appendix h8_salt balance of northern spencer gulf

8/12/2019 Appendix H8_Salt Balance of Northern Spencer Gulf

1/53Olympic Dam Expansion Supplementary Environmental Impact Statement 20101

Salt balance of Northern Spencer Gulf

APPENDIX H8


2/53Olympic Dam Expansion Supplementary Environmental Impact Statement 20101

Long-term effects of desalination and climate change

APPENDIX H8.1


3/53

Long-Term Salt Balance of Northern Spencer Gulf R. A. Nunes-Vaz

TheEffectsofDesalinationandClimateChange

ontheLongTermSaltBalanceofNorthernSpencerGulfRICHARD.A.NUNES-VAZ

EXECUTIVE SUMMARY

Thisreportexaminesthepotential impactsofdesalinationandclimatechangeon

thelargescale,longtermsaltbalanceofnorthernSpencerGulf(NSG),addressing

thefollowingthreeissues:

how rapidly Spencer Gulf responds to influences that disturb its

salinity,andhowlongittakesto recover followingadisturbance

the degree or extent to which salinity is expected to change in

responsetoadisturbance,and

whetheranticipatedperturbationsassociatedwithdesalinationand

climate change will be large enough to shift the Gulfs salinity,

irreversibly, toanewequilibriumstate.

SpencerGulfexperienceshighersalinitiesthantheadjacentoceanduetoanetloss

offreshwaterfromevaporation.Thisfluidlossisreplacedbyaninfluxofseawater

atameanvelocity~103ms

1 (Nunes&Lennon,1986)whichbringsadditionalsalt

thatmustbeejectedbyoceanographicmechanisms.

Theannualand interannualvariationsofsalinityofthewholebodyofNSGwere

determined by spatial interpolation of field data collected on twelve occasions

duringthe1980s (Nunes,1985)and fromtwonewsurveys in2008and2009.All

salinitydatawerecollectedandcrossreferencedby thesame investigator,using

thesamecalibration instrumentsandstandards,therebyproducingan internally

consistentdatasetspanning27years.

Assumingthat themeansalinityofNSG isgovernedbythecumulativeeffectsof

net evaporation, a correlation between NSG salinity in the 1980s, and net

evaporationmeasuredbytheBureauofMeteorologyatPrice(~250kmfromNSG)

wasdetermined. The correlationwas greatest (with a coefficient of 0.89)when

dailynetevaporationwasaveragedover the sixmonths (192days)prior toany

salinityobservation.ContinuingtheaveragingofPricenetevaporationthroughto

2009, it was shown that themore recent observations of NSG salinity remain

consistentwiththe1980srelationship.

Thus,despite large interannualanddecadaltrends inthe records, it isapparent

that NSG mean salinity on any particular day is primarily determined by the

cumulative effectsofnet evaporationduring the prior sixmonths, and that the

Gulfhasessentiallynomemoryof longer(e.g.,decadal)influences.

Regression betweenmeanNSG salinity, and 192day averaged net evaporation,

indicated thata sustainedchangeof1x103mday

1ofnetevaporation induceda

salinitychangeof0.23gl1.Withaveragednetevaporationasaproxy indicatorof

NSGsalinity,themeansalinityofNSGduringtheperiod19822009wasinferredto

be41.56gl1;themeanpeaktotroughannualvariationwas1.38gl

1;meansalinity

shiftedupwardsby0.77gl1duringthetenyearperiodto2007/8and,onaverage,

theannualsalinitymaximumwasreachedon12thApril,andtheminimumon14th

October.

Page 1 of 49 July 2010


4/53


Assumingthatdesalination(byremovingfreshwater)isequivalenttoevaporation,

theregressionresultwasusedtoestimatepotentialdesalination impactsonNSG

salinities.Theworstcaseassumptioninwhichallreturnwaterassociatedwithan

averagedesalinationrateof251megalitresday1,isabsorbedentirelywithinNSG,

impliesanincreaseofNSGsalinitybyupto0.11gl1,toanewmeanof41.67gl

1.

TheeffectsofclimatechangeonNSGsalinityweresimilarlyassessedbyestimating

changesofnetevaporationfromtheprojectionsofSuppiahetal(2006), implying

increases ofmean NSG salinity of up to 0.07gl1 in 2030, and 0.22gl

1 in 2070.

Combining theseestimatesof climate changeanddesalination impactswith the

highest inferredsalinityof thepast27years (obtained from thenetevaporation

proxy), implies an upward shift ofmean NSG salinity of 0.61gl1 in 2030, and

0.76gl1in2070comparedwiththe19822009mean.

Finally,theoceanographicmechanismsofsaltdischargewereexaminedinorderto

assesswhethertheymaysufferirreversibledisruptionbyanticipatedperturbations

associatedwithdesalinationandclimatechange.Regardlessofwhichmechanism

is

primarily

responsible

for

maintaining

the

long

term

salt

balance

of

NSG

(that

is,

gravitational circulation or turbulent dispersion) each is expected to adapt

smoothly,modestlyandreversiblytoanticipateddisturbances.



5/53


1 Introduction

On 1May 2009 BHP Billiton released the Olympic Dam Expansion Draft Environmental

ImpactStatement (BHPBilliton,2009)detailingtheproposaltoexpand itsmining facilities

and related infrastructure. Included within the planned development was a coastal

desalinationplantatPointLowlyinUpperSpencerGulfwhichisrequiredtoyieldanaverage

of 251 megalitres day1 of freshwater for the Olympic Dam expansion. In community

discussionabouttheproposal,questionshavebeenraisedaboutthepotentialforsignificant

longterm change of salinity in Upper Spencer Gulf, particularly when combined with

anticipatedeffectsofclimatechange.

ThisreportaddressesquestionsrelatingspecificallytothelongtermsaltbalanceofSpencer

Gulf. It does not address localscale issues associatedwith the dilution and discharge of

returnwater inthevicinityofthedesalinationplant itself.Thepurposeofthisreport isto

examine the factors thatareprimarily responsible fordetermining the largescalebalance

and distribution of salt in the Gulf, and to assess the potential for both natural and

anthropogenicinfluencestoperturbtheirextantcharacter.

Basedprimarilyon longtermmeasurementof salinityand environmental conditions, this

reportquantifiestherelationshipbetweennaturalevaporativeforcingandsalinitychangein

northernSpencerGulf.Assuming(asdiscussedbelow)thatdesalinationandclimatechange

may be treated as perturbations of evaporation (freshwater removal) and precipitation,

quantitative assessment of their separate and combined effects on the northern Gulfs

salinityarederived.

1.1 Background and Context

Figure1.TheSouthAustralianGulfs



6/53


SpencerGulf(Figure1)isaninverseestuary(seeforexample,Bye,1981;Nunes&Lennon,

1986,hereafter calledNL86;andNunesVazetal,1990,hereaftercalledNV90), that is,a

semienclosedseainwhichseawaterismeasurablyconcentrated(ratherthandiluted,which

isthecaseinaclassicalestuary)bytheremovaloffreshwaterduetoevaporation.Annual

meansalinity rises fromvaluesofapproximately36gramsper litre1 (gl

-1)on theadjacent

continental

shelf,

to

more

than

45gl1

near

the

head

of

the

Gulf,

at

Port

Augusta.

To

avoid

confusionwithotherstudies,theareaofUpperSpencerGulfnorthof33Swillbereferred

toasnorthernSpencerGulf(NSG),asidentifiedinFigure2.

Figure2.Theareaofinterest,termednorthernSpencerGulf

Figure3showsasimpleschematicrepresentationofSpencerGulfand itsonedimensional

fluxesof salt.Annualmeannetevaporation2over theentire surface, removes freshwater

whichmustbereplacedbyanetinfluxofseawater.Thisatmosphericallyinducedlandward

salt flux is representedbySa inFigure3. It is theoriginoftheGulfselevatedsalinitiesor

inverse

estuarine

character

because

the

freshwater

loss

can

only

be

replaced

by

seawater.

Increasingly remotepartsof theGulf (equivalent to increasingdistance from theocean in

this onedimensional representation) are characterised by higher salinities, because the

additionofsalt iscumulativeonmovingnorthwiththe landwardflow.Theneteffect isto

createaonedimensionalsalinitygradientalongtheGulfwithhighestsalinitiesatthehead,

asindicatedinFigure3.

1 Despitetheoceanographicconventionwhichhasnowremovedallreferencetounits,salinitywillbe

expressedinunitsofgramsperlitrefromhereonwards,inkeepingwithparallelmodellingand

ecology

reports

associated

with

the

BHPB

desalination

proposal. 2 Thecycleofnetevaporationisnegativeatvarioustimesinthewinterwhenprecipitationexceeds

evaporationbut,averagedthroughtheannualcycle,netevaporationispositive.



7/53


Figure3.Schematicinverseestuaryshowingonedimensionalsaltfluxes

Iftherewerenootherprocessesoperating,thesalinityofSpencerGulfwouldcontinueto

rise year after year (almost) without bounds. However, observations (and common

experience)indicatethatsalinitiesdonotcontinuallyrise,andthisisbecausethereisanet3

seaward fluxof saltofsimilarmagnitude to the landward flux.The seawardsalt flux isof

oceanographic origin, and is indicated as So in Figure 3. Preciselywhich oceanographic

mechanismsgenerate the seaward salt flux isactuallyof little importance in thisanalysis,

butanumberof studies (forexample,Lennonetal,1987;NV90)have indicated that it is

associatedwiththecombinedeffectsofturbulentdispersion,frominternalshearandwind

andtideinducedstirringatthesurfaceandseabed,andgravitationalcirculationdrivenby

thehorizontalgradientsofdensity (largelydue to salinity).Both turbulentdispersionand

gravitational circulation always act tomove salt down the salinity gradient, that is, fromhighersalinitiestolowersalinities.Thus,theoceanographicsaltfluxwillalwaysalignitselfto

counter theeffectsof the landward fluxdue tonetevaporation. It isalsoassumed4,and

examinedindetailinsection6.6,thattheseawardoceanographicsaltfluxisapproximately

constantallyearroundtothesameextentthatthesalinitygradientcanalsobeconsidered

approximatelyconstant.

Inadditiontothemeansalinitygradient,observationsduringtheearly1980s(NL86)showed

that there isanannualsalinitycyclewithanamplitude that increases fromapproximately

0.3gl1on the shelf tomore than5gl

1near thehead.Theannual salinity cycle isdelayed

relative to the annual cycle of net evaporation. Salinity was seen to reach its annual

maximum

in

early

to

mid

March

in

northern

Spencer

Gulf

(and

not

until

early

June

at

the

mouth)comparedwith thenetevaporationcyclewhichgenerally reaches itsmaximum in

January.The lagof salinitiesbehind thenet evaporation cyclemaybe seenasa formof

halineinertia,akintothethermalinertiawhichcausesthelagofGulftemperaturesbehind

theinsolationcycle.Thetemperaturelagduetothermalinertiaalsoincreasesfromlessthan

3 annualmean 4 NunesVaz&Lennon(1987),forexample,establishedthatseawardsaltfluxinSpencerGulfcanbe

episodic,thatisunsteady,showingaseasonalbiaswithmaximumsaltdischargetendingtooccur

aroundautumn.Thetendencytostratifyandgeneratedensityoutflows,andtheirresistanceto

turbulentdestruction,however,wereshowntobeinverselyrelatedtowaterdepth.Thus,theupper

(shallower)

parts

of

Spencer

Gulf

are

expected

to

display

a

tendency

to

more

frequent

(daily

to

weekly,ratherthanseasonal)shortlived(oftheorderof1hour)dischargeevents,andthereforea

relativelysteadyseawardsaltfluxcomparedwithmoresoutherlyregionsoftheGulf.



8/53


amonthinthenorthernreachesofNSGtonearlythreemonthsatthemouth(seeNL86for

details)5.

Inanysingleyear(seeFigure4)therewillbetimeswhenthelandwardsaltflux(represented

by theblue curve inFigure4) is larger than itsannualmean (during the summermonths

whennetevaporationisrelativelyhigh)sothatthemagnitudeofthelandwardfluxexceeds

thatoftheseawardflux(representedbythestraightblue line inFigure4).Atthesetimes,

theGulfaccumulatessalt,andsalinities(red)showarisingtrend.Atothertimesoftheyear

(particularlyinwinter),whenthelandwardfluxislessthanitsannualaverage,theseaward

flux exceeds the landward fluxand salinities fall. Thenet salt flux (magenta),due to the

combinationofthetwoinfluencesshowsanannualcyclewithameanclosetozero.

Figure4.RepresentationofonedimensionalsaltfluxesinSpencerGulf.Thefluxofsalt

inducedbyevaporation(bluecurve)hasapositive(landward)meanwhichisbalancedby

anequalnegative(assumedconstant)seawardflux(blueline)causedbyvarious

oceanographicexchangeprocessesdiscussedinthetext.Thenetfluxofsalt(purple),

whichispositiveinsummerandnegativeinwinter(withameanofzero),causesasalinity

cycle(red)thatlagsthesaltfluxcycle,causingmaximumsalinitytooccurlaterthanthe

peakofevaporation.

On a longterm basis, the landward atmosphericallyinduced salt flux is approximately

balancedbytheseawardoceanographicsaltfluxsothatthemeansalinityofSpencerGulf

varies only modestly according to the slight, interannual imbalance between the two

influences. Why is this so why are the two influences approximately balanced? The

strengthoftheseawardsaltflux,whethergeneratedbyturbulentdispersionorgravitational

circulation, is related to themouthtohead salinity gradient. If net evaporationwere to

increaseforasustainedperiod,thiswouldenhancethelandwardsaltflux,Sa,causingboth

5 Itisinterestingtonote(NV90;Petrusevics,1993;andWhitehead,1998)thatthenaturaldeclineof

thelongitudinal salinitygradientfromtheheadtothemouth,permitssummertimetemperature

gradientstoswitchtheGulffromitsinverseestuarinecharactertoclassicalestuarinecharacterinthe

vicinityofthemouth.Itseffectistoblockthedischargeofaccumulatedsaltforamonthortwoduring

summer,

but

this

arrangement

is

necessarily

broken

during

the

cooler

months

when

longitudinal

gradientsoftemperaturereinforcethoseofsalinityallowingtheGulftofreelyexchangesaltwiththe

adjacentshelf.



9/53


thesalinityattheheadandthealongGulfsalinitygradient,toincreasewhich,inturn,would

increase the seaward (counter) flux.Thisnegative feedback systemproducesa longterm

stableequilibriuminwhichthemeanmagnitudeofthesalinitygradient,drivingtheseaward

flux,adaptstothemeanstrengthofthelandwardflux.

However,threequestionsremainwithregardtotheGulfsequilibriumdynamics,whichthis

reportaimstoanswer,namely:

1. how rapidlySpencerGulf responds to influences thatdisturb its salinity,andhow

longittakestorecoverfollowingadisturbance?

2. the degree or extent to which salinity is expected to change in response to a

disturbance,and

3. whetheranticipatedperturbationsassociatedwithdesalinationandclimatechange

willbe largeenough to shift theGulfs salinity, irreversibly, toanewequilibrium

state.

In answering these questions, the recent history (1982 to 2009) of precipitation and

evaporation is firstexamined tounderstandmeanatmospheric forcing,and theextentofanysignificanttrends. ThesalinityresponseofnorthernSpencerGulfisthenexaminedover

thesame27yearperiod,todeterminethenatureoftherelationshipbetweenforcingand

response.Oncecalibratedinthisway, expectedperturbationsofmeanNSGsalinitydueto

desalinationand climate changearederived, followedbya theoreticalassessmentof the

dynamicsof the seaward salt fluxandwhether itmay beoverstressedby (andpossibly

unstableto)anticipatedchanges.

2 Long-Term Data Trends and Relationships

2.1 Atmospheric Influences

The Bureau of Meteorology operates many data collection stations in South Australia.

However,theregularityand longevityoftheserecordsvarysubstantiallywhichmeansthat

very fewarepotentiallyuseful to this study.Figure5 shows the locationsof four stations

(Ceduna,Spalding,PriceandAdelaideAirport), closest toSpencerGulf, forwhich records

span the period 1982 to 2009, and for which both precipitation and evaporation were

measured.



10/53


Figure5.AdelaideAirport,Price,SpaldingandCedunameteorologicalstationsproviding

longtermrecordsofprecipitationandevaporation.

Evaporation measurements reported by the Australian Bureau of Meteorology are

unadjustedUSClassAPanEvaporimetermeasurements, representing themeasureddaily

lossofwaterfromametalpanofstandarddimensions.Discussionoftheinterpretationand

adjustment of pan evaporation data is held over until section 2.1.1 below, but for the

purposesofthisreport,panevaporationmultipliedbyafactor(calledthepanfactororpan

coefficient,seeHounam,1973)of0.7calledadjustedevaporation isusedasalongterm

proxyindicatorofenvironmentalevaporation.

Net evaporation is evaluated as adjusted evaporationminus precipitation (measured in

millimetres day1). Figure 6 shows the daily, net evaporation data for the four stations,

plottedonequivalentverticalscales.



11/53


Figure6.DailynetevaporationatSpalding,Price,AdelaideAirportandCedunastations

From approximately 2000 onwards, data from Spalding show anomalous behaviour,

suggestingthatSpaldingmaynotprovidelongtermdataofsuitablequality.

Figure7.DailynetevaporationatSpalding,Price,AdelaideAirportandCedunastations

Figure

7

shows

the

same

data

plotted

for

the

shorter

period

1982

to

1986,

exposing

the

fact

thatthePricerecordhasagreatmanymissingdatapoints,whichisthecasethroughoutthe



12/53


record. In fact themissingdata reflectadifferentdata collection regimewhereby several

days(usuallyincludingweekends)mayelapsebeforeanotherreadingistaken.Replacingall

missingdatawithzerosprovidesadataset that isuseful forassessing longtermaveraged

behaviour,smoothingoutthelargedaytodayfluctuations.

Arunningmeanforanydatasetmaybecalculatedasfollows.Therunningmeanhasafilter

orwindow lengthwhichdescribes thespanof thedata itassesses.Fora1year (365day)

runningmean,365dailyvaluesarereadin,themeaniscalculatedandthevalueisassigned

toaparticularday.Foracentral runningmean, thealgorithm takesadatapoint together

with182pointsfromeitherside,andassignsthemeanofall365valuestothecentralpoint.

Itthenmovesto thenextpointand repeats theprocess.A365daycentral runningmean

cannotbeevaluateduntilthe183rddatapointfromthestartofthedataset,andsimilarlyit

yieldsnoresultsfordatacloserthan183datapointsfromtheend.Thusthefilterproducesa

resultwhichishalfthefilterlengthshortateachend(foracentralisedfilter).

Figure8.OneyearrunningmeansofnetevaporationatSpalding,Price,AdelaideAirport

andCeduna.

Missing

data

points

in

the

Spalding

and

Price

records

shown

in

Fig

7,

were

filledwithzeros.

Figure8 shows the resultof thisprocessapplied to the fournetevaporationdatasets.All

filtered results are visually correlated (moreeasily seen in figure 9)which isencouraging

withregardtodataquality.Notethatallrecords indicate increasednetevaporation inthe

late2000s,butPrice showsaverymarked changenotduplicatedatSpalding (the closest

stationtoNSG).



13/53


Figure

9.

One

year

running

means

of

net

evaporation

at

Spalding,

Price,

Adelaide

Airport

andCedunastations

Figure9 indicates that therewas significantagreementbetweennet evaporationdata at

AdelaideAirport,PriceandSpalding,while theCeduna record showed similar interannual

structurearoundasubstantiallyhighermean.Duetothehighfrequencyincompletenessof

the Spalding and Price records, their filtered results show significantlymore noise. The

following analysiswill focus on Spalding, Price and Adelaide Airport records, as Ceduna

appearstoshowdisparatebehaviourscomparedwiththethreestationsclosesttonorthern

SpencerGulf.

Figure10showsthelongtermtrendsofatmosphericforcing,thatis,precipitation,adjusted

evaporationandnetevaporationforAdelaideAirportfortheentire1982to2010period,as

1year,3yearand5yearcentralrunningmeans.Notethattheyaxisscalerangeisthesame

forprecipitation(shownasnegative)andevaporation,butthesedifferfromthescalingused

fornetevaporation(whichhasalargerrange).



14/53


Figure10.1year,3yearand5yearrunningmeansofmeteorologicaldatafromAdelaide

Airport

From the early 1980snet evaporation atAdelaideAirport,which shows significant inter

annualvariability,experiencedadownwardtrenduntilthemid1990sand,sincethattime

has risento levelsabovethoseof theearly1980s.Thedownwardtrend isconsistentwith

theaveragedbehaviourofallAustralianClassApans from theearly1980s (Giffordetal,

2005;Rodericketal,2009).Giffordetal(2005)areambiguousindescribingtrendsafterthe

mid1990s,althoughtheydostatethat from1994thedata forAustraliahaveshowna

slowdown and reversal of the downward trend especially in the eastern part [of

Australia].This isconsistentwiththedataforAdelaideAirport(andPriceandSpalding),

wherenetevaporationhasrisenconsistentlyformorethanadecade,andmaystillberising

(accordingtothe5yearrunningmean).

The trends of Adelaide Airport net evaporation are evident in both precipitation and

evaporation data (precipitation is plotted on a negative scale tomake the additionwith

evaporation

visually

simpler).

The

long

term

trend

of

increasing

evaporation

was

accompanied by decreasing precipitation (and vice versa). This correlation, which is

commonly seen in such data (Gifford et al, 2005), extends well into shorter period

fluctuations.

These conclusions are also consistent with the behaviour of meteorological data from

SpaldingandPricestations.

2.1.1 Extrapolating Pan Evaporation to Evaporation from Northern SpencerGulf

Muchhasbeenwrittenabout thevalidityorotherwiseofusingdata fromUSClassAPan

Evaporimeters to infer environmental evaporation rates (see forexample, Linacre, 2005).

Evaporationpanssufferfrommanysourcesoferror(Linacre,2005) includingalgalgrowth,

shadeorwindshelterduetotreesandbuildings,thechangingalbedoofsediment,animalor



15/53


birddrinking(althoughmeshguardsaregenerallyusednowadays),heatappliedtothesides

ofthepanfromdirectanddiffuseradiation,andbothshortwavereflectionandlongwave

radiationfromthechangingsurrounds.

However, despite these difficulties, pan evaporation provides data which is generally

consistentwiththePenmanequation,whichpredictspotentialevaporationbasedondaily

mean temperature, wind speed, relative humidity and solar radiation, and with

measurementsoflakeevaporation,determinedbywaterbalancemethodswherethelake

is small and shallow enough to have no horizontal or vertical currents nor to alter

appreciably the atmosphere above it (Linacre, 2005). Such correlations improve when

considering longerterm averaged data, such that Johnson & Sharma (2007) obtained a

correlationof0.97betweenmonthlymeansofpanevaporationandevaporation inferred

fromthePenmanequation.

ComparisonbetweenevaporationmeasuredusingClassApans,andestimates frommass

balancecalculationsinlakes,andfromthePenmanequationindicatethatClassApansover

estimateevaporation,ingeneralbecausepansaresubjecttoadditionalheateffectsthrough

their

sides.

For

this

reason,

a

correction

factor,

called

the

pan

factor

or

pan

coefficient,

is

commonlyappliedtopanevaporationmeasurements inordertoestimatetheevaporation

which is expected to occur from small water bodies such as lakes. Estimates of pan

coefficientsvaryfrom0.7to0.8accordingtocontextanddataset,however,Linacre(1994),

in Australian conditions, found that lake evaporation is approximately 0.7 times the

evaporationfromaClassApan.

Evaporation from the oceans is notwell represented by evaporation from Class A pans

(Farquhar&Roderick,2005),firstlybecausetheirthermalinertiasareverydifferent,thatis,

theoceansabilitytoabsorbheatwithoutappreciablychangingitstemperaturemeansthat

the pan is not a goodmodel of oceanic heat budgets, and secondly because amarine

boundarylayerformsoverlargebodiesofwater,whichgreatlymodifiesproperties,suchas

humidity, thatplay a large role indriving evaporation.Thuspan evaporation from South

Australianmeteorological stations is not expected to provide a valid proxy indicator of

evaporationfrom,forexample,SouthAustralianshelfwaters.

However,theupperpartofSpencerGulf,northof33S isarelativelynarrowandshallow

water body. A significant fraction of its evaporation is anticipated to occur around its

margins where thermal inertia is relatively small and diurnal temperature changes are

significant.Furthermore,amarineboundarylayermayremainundevelopedforallbutwinds

withpredominantly southerlyornortherly components,although theseare thedominant

windsinsummer.

Thus,althoughevaporationinferredfromClassApansremainsanimperfectproxymeasure

of

evaporation

from

northern

Spencer

Gulf,

it

is

expected

that

it

should

provide

useful

insightsintopropertiesthataresignificantlyinfluencedbyevaporation.Onthebasisofsuch

caveats, the analysisproceedsusing evaporationmeasuredbyClassApanevaporimeter,

adjustedusingapan factorof0.7,asan indicatorofevaporation fromnorthern Spencer

Gulf: a value at the low end of the normal range of pan coefficients because this will

exaggerate or lead to upperbound estimates of the potential environmental (and

desalination)effectsonNSG.

2.2 The Salinity Response of Northern Spencer Gulf

NunesVaz (1985)andNL86reportedthe resultsof fourteen largescalethreedimensional

surveys of temperature and salinity in SpencerGulf frommid1982 to early 1985. These

surveys (using an internallyrecording Applied Microsystems electronic conductivity



16/53


temperaturedepth,orCTD,probe)profiledconditionsatmanystationsbetweenthehead

andpointsasfarsouthas3420S.

In NSG, where salinities often rise above the measurable limit (42gl-1) of typical

oceanographicinstrumentation,aperspexsleevewasinsertedintotheprobesinductivecell

to reduce the cells sample volume and thus reduce its sensitivity, to bring the highest

salinitieswithinmeasurable range. Thispracticedestroyed themanufacturers calibration

betweenactualandmeasuredsalinityand,forthisreason,bothsurfaceandbottomsamples

werecollected (usingNansenorNiskinbottles)ateverystation, formeasurementback in

thelaboratoryusinganaccuratebenchsalinometer.

When measuring salinity using the (Yeokal) bench salinometer (actually measuring

temperature,andconductivityratiorelativetoapreciselyknownseawaterstandard),itwas

necessary to dilute the samples using a precisely known amount of deionisedwater, in

ordertolowerthesalinitybelow42gl-1,againintomeasurablerange.Oncedetermined,the

measurementwas adjustedupwards according to theamountofdilution.These samples

consequently provided a separatemulti (typically 50plus) point calibration of each CTD

survey.

Twelve

such

surveys

provided

substantial

data

in

NSG

during

the

1980s.

TwofurthersurveysofNSGwereconductedon6August2008and2March2009(thedata

areprovidedinAppendixA).However,aprofilingCTDprobewasunavailable,andsosalinity

at each station was sampled using only nearsurface and nearbottom water samples

collected using a Nansen bottle. Nevertheless, the samples were analysed at Flinders

Universityusingthesamebenchsalinometerastheanalysesconducted inthe1980s.New

salinitystandardswereshipped6fromtheUKandacrosscheckwasperformedbetweenthe

newstandards,andthreeoftheglasssealedsalinitystandardsshippedfromtheUK inthe

1970s:thesamebatchofstandardsusedtoreferencetheanalysesofNSGsalinityatFlinders

Universityduringthe1980s.Thisintercalibrationshowedthatthe1970sstandardshadnot

measurablydrifted,andpermittedthenewsurveystobepreciselyreferencedtothe1980s

data(towithin


17/53


Figure11.NorthernSpencerGulfwithNovember1982surveystations(redcircles)and a

2kmby2kminterpolationgrid(bluedots).

Figure 11 shows the location of stations in NSG from a typical survey, in this case 24th

November1982.Foranyparticularsurveysuchasthis,thedetailedverticalprofileofsalinity

ateachstationyieldedadepthaveragedsalinity,andthusallstationsinthesurveyallowed

the region tobe contoured forhorizontaldepthaveraged salinity.Using these contoured

distributions, salinitieswere then interpolated or extrapolated onto the regular (2km by

2km)grid,showninFigure11asbluedots.Ateachgridpoint,themassofsalt(thatis,the

interpolatedsalinitymultipliedbythedepth,thenmultipliedby4km2)wasdetermined,and

summedforall(130)gridpoints.Thissaltmasswasthendividedbythemeandepthofall

gridpoints(andalso4km2)togivethemeansalinityofNSG.Thismethodwasthenrepeated

forallothersurveysofNSG.MeansalinitiescalculatedinthiswayarecollectedinTable1.

# SURVEYDATE

MEAN

SALINITY

1 28 Jul 1982 42.04gl-1

0.05gl-1

2 24 Nov 1982 41.67gl-1

0.05gl-1

3 15 Jan 1983 41.92gl-1

0.05gl-1

4 9 Feb 1983 42.21gl-1

0.05gl-1

5 20 Apr 1983 42.57gl-1

0.05gl-1

6 22 Jun 1983 41.68gl-1

0.05gl-1

7 11 Aug 1983 40.90gl-1

0.05gl-1

8 1 Nov 1983 40.99gl-1 0.05gl -1



18/53


9 6 Mar 1984 42.33gl-1

0.05gl-1

10 26 Jun 1984 42.04gl-1

0.05gl-1

11 10 Oct 1984 41.04gl-1

0.05gl-1

12 23 Jan 1985 41.57gl-1

0.05gl-1

23 year gap

13 6 Aug 2008 41.97gl-1 0.08gl -1

14 2 Mar 2009 42.70gl-1

0.08gl-1

Table1.MeansalinitiesofNSGdeterminedaccordingtothemethoddescribedinthetext.

Minimumandmaximumvaluesfromthe1980ssurveysarehighlightedinboldfacetype.

Estimation of the error associated with the mean salinity from 1980s surveys was

determined by adjusting the arrangements of salinity contours to the highest salinity

(contoursmovedasfarsouthaspossible)and lowestsalinity(contoursmovedasfarnorth

aspossible)distributionsusingnonlinearinterpolation,whileremainingconsistentwiththe

measureddata.Themeansalinitywasthencalculatedfromeachextremeintheusualway,

givinghighestandlowestestimatesforeachsurvey.

TheerrorassociatedwithestimatesofmeanNSGsalinityfromthe2008and2009surveys

was larger than that of the 1980s because the data lacked information about the depth

variationof salinityateach station. For this reason,oneextremearrangementof salinity

contourswasderivedbyassuming that thehigherof the surfaceandbottom salinitiesat

eachstationappliedtothewholedepth,andcontourswerepushedasfarsouthaspossible

consistentwiththestationdata, leadingtothehighestestimateofNSGmeansalinity.The

otherextremeinvolvedassumingthatthelowerofsurfaceandbottomsalinitiesappliedto

the

whole

depth

at

each

station,

and

contours

were

pushed

as

far

north

as

possible

consistentwithdata,givingthelowestmeansalinityestimate.

Figure12showsthe30monthvariationofNSGmeansalinitywithacubicsplinefittothe

data.Figure13enablesmoredirectcomparison,givingasenseoftheinterannualvariation

of the seasonal salinity response of NSG during 1982/83 (blue), 1983/84 (green) and

1984/85 (magenta).The latesummermaxima inApril1983andMarch1984weresimilar,

even though theywere preceded by very differentwinter/springminima. Theminimum

observedin1982wassubstantiallyhigherthanthatof1983,withtheminimumof1984lying

betweenthetwo.



19/53


Figure12.ThevariationofmeanNSGsalinitythroughoutthe30monthperiodofdetailed

measurement

Figure13.

The

mean

salinity

of

NSG

as

separate

annual

cycles

overlain

for

ease

of

comparison

Themean salinityofNSGbasedon2March2009data (42.70gl1)was thehighestof the

fourteensurvey, 27year dataset, although it was only 7.8% outside the salinity range

(40.90gl1 to 42.57gl

1)seenintheearly1980s.Themeansalinityof6August2008laywithin

thevariationsseenatsimilarseasonaltimingsinthe1980s.Thus,therecentdatacouldnot

beinterpretedasinanywayanomalouswhenreferencedagainst1980sdata,althoughthis

interpretation is subject topotential revisionawaitingassessmentofatmospheric forcing,

anditsinfluenceoverequivalentperiods.Thisisthesubjectofthenextsection.



20/53


3 The Relationship Between Atmospheric Forcing and the Salin ityResponse of Northern Spencer Gulf

Theargumentsof section1 implyahypothesiswhichmaynowbe tested.Thehypothesis

canbestatedasfollows:

the mean salinity of northern Spencer Gulf reflects, in a consistentmeasurableway, the integrated influencesofnetevaporationduringa

precedingperiodwhoselengthmaybedetermined.

In other words, anomalously high and sustained net evaporation is expected to raise

salinities inNSG relative tomeanbehaviour.Similarly, sustained lowernetevaporation is

expectedtoleadtolowerNSGmeansalinities,notingthatthemeaningofsustainedmust

alsobeexamined.

The hypothesis implies that the salinity on any particular day (say day-n) reflects the

cumulative effects of net evaporation for x days preceding day-n. Unlike the central

running mean filter, a lagged filter is now required: one that averages xdays of net

evaporationdata,andassignstheresulttothefinaldayofthedatawindow.Thefilterthen

movesforwardonedayandrepeatstheprocess,continuingtotheendofthedataset.

Figure 14 shows raw (daily)net evaporationdata fromAdelaideAirportduring the early

1980s (light blue, highly variable dataset), together with data obtained using a 90day

centralrunningmeanfilter(darkblue),a90daylaggedmeanfilter(magenta)anda180day

laggedmean filter (red).Notice that the90daycentral runningmean,and90day lagged

filterseriesareidentical(astheyshouldbe)exceptfora90dayrelativeshiftalongthetime

axis.Notealsothatthelongerthefilter,thesmootherthedata,andthesmallertherange.

Forlaggedfilters,allofthedatalossassociatedwiththefilterlengthoccursatthebeginning

ofthedataset,whichiswhyeachfilteredresultstartsonadifferentday.

Figure14.NSGproxynetevaporation,withvariousfilteredderivatives

In order to compare lagged mean net evaporation (LMNE) with mean NSG salinity

(MNSGS),bothdatasetsmustbenormalised(rescaledtoarangefromzerotoone)sothat

they can be plotted on the same axes. The assessment of the relationship between net

evaporationandsalinityresponseinitiallyfocusesonthe1980sdataalone,sothatdatafrom

themorerecentsurveysmaybereservedasindependenttestsofanyinferredrelationship.Mean NSG salinities were normalised using the minimum (survey #7 of Table 1) and



21/53


maximum(survey#5ofTable1)measuredvalues.Similarly,after filteringnetevaporation

data, the resultwasnormalizedusing theminimumandmaximumLMNEobservedwithin

thesameperiod.

Figure15 showsnormalised laggedmeannetevaporationusing sixdifferent filter lengths

(90daysupto208days)plottedwithnormalisedmeanNSGsalinity.A filter lengthof208

dayswasthemaximumthatcouldbeappliedwithoutlosingtheabilitytocomparewiththe

firstsalinitydatapointwhichwasobtained209daysinto1982.

Figure15.LMNE(blackline)atAdelaideAirportversusMNSGS(reddatapoints:seetext)

forvariousdifferentfilterlengths

Thevariance(sumofthesquareddifferencesbetweenthediscretesalinitydatapointsand

thespecificvaluesofLMNEonequivalentdays)isalsoshownineachoftheplotsofFigure

15.Byvarying the lagperiodonedayata time, thevariancewas seen to followabroad

minimumforlaggedfilterlengthsbetween190daysand200days,withtheoptimumlagat

190days.

Figure16showsthefilteredresultsusingoptimumlagsfromeachofAdelaideAirport(lagof

190days),Price(192days)andSpalding(192days),andtheirvariances,fittedtothe1980s

NSG salinity results. The resultwith the smallest variance (0.16)was obtained using net

evaporationfromPrice,althoughallthreeresultsaresimilar.



22/53


Figure16.LMNE(solidlines)fromAdelaideAirport,PriceandSpaldingstations,forlags

thatachievethesmallestvariancewhenfittedtoMNSGSdata(discretepoints).The

optimumfilterlag,indays,isindicatedoneachpanel,togetherwiththecorresponding

variance.

The message from these results is significant: that, at least during the 30months of

observations in the early 1980s, themean salinity of NSG on any particular day closely

reflectedtheaccumulatedeffectsofnetevaporationduringthepreceding190or192days

(approximately six months). This goes some way to answering the question about the

responsetimescaleofNSG: itsuggeststhattheGulf largelyrespondedtovariationsof less

thanannualperiod,althoughthislimited(30month)datasetisnotdefinitivewithrespectto

longerterm(decadal)influences.

Figure17providesadifferentviewofthesamedatashowninFigure16.Notethattheinter

annual differences between themaxima of 1983 and 1984 have the same sense as the

changesinobservedsalinityinthesameyears.Similarly,theminimaofLMNEin1982,1983

and1984mirror thechanges in thedepthof the salinityminima (other thanSpalding for

1983and1984).



23/53


Figure17.LMNEfromAdelaideAirport(black),Price(red)andSpalding(blue),forlagsthat

achievethesmallestvariancewhenfittedtoMNSGSdata(discretepoints).

3.1 The Long-Term Behaviour of Northern Spencer Gulf

ResultsdiscussedabovesuggestthatNSGsalinityonanyparticulardaywasdetermined,toa

significantdegree, by the cumulative effectsof net evaporation during thepreceding six

months.Thequestionremains,however,astowhetherthereisalongertermcomponentof

theresponsethataccumulatestheeffectsofmuch lowerfrequencies intheforcing,which

mightdrivesignificantshiftsofmeanNSGsalinity.

Given

that

net

evaporation

sustained

a

consistent

increasing

trend

for

more

than

a

decade

(Figure10), ifNSGsalinity issensitivetothe longertermrisingtrendthenextensionofthe

simple1980s sixmonthmemory relationship, should revealwhetherashiftof themean

salinitydatumhasoccurred.

Figure 18 shows this result using net evaporation from Price. It was constructed by

continuing the 192day lagged mean filter for net evaporation, and maintaining the

normalisationusedforthe1980sdata,that is,bothMNSGSandLMNEwerenormalisedto

theearly1983summermaximumandlate1983winterminimum.



24/53


Figure18.LMNEforPrice(solid),using192dlagandthesame1980snormalisation

establishedabove,extendedthroughto2009,comparedwiththefullMNSGSdataset

(discretepointswitherrorbars).

Figure 19 shows the results of Figure 18 replotted to include only those years when

observationsof salinityweremade.Again, theFigurewasnot rescaledor renormalised,

and thus theupwardshiftofobservedNSG salinitybetween1980sand late2000sdata is

mirroredintheupwardshiftoflaggednetevaporationmeasuredatPrice.

Figure19.DatafromFigure18replottedwithyears1986to2007removed,toshow

detail.

Haddecadalshiftsofnetevaporationcausedirreversible(ratcheted)responsesofsalinity,

thenthenetevaporationshifts thathaveoccurredbetweenthe1980sandthe late2000s

wouldhavedestroyedtheearlierrelationship,andthefittorecentdatawouldbepoor.Itis

apparent in these results that themean salinityofnorthernSpencerGulfhas tracked the

1980scalibrationquitecloselyacrossaperiodofnearlythreedecades,despitethepresence

ofsubstantialintra andinterdecadaltrendsevidentinthenetevaporationdata(Figure18).

FromthisitisdeducedthatmeanNSGsalinityatanyparticulartimeisprimarilydetermined

bythecumulativeeffectsofnetevaporationduringtheprecedingsixmonths,andthatthe

Gulfisnotsignificantlyinfluencedbyenvironmentalforcingonlongertimescales.Thisresult

maybeusedtodetermineacalibrationofchangesinNSGsalinityrelativetochangesofnet

evaporation,inthenextsection.

4 Calibration of Salin ity Change in Northern Spencer Gulf to Net

Evaporation

4.1 Behaviour from 1982 to the present

Theresultsabove implythatachangeofnetevaporation (measuredataSouthAustralian

land station) is correlatedwith a change ofmean salinity in northern SpencerGulf. The

relationship can be deduced by carefully comparing the variations of both properties

without normalising either. However, some consideration of the effects of filtering a

periodictimeseriesisrequired.



25/53


Figure20.Theeffectsoffilteringaperiodictimeseries,asdiscussedinthetext.

Figure20showsaperiodictimeseries(inblack)withanamplitudeofoneunitandanotional

periodof365timestheintervalbetweenindividualvalues,representinganannuallyvarying

quantitysampledonadailybasis.Theseries (inblue)withanamplitudeofapproximately

0.6

units

is

the

result

of

applying

a

(central)

filter

of

190

samples

(or

days)

in

length,

showing

thattheamplitudeofthefilteredresultissubstantiallysmallerthanitsunfilteredequivalent.

Thereductiondependsonthefilterlengthanditsrelationshipwiththeperiodoftheseries.

Foranannualperiodvariable,filteredusinga190dfilter,theamplitudeisreducedto0.61of

the unfiltered amplitude. A filter length of 192d, produces a series with an amplitude

reducedbythefactor0.60.

Figure21shows theresultofthisprocessapplied tonetevaporationmeasurements from

AdelaideAirport7. Individual points representdailymeasured values (with ameanof 2.4

millimetres perday). The 190d filtered result is shown as a solid red line. Increasing the

amplitudeofthefilteredresultbydividingbythefactor0.61,alsoensuringthatthemeans

coincide,produces the result shown as theblue solid line in Figure 20. The filtered and

adjustedresultrepresentsthesmootheddailyvariationofnetevaporation.

Figure21.NetEvaporationatAdelaideAirport,with190dfilterapplied(redsolidline)and

thefilteredresultadjustedtocompensateforamplitudereductionduetofiltering(blue

solidline).

Figure22showsthemeansalinityofNSGasmeasuredonfourteenoccasionsbetweenJuly

1982andMarch2009,withfilteredandadjustedvaluesofnetevaporationfromPriceon

7 Adelaide Airport was chosen for illustration because Price and Spalding data suffer from many

missing daily values.



26/53


thesamedays.Theleastsquaresregressionresult(withacorrelationcoefficientof0.89) is

alsoplottedontheFigure.Theresultindicatesthatasustainedchangeofonemillimetreper

day of net evaporation corresponded to a change ofmeanNSG salinity of 0.230gl1. For

completeness, the regressionresult fittedto1980sdataalonegaveagradientof0.228gl1

andacorrelationcoefficientof0.87.

Figure22.MeanSalinityofNSGagainstcorrespondingvaluesoffilteredandadjustednet

evaporationfromPrice(usinga192dfilter).Theleastsquaresregressionresultisplotted

togetherwith95%confidenceintervals.

Based on this relationship, by using net evaporation as a proxy (hindcast) indicator, the

historical behaviour of salinity inNSG during the period 1982 to 2009may be inferred.

Figure23 shows theNSGmean salinity time series inferred from its relationshipwithnet

evaporation,togetherwith3yearand5yearcentralrunningmeans.



27/53


Figure23.InferredbehaviourofNSGsalinityduringthepast27years,with3yearand5

yearcentralrunningmeans.DiscreteobservationsofmeanNSGsalinityarealsoshown

togetherwitherrorbars.

Results inferred from thisproxy suggest that,during theperiod 1982 to 2009, themean

salinityofNSGwas41.56gl1; theannualmeanpeaktotroughvariationwas1.38gl

1,with

themean annual peak being 42.25gl1; the lowest annual average salinity occurred from

1July1997 to 30June1998 with a value of 41.25gl1; the highest annualmean salinity

occurredintheyear1July2007to30June2008withameanof42.02gl1;thustheannual

mean salinity ofNSG rose by 0.77gl1 during the ten year period to 2007/8. The annual

salinitymaximumwasreached,onaverage,on12thApril,andtheannualminimumon14

th

October.

5 Natural & Anthropogenic Influences on Northern Spencer GulfSalinity

5.1 Potential Desalination Impacts

Theprocessofdesalination involves the intakeofacontinuous flowofambient seawater

fromamarinezone,theremovalofsaltfromafraction(typically40%to45%)ofthe flow

whichistappedoffasfreshwater,anddischargeoftheremainingflow,whichcarriesaway

the sameamountof saltas theoriginal inflow,atanelevated (typicallyninefifthsof the

original)salinity.Thus,theneteffectofdesalination istheremovaloffreshwater,which is

alsotheneteffectofevaporation,althoughtheformeractsatapointwhilethe latteracts

(approximately)uniformlyoveralargearea.

This distinction in theway desalination and evaporation act is ofparamount importance

whenconsideringpotentialeffectsoflocalisedbrinedischarge.However,theaimhereisto

assessitseffectsonalargevolumeofSpencerGulf,thatisnorthernSpencerGulf,overthe

long

term.

Figure24.Schematicgulfsubjecttoevaporationanddesalination.Notethatthereisno

intentiontoprovidemeaningfulindicationofaregionofinfluenceassociatedwith

withdrawalfrom,anddischargeoffluidsintotheGulf.Itisalsoimportanttonotethatthe

schematiclocationofthedesalinationplantputsitentirelywithintheconfinesofNSG,

whichdiffers

from

its

proposed

location

at

the

intersection

of

NSG

and

the

mid

Gulf.

For



28/53


thisreason,thediscussionsurroundingthisarrangement,inthetext,maybeconsideredto

overestimatepotentialimpactsonNSG.

Figure24showsaschematicgulf(thesameasinFigure3)withtheadditionofadesalination

plant on its shore. The plant draws in seawater from an undefined zone and discharges

brines, as shown here, entirely to northern Spencer Gulf. Thus, freshwater removal by

desalination increases the total lossof freshwater fromNSG (representedby the vertical

arrowsintheFigure),whichenhancesthecompensatinginflow, Sa,andinturn,raisesthemeansalinityofNSG(asdescribedearlier).

RegardlessofthedegreeofmixingofdesalinationreturnwaterintoNSG,themeansalinity

will reflect the gross removal of freshwater (whether localised or not) which drives a

compensating landward salt flux, so that desalination acts as a small but equivalent

mechanism to evaporation. This report explicitly excludes consideration of any localised

effects focusing, instead,on the longterm (measured inmonths)grossbehaviourofNSG,

whichmust respond todesalination in the sameway that it responds toevaporation,by

elevatingmeansalinity.

ItisalsoassumedinthefollowinganalysisthatthedesalinationplantissitedalongsideNSG,

anditseffectsareentirelyappliedtoNSG.Thisassumptionexaggeratespotentialimpactson

NSGbecausethe intake isproposedtobesitedveryclose (oftheorderof100m)toPoint

Lowly(seeFigure2)withthedischargeonthePointitselfwhichmeans,giventhegenerally

northsouth orientation of tidal flows and themean effects of the deeper gravitational

circulation (discussed in section 6.2), that some of the effectswill actually be applied to

watersoutsideNSG.

5.1.1 Quantitative Estimate of Desalination Impact

ReferringagaintoFigure24,weusethefollowingnumerics:

Proposeddesalination(orfreshwaterremoval)rate: 251megalitresday1

=2.51x108litresday

1

=2.51x105m

3day

1

NorthernSpencerGulf:

Volume 4.4x109m

3

SurfaceArea 0.52x109m

2

MeanDepth 8.4m

WholeofSpencerGulf

SurfaceArea 2.2x1010m

2

Assumingthatdesalination isgrosslyequivalenttotheremovalofa freshwater layer from

theentiresurfaceareaofNSG,thelayerthicknessisdeterminedas:

(totalvolumeremovedbydesalinationperunittime) / (surfaceareaofNSG)

= (2.51 x 105m3day

1) / (0.52 x 10

9m2)

= 0.00048 m day1

= 0.48 x 10-3

m day-1

= 1.92 x 10-6x DR m day-1



29/53


whereDRisthedesalinationrateinmegalitresperday.Thus,theevaporationequivalentof

desalination is a loss of approximately one half of amillimetre per day from the entire

surface of NSG, representing about 15% of the (adjusted) mean daily evaporative loss

measuredatSpaldingandPricebetween1982and2009.For comparison, the best case

alternativescenarioinwhichthefreshwaterlossisconsideredtobedrawnfromtheentire

surface

area

of

Spencer

Gulf

implies

that

desalination

effectively

removes

a

surface

film

of

1.1x 10-5

m day-1,orapproximately0.3%ofnaturalevaporation.

Asdeducedinsection4.1(andFigure22),afreshwaterlossof1x103mday

1(inadjusted

LMNE at Price) corresponded to a change ofMNSGS of 0.23gl1 implying that the gross

effects of desalination, if assumed to impact entirely onNSG, are expected to cause an

increaseofmeanNSGsalinityof(0.48x0.23gl1 or)~0.11gl

1 (or4.4x104xDRgl1).This

representsariseequivalenttoapproximately8%ofthemeanannualsalinityrange(1.38gl1)

observed in NSG from 1982 to 2009, implying a rise of annualmean NSG salinity from

41.56gl1 to 41.67gl

1.Italsorepresentsachangeequivalentto14%oftheobservedinter

decadalshift (0.77gl1)ofmeanannualNSGsalinitybetween1997and2007, inferredfrom

thenetevaporationproxy.

Forcomparison,usingtheNSGcalibrationofsalinitywithnetevaporation,whichisexpected

tooverestimate the salinity responseof thewholeof SpencerGulf, the gross impact of

desalinationonthewholeGulfisinferredtobeanincreaseofmeansalinityof~0.003gl1(or

1x105xDRgl1).

5.2 Potential Climate Change Impacts

5.2.1 Past Trends in Climate Variables

Suppiahetal(2006,hereaftercalledS06)reportobservedtrendsandmodelledpredictions

of climate change effects on South Australian atmospheric andmarine conditions. This

report

does

not

question

the

veracity

or

validity

of

these

predictions

in

the

context

of

debateaboutclimate change,but simplyaccepts the figures inorder todeducepotential

impacts.

Since the 1950s, South Australias average air temperature has increased by 0.20C per

decade(theaveragemaximumhasincreasedby0.21Candtheaverageminimumby0.18C

per decade). Sea surface temperatures (SST) have also risen, but by smaller amounts.

Specificallybetween1900and2005,theSSTofwatersnearthemouthofSpencerGulfhas

increasedby0.05Cperdecade,oratahigherrateof0.11Cperdecadesince1950(S06).

Over the whole State, rainfall trends were generally weak, showing a small tendency

towardswetter conditions in thenorthof theState,anddrier conditions in the southern

coastalareas,changingbyapproximately5x103mdecade

1from1900to2005.Theperiod

1950 to2005 showeda slightly strongerdrying trend for coastal regionsofup to102m

decade1,althoughthedetailintheseobservationssuggeststhattheregionofNSGmayhave

experienced no significant longterm trend (see Figure 6 of S06). Longterm changes of

rainfall were very small, although there were very large interannual and interdecadal

variationsbetween1956and2009froma lowof234.6millimetresannualrainfall in2006,

up to730.8millimetres in1992,aroundameanof441.7millimetres (atAdelaideAirport,

BureauofMeteorologyonlineclimatestatistics8).

Evaporationmeasurementswereassessed fromUSClassApanevaporationdata,which is

acknowledgedtobepronetosignificanterrors(S06)atthreesites,Woomera,Cedunaand

8

http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=139&p_display_type=dataFile&p_startYear=&p_stn_num=023034

http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=139&p_display_type=dataFile&p_startYear=&p_stn_num=023034http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=139&p_display_type=dataFile&p_startYear=&p_stn_num=023034http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=139&p_display_type=dataFile&p_startYear=&p_stn_num=023034http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=139&p_display_type=dataFile&p_startYear=&p_stn_num=023034


30/53


Mount Gambier. Only Woomera showed a statistically significant trend, rising by

11.5x103myear

1,eachyeararoundameanof3.21myear

1.

5.2.2 Projected 2030 and 2070 Air Temperatures

S06wenton toassess the resultsof25climatemodels frommanydifferent international

sources.Thesemodelsshowsubstantialinconsistenciesintheirprojectionsofchangewhich,

whencombinedwiththeIntergovernmentalPanelonClimateChangealternativescenarios,

lead topredictionsofchange that fall intobroad ranges.With regard toair temperature,

SouthAustraliaisassessedinthreeregions:lessthan200kmfromthecoast(whichwerefer

toasthe coastalband);200to600kmfromthecoast;andgreaterthan600kmfromthe

coast. We will assume that Spencer Gulf belongs to the coastal band, as continental

conditionsmayoverwhelmthesmallmaritimecontributionofNSG inthesecondband.By

2030, air temperaturewithin the coastalband isprojected to riseby0.2C to 1.6C,and

2070projectionsindicateariseof0.5Cto4.7C.

S06alsoprovidedprojectionsfortheNationalResourceManagement(NRM)regions.The

two NRM regions thatmost closely straddle NSG are those of the Northern and Yorke

Region and Eyre Peninsula. Summaries are provided for these regions and are largely

consistentbetweenthetwo.By2030airtemperature isexpectedtoriseby0.4Cto1.2C,

and by 2070 this is projected to be 1.0C to 3.8C, so the NRM forecasts are more

conservativethanthoseofthecoastalband.

5.2.3 Projected 2030 and 2070 Precipi tation

Developingforecastsforrainfallissaid(S06)tobemorecomplexandlessconsistentthanfor

temperatures.Using the30yearannualaverage rainfall centredon1990as thebaseline,

projectionsforthecoastalbandshowachangeof 15%to0%by2030,and 45%to0%by

2070.ProjectionsfortheNRMregionsindicatechangesof 10%to0%by2030,and 30%to

1%

by

2070

which

are,

again,

more

conservative

than

those

of

the

coastal

band.

5.2.4 Projected 2030 and 2070 Evaporation

S06 did not report evaporation trends, primarily because direct measurement of

evaporation is difficult, and evaporation is usually assessed as the residual term in heat

budgetcalculations.

5.2.5 Estimating 2030 and 2070 Climate Change Impacts on the Salinity ofNorthern Spencer Gulf

InordertodeduceclimatechangeeffectsonNSGsalinity, it isnecessarytotranslatetheir

effectsintoanoverallassessmentofthechangetonetevaporation.Estimatingevaporation

astheresidualterminaheatbudgetcalculationwillnotsufficeherebecausesuchestimates

wouldinherittheuncertaintiespresentinallotherterms.

Twoalternativemethods toassessevaporationareofferedbelow.Firstly,asnoted in the

discussion of Figure 10 and section 2.1, evaporation and precipitation appear to be

negatively correlated and thus, projections of changing rainfallmay be used to deduce

evaporationprojections.Alternatively,projectedchangesofairandseatemperature (S06)

allowevaporationtobeestimatedusingbulkaerodynamicformulae.Bothmethodswillbe

usedtoestimateevaporationin2030and2070which,combinedwiththeS06projectionsof

precipitation, yield estimates of net evaporation fromwhich changes toNSG salinity are

deduced.



31/53


5.2.5.1 EstimatingImpactsfromProjectionsofPrecipitation

The3year runningmeansofevaporationandprecipitationatAdelaideAirport (shown in

Figure10) implya relationshipbetweenevaporationandprecipitation trendswhichmight

bededucedthroughsimpleregression. Suchaprocedure,usingannualmeans,showsthat

thetwoparametersarecorrelatedaccordingto:

(Evaporation)365d mean = -0.75 x (Precipitation) 365d mean + 4.6

wherebothevaporationandprecipitationaremeasured inmillimetresperday.Thisresult

supports translation of the climate projections of precipitation into estimates of the

correspondingevaporationtrends.

2030: Suppiah et al (2006) projected precipitation changeswithin the coastal band for

2030 and 2070 from the 30year baseline centred on 1990, indicating that a

reductionofbetween0%and15%isanticipatedin2030(amoreextremeworstcase

thanwasobtainedbyconsideringthetwoNRMregions).

SinceAdelaideAirport showsausefulprecipitationevaporationcorrelation,and it

belongswithinthecoastalband,theresultisusedasfollows.Forthe30yearperiodcentredon1990,theannualaverageprecipitationatAdelaideAirportwasmeasured

tobe448x103myear

1(BureauofMeteorology

9).

The2030annualaverage rainfall is consequentlyprojected (S06) to liewithin the

range381 x103myear

1 to448x10

3myear

1.From the regression relationship

above, the 2030 annual evaporation is therefore inferred to lie between

1.38myear1and1.33myear

1.Thus,theworstcasenetevaporation(evaporation

minus precipitation) is inferred to be 1.00m year1, representing an increase of

0.12myear1 or0.32x10

3mday

1over1990centredaverageconditions

2070: S06projecteda reductionofprecipitation in thecoastalbandofbetween0%and

45%

by

2070,

thus

precipitation

at

Adelaide

Airport

is

expected

to

lie

within

the

range246x103myear

1and448x10

3myear

1in2070(notingthatprojectionsfor

theNRMregionswerelessextremethanthis).

Evaporation is consequently projected to lie between 1.48 m year1 and

1.33 m year1, leading to a worst case increase of net evaporation in 2070 of

0.35myear1,or0.97x10

3mday

1over1990centredvalues.

5.2.5.2 EstimatingNetEvaporationChangesfromProjectionsofAirTemperatureand

Precipitation

Estimates of evaporation may be obtained using bulk aerodynamic formula (see, for

example,Brutsaert,1982;andHolloway,1980),as follows.Thebulkaerodynamic formula

forthelatentheatflux(He)associatedwithevaporationisgivenas:

He= aL C Eu 10(q 0 q 10)

whereaisthedensityofair,Listhelatentheatofvapourisationofwater,CE isthebulktransfer coefficient for moisture (~0.0019), u10 is the wind speed at the standard

meteorologicalheightoftenmetres,andq isthespecifichumidityatthesurface(suffixof

zero)andattenmetres(suffixof10).

Thespecifichumidityisgivenby:

q=(0.622e)/(P0.378 e ) 0.622e/P

9 http://www.bom.gov.au/climate/averages/tables/cw_023034.shtml

http://www.bom.gov.au/climate/averages/tables/cw_023034.shtmlhttp://www.bom.gov.au/climate/averages/tables/cw_023034.shtml


32/53


where P is atmospheric pressure, and e is the vapour pressure, which is temperature

dependentaccordingtothefollowingpolynomial:

es(T) = a0+ a 1.T + a2.T2+ a 3.T

3+ a 4.T

4+ a 5.T

5

inwhicha0=6.11,a1=4.44x101 ,a2=1.42x10

2,a3=2.71x104,a4=2.74x10

6 and

a5

=

2.73

x

10

8

.

Evaporation isdrivenby thegradientof specifichumiditybetween thewater surfaceand

10mheight,whichisgivenby:

q0 q 10 [e (T0) e (T10)] / (1.61 P) = [es(T 0) R.es (T10)] / (1.61 P)

where thevapourpressureat10mheight isequal tothesaturatedvapourpressure times

therelativehumidity(R)at10m, leadingtoestimationofevaporationchangesaccordingto

projectedchangesinairandseasurfacetemperatures.

2030: Projectionsofmeanatmospherictemperatureriseinthecoastalbandforthe2030

timeframe (S06)werebetween0.2Cand1.6C.Basedonhistoricdataforthe last

halfcenturyS06alsonoted thatseasurface temperature (SST)near themouthof

SpencerGulfhadrisenatapproximatelyhalftherateofairtemperature.Itmightbe

argued thatSSTwithinSpencerGulf,particularlyNSG, shouldmore closelymatch

the fullmagnitude of the rise in air temperature.However, using different rates

produces a larger atmospheric vapour pressure gradient and hence greater

evaporation,representingaworstcasescenario.Onthisbasis,itisinferredthatSST

inNSGin2030willhaverisenby0.1Cto0.8C.

MeansummertimedailymaximumairtemperaturesatPortAugustaPowerStation

(the top of NSG) between 1962 and 1997 were approximately 31C (Bureau of

Meteorology),whileSSTsinNSGweremeasured(NL86)atapproximately24C.Daily

peak summertime latent heat fluxes using these temperatures (and assuming a

relative

humidity

of

75%,

the

approximate

average

summertime

value

coincident

withpeakdailyairtemperatures,recordedatYarravilleShoal,3317S13735.6E,

approximately35kmsouthsouthwestofPointLowly,NunesVazdata)givesapeak

evaporationrateof2.7x108ms

1.Assumingthat,in2030,meandailymaximumair

temperatureswill have risen to 32.6C,withNSG SSTs at 24.8C leads to a peak

evaporationrateof4.0x108ms

1,representinga45%increaseinpeakevaporation.

Making the two furtherassumptions thatdailyevaporation is sinusoidalbetween

zeroand thepeakdaily value,and that theannualevaporation isalso sinusoidal,

withanannualmeanofapproximatelyhalfthesummertimepeak(bothofwhichare

approximately true from observations), implies that the 2030 estimate of

evaporation isapproximately11% largerthan1990values, reaching1.47myear1.

Combiningthiswiththeprojectedchangeofprecipitationimpliesthatmeanannual

net evaporation in 2030 will reach 1.09 m year1, an increase of

0.21myear1.

2070: A similar procedure, assuming that air temperature rises to 35.7C and the sea

temperature rises to 26.3C implies that annual evaporation increases to

1.83myear1and meanannualnetevaporationin2070willreach1.59myear1,an

increaseof0.71myear1.

5.2.5.3 ImpliedEffectsonNorthernSpencerGulfSalinity

Withtheseestimatedchanges,itisnowpossibletoprovideinferredimpactsonNSGsalinity,

based

on

the

result

of

section

4.1.



33/53


2030: Projections of precipitation changes for 2030 (S06) indicate an increase in net

evaporation of 0.32 x 103m day

1,which implies an increase in NSG salinity of

0.07gl1.

Air and sea temperature projections, combined with projections of precipitation

changes,indicateanincreaseinnetevaporationof0.21x103mday

1,whichimplies

anincreaseinNSGsalinityof0.05gl1.

2070: Projections of precipitation changes for 2070 (S06) indicate an increase in net

evaporation of 0.97 x 103m day

1,which implies an increase in NSG salinity of

0.22gl1.

Air and sea temperature projections, combined with projections of precipitation

changes for 2070, indicate an increase innet evaporationof 0.71 x 103mday

1,

whichimpliesanincreaseinNSGsalinityof0.16gl1.

These results are collected, togetherwith the estimated impacts of desalination and the

observednaturalvariationofsalinitybetween1982and2009,inTable2.

Influence Basedon Changeof

Net

Evaporation

(x103mday

1)

Changeof

NSGmean

salinity

g l-1

Inferred

Mean

Salinity

g l-1

Impacts,

Relativeto

Natural

Variation

Annual

Variation

Mean1982to2009 6.0 1.38 41.56 100%

Desalination 180x106litresday

1


34/53


Figure25.MeanannualsalinityvariationinNSGinferredfromthePriceLMNEproxy(blue)

togetherwithinferredsalinitycyclesthatincludetheeffectsof:desalination(green);

climatechangein2030(orange);andclimatechangein2070(red).TheinferredNSG

salinitiesduring199798and200708(inblack)areincludedtoindicatetheextentofinter

annualvariability,notingthatthesetracesincludetheeffectsoflargeprecipitation

variabilityandirregulardigitisationassociatedwiththecollectionofmeteorologicaldata.

Figure26.Fourpanelsshowingcomparativeestimatesofbothclimatechangeand

desalinationimpactsonNSGsalinity.Theupperleftpanelcompares2070climatechange

alone,andcombinedwithdesalination,onthemeanannualsalinitycycleinferred

between1982and2009.Theupperrightpanelshowstheseperturbationsasadditionsto

themostextremeannualsalinitycycleinferredbetween1982and2009.Thelowerleftand

lowerrightpanelsareequivalenttotheupperpanels,wherepredictionsoftheimpactof

2030climatechangehavebeensubstitutedforthoseof2070.

ThestrongbluelinesinFigure26areallequivalentandshowtheannualmeansalinitycycle

inferredfromnetevaporationbetween1982and2009(mean41.56gl-1,amplitude0.69gl

-1).



35/53


Irregular (black) traces (and their smoothed interpretation) show the inferred cycle for

200708(mean42.02gl-1)whichwasthehighestmeansalinitycycleinferredfromtheproxy

between1982and2009.Theimpactsofclimatechange(red),andthenincombinationwith

desalination(theadditionalgreyband)arerepresentedasperturbationstothemeansalinity

(leftpanels)and to thehighestextreme salinity (rightpanels) forboth2070 (upper)and

2030

(lower)

climate

change

predictions.

With respect toworst case scenarios, combining the impactsofdesalination and climate

changewiththemeanofthehighestannualcycleofsalinityinferredfromthepast27years

(in200708),themeansalinityofNSG isexpected to riseto42.20gl-1 in2030 (lowerright

panelofFig25),and42.35gl-1in2070(upperrightpanel).Theseresultsrepresentincreases

of 46% and 57%, respectively,of themean annual salinity variation (of 1.38gl-1) inferred

between 1982 and 2009. The peak salinities for the extreme salinity cycles shown in

Figure26wouldbe42.92gl1and43.07gl

1,for2030and2070respectively,comparedwith

theannualmeanpeakof42.25gl1(asshownbythebluelineinFigure26).

Atthispoint,thefirsttwodrivingquestionshavebeenresolved,thatis:

1. howSpencerGulf responds to influences thatdisturb its salinity,andhow long it

takestorecoverfollowingadisturbance,and

2. the degree or extent to which salinity is expected to change in response to a

disturbance.

Thenextsectionaddressesthelastofthethreequestions,namely:

3. whetheranticipatedperturbationsassociatedwithdesalinationandclimatechange

willbe large enough to shift theGulfs salinity, irreversibly, to anewequilibrium

state.

6 Equilibrium Dynamics of Oceanographic Salt Exchange in North

Spencer Gulf

Toanswer the thirdquestionabove requiresexaminationof thenatureofoceanographic

saltexchangeprocessesoperatinginSpencerGulf.Asbrieflydiscussedinsection1.1,there

are twoprocesses thatcontribute to the transportof saltdown theGulfsnaturalsalinity

gradient, namely gravitational circulation, and turbulent diffusion and dispersion. Both

contributetothemaintenanceofalongtermequilibriumandeachdominatestheother in

different(wind,tideandcurrent)conditions,locationsandtimes(forexample,seasons).

However, the assessment may be made relatively straightforward by examining each

mechanisminturn,toaskwhethertheremightbeanaturallimitationontheexchangethat

eachcanachieveindividually.

6.1 Horizontal Gravitational Circulation

NV90usedasimpletwodimensional(flat)representationofUpperSpencerGulf(seeFigure

27) characterised by two contributions to flow, namely: a net inflow, at speed uE, re

supplyingthefluidlostbynetevaporation(whichwasalsoshownschematicallyinFigure3);

andahorizontalcirculationat speeduC,carrying lowersalinity fluid (si)northwardon the

westernside,andhighersalinityfluid(so)southwardontheeasternside,toaccountforthe

observed twodimensional orientation of isohalines. In this arrangement, NV90 showed

(theirequation3)thatthecirculationspeedisgivenby:

)(.

).(2

io

i

c ss

s

bh

APEu

= (1)



36/53


where(E P)isthenetevaporationfromsurfaceareaA,bisthehorizontalwidthandhis

the depth of the southern section through which the flows occur. Assuming that all

quantitiesare fixed, including the salinityof the inflow (si), the salinityof theoutflowwill

increase as the speed of the circulation decreases. Intuitively, thismeans that amore

sluggish circulation is more affected by evaporation causing the outflow salinity to be

correspondingly

higher.

Figure27.SchematicTwoDimensionalSaltExchangesinSpencerGulf

(reproducedfromNunesVazetal,1990)

Figure28.Therelationshipbetweencirculationspeedandinflowoutflowsalinitycontrast

definedbyequation(1)

Takingthesouthernboundarytobeat33S(thesouthernlimitofNSGasdefinedhere),and

using thedataofNV90 (i.e., (E P) 4.4 x 10-8

ms-1, A ~ 6 x 10

8 m

2, b ~ 1.6 x 10

4 m,

si~ 41gl-1

) with h ~ 8.4m Figure 28 showstherelationshipbetweencirculationspeedand

the inflowoutflow salinity contrast. Assuming, as in NV90, that the outflow salinity

so~ 42gl-1impliesthatuC isapproximately0.02ms

-1

.Thus,arathermodestnetcirculationisrequiredtoaccountfortheflushingofexcesssaltfromNSGproducedbynetevaporation.



37/53


As these flowsarehorizontallydistinct, there isonlyoneconceptual limitationon thenet

fluxofsalttheyareabletoachieve,which issetbythehydraulicsofthechannel inwhich

the flows takeplace.Thehydraulic limit is representedbyaFroudenumber,Fr, (seeany

hydraulicstext):

gh

u

Fr= (2)

whereg istheaccelerationofgravity,u isavelocityandh isadepth,whichexpressesthe

notion that the gravitational potential energy of the system (represented by the

denominator)cansustainaflowatthecontrolsection(whichwetaketobe33S)toproduce

aFroudenumberno greater thanone.This limit isachievedwhereu =gh implyinganupperlimitonthecirculationspeedofapproximately8ms-1. Clearlythisdoesnotconstitute

asignificantconstraintonhorizontallydistinctexchangeflows.

However, there isnodynamicmechanismable to sustain the complete lateral separation

between inflowandoutflow in thepresenceof gravity. Ignoring the Earths rotation, the

flows

would

instead

be

vertically

differentiated

(with

a

low

density

inflow

above

a

high

densityoutflow),whichisconsiderednext.

6.2 Vertical Gravitational Circulation

Spatially differentiated exchange flows influenced by gravity in a nonrotating frame of

reference are verticallyjuxtaposed, separated by a horizontal interface with inflow and

outflowlayerseachoccupyingapproximatelyhalfthedepth.Thetwo layers(withdensities

iando,wherethesubscriptsrepresentthelessdenseinflow,i,andmoredenseoutflow,

o)remainstableinthepresenceofshearacrosstheinterfaceuntilthegradientRichardson

number,Ri, decreases to approximately 0.25 (Turner, 1973). The gradient Richardson

numberisgivenby:

2)(

'

io

iuu

hgR

= (3)

inwhich h is the vertical scale of the interface, and g is the reduced gravity (given by

oiog ).( whichassumesthattheBoussinesqapproximation isvalid,that is,densitydifferences are small). Large Richardson numbers represent stable conditions. As the

Richardsonnumberapproaches0.25,theenergyimpartedbytheshearacrosstheinterface

becomessufficienttoovercomethestabilityofthestratification,andtheinterfacetypically

forms large rolling (KelvinHelmholtz)billows that rapidlyachieve verticalmixingof scalar

properties(suchassalinityordensity)andmomentum.

Thus, an organised vertically differentiated inflow/outflow is able to support the net

exchangeof saltuntileither thedensitycontrast falls,or the shear increases such thatRi

approaches0.25.Whensuchan interfacebecomesunstable,boththedensitycontrastand

the shear are suddenly reduced in a catastrophic breakdown of the exchange flow

arrangement.This is the limit thatmustbeunderstood because it representsapotential

limittothemaintenanceofanequilibriumbetweentheprocessesthatgenerateanexcess

ofsaltinNSG,andthosethatremoveit.

An alternative stability criterion is represented by the internal or densimetric Froude

number,Fri,givenby

hg

u

Fri '=

(4)



38/53


noting that reduced gravity replaces normal gravity in equation (1).While the use of a

Richardsonnumberstabilitycriterionwouldbepreferred,itisnotpossibletoestimatepre

and postmixing interface thicknesses. The internal Froude number uses a characteristic

verticalscaleof,h,thelayerdepthratherthantheinterfacethickness,andisthereforemore

easilyapplied.AstratifiedshearflowlosesstabilityastheinternalFroudenumberincreases

and

approaches

unity.

Reinterpreting the horizontal net circulation discussed in section 6.1 as a vertically

differentiated exchange flow with each layer occupying half the depth, equation (1)

describestherelationshipbetweentheverticalshear(twicethecirculationvelocity)andthe

verticalsalinitycontrast.Thus,forgivenenvironmentalconditions,equation(4)canbeused

toquantifytheinternalFroudenumber.

Figure29.Theoreticalspeedofthesouthwardcirculationcomponent,asafunctionofits

salinity,requiredtomaintainthelongtermsaltbalanceofNSG(assumingthatthesalinity

ofthenorthwardcomponentis41.0gl1).TheinternalFroudenumberofthestratified

exchangeflowisalsoshown.

Assuming that the balance expressed by equation (1) applies to a vertically separated

exchange circulation, and that the inflow salinity (si) remains at 41.0gl1 (anchored by

conditions furthersouth) therelationshipbetween theoutflowsalinityandthecirculation

speed(Figure28)impliesthatasthesalinitycontrastbetweeninflowandoutflowdecreases,

boththecirculationspeedand,throughequation(4),theinternalFroudenumber,increase.

Figure

29

indicates

that

internally

critical

flow

(Fri~ 1 )

is

reached

when

the

outflow

salinity

falls to 41.35gl-1 which implies that, if the exchange circulationwere entirely vertically

separatedthenthedischargeofexcesssaltfromNSGcouldnotbemaintainediftheaverage

salinitycontrastbetweeninflowandoutflowfellbelow0.35gl-1.Limiteddataobtainedfrom

NSGdonothaveadequatespatialandtemporalresolutionto indicatewhether in factthe

sectionmaintains stratification to this degree, but this appears to be quite a stringent

conditionwithregardtothedischargeofexcesssalt.

Continuingthislineofthought,assumingthatthiscrosssectionremainsacontrolsectionfor

the internal hydraulics as additional stresses associated with desalination and climate

changeareapplied,manipulationofequations (1)and (4) implies that theverticalsalinity

stratificationmustincreasewith .Thus,ifnetevaporationincreasesby10%,the

salinitystratificationat33Smustriseby6.5%implyingthatthedeepoutflowsalinitymust

3/2)( PE



39/53


rise from 41.35gl-1 to 41.37gl

-1, a verymodest adjustment, tomaintain the equilibrium

exchange.

6.3 Gravitational Circulation Including Earths Rotation

In fact neither the horizontallydifferentiated nor verticallydifferentiated flows are

physicallyrealisticsolutionsbecausetheEarthsrotational(Coriolis)effectsareimportantat

thisscale.Theeffectofrotationistotilttheinterfaceawayfromthehorizontal,moresoas

the rotational effects increase, so that the horizontal differentiation between the flows

increases.Ou(1984)showedthatthedistancebetweentheinterfaceintersectionswiththe

surfaceand theseabed isapproximately twicethe internalRossbyradiusofdeformation,

Roi,givenby:

f

hgRoi

'= (5)

wherefistheCoriolisparameter2..sin,inwhichistherotationrateoftheEarth(7.29x

10

5

rad

s

1

)

and

is

the

latitude.

Thus

as

f

increases

the

Rossby

radius

decreases

and

the

interfacebecomesmorevertical.

Thestratificationrangeshown inFigure29 impliesan internalRossbyradiusofbetweena

fewhundredmetresandapproximatelythreekilometres,comparedtothewidthofNSGat

33Swhich ismorethan7kilometres.Thus,the interfacebetweeninflowandoutflow(see

Figure 30) is inferred to bemore vertical than horizontal implying that the horizontally

separatedexchangecirculation(discussedinsection6.1)isabettermodelthanthevertically

separatedcase(discussedinsection6.2).Consequently,theexchangecirculationisunlikely

tobelimitedtoanysignificantextentbyaninterfacialstabilitycriterion.

Figure 30. (a) Smoothed version of the steadystate density field after geostrophic

adjustmentfrom

an

initial

vertically

uniform

non

linear

horizontal

gradient.

The

horizontal

scale, y, is measured in units of the internal Rossby radius of deformation. (b) The

correspondinggeostrophicvelocityfieldperpendiculartotheplaneofthediagraminthe

caseofSpencerGulf,anegativevelocity shouldbe interpretedasa southwardoutflow.

ReproducedfromOu(1984).

6.4 The Role of Temperature in Driving Gravitational Circulation

Given that the effects of climate change are expected to include elevation of Gulf

temperatures, it is important toassess the roleof temperature in theGulfsgravitational

exchangeprocesses,andwhetherhighertemperaturesmayhaveanadverseeffectonthe

exchange mechanisms. The gravitational exchange is driven by alongGulf density

differences,andbothtemperatureandsalinityaffectdensity,althoughbydifferingamounts.

Temperature affects density through the thermal expansion coefficient, which is



40/53


approximately1.7x104perC,that is,thedensitydecreasesbyapproximately0.017%for

each degree of temperature rise. Similarly, the haline contraction coefficient is

approximately7.8x104pergl

1ofsalinitychange,ordensity increasesby0.078%foreach

gl1of salinity increase.Thus theeffectof salinity (ingl

1)ondensity isapproximately4.6

timesthatoftemperature(inC)10.

NL86reportedtemperatureaswellassalinityfromeachofthetwelve1980sGulfsurveys.

Theyshowed(NL86,Figure4)thatthetemperaturedifferencefromtheheadoftheGulfto

33S(i.e.,alongNSG)reachesamaximuminJanuaryofapproximately2C.Equivalently,the

maximumsalinitydifferencereachesapproximately7gl-1,inautumn.Thesetemperatureand

salinitygradients(bothwithhighervaluesatthehead)actondensityintheoppositesense,

such that the temperaturecontribution reduces thecontributionof salinity to thedensity

gradient, by approximately 6%. At other times of the year,when the temperatures are

relatively low at the head, the temperature gradient reinforces the salinity gradient and

strengthensthedensitygradient.

Thus,climatechangemustproducearather largedifferencebetweenthetemperaturesat

Port

Augusta

and

those

at

Point

Lowly

if

the

temperature

gradient

is

to

play

a

significant

role

inreducingthedensityinducedforcingofgravitationalexchange.

6.5 Turbulent Diffusion

The alternative to an organised gravitational exchange circulation involves turbulent

diffusionanddispersion(Fischeretal,1979),represented(seeNL86)by:

2

2

.)(

x

sKs

h

PE

t

sx

+

=

(5)

wherevariablesretaintheirformermeanings,xisanalongGulfcoordinatewithitsoriginat

the head, and Kx is the alongGulf eddy diffusion coefficient associated with boundary

induced and internal sources of turbulence. Using the observed annual mean salinity

gradient, NL86 (see their Figure 7) deduced that the diffusion coefficient required to

maintain a southward diffusive salt flux equal to the northward flux associated with

evaporation,ischaracterisedbyvaluesof~10 m2s

-1nearthehead,increasingto~100 m

2s

-1

atthesouthernlimitofNSG(at33S).

AssumingthatNSGmaintainsasteady(repeatableannualmean)salinitygradientalong its

axisremovesthetimedependency,andleadsto:

2

21.)(

x

s

sKhPE x

= (6)

whichshowsthatachangeofnetevaporationrequiresanadjustmentofthecurvatureofthealonggulfsalinitygradient,assumingthatlevelsofturbulence(generatedbywinds,for

example)donotalsochange.

Using a simple inverse logarithmic representation of the alonggulf salinity variation

indicatesthata10% increase innetevaporationrequiresan increaseofNSGmeansalinity

from41.56gl1to41.74gl

1tomaintaintheequilibriumsouthwarddiffusivefluxofsalt.Thus,

again,ifthisweretheonlymechanismresponsibleformaintainingthesaltbalanceofNSG,

thenitisinferredthatrelativelymodestadjustmentofthealonggulfsalinitygradientwould

besufficienttorespondtoprojectedenvironmentalchan

appendix h8_salt balance of northern spencer gulf

Documents