March 2000) andLaguna Ibera with a grid of 40 points (23 June 2000).Samplings lasted 4 h in Laguna Galarza and 8 h inLaguna Ibera. The water samples were collected at0.5 m depth in polyethylene 2 L-bottles. A variablevolume of sample ranging from 250 to 600 mL wasimmediately filtered under low vacuum pressure. Pre-viously, the WhatmanGF/F filters (47 mm-diameter)were dried (60◦C, 5 h) and weighed. The filters withthe retained residue were again dried and weighed.The net weight of the residue and the filtered watervolume were used to calculate the TSS concentration.

The mean sinking velocity of the resuspended sedi-ment in the lakes was estimated through sedimentationcolumns (SETCOL) after Bienfang (1981). The sam-ples for this analysis were acquired after artificial resus-pension events of the superficial lake sediment (3 cm)in microcosms. Sinking velocities of the organic andinorganic matter were discriminated.

4. Correlation between wind and turbidity

The continuum series from 1 to 20 September 1999c yto-p aryp ofw nge

of wind of the region was registered during this pe-riod (Fig. 2). The wind blew basically from the south,coming exclusively from this direction when the hourlyspeeds were higher than 5 m/s. The temporal seriesof turbidity was driven by wind, however some daysshowed a smooth turbidity increment during calm pe-riods and a fall when the wind speed increased again.This process was clearly observable twice, once duringthe first and second days and again during the seventhday. Both periods showed similar inverse hourly rela-tionships between wind speed and turbidity (Fig. 3).A possible explanation for this apparent paradox couldbe a physical or biological accumulation of particlesin the superficial layer and a subsequent dispersionwhen the wind begins to blow. Before the formation

F idity.T d thes

orresponded to the studied period of lowest phlankton biomass, and probably the lowest primroductivity (austral winter). A strong storm (peakind speed of 13.6 m/s) that covered the whole ra

ig. 3. Inverse hourly correlation between wind speed and turbhe empty circles correspond to 1 and 2 September 1999 anolid circles to 7 September 1999.


of well-developed waves, the wind-induced motion isinsufficient for sediment resuspension but not for thedispersion of the particles accumulated in the superfi-cial layer, near the sensor. The superficial accumulationcould be due to the buoyancy of some cells likeMicro-cystis aeruginosa(Reynolds and Rogers, 1976), a dom-inant species in Laguna Galarza. Although these localincrements of turbidity do not modify significantly thegeneral wind-turbidity correlation, we have removedthese days in order to obtain a more precise modellingof the resuspension phenomenon.

The acquisition of simultaneous hourly series ofwind and turbidity allowed a suitable daily aver-age for modelling. The traditional samplings of TSSmake it difficult to acquire an hourly data series(Somlyody, 1982; Aalderink et al., 1985; Kristensenet al., 1992). The use of wind speed on a dailybasis presents two considerable advantages. (i) Thevelocities of sedimentation measured in the settlingcolumns were 12.4± 0.9 m/day for inorganic matterand 6.5± 0.7 m/day for the organic matter. The conse-quent short-residence times agree with other resuspen-sion studies in shallow lakes (Kristensen et al., 1992;Bailey and Hamilton, 1997; Ogilvie and Mitchell,1998). A dynamic resuspension model implies the con-sideration of the settling process of the particles resus-pended by previous events. Nevertheless, according tothese results and the lake shallowness, the resuspendedparticles are not likely to remain suspended in the watercolumn for periods longer than a day. Consequently, thed ionp on-c ,1 en-s thepe hee ave( ngw sure-m sionm ents ationo Dueta inte-g ycleo

Reported mathematical functions linking TSS (ortotal turbidity, TT) with wind speed (W) use thefollowing equation: TT =αWβ +T0 (Somlyody, 1982;Aalderink et al., 1985; Kristensen et al., 1992). TheT0coefficient would correspond to the background tur-bidity, which does not depend on wind resuspension.The plankton, due to its mobility, buoyancy or smallsize, can remain in suspension making a significant parton this background turbidity. Theα andβ coefficientsdepend on the characteristics of the process of remo-bilization of the particles from the bottom. Our dataalso followed this pattern (Fig. 4). A saturation levelof resuspended particles was not observed despite thehighest registered wind speed was measured during thecalibration period (8 September 1999). The lack of sat-uration level of resuspended solids has occurred in oth-ers studies like Lake Arresø (mean depth 3 m), whereup to 2 cm of the superficial sediment was resuspendedin a mid-lake station (Kristensen et al., 1992). Using aparticle concentration of bottom sediments from sim-ilar ecosystems (Carignan and Vaithiyanathan, 1999)or from general resuspension models (Weyhenmeyeret al., 1997), the maximum registered turbidity peakwould erode only 0.2–0.6 cm thick layer. Therefore, itseems probable that the lack of a saturation level is the

Fs m-bW

aily periodicity avoids the modelling of the depositrocess, which complicates the modelling of the centration of suspended solids (Bailey and Hamilton997). The turbulent conditions during the resuspion event may change some initial properties ofarticles, for instance, the tendency to flocculate (Lickt al., 1993). (ii) A daily averaged wind reduces trror due to generation time of a well-developed wwith defined direction, long crests and relatively loavelengths). For the same instantaneous meaents of wind speed and direction, the resuspenay even be different according to the developm

tate of the waves. The time necessary for the formf a well-developed wave is on the order of hours.

o the day–night periodicity of the wind speed (Fig. 2),daily average is the most suitable because it

rates the matter resuspended into single daily cf wind.

ig. 4. Empirical correlation (TT =αWβ +T0) between daily windpeed (W) and mean daily turbidity (TT) from 1 to 20 of Septeer 1999 in Laguna Galarza: (a)W= mean daily wind,Wmed; (b)= maximum daily wind,Wmax.


general feature of the resuspension regime in the softsediments of Ibera wetland.

Diverse possible estimations of the daily wind wereexamined. The mean daily turbidity was adequatelycorrelated with the wind through the mean daily speed(Wmed) and the maximum daily speed (Wmax). Otherpossible quantifications of daily wind were also ex-plored through the combination of bothWmedandWmax(e.g.WmaxWmed, R= 0.9587). The general correlationin the flat segment of the curve is probably due to thevariability of theT0 when resuspension does not oc-cur. The generally good correlations suggest a goodreliability in the use of the day–night wind cycle forresuspension modelling in shallow lakes. The use of amoving 24 h-average of wind speed did not show goodresults.

5. Building of the model

The wave-affected depth depends on the wave-length, which in turn depends on wind speed and fetchline. The dependence of the energetic wave-affectedlayer on the fetch determines the spatial variability ofwind-induced resuspension. We have used an empiri-cal approach of the wave theory developed in coastaloceanography but widely used in the qualitative spa-tial analyses of the resuspension in shallow lakes (e.g.Carper and Bachmann, 1984; Scheffer, 1998; Nagidet al., 2001). The estimation of the wavelength (L )lf

L

R wave-a cor-r eedst sb umw shal-l n,1E ro-l ionp thew

2.3 m/s. The calibration curve shows that resuspen-sion was only registered when this wind speed value isexceeded.

The obtained empirical correlations between TT andWare only valid at the point where the hydrological sta-tion is located and for winds coming from south. As-suming a homogeneous layer of sediment in the lake,we have used the calibrated resuspension curve accord-ing to the respective value ofW0 in each point and foreach wind direction (from Eq.(1)). The proposed quan-titative spatial model is based on the discontinuous typ-ical response of the sediment layer to a gradual increaseof the wind. In this way, the turbidity by wind resus-pension (TR) at any point of the lake can be expressedas:

TR(NTU) = 2.2

Dα(W + (2.3 − W0))βδ(W, W0) (2)

where “α(W+ (2.3−W0))β” determines the turbidityfrom W (m/s) and the resuspension curve referred tothe each value ofW0; δ(W, W0) a step-function thatdetermines when the wind-induced wave begins to re-suspend sediment,δ = 0 if W<W0 andδ = 1 if W≥W0;the “2.2” factor indicates the depth (m) where thewind-turbidity relation was originally determined andD the depth (m) of the new site where it is applied.Dis in the denominator because resuspended solids be-come more diluted when the water is deeper than 2.2 mdepth.

The total turbidity (TT) is obtained addingTRt yt thew -m xisa s-p deds ent perN

S

6

am-p ds.W ion of

w

inks wind speed (W) and fetch (F) in the followingorm (Pond and Pickard, 1983):

w = 1.56

[0.77W tanh

[0.077

(9.8F

W2

)0.25]]2

(1)

esuspension occurs when, at least, the energeticffected layer reaches the bottom. This conditionesponds to the case in which the wavelength excwice the water depth (Lw ≥ 2D). This approach haeen used to make spatial estimates of the minimind speed necessary for resuspension in prairie

ow lakes like Little Wall Lake (Carper and Bachman984) or Lake Newman (Nagid et al., 2001). Fromq. (1) and considering the depth at the hyd

ogical station and the fetch during the calibrateriod (Fig. 2), the wind speed necessary forave-affected layer to reach the bottom (W0) is

o the background turbidity (T0). In the same wahat the resuspension curve is displaced alongind axis according to theW0 value, a displaceent is also performed along the turbidity according toT0 value. The turbidity due to resuension (Eq.(2)) can be converted into resuspenolids (SR) by using the empirical relation betweurbidity and suspended solids (1.584 mg/LTU; Suk et al., 1998):

R(mg/L) = 3.48

Dα(W + (2.3 − W0))βδ(W, W0) (3)

. Model validation

The model was validated using whole lake slings after extended period (>48 h) of regular wine have assumed that the measured concentrat


Fig. 5. Transect of water temperature (◦C) including five vertical profiles on 9 March 2000 from the northwest coastline to the southeast coastlineof Laguna Galarza (seeFig. 6).

TSS in the 2 L-averaged samples at 0.5 m depth is agood estimate of the daily turbidity. The daily windvelocities were estimated 24 h prior to the sampling.The determination of the background concentrationof suspended solids (S0) was resolved by obtainingan averaged concentration of suspended solids in thelake area that did not show resuspension. We averagedthe TSS concentration of the sampling sites in whichW0 < 3 m/s in order to obtain a better spatial estimationof theS0 (four sampling sites in Laguna Galarza andtwo sampling sites in Laguna Ibera). Measurementsof the temperature profile along the fetch directionshowed the tilting of the mixing depth along the winddirection. A very slight thermal stratification can beobserved to the leeward side and a well-mixed water

column in the windward side (Fig. 5). The sampledresuspension events can be classified as peripheralwave attacks (Evans, 1994).

During the resuspension event of Laguna Galarza(9 March 2000), the wind blew from E–SE direction(112.5◦). The daily mean and standard deviationof the wind speed was 3.34± 0.39 m/s. The valueof S0 was 4.7± 2.5 mg/L (n= 4). The mean dailyspeed of wind (Wmed) was the best estimator of thedaily wind for the model (Fig. 6). The modelledand empirical TSS showed a good linear correlationwhen the model and measured values are compared(TSSmeasured= 1TSSmodel+ 0), R= 0.8945. The re-mainder estimators of the daily wind did not showacceptable modelled results, overestimating and

F (9 Mar lids (mg/L).R ). The e indicatest

ig. 6. Resuspension event by E–SE wind in Laguna Galarzaight: modelled concentration of total suspended solids (mg/L

he drawn transect inFig. 5.

ch 2000). Left: measured concentration of total suspended sospatial interpolation has been performed by kriging. Dash lin


underestimating TSS concentrations in calm andresuspended areas, respectively.

The model was also applied to another shallow lakein Esteros del Iberá, Laguna Ibera. The artificial inletthat crosses in the narrow pass of Laguna Ibera impedesthe propagation of the waves between both basins. Con-sequently, the model was computed independently forsouthern and northern basins. During the resuspen-sion event of Laguna Ibera (23 June 2000), the windblew from E–NE (67.5◦). The daily mean and standarddeviation of the wind speed was 4.62± 1.32 m/s forthe northern basin and 4.88± 1.81 m/s for the south-ern basin. The value ofS0 was 6.7± 1.2 mg/L (n= 2)for the northern basin and 3.1± 1.3 mg/L (n= 3) forthe southern basin. The relationship between mod-elled and empirical values of TSS showed also arelatively good linear correlation for the equationTSSmeasured= 1TSSmodel+ 0,R= 0.7611 (Fig. 7).

The effect of the horizontal advection seemed tocause an elongation of the area of abundant suspendedsolids in the NW sector of the wave attack zone inLaguna Galarza, and towards the eastern shore ofthe narrow pass of Laguna Ibera, which is a wind-protected area in the simulation. On the other hand,the eutrophic Laguna Ibera shows more heterogeneousdistribution of plankton than Laguna Galarza. Thecentral area of the northern basin of Ibera usuallyaccumulates a higher concentration of plankton.These factors induce an error in the modelled spatialdistribution. The coupling of a hydrodynamic model

would help to increase the accuracy of the simulationthrough the consideration of the redistribution ofmatter resuspended by the wave action.

7. Model simulation

In addition to using a homogeneous background(S0) and ignoring horizontal advection, other restric-tions must be taken into account when using themodel for simulation purposes. As the resuspensioncurve does not have a limit for maximum resuspendedmatter, the model could overestimate the concen-tration of SR in very shallow waters. The minimumdepth to which the model was applied in the modelvalidation was 1.6 m. The percentage of the lakearea exceeding this depth is 73% in Laguna Galarzaand 94% in Laguna Ibera. Other limitations to themodel simulation are the existence of a heterogeneousbottom (i.e. rock, sands) or with beds of benthicmacrophytes. These effects could be modelled byassigning different resuspension calibration curves tothem.

Resuspension models offer wide practical possi-bilities from the temporal or spatial analysis of lightattenuation or amount of material that is broughtinto suspension. The impacts of the modificationsto the lake morphology on the resuspension regimecan be also described. These modifications can resultfrom changes in the water level, alterations of the

F ne 20 mg/L).R ). The s

ig. 7. Resuspension event by E–NE wind in Laguna Ibera (27 Juight: modelled concentration of total suspended solids (mg/L

00). Left: measured concentration of total suspended solids (patial interpolation has been performed by kriging.


lake contour or by the construction of infrastructureslike the bridge foundation that actually separates thenorthern and southern basins of Ibera.

7.1. Littoral zone

A simple application is the analysis of the im-pacts of resuspension events on the littoral vegetation.The northwestern and southeastern shores of LagunaGalarza showed appreciable differences in their vege-tation patterns (Fig. 8). The SE shore was colonized bypatches of marsh vegetation, which do not occur in theNW shore. We have compared the resuspension regimein a site located in the northwestern shore (NW station)and another in the southeastern shore (SE station). Bothstudied sites were 1.7 m depth. The straight-line fetchfor each site was computed for each octant of wind di-rection (north, northeast, east, etc.) and the daily evolu-tion of theSRwas modelled with the available registersof wind for 1999 and 2000.

F uringF north-w n thel firmera es acro-p sw nti ensep

The resuspension dynamics showed a relation withthe situation in the lake (Fig. 9). The averaged flux of re-suspended particles at NW station was 19.6 g m−2 d−1

while at SE station was 8.9 g m−2 d−1. The estimationof the potential energy of the resuspended matter couldbe interpreted as a minimum limit of the energy usedto resuspend the sediment. Assuming complete mix-ing of the water column, the median potential dailyenergy at SE station would be 0.53 J/m2, while at NWstation would be 1.63 J/m2. Macrophytes genera thathave been reported as susceptible to the wave actionare found in the shoreline of Ibera wetland lakes (e.g.Scirpus maritimus; Foote and Kadlec, 1988). Imma-ture developing plants are particularly sensitive and thespring growth period is considered the most vulnerablephase due to the easy unearthing of young plants (Leeand Stewart, 1981). In this sense, it can be observedthat the spring (October–December) is the season ofhighest calm in the SE littoral area, facilitating theyearly re-colonization of this shoreline sector. A rel-atively calm period during spring does not occur in theNW littoral. With the actually existing conditions inLaguna Galarza, the complexity of the interaction be-tween the positive/negative effects of the wave actionon the macrophytes stands (e.g. water renewal, lightattenuation, sediment erosion) leads to a reduction ofthe vegetal colonization of the wave-exposed littoral.

The floating mats of organic matter and vegetation(embalsados) that compose much of the shoreline ofLaguna Galarza also show marked spatial differences.T lessvT nds sly,t ors,b rtantf cesw malp unas etrew int ers.c th hests herem ct oft acro-p s to

ig. 8. A landsat-5 image in the infrared waveband obtained debruary 1999. The location of the southeastern station andestern station are indicated. Differences in the vegetation o

ake coasts can be observed. The NW shoreline is composed bynd less vegetatedembalsados(light grey). The white spots of thouthern littoral of the lake correspond to patches of floating mhytes, mainly water hyacinths (Eicchornia azurea). Some plantith ribbon-like leaves (S. californicus, T. latifolia) are also prese

n sparse patches along the SE littoral, although these little datches were not detectable by remote sensing.

he NW shoreline is composed by firmer andegetatedembalsadosthan the SE shoreline (Fig. 8).he embalsadosseem to support higher erosion atack in the NW coast by wind and waves. Obviouhe plant colonization depends on multiple factut the wind-induced waves seem to be an impoactor in Laguna Galarza. Likewise, great differenere also observed in the densities of littoral aniopulations such as the caiman. In the Galarza-Lystem, the number of caimans per coast kilomas 2.9-fold higher in the southern coast than

he northern coast for the year 2000 (Waller, pommun.). Bonetto et al. (1981)pointed out thaighest concentrations of fish biomass and the higpecies diversity were found in the lakes sectors wacrophytes beds occur. Thus, the cascading effe

he resuspension on the ecosystem structure (mhytes beds > benthic and fish communities) seem


Fig. 9. Modelled daily concentrations of resuspended solids during an averaged Julian year estimated from available winds registers during1999 and 2000. Results have been modelled for SE station (black line) and NW station (grey line) of Laguna Galarza. The dash line shows theupper limit of daily concentration of resuspended solids in the resuspension calibration curve. The peaks that exceed this limit assume the lackof a saturation level of resuspended solids.

influence the spatial distribution of a top-predator likethe caiman.

An interesting application inEsteros del Iberá isthe analysis of the effect of the water level on thewind-induced resuspension. We have modelled thevariability of the resuspension frequency along a watergradient. The modelling of the number of resuspensionevents per time unit allows the resuspension analysisin shallower waters without a possible overestimationof the resuspended matter. The difference between

Fig. 10. Predicted number of daily events per week able to resuspendmore than 5 mg/L in function of the depth of the water column.Predictions correspond to SE station (solid line), a mid-lake station(dotted line) and NW station (dashed line) of Laguna Galarza. Thel

the resuspension regime of the NW and SE shoresincreases more towards shallower waters (Fig. 10). Itis interesting to note that the relation between resus-pension frequency and water level is not linear. Thereis a sharp increase of the resuspension frequency whenthe water depth is lower than 1 m (typical macrophyteshabitat). Thus, the negative impact of the waves on thelittoral ecosystem rapidly increases when new areasare flooded (with a thin water layer). The physicaldamage is not the only negative impact acting in theseareas. Probably, the deposition of resuspended matterabove the new flooded areas plays also an importantrole. Galinato and van der Valk (1986)reported thatthe seeds may be prevented from germinating if theywere covered by as little as 1 cm of sediment.

7.2. Limnetic area

In the limnetic area, the permanent increase in thewater level registered in the wetland should also haveimpacts on the trophic state due to a lower mobilizationof the matter accumulated in the sediment. Accordingto Fig. 10, 63 resuspension events (>5 mg/L) per yearoccur with the current water level, and 90 events witha water level 1 m lower. Although historic data are notavailable, we would expect wind-induced increases inthe trophic state with shallower waters as it has oc-curred in similar systems.Nagid et al. (2001)observedan increase of the total nitrogen and phosphorous con-c ase
ines have been calculated from wind registers of 1999–2000. entrations in Lake Newman (Florida) with a decre


of the water level because the particulate matter of thebottom is more frequently brought to the water column.

The background suspended particles concentrationin the lakes (S0) would depend on the autochthonousproduction and the allochtonous supply, mainlythrough inflowing rivers and streams. The resus-pension events can alter the temporal evolution ofthese latest processes with sudden increases of thesuspended particulate matter. The contribution ofthe resuspension to TSS must be estimated throughthe daily variation of resuspended matter during theseasonal cycle. We estimate the matter that can be sup-plied by autochthonous production from the seasonalbiovolume-size spectra of the planktonic community,from bacteria to zooplankton (Cozar et al., 2003). Thedry weight can be obtained through the conversionfactors proposed byMoloney and Field (1989). Theparticle concentration in the mouth of Arroyo Yacarevaried from 0.4 to 1.1 mg/L from winter to summer.The water drainage through the huge vegetation masssurrounding the lakes causes the natural filteringand decantation of most of the particulate mattertransported towards the lakes. The allochtonous par-ticulate matter measured in the mouth of the streamsis mostly resuspended from the flux of the streams.Once the allochtonous particulate matter arrives tothe open waters, it settles in a few days, adding to thepotentially resuspendibile matter. The majority of theparticulate matter settles during≤5 days accordingto the lakes depth. Considering the renewal time oft pplyf than0 thel latem ingt andt ewal(c r tot thes netica % oft agedT to6 Thisc thei rticalfl ly,

the aesthetic aspect of the lake. If the water level inLaguna Galarza were 1 m less than the present, theSRcontribution to TSS would increase 82% in the limneticarea, increasing the annual averaged TSS to 9.5 mg/L.

The differences of trophic state and morphologyof Laguna Ibera lead to a different composition of thesuspended matter with respect to Laguna Galarza. TheSR contribution in the limnetic area was 6% in southernbasin (annual averaged TSS∼ 3.9 mg/L) and 26% innorthern basin (annual averaged TSS∼ 6.4 mg/L).Applying a decrease in the lake water level of 1 m, theSR contribution would increase to 8% in the southernbasin (TSS∼ 4.0 mg/L) and to 34% in the northernbasin (TSS∼ 7.2 mg/L).

8. Conclusion

The morphology of the southern basin of LagunaIbera determines a relatively low concentration ofresuspended particles and TSS. Despite the fact that theannual averages of TSS are similar in Laguna Galarzaand in the northern basin of Ibera, a significant differ-ence occurs with respect to the particulate compositionand, consequently, the frequency of appearance ofthese particles. In the northern basin of Ibera, most ofthe particulate matter is composed by plankton (74%).Thus, the variation of suspended matter concentrationshows a strong seasonal component. Laguna Galarza,on the other hand, reaches similar annual averaged TSSd ation.

etrya atterd istica hed gimea Sc ouldh enta

A

nions ableM sur”( d

he studied lakes (>60 days), the allochtonous surom streams would constitute on average less.1 mg/L when considering the whole volume of

ake. The contribution of the allochtonous particuatter to the TSS is practically negligible consider

he average plankton concentration (1.7 mg/L)he resuspension frequency during one lake renmore than 10 events withSR> 5 mg/L, Fig. 10). Theontribution of the allochtonous particulate mattehe TSS of the whole lake represented less 3% intudied basins. The resuspended matter in the limrea of Laguna Galarza represented about 72

otal suspended particulate matter (annual averSS∼ 6.2 mg/L). This percentage varied from 87%8% between the NW and SE shores, respectively.omposition of the particulate matter highlightsmportance of the resuspension events on the veux of particles, the light environment, or simp

ue to intense resuspension pulses of shorter durThe use of resuspension models and turbidim

ppear as a useful tool to understand particulate mynamics, especially in ecosystems where lognd accessibility difficulties hamper their study. Tescribed relations between the resuspension rend the distribution of littoral communities, TSoncentrations, water level and lake morphology celp to improve the understanding of managemctions onEsteros del Iberá reserve.

cknowledgements

This study was conducted under a European Uponsored research program entitled “The Sustainanagement of Wetland Resources in Merco

ERB I18 CT98 0262). Mr. Tomas Waller provide


valuable advice during the study as well as someunpublished information. The authors are also gratefulfor assistance provided by the parkguards of ReservaProvincial del Ibera and Universidad del Salvador(Argentina).

References

Aalderink, R.H., Lijklema, L., Breukelman, J., Van Raaphorst, W.,Brinkman, A.G., 1985. Quantification of wind induced resuspen-sion in a shallow lake. Water Sci. Technol. 17, 903–914.

Bailey, M.C., Hamilton, D.P., 1997. Wind induced sediment resus-pension: a lake-wide model. Ecol. Mod. 99, 217–228.

Bienfang, P.K., 1981. A technologically simple and reliable methodfor measuring phytoplankton sinking rates. Can. J. Fish. Aquat.Sci. 38, 1289–1294.

Bonetto, A.A., Roldan, D., Canon Veron, M., 1981. Algunos aspectosestructurales y ecologicos de la ictiofauna del Sistema de Ibera(Corrientes, Argentina). Ecosur 8 (15), 79–89.

Carignan, R., Vaithiyanathan, P., 1999. Phosphorus availability in theParana floodplain lakes (Argentina): Influence of pH and phos-phate buffering by fluvial sediments. Limnol. Oceanogr. 44 (6),1540–1548.

Carper, G.L., Bachmann, R.W., 1984. Wind resuspension of sedi-ments in a prairie lake. Can. J. Fish. Aquat. Sci. 41, 1763–1767.

Cotner, J.B., Johengen, T.H., Biddanda, B.A., 2000. Intense win-ter heterotrophic production stimulated by benthic resuspension.Limnol. Oceanogr. 45 (7), 1672–1676.

Cozar, A., Garcıa, C.M., Galvez, J.A., 2003. Analysis of planktonsize spectra irregularities in two subtropical shallow lakes (Es-teros del Ibera, Argentina). Can. J. Fish. Aquat. Sci. 60, 411–420.

de Anda, J., Shear, H., Maniak, U., Riedel, G., 2001. Phosphates inLake Chapala, Mexico. Lakes Reserv. Res. Manage. 6, 313–321.

D sed-.

D

E in a

E ment

F mics

F lantol. 4,

G ts ofDelta

Grant, W.D., Madsen, O.S., 1986. The continental-shelf boundarylayer. Ann. Rev. Fluid Mech. 18, 265–305.

Hakanson, L., Jansson, M., 1983. Principles of Lake Sedimentology.Springer-Verlag, Berlin.

Hellestrom, T., 1991. The effect of resuspension on algal productionin a shallow lake. Hydrobiologia 213, 183–190.

Kirk, J.T.O., 1994. Light and Photosynthesis in Aquatic Ecosystems.Cambridge University Press, Cambridge.

Kristensen, P., Søndergaard, M., Jeppesen, E., 1992. Resuspensionin a shallow eutrophic lake. Hydrobiologia 228, 101–109.

Lee, P.F., Stewart, M.J., 1981. Ecological relationships of wild rice,Zizania aquatica. Can. J. Bot. 59, 2140–2151.

Leuttich, R.A., Harleman, D.R.F., Somlyody, L., 1990. Dynamicbehaviour of suspended sediment concentrations in a shallowlake perturbed by episodic wind events. Limnol. Oceanogr. 35,150–167.

Lick, W., Huang, H., Jepsen, R., 1993. Flocculation of fine-grainedsediments due to differential settling. J. Geophys. Res. 98 (C6),10279–10288.

Moloney, C.L., Field, J.G., 1989. General allometric equations forrates of nutrient uptake, ingestion, and respiration in planktonorganisms. Limnol. Oceanogr. 34 (7), 1290–1299.

Nagid, E.J., Canfield, D.E., Hoyer, M.V., 2001. Wind-induced in-creases in trophic state characteristics of a large (27 km2), shal-low (1.5 mean depth) Florida lake. Hydrobiologia 455, 97–110.

Ogilvie, B.G., Mitchell, S.F., 1998. Does sediment resuspensionhave persistent effects on phytoplankton? Experimental studiesin three shallow lakes. Fresh. Biol. 40, 51–63.

Pond, S., Pickard, G.L., 1983. Introductory Dynamical Oceanogra-phy. Pergamon Press, Oxford, 329 pp.

Reynolds, C.S., Rogers, D.A., 1976. Seasonal variations in the ver-tical distribution and buoyancy ofMicrocystis aeruginosaKutz.emend. Elekin in Rostherne Mere, England. Hydrobiologia 48,17–23.

Scheffer, M., 1998. The Ecology of Shallow Lakes. Chapman & Hall,

S ofMod.

S ime-arsh

V Ibera.mo

W nk-

W don(7),

ouglas, R.W., Rippey, B., 2000. The random redistribution ofiment by wind in a lake. Limnol. Oceanogr. 45 (3), 686–694

oyle, R.D., 2001. Effects of waves on the early growth ofVallisneriaamericana. Fresh. Biol. 46, 389–397.

ngel, S., Nichols, S.A., 1994. Aquatic macrophytes growthturbid windswept lake. J. Fresh. Ecol. 9, 97–109.

vans, R.D., 1994. Empirical evidence of the importance of sediresuspension in lakes. Hydrobiologia 284, 5–12.

errati, R., Canziani, G.A., 2005. An analysis of water level dynain Esteros del Ibera wetland. Ecol. Mod. 186, 17–27.

oote, A.L., Kadlec, J.A., 1988. Effects of wave energy on pestablishment in shallow lacustrine wetlands. J. Fresh. Ec523–532.

alinato, M.T., van der Valk, A.G., 1986. Seed germination traiannuals and emergents recruited during drawdowns in theMarsh, Manitoba, Canada. Aquat. Bot. 26, 89–102.

London, 357 pp.omlyody, L., 1982. Water-quality modelling: a comparison

transport-oriented and ecology oriented approaches. Ecol.17, 183–207.

uk, N.S., Guo, Q., Psuty, N.P., 1998. Feasibility of using turbidter to quantify suspended solids concentration in a tidal saltmcreek. Estuar. Coast Shelf Sci. 46, 383–391.

arela, M.E., Bechara, J.A., 1981. Bentos del macrosistemaIn: Bonetto, A.A. (Ed.), Estudio del Macrosistema Ibera, ToV (1). Ecologıa. ICA-CEOCAL, Corrientes, Argentina.

ebster, I.T., 1990. Effect of wind on the distribution of phytoplaton cells in lakes. Limnol. Oceanogr. 35 (5), 989–1001.

eyhenmeyer, G.A., Hakanson, L., Meili, M., 1997. A validatemodel for daily variations in the flux, origin and distributiof settling particles within lakes. Limnol. Oceanogr. 421517–1529.

Documents

Sediment resuspension by wind in a shallow lake of …F...Ecological Modelling 186 (2005) 63–76 Sediment resuspension by wind in a shallow lake of Esteros del Iber´ a (Argentina):