usermanual swat cup

7/18/2019 Usermanual Swat Cup

1/100

SWATCUP

SWATCalibrationandUncertaintyPrograms

KarimC.Abbaspour

[email protected]


2/100

2

SWATCUP:SWATCalibrationandUncertaintyPrograms AUserManual.

Eawag2015


3/100

3

DISCLAIMER

ThisreportdocumentsSWATCUP,acomputerprogramforcalibrationofSWATmodels.SWATCUP4is

apublicdomainprogram,andassuchmaybeusedandcopied freely.Theprogram linksSUFI2,PSO,

GLUE,ParaSol,andMCMCprocedures toSWAT. Itenables sensitivityanalysis, calibration, validation,anduncertaintyanalysisofSWATmodels.SWATCUP2012hasbeentestedforallprocedurespriorto

release. However, no warranty is given that the program is completely errorfree. If you encounter

problemswiththecode,finderrors,orhavesuggestionsfor improvement,pleasewritetoSWATCUP

[email protected]


4/100

4

Content Page

Foodforthoughtincalibrationandapplicationofwatershedmodels 6

SUFI2 16

ConceptualbasisoftheSUFI2uncertaintyanalysisroutine 17

SUFI2asanoptimizationalgorithm 19

SWATCUP 20

StepbystepCreatingofSWATSUFI2InputFiles 21

ParameterizationinSWATCUP 51

Objectivefunctiondefinition 55

Sensitivityanalysis 59

ParallelProcessing 63

ValidationinSUFI2 64

Thesequenceofprogramexecution 65

Howto.. 66

PSO 70

IntroductiontoPSO 71

GLUE 74

IntroductiontoGLUE 75

CouplingGLUEtoSWATCUP 77

ValidationofGLUE 78

FileDefinition 79

ParaSol 81

IntroductiontoParaSol 82

CouplingParaSoltoSWATCUP 83


5/100

5

ParaSol:Optimizationanduncertaintyanalysis 85

MCMC 92

IntroductiontoMCMC 93

StepbysteprunningofMCMC 96

References 98


6/100

6

Foodforthoughtincalibrationandapplicationofwatershedmodels

Calibrationanduncertaintyanalysisofdistributedwatershedmodels isbesetwithafewserious issues

thatdeserve theattentionandcarefulconsiderationofresearchers.Theseare:1)Parameterizationof

watershedmodels.2)Definitionofwhatisacalibratedwatershedmodelandwhatarethelimitsofits

use.3)Conditionalityofa calibratedwatershedmodel.4)Calibrationofhighlymanagedwatersheds

wherenaturalprocessesplayasecondaryrole,and5)uncertaintyandnonuniquenessproblems.These

issuesarebrieflydiscussedhere.

1)ModelParameterization

Shouldasoilunitappearinginvariouslocationsinawatershed,underdifferentlandusesand/orclimate

zones,havethesameordifferentparameters?Probablyitshouldhavedifferentparameters.Thesame

argument could be made with all other distributed parameters. How far should one go with this

differentiation?Ontheonehandwecouldhavethousandsofparameterstocalibrate,andonotherwe

maynothaveenoughspatialresolution inthemodeltoseethedifferencebetweendifferentregions.

This balance is not easy to determine and the choice of parameterization will affect the calibrationresults.Detailed informationon spatialparameters is indispensable forbuilding a correctwatershed

model.Acombinationofmeasureddataandspatialanalysis techniquesusingpedotransfer functions,

geostatisticalanalysis,andremotesensingdatawouldbethewayforward.

2)Whenisawatershedmodelcalibrated?

Ifawatershedmodeliscalibratedusingdischargedataatthewatershedoutlet,canthemodelbecalled

calibrated for that watershed? If we add water quality to the data and recalibrate, the hydrologic

parametersobtainedbasedondischargealonewillchange. Is thenewmodelnowcalibrated for that

watershed?Whatifweadddischargedatafromstationsinsidethewatershed?Willthenewmodelgivecorrect loadsfromvarious landuses inthewatershed?Perhapsnot,unlesswe includethe loads inthe

calibrationprocess (seeAbbaspouretal.,2007).Hence,an importantquestionarisesas to:forwhat

purpose can we use a calibrated watershed model? For example: What are the requirements of a

calibratedwatershedmodelifwewanttodolandusechangeanalysis?Or,climatechangeanalysis?Or,

analysis of upstream/downstream relations in water allocation and distribution? Can any single

calibratedwatershedmodeladdressall these issues?Canwehave several calibratedmodels for the

samewatershedwhereeachmodel isapplicable toa certainobjective?Note that thesemodelswill

mostlikelyhavedifferentparametersrepresentingdifferentprocesses(seeAbbaspouretal.1999).

3)Conditionality

of

calibrated

watershed

models

Conditionality isan important issuewithcalibratedmodels.This isrelatedtothepreviousquestionon

thelimitationontheuseofacalibratedmodel.Calibratedparametersareconditionedonthechoiceof

objectivefunction,thetype,andnumbersofdatapointsandtheprocedureusedforcalibration,among

other factors. Inapreviousstudy (Abbaspouretal.1999),we investigatedtheconsequencesofusing

different variables and combination of variables from among pressure head, water content, and

cumulativeoutflowontheestimationofhydraulicparametersby inversemodeling.The inversestudy


7/100

7

combinedaglobaloptimizationprocedurewithanumerical solutionof theonedimensionalvariably

saturated Richards flow equation. We analyzed multistep drainage experiments with controlled

boundaryconditions in large lysimeters.Estimatedhydraulicparametersbasedondifferentobjective

functionswerealldifferentfromeachother;however,asignificanttestofsimulationresultsbasedon

these parameters revealed that most of the parameter sets produced similar simulation results.

Notwithstanding the significance test, rankingof theperformancesof the fittedparameters revealed

that theywerehighlyconditionalwith respectto thevariablesused intheobjective functionand the

typeofobjectivefunctionitself.Mathematically,wecouldexpressacalibratedmodelMas:

,....),,,,,( mvbwgpMM

whereisavectorofparameters,p isacalibrationprocedure,g istheobjectivefunctiontype,wisa

vectorofweights intheobjectivefunction,b istheboundaryconditions,v isthevariablesused inthe

objectivefunction,misthenumberofobservedvs,etc.Therefore,acalibratedmodelisconditionedon

the procedure used for calibration, on the objective function, on the weights used in the objective

function,onthe initialandboundaryconditions,onthetypeand lengthofmeasureddataused inthe

calibration,etc.Suchamodelcanclearlynotbeappliedforjustanyscenarioanalysis.

4)Calibrationofhighlymanagedwatersheds

Inhighlymanagedwatersheds,naturalprocessesplayasecondaryrole.Ifdetailedmanagementdatais

notavailable,thenmodelingthesewatershedswillnotbepossible.Examplesofmanagementsaredams

andreservoirs,watertransfers,andirrigationfromdeepwells.InFigure1atheeffectofAswandamon

downstreamdischargebeforeandafteritsoperationisshown.Itisclearthatwithouttheknowledgeof

damsoperation,itwouldnotbepossibletomodeldownstreamprocesses.Figure1bshowstheeffect

ofwetlandondischargeupstream,inthemiddle,anddownstreamofNigerInlandDelta.

Figure1.left)EffectofAswandamondownstreamdischargebeforeandafteritsoperationin1967.

right)TheinfluenceofNigerInlandDeltaonthethroughflowatupstream,within,anddownstreamof

thewetland.(AfterSchuoletal.,2008a,b)


8/100

8

InFigure2theeffectofirrigationonactualETandsoilmoistureisillustratedinEsfahan,Iran.Esfahanis

aregionofhighirrigationwithanegativewaterbalanceforalmosthalfoftheyear.

Figure2.

Illustration

of

the

differences

in

predicted

actual

ET

(a)

and

soil

moisture

(b)

with

and

without

consideringirrigationinEsfahanprovince,Iran.Thevariablesaremonthlyaveragesfortheperiodof

19902002.(AfterFaramarzietal.,2009)

Inthestudyofwaterresources in Iran,Faramarzietal., (2008)producedawatermanagementmap

(Figure3)inordertoexplainthecalibrationresultsofahydrologicmodelofthecountry.

Figure3.WatermanagementmapofIranshowingsomeofmansactivitiesduring19902002.Themap

showslocationsofdams,reservoir,watertransfersandgroundwaterharvest(backgroundshows

provincialbasedpopulation).AfterFaramarzietal.,(2009).

0

10

20

30

40

50

60

Jan Feb

Mar Ap

rMa

y Jun Ju

lAu

gSep

Oct

Nov

Dec

Month

ActualET(mmm

onth-1)

0

10

20

30

40

50

6070

80

90

100

Jan Feb

Mar

Apr

May Ju

n Jul

Aug

Sep

Oct

Nov

Dec

Month

Soilwater(mm

(a) (b)

With irrigation Without irrigation


9/100

9

5)Uncertaintyissues

Anotherissuewithcalibrationofwatershedmodelsisthatofuncertaintyinthepredictions.Watershed

modelssufferfromlargemodeluncertainties.Thesecanbedividedinto:conceptualmodeluncertainty,

input uncertainty, and parameter uncertainty. The conceptual model uncertainty (or structural

uncertainty) couldbeof the following situations:a)Modeluncertaintiesdue to simplifications in the

conceptualmodel,

b)

Model

uncertainties

due

to

processes

occurring

in

the

watershed

but

not

included

in the model, c) Model uncertainties due to processes that are included in the model, but their

occurrencesinthewatershedareunknowntothemodeler,andd)Modeluncertaintiesduetoprocesses

unknowntothemodelerandnotincludedinthemodeleither!

Inputuncertaintyisasaresultoferrorsininputdatasuchasrainfall,andmoreimportantly,extension

ofpointdatatolargeareasindistributedmodels.Parameteruncertaintyisusuallycausedasaresultof

inherentnonuniquenessofparameters in inversemodeling.Parametersrepresentprocesses.Thefact

thatprocessescancompensateforeachothergivesrisetomanysetsofparametersthatproducethe

sameoutputsignal.Ashortexplanationofuncertaintyissuesisofferedbelow.

5.1)Conceptualmodeluncertainty

a)Modeluncertaintiesduetosimplificationsintheconceptualmodel.Forexample,theassumptionsin

the universal soil loss equation for estimating sediment loss, or the assumptions in calculating flow

velocityinariver.Figures4aand4bshowsomegraphicalillustrations.

Fig.4.(left)Asimplifiedconceptualmodelofhydrologyinawatershedwhererevapisignored.(right)A

naturalprocessnearthesourceofYellowRiverinChinaplayinghavocwithriverloadingbasedonthe

USLE!

b)Modeluncertaintiesduetoprocessesoccurringinthewatershedbutnotincludedinthemodel.For

example,winderosion(Fig.5 left),erosionscausedby landslides(Fig.5right),andthesecondstorm

effecteffectingthemobilizationofparticulatesfromsoilsurface(seeAbbaspouretal.,2007).


10/100

10

Figure5.Naturalprocessesnotincludedinmostwatershedmodelsbutwithalargeimpactonhydrology

and

water

quality

of

a

watershed,

albeit

for

a

short

period

c)Modeluncertaintiesdue toprocesses thatare included in themodel,but theiroccurrences in the

watershed areunknown to themodelerorunaccountable; for example, various formsof reservoirs,

watertransfer,irrigation,orfarmmanagementaffectingwaterquality,etc.(Fig.6,7).

Fig.6.Agriculturalmanagementpracticessuchaswaterwithdrawalandanimalhusbandrycanaffect

waterquantityandquality.These,maynotalwaysbeknowntothemodeller.


11/100

11

Fig.7.Watercontrolandwaterdiversionsmaychangetheflowinwaysthatareunknowntothe

modellerand,hence,cannotbeaccountedforinthemodel.

d)Modeluncertaintiesduetoprocessesunknowntothemodelerandnotincludedinthemodeleither!

These includedumpingofwastematerialandchemicals intherivers,orprocessesthatmay lastfora

numberofyearsanddrasticallychangethehydrologyorwaterqualitysuchaslargescaleconstructions

ofroads,

dams,

bridges,

tunnels,

etc.

Figure

8shows

some

situations

that

could

add

substantial

conceptualmodelerrortoouranalysis.

Fig.8Largeconstructionprojectssuchasroads,dams,tunnels,bridges,etc.canchangeriverflowand

waterqualityforanumberofyears.Thismaynotbeknownoraccountablebythemodellerorthemodel


12/100

12

5.2)InputUncertainty

Inadditiontomodeluncertainty,thereareuncertaintiesduetoerrorsininputvariablessuchasrainfall

andtemperature,aspointmeasurementsareusedindistributedmodels.Itisquitedifficulttoaccount

forinputuncertainty.Someresearchersproposetreatinginputsasrandomvariable,whichallowsfitting

them togetbetter simulations.Asmodeloutputsarevery sensitive to inputdata,especially rainfall,

caremust

be

taken

in

such

approaches.

In

mountainous

regions,

input

uncertainty

could

be

very

large.

5.3)Parameternonuniqueness

Asinglevaluedparameterresults inasinglemodelsignal indirectmodeling.Inan inverseapplication,

anobservedsignal,however,couldbemorelessreproducedwiththousandsofdifferentparametersets.

Thisnonuniquenessisaninherentpropertyofinversemodeling(IM).IM,hasinrecentyearsbecomea

verypopularmethodforcalibration(e.g.,BevenandBinley,1992,2001;Abbaspouretal.,1997,2007;

Duanetal.,2003;Guptaetal.,1998). IM isconcernedwith theproblemofmaking inferencesabout

physical systems from measured output variables of the model (e.g., river discharge, sediment

concentration).This

is

attractive

because

direct

measurement

of

parameters

describing

the

physical

system is time consuming, costly, tedious, and often has limited applicability. Because nearly all

measurementsaresubjecttosomeuncertainty,theinferencesareusuallystatisticalinnature.

Figure9.Exampleofparameternonuniquenessshowingtwosimilardischargesignalsbasedonquite

differentparametervalues

Furthermore,becauseone canonlymeasure a limited numberof (noisy) data and becausephysical

systems are usually modelled by continuum equations, no hydrological inverse problem is really

uniquely solvable. Inotherwords, if there isa singlemodel that fits themeasurements therewillbe

GW_DELAY CH_N2 CN2 REVAPMN Sol_AWC g

3.46 0.0098 50 0.8 0.11 0.010

0.34 0.131 20 2.4 0.2 3 0.011


13/100

13

manyofthem.AnexampleisshowninFigure9wheretwoverydifferentparametersetsproducesignals

similartotheobserveddischarge.Ourgoalininversemodellingisthentocharacterizethesetofmodels,

mainlythroughassigningdistributions(uncertainties)totheparameters,whichfitthedataandsatisfy

ourpresumptionsaswellasotherpriorinformation.

TheSwisscheeseeffect

Thenonuniquenessproblemcanalsobelookedatfromthepointofviewofobjectivefunction.Plotting

the objectivefunction response surface for two by two combinations of parameters could be quite

revealing.Asanexample,seeFigure10wheretheinverseofanobjectivefunctionisplottedagainsttwo

parameters,hence,localminimaareshownaspeaks.Sizeanddistributionofthesepeaksresemblesthe

mysteriousholesinablockofSwissEmmentalercheesewherethesizeofeachholerepresentsthelocal

uncertainty.Ourexperienceshowsthateachcalibrationmethodconvergestoonesuchpeak (see the

papers by Yang et al., 2008, Schuol et al., 2008a, and Faramarzi et al., 2008). Yang et al., (2008)

comparedGeneralized LikelihoodUncertaintyEstimation (GLUE) (BevenandBinley,1992),Parameter

Solution (ParaSol) (Van Griensven and Meixner, 2003a), Sequential Uncertainty Fitting (SUFI2)

(Abbaspouretal.,2004;2007),andMarkovchainMonteCarlo(MCMC)(e.g.,KuczeraandParent,1998;

Marshalletal.,2004;Vrugtetal.,2003;Yangetal.,2007)methodsinanapplicationtoawatershedin

China. They found that these different optimization programs each found a different solution at

different locations in theparameter spaceswithmore less the samedischarge results.Table1hasa

summaryofthecomparison.

To limit thenonuniqueness, theobjective function shouldbemadeascomprehensiveaspossibleby

includingdifferent fluxesand loads (seeAbbaspouretal.,2007).Thedownsideof this is thata lotof

datashouldbemeasured forcalibration.Theuseofremotesensingdata,when itbecomesavailable,

couldbeextremelyuseful.Infactwebelievethatthenextbigjumpinwatershedmodelingwillbemade

asaresultofadvancesinremotesensingdataavailability.

Furthererrorscouldalsoexist intheverymeasurementsweusetocalibrate themodel.Theseerrors

couldbe very large, forexample, in sedimentdataandgrab samples ifused for calibration.Another

uncertainty worth mentioning is that of modeler uncertainty! It has been shown before that the

experienceofmodelerscouldmakeabigdifference inmodelcalibration.Wehopethatpackages like

SWATCUPcanhelpdecreasemodeleruncertaintybyremovingsomeprobablesourcesofmodelingand

calibrationerrors.

Onafinalnote, it ishighlydesirabletoseparatequantitativelytheeffectofdifferentuncertaintieson

model outputs, but this is very difficult to do. The combined effect, however, should always be

quantifiedonmodeloutputs.


14/100

14

Figure10.Amultidimensionalobjectivefunctionismultimodalmeaningthattherearemanyareasof

goodsolutions

with

different

uncertainties

much

like

the

mysterious

holes

in

aslice

of

Swiss

cheese.


15/100

15

Table1.Summarystatisticscomparingdifferentcalibrationuncertaintyprocedures.

Criterion GLUE ParaSol SUFI2 Bayesianinferencewithcont.

autoregr.errormodel

MCMC IS

Goalfunction NashSutcliffe NashSutcliffe NashSutcliffe post.prob.

density

post.prob.

density

a__CN2.mgt

v__ESCO.hru

v__EPCO.hru

r__SOL_K.sol

a__SOL_AWC.sol

v__ALPHA_BF.gw

v__GW_DELAY.gw

r__SLSUBBSN.hru

a__CH_K2.rte

a__OV_N.hru2dry

2

wet2dry

2wet

16.8 (29.6,

9.8)1

0.76 (0.02,0.97)

0.22 (0.04,0.90)

0.16 (0.36,

0.78)

0.11 (0.01,0.15)

0.12 (0.06,0.97)

159.58 (9.7,

289.3)

0.45 (0.56,

0.46)78.19 (6.0,144.8)

0.05 (0.00,0.20)

21.0 (21.9,

20.1)

0.67(0.65,0.69)

0.16(0.13,0.20)

0.37(0.41,

0.34)

0.07(0.08, 0.08)

0.12(0.08, 0.13)

107.7(91.2,115.2)

0.59(0.60,

0.58)

35.70(27.72,37.67)

0.11(0.07,0.10)

26.9(30.0,7.2)

0.82(0.43,1.0)

1(0.34,1.0)

0.1(0.58,0.34)

0.07(0.05,0.15)

0.51(0.23,0.74)

190.07(100.2,300)

0.52(0.60, 0.03)

83.95(69.4,150.0)

0.06(0.00,0.11)

14.2(16.8,

11.6)

0.74(0.63,0.75)

0.94(0.39,0.98)

0.29(0.31,0.78)

0.12(0.1,0.13)

0.14(0.11, 0.15)

25.5(17.8,33.3)

0.55(0.56,

0.15)

78.3(68.0,86.2)

0.12(0.00, 0.19)0.93(0.81,1.10)

2.81(2.4,3.9)

38.13(29.5,53.8)

3.42(2.4,8.0)

19.60

0.62

0.27

0.01

0.05

0.91

33.15

0.58

147.23

0.08

0.87

2.3028.47

0.92

NSforcal(val) 0.80(0.78) 0.82(0.81) 0.80(0.75) 0.77(0.77) 0.64 (0.71)

R2forcal(val) 0.80(0.84) 0.82(0.85) 0.81(0.81) 0.78(0.81) 0.70(0.72)

LogPDFforcal(val) 1989(926) 2049(1043) 2426 (1095) 1521(866) 1650(801)3pfactorforcal

(val)

79%(69%) 18% (20%) 84% (82%) 85%(84%)

4

dfactorforcal

(val)0.65(0.51) 0.08 (0.07) 1.03 (0.82) 1.47(1.19)

Uncertainty

describedby

parameter

uncertainty

Allsourcesof

uncertainty

Parameter

uncertaintyonly

Allsourcesof

uncertainty

Parameter

uncertaintyonly

Parameter

uncertainty

only

Difficultyof

implement.

veryeasy easy easy morecomplicated more

complicated

Numberofruns 10000 7500 1500+1500 5000+20000+

20000

100000

1c(a,b)foreachparametermeans:cisthebestparameterestimate,(a,b)isthe95%parameter

uncertaintyrangeexceptSUFI2(inSUFI2,thisintervaldenotesthefinalparameterdistribution).2

thedry,wet,dry,andwetusedtocalculatetheCalculatethelogarithmoftheposteriorprobabilitydensityfunction(PDF)arefromthebestofMCMC.3pfactormeansthepercentageofobservationscoveredbythe95PPU

4dfactormeansrelativewidthof95%probabilityband(AfterYangetal.,2008)


16/100

16

SUFI2

SequentialUncertaintyFitting

version2

Discharge Calibration

0

100

200

300

400

500

600

700

01.01.91 01.07.91 01.01.92 01.07.92 01.01.93 01.07.93 01.01.94 01.07.94 01.01.95 01.07.95 01.01.96

Date

Dailydischa

rge(m3

s-1)

measured data bracketed by the 95PPU = 91%

d-factor= 1.0


17/100

17

ConceptualbasisoftheSUFI2uncertaintyanalysisroutine

The deterministic approach to calibration is now outdated and unacceptable. Example of a

deterministicoptimization is trial and error.Meaning you keep adjustingparametersuntil you get

somekindofa reasonablematchbetween simulationandobservation.Reporting thisasacalibrated

model,inmyopinioniswrong,andwillnotstandinanycourtoflaw,ifitcomestothat.Here,wewill

not furtherdiscuss thedeterministicapproaches that result ina single setofparameters claiming to

representthebestsimulation.

Instochasticcalibration,werecognizetheerrorsanduncertainties inourmodelingworkand try to

capture,tosomedegree,ourignoranceandlackofunderstandingoftheprocessesinnaturalsystems.

There is an intimate relationship between calibration and uncertainty (Abbaspour, et al., 2015).

Reporting the uncertainty is not a luxury in modeling, it is a necessity. Without the uncertainty,

calibration ismeaninglessandmisleading. Furthermore,anyanalysiswith the calibratedmodelmust

includetheuncertaintyintheresultbypropagatingtheparameteruncertainties.

In SUFI2, uncertainty in parameters, expressed as ranges (uniform distributions), accounts for all

sources of uncertainties such as uncertainty in driving variables (e.g., rainfall), conceptual model,parameters, and measured data. Propagation of the uncertainties in the parameters leads to

uncertainties inthemodeloutputvariables,whichareexpressedasthe95%probabilitydistributions.

Thesearecalculatedatthe2.5%and97.5% levelsofthecumulativedistributionofanoutputvariable

generatedby thepropagationof theparameteruncertaintiesusing Latinhypercube sampling.This is

referred to as the 95% prediction uncertainty, or 95PPU. These 95PPUs are the model outputs in a

stochastic calibration approach. It is important to realize that we do not have a single signal

representing model output, but rather an envelope of good solutions expressed by the 95PPU,

generatedbycertainparameterranges.

InSUFI2,wewant thatourmodel result (95PPU)envelopsmostof theobservations.Observation, is

whatwehavemeasuredinthenaturalsystem.Observationisimportantbecauseitistheculminationof

alltheprocessestakingplaceintheregionofstudy.Theargument,howevernave,isthatifwecapturetheobservationcorrectlywithourmodel,thenwearesomehowcapturingcorrectlyall theprocesses

leadingtothatobservation.Theproblem,ofcourse, isthatoftenacombinationofwrongprocesses in

ourmodelmayalsoproducegoodsimulationresults.Forthisreason,themorevariables(representing

differentprocesses)we include in theobjective function, themore likelyweare toavoid thewrong

processes.

Toquantifythefitbetweensimulationresult,expressedas95PPU,andobservationexpressedasasingle

signal (withsomeerrorassociatedwith it)wecameupwith two statistics:PfactorandRfactor (see

Abbaspouretal.,2004,2007 referencesprovided in the reference listof SWATCUP).Pfactor is the

percentageofobserveddataenvelopedbyourmodelingresult,the95PPU.Rfactor isthethicknessof

the95PPUenvelop.InSUFI2,wetrytogetreasonablevaluesofthesetwofactors.Whilewewouldlike

tocapturemostofourobservations inthe95PPUenvelop,wewouldatthesametime liketohavea

smallenvelop.Nohardnumbersexistforwhatthesetwofactorsshouldbe,similartothefactthatno

hardnumbersexistforR2orNS.The largertheyare,thebettertheyare.ForPfactor,wesuggesteda

valueof>70%fordischarge,whilehavingRfactorofaround1.Forsediment,asmallerPfactoranda

largerRfactorcouldbeacceptable.

SUFI2operatesbyperforming several iterations,usuallyatmost


18/100

18

previousiteration.Naturally,asparameterrangesgetsmaller,the95PPUenvelopgetssmaller,leading

tosmallerPfactorandsmallerRfactor.Aseachiterationzooms intoabetterregionoftheparameter

space,obtainedbythepreviousiteration,itisgoingtofindabetterbestsolution.So,ifyouhaveNSas

yourobjectivefunction,thenyouwillgetabetterNS insubsequent iterations,butthePfactorandR

factorwilldecreasebecauseofnarrowerparameter ranges.But the idea isnot to find that socalled

bestsimulation.Because,1)therearealwaysbettersimulations,and2)thedifferencebetweenthe

best simulation and the next best simulation and the next next best simulation is usually

statisticallyinsignificant(e.g.,NS=0.83vsNS=0.81areprobablynotsignificantlydifferent),meaningthat

theycouldbothbeidentifiedasthebestsimulations.Butwhilethedifferencesareinsignificantinterms

oftheobjectivefunctionvalue,theyareverysignificant intermsofparametersvalues.Therefore,the

nextbestsolutioncannotbeignored.

TheconceptbehindtheuncertaintyanalysisoftheSUFI2algorithmisdepictedgraphicallyintheFigure

below.ThisFigure illustratesthatasingleparametervalue(shownbyapoint) leadstoasinglemodel

response (Fig.a),whilepropagationof theuncertainty inaparameter (shownbya line) leads to the

95PPU illustratedby the shaded region in Figureb.Asparameteruncertainty increases, the output

uncertaintyalso increases (notnecessarily linearly) (Fig. c).Hence,SUFI2 startsbyassuminga large

parameteruncertainty (withinaphysicallymeaningful range),sothatthemeasureddata initially fallswithin the95PPU, thendecreases thisuncertainty in stepswhilemonitoring thePfactorand theR

factor. In each step, previous parameter ranges are updated by calculating the sensitivity matrix

(equivalenttoJacobian),andequivalentofaHessianmatrix,followedbythecalculationofcovariance

matrix, 95% confidence intervals of the parameters, and correlation matrix. Parameters are then

updated in such a way that the new ranges are always smaller than the previous ranges, and are

centeredaroundthebestsimulation(formoredetailseeAbbaspouretal.,2004,2007).

A conceptual illustration of the relationship between parameter uncertainty and prediction

uncertainty

Thegoodnessof fitand thedegree towhich the calibratedmodelaccounts for theuncertaintiesare

assessedbytheabovetwomeasures.Theoretically,thevalueforPfactorrangesbetween0and100%,

whilethatofRfactorrangesbetween0andinfinity.APfactorof1andRfactorofzeroisasimulation

thatexactlycorrespondstomeasureddata.Thedegreetowhichweareawayfromthesenumberscan

beusedtojudgethestrengthofourcalibration.A


19/100

19

largerPfactorcanbeachievedattheexpenseofalargerRfactor.

Hence,oftenabalancemustbereachedbetweenthetwo.WhenacceptablevaluesofRfactorandP

factor are reached, then the parameter uncertainties are the desired parameter ranges. Further

goodness of fit can be quantified by the R2 and/or NashSutcliff (NS) coefficient between the

observations and the final best simulation. It should be noted that we do not seek the best

simulationasinsuchastochasticprocedurethebestsolutionisactuallythefinalparameterranges.

Ifinitiallywesetparameterrangesequaltothemaximumphysicallymeaningfulrangesandstillcannot

finda95PPUthatbracketsanyormostofthedata,forexample,ifthesituationinFiguredoccurs,then

theproblemisnotoneofparametercalibrationandtheconceptualmodelmustbereexamined.

SUFI2asanoptimizationalgorithm

ForadescriptionofSUFI2seeAbbaspouretal.,(2004,2007).ReferencesareprovidedintheReference

directoryofSWATCUP.

ForacalibrationprotocolseeAbbaspouretal.,(2015).

http://www.sciencedirect.com/science/article/pii/S0022169415001985


20/100

20

SWATCUP

Automatedmodelcalibrationrequiresthattheuncertainmodelparametersaresystematicallychanged,

themodel is run,and the requiredoutputs (corresponding tomeasureddata)areextracted from the

modeloutputfiles.Themainfunctionofaninterfaceistoprovidealinkbetweentheinput/outputofa

calibrationprogramandthemodel.Thesimplestwayofhandlingthefileexchange isthroughtextfile

formats.

SWATCUP is an interface that was developed for SWAT. Using this generic interface, any

calibration/uncertaintyorsensitivityprogramcaneasilybe linkedtoSWAT.Aschematicofthe linkage

betweenSWATandfiveoptimizationprogramsisillustratedintheFigurebelow.


21/100

21

StepbystepCreatingofSWATSUFI2InputFiles

SWAT_Edit.exeBACKUP

SUFI2_LH_sample.exe

SUFI2_new_pars.exepar_val.txt

swat.exe

ModifiedSWAT inputs

SWAT

outputs

par_inf.txt

SUFI2_extract_rch.exe

SUFI2_goal_fn.exe

SUFI2_95ppu.exe

*.out

Is

calibration

criteriasatisfied?

observed.txt

goal.txt

yes

stop

no

SUFI2_extract_rch.def

SUFI2_swEdit.def


22/100

22

1.BeforeSWATCUP

BecomefamiliarwithSWATparameters.TheyareallexplainedintheSWATI/Omanual.Also,readthe

theoryandapplicationofSWATSUFI2atthebeginningofthismanualandinthefollowingpapers:

Thurwatershedpaper(Abbaspouretal.,2007)

Landfill

application

in

Switzerland

(Abbaspour

et

al.,

2004)

ThecontinentalapplicationinAfrica(Schuoletal,2008a,b)andEurope(Abbaspouretal,2014)

ThecountrybasedapplicationinIran(Faramarzietal.,2008)

Thecomparisonofdifferentoptimizationprograms(Yangetal.,2008)

Theparallelprocessingpaper(Rouholahnejadetal.,2013)

TheBlackSeaapplicationpaper(Rouholahnejadetal.,2014)

TheapplicationtoentireEurope(Abbaspouretal.,2015)

(http://www.sciencedirect.com/science/article/pii/S0022169415001985)

AndothersprovidedintheC:\SWAT\SWATCUP\ExternalData\References

2.StartSWATCUP

InstalltheSWATCUPinC:\SWAT\SWATCUP,thesamedirectoryastheSWATandstarttheprogramby

pressingtheSWATCUPicononthedesktop:

3.OpenaProject

Toopenaneworoldproject:Pressthe symbolatthetopleftcorner

andchooseaNeworOpenanoldproject


23/100

23

Foranewproject locatea SWAT TxtInOutdirectory.Any filewith TxtInOut in thename string

wouldbeacceptable

ChooseSWATandprocessorversions

Selectaprogramfromthelistprovided(SUFI2,GLUE,ParaSol,MCMC,PSO).Adetailedexplanationof

theSUFI2procedureisofferedhere,butallprogramsfollowthesameformat.

GiveanametotheprojectandalocationwhereSWATCUPprojectcanbesaved.


24/100

24

NotethedefaultadditiontothenameprovidedinthewindowtotherightofProjectNamewindow.

For a SUFI2 project, the full SWATCUP directory name in the example below would be

test_1.Sufi2.SwatCup,whichwillresideinc:\ArcSWATprojects\Black_Seadirectory.

AtthispointTheprogramcreatesthedesiredprojectdirectoryandcopiesthereallTxtInOutfilesfrom

theindicatedlocationintotheSWATCUPprojectdirectory.ItalsocreatesadirectorycalledBackupin

thesameSWATCUPprojectdirectoryandcopiesallSWATTxtInOutfilethere.Theparameters inthe

filesintheBackupdirectoryserveasthedefaultparametersanddonotchangedduringthecalibration

process.TheBackupdirectory isalwaysneeded as itsoriginal formbecause,relativechangesthat

havebeenmadetotheparametersduringcalibration,weremaderelativetotheparametervalues in

theBackupdirectory.Therefore,itisimportantthattheBackupdirectoryisneverchanged.

4.SWATOutputFiles

Youcancalibratethemodelbasedonthevariablesfromoutput.rch,output.hru,output.sub,output.res,

output.mgt,andnowalsohourlygeneratedfiles.However,theinterfaceonlyshows.rch,.hru,and.sub

files.Asmostoftenwehaveeitherdischarge,nutrientsdata,orsedimentdata,onlytheseareshownin

the interface.Justclickandactivatewhich fileyouwanttouse (i.e.,yourobservedvariablesreside in

whichSWATfile(s)).


25/100

25

5.CalibrationInputs

UndertheCalibrationInputseditthefollowingfiles:

Par_inf.txt

This file resides in the project directory in SUFI2.IN directory. It contains input parameters to be

optimized.Anexample isprovided,whichneeds tobe edited by theuser. The examples shows theformatofthefile.Editthistoyourneeds.Atextviewandaformviewisprovided.Theformviewhelps

with finding the correct parameter syntax or expression. All SWAT parameters (up to the time of

compilationofthe lastswatcupversion)canbefoundhere.TheformviewfollowsstandardWindows

protocol and the users are advised to familiarize themselves with this module by testing different

featuresof it.Whatyoudo in theformview,appears in the textviewandviceversa.Usersarealso

encouraged to trydifferent things in theformviewand lookat the them in the textview tobecome

familiarwiththisimportantanduniquefeatureofSWATCUP.

Thisfilecontainsthenumberofparameterstobeoptimizedandthenumberofsimulationstomakein

thecurrent iteration.SUFI2 is iterative,each iteration containsanumberof simulations.Around500

simulations are recommended in each iteration. But if a swat project takes too long to run, fewer

numberofsimulations (200300) ineach iterationcouldbeacceptable.Parametersaresampledusing

Latinhypercubeschemeexplainedlaterinthemanual.Usually,notmorethan4iterationsaresufficient

to reach an acceptable solution. A parallel processing module is also available to speed up the

calibrationprocess.

To learn more about parameterization and parameter qualifiers r__, v__, and a__ please see the

sectiononparameterizationbelow.


26/100

26

What parameters to use depend on the objective function. Initially, in every case, flow should be

calibratedand thenwaterqualityvariablesaddedoneat time (seeAbbaspouretal.,2007,2015 for

parameterchoicesandcalibrationprotocol).

SUFI2_swEdit.def

This file contains the beginning and the ending simulation years. Please note that the beginning

simulationdoesnot include thewarmupperiod.SWATsimulates thewarmupperiodbutdoesnot

printanyresults,therefore,theseyearsarenotconsideredinSWATCUP.Youcancheckoutput.rchfile

ofSWATtoseewhenthebeginningandtheendingsimulationtimesis.


27/100

27

File.cio

ThisisaSWATfile.Itisputhereforconvenience.Whatyouneedfromthisfilearethesimulationyears

andthenumberofwarmupyears(NYSKIP)tocorrectlyprovideSWATCUPwithbeginningandendyear

ofsimulation.Itisrecommendedthatyouhave23yearsofwarmupperiod.

.

.

Missspecifyingthecorrectdatesisthecauseofbiggestusererror!Pleasenotethefollowing:

- Intheaboveexample,beginningyearofSWATsimulationis1987,endyearis2001

- There are 3 years of warm up period as indicated by NYSKIP. Therefore, SWAT output files

containdata from1990 to2001.Thesedatesareof interest toSWATCUP.So, inSWATCUP

beginningyearis1990andendyearis2001.

- AlsonotethatSWATCUPrequiresthe IDAFtobeatthebeginningof theyear (always1)and

IDALtogototheendoftheyear(365or366forleapyears).ThereforeSWATsimulationshould

alwaysbe from thebeginning to theendof theyear.Soyourclimatedatamustbe from thebeginningtotheendoftheyear.

Absolute_SWAT_Values.txt

Allparameterstobefittedshouldbeinthisfileplustheirabsoluteminandmaxranges.Currentlymost,

butperhapsnotallparametersareincludedinthisfile.Simplyaddtoittheparametersthatdontexist.

TheSWAT_Edit.exeprogram, that replacesparameters in theSWAT files,doesnotallowparameters

beyondthisrangeintoSWATfiles.

etc.


28/100

28

6.Observation

UnderObservationarethreefilesthatcontaintheobservedvariables.Observedvariablescorrespond

to thevariables inoutput.rch,output.hru,andoutput.sub,output.res,andoutput.mgt files,although

thelattertwodonotappearinSWATCUP.

Initially, all options are deactivated. To activate you need to choose which SWAT file contains the

simulateddata(step4above).Variablesfromdifferentfilescanbeincludedtoformamulticomponent

objectivefunction.Simplyonlyeditthefile(s)thatappliestoyourprojectanddonotworryaboutthe

onesthatdont.Theformatshouldbeexactlyasshown intheexamplesprovided intheprogram.The

three files Observed_rch.txt, Observed_hru.txt, and Observed_bsn.txt can be edited here, but

observed_res.txtforreservoirdataandobserved_mgt.txtforcropyieldarealsoavailablethatcouldbe

editeddirectlyinthe.\SUFI2.INdirectoryintheSWATCUPprojectdirectory.

Missingvalues

Theformatoftheobservationfilesareasshown intheexamplesprovided.Thesefilescouldeasilybe

madeinExcelandpastedhere.IntheobservedfilesYoumayhavemissingdatathatcanbeaccounted

forasshownintheexamplefilesandexplainedbelow.

Thefirstcolumnhassequentialnumbersfromthebeginningofsimulationtimeperiod.Intheexample

below,thefirst10monthsaremissingsothefirstcolumnbeginsfrom11.Also,months18,19,and20

aremissing.


29/100

29

Thesecond

column

has

an

arbitrary

format

but

it

should

be

one

connected

string.

Here,

it

is

showing

thevariablename,month,andyear.Thirdcolumnisthevariablevalue.

Ifbaseflow isseparated,anddynamicbaseflow isused,thenaforthcolumn indicatingthebaseflow

should also be added. The example of an observation file with base flow is given in observed+.txt

in.\SUFI2.INdirectory.

Allotherobservation_*.txtfileshavethesameformat.ThisfiletellstheextractprogramsofSWATCUP

whattoextractfromtheSWAToutputfiles.

Theobserved_rch.txt filescancontainmanyvariablessuchasdischarge,sediment,nitrate,etc.which

appearintheSWAToutputfileoutput.hru.Simplyusethesameformatforallvariablesasshowninthe

examples.Also,

for

the

variables

name,

be

consistent

in

all

SWAT

CUP

files.

7.Extraction

UnderExtractionyouwillfindtwotypesoffiles.txtand.defcorrespondingagaintoSWAToutputfiles

output.rch,output.hru, andoutput.sub. If youhaveobservations corresponding to variables in these

files,thenyouneedtoextractthecorrespondingsimulatedvaluesfromthesefilesonly.


30/100

30

.txtfilessimplycontainthenamesofthefilesthatextractedvaluesshouldbewrittento,and.deffiles

define which variables need to be extracted from which subbasins. These files are relatively self

explanatory.Hereagainonlyedittheneededfiles.

Intheexampleprovided,wehave4measuredvariables,3dischargesfromsubbasins1,3,7,and1nitrate

from subbasin 7. The extract files of SWATCUP extract the corresponding simulated data from

output.rchfileandwritethemtothefilesindicatedhereforeverysimulation.

BelowisanexampleofSUFI2_extrcat_rch.def


31/100

31

Thefileisselfexplanatory.Weareextracting2variables:dischargeandnitrate.Theyare incolumns7

and18 in theoutput.rch file.Thereareatotalof20subbasins in theproject.Fordischarge,wehave

flow measurements from subbasins 1, 3, and 7. For nitrate we have measurement from subbasin

number7only.Thebeginningdatedoesnotincludewarmupperiod.

8.Objectivefunction

Next,ObjectiveFunctionisdefined.InthissteptwofilesObserved.txtandVar_file_name.txtshouldbe

edited. The Observed.txt file contains all the information in observed_rch.txt, observed_hru.txt,

observed_sub.txtfiles,plussomeextrainformationforthecalculationoftheobjectivefunction.

Var_file_name.txt contains the names of all the variables that should be included in the in the

objectivefunction.Thesenamesaresimilartothenamesinthevar_file_*.txtintheExtractionsection.

Observed.txtfilealsoisquiteselfexplanatory.


32/100

32

Inthisexamplewehave4variables.Wehaveanoptiontochoosefrom10differentobjectivefunctions.

ReadthesectiononObjectiveFunctionbelowformoreexplanationofthefunctions.

Thethird line isoptional torun,butthereshouldbeanumberhere. Ifyouexpressathresholdvalue

here,thenallsimulationswithobjectivefunctionvaluebetterthanthethresholdarecollectedandthe

95PPUcalculatedbasedon thosesimulations.Here, theSUFI2becomessimilar toGLUE.Thepfactor

andthe

rfactor

for

agiven

threshold

may

be

different

from

the

solution

where

threshold

is

not

considered. Pleasenote that the threshold valuemust correspond to the typeofobjective function

beingused.

Baseflowseparation

Two options are provided to considerbaseflow separation: static and dynamic. In the static case, a

constantthresholdvalue forbaseflow isused.Thisvalues,dividesthedischargesignal intotwoparts.

Valuessmallerthanthethresholdandvalueslargerthanthethresholdaretreatedastwovariablesand

carrytwodifferentweights.This istoensurethat, forexample,base flowhasthesamevaluesasthe

peakflows.

Without

this

division,

ifyou

choose

option

2for

objective

function,

i.e.,

mean

square

error

(see objective function section below), then the small flows will not have much effect on the

optimization.Hence, peak flowwill dominate the processes.With the static threshold option, small

flows can be given a larger weight in the objective function so that they have almost the same

contribution to theobjective function as thepeak flows. Thebase flow separation ismost effective

whenoption2ischosenforobjectivefunction.Separatingthebaseflowdoesnotbecomeverycriticalif

R2orbR2isusedforobjectivefunction.

Tonotusethisoption,simplysetconstant flowseparationthresholdtoanegativevalue (say 1 fora

variablethatisalwayspositive)andtheweightsforsmallerandlargerflowsto1.

Threshold=35


33/100

33

In the dynamic case, a flow separation program should be used to calculate the base flow. Both

observedflowandbaseflowmustthenappearintheobserved.txtfileastwoseparatecolumns,column

3andcolumn4,respectively,asshownintheobserved+.txtexamplefilein.\SUFI2.INdirectory.

Percentageof

measured

error

Thisvaluereferstothemeasurementerror.Adefaultvalueof10%isspecified,buttheuserscanchange

thisbasedontheirknowledge.Thisvalueisreasonableforflow,butshouldbehigherforothervariables

suchassedimentandnitrate,etc.

9.No_Observations

TheNo_Observation section isdesigned for the extraction and visualization of uncertainties for the

variables for which we have no observation, but would like to see how they are simulated such as

variousnutrientloads,orsoilmoisture,ET,etc.The.txtfilesareinactive.

The.deffileshavemoreorlessthesameformatastheExtractionsection.

Extract_rch_No_Obs.def Extract_hru_No_Obs.def Extract_sub_No_Obs.def

This files isalso selfexplanatory.Thenumberof variables to get, columnnumbers (sequential),and

representativevariablenamesarespecified(R isusedheretoindicatethesearefromoutput.rchSWAT


34/100

34

file) in rows 35. These names are used to build files where simulated values are collected for all

simulations.Thentotalnumberofsubbasinsintheprojectisspecified.

Foreachvariable,weidentifythenumberofsubbasinstoget,andthesubbasinnumber(s).Ifwewant

to getall subbasin values, forexample forplottingmaps, then simply indicateALL.This followedby

beginningandendyearofsimulation.Again,beginningyearofsimulationshouldnot includewarmup

period.

95ppu_No_Obs.def

Finally,forNO_Observationoptionweneedtoeditthe95ppu_No_Obs.deffile.Thisisafileusedforthe

calculationofthe95ppufortheextractedvariableswithnoobservation.

95ppu_No_Obs.def

Thisfileagain isquiteselfexplanatory.Thenumberofvariablesforwhich95PPU istobecalculated is

giveninthesecondrow.Variablesnamesarethenprovided.Thesenamesshouldbeexactlythesame

astheonesgiveninthe.deffile(s).Finally,thenumberofsimulationtimestepsaregiven.For12years

ofmonthlysimulationthiswouldbe144.


35/100

35

10.Executables

ThesectiononExecutableFilesplaystheroleofengineinSWATCUP.Thefourbatchfilesindicatewhat

shouldorshouldnotberun.

SUFI2_pre.bat

This batch file runs the preprocessing procedures. It include running the Latin hypercube sampling

program.Thisbatchfileusuallydoesnotneedtobeedited.

NotethatmanyfilesattheendhaveFormViewandTextView.IntheTextViewyouhavethetextfile,

whichappears inyouSWAZCUPprojectdirectory.Youcaneasilyeditthistextfileasneededwiththe

sameformatasshown.


36/100

36

SUFI2_run.bat

ThisprogramexecutesSUFI2_execute.exe program,which runs theSWAT_Edit.exe,extractionbatch

files,aswellasSWAT.exe.

SUFI2_post.bat

runs thepostprocessingbatch file,which runs theprograms forobjective functioncalculation,new

parametercalculation,95ppucalculation,95ppuforbehavioralsimulations,and95ppuforthevariables

withnoobservations(optional).IntheTextFormonecanuncheckaprogramifitisnotneededtorun.


37/100

37

SUFI2_Extract.bat

Thisbatchfilecontainsthenamesofalltheextractprogramswithorwithoutobservations.Currently8

programsaresupported.Thisfilemustbeeditedandtheprogramsthatarenotdesiredtorunshouldbe

remarked oruncheckedasshownbelow:

11.BeforeCalibration

Atthispointtheinputfilesarecompleteandtheprojectisreadytobecalibrated.Butbeforestartingan

iteration,youneedtomakesurethatthemodelyouhavebuiltinfeasibleinitially.So,youshouldmake

afirstrunwiththeinitialmodelstructureandinitialmodelparameters.

TochecktheinitialmodelwithSWATCUPdothefollowing:

1InPar_inf,putthenumberofsimulationsandthenumberofparametersto1

2InSUFI2_swEditputthebeginningandtheendingsimulationalsoto1

3Setupadummyparametersuchas

r__SFTMP.bsn 0 0 (thisdoesnotchangeanything)

4Thenexecuteinorder:Pre,Run,andPostprocessing.

Nowlookatthe95PPUresultofyourdefaultorinitialmodelrun.Ifsimulationsandobservationsaretoodifferent,thenyouneedtotakeacloserlookatyourswatmodel,includingrainfall,etc.Else,lookatthe

calibrationprotocolin(http://www.sciencedirect.com/science/article/pii/S0022169415001985)to

adjusttheparameterinawayastoachievethebestsimulationresultateachobservedoutlet.

Forthisinitialsimulation, youshouldalsolookattheoutput.stdfiletomakesuretheoverallwatershed

flowcomponentsarecorrectornot.


38/100

38

12.Calibration

Next,aftereditingallthe inputfiles,performSaveAllandCloseAlltasks.Theruntheprograms in

theCalibrationwindowintheorderthattheyappear.Inthissectionthreestepsareperformed:

i) Sufi2_pre.bat ThiscommandrunstheSufi2_pre.batfile.Thisfilemustberunbeforethestart

ofeverynewiteration.

ii) SUFI2_run.bat Thiscommandexecutestherunbatchfile.

iii) SUFI2_post.bat

Afterall

simulations

are

finished,

this

command

executes

the

post

processing

batchfiledescribedabove.


39/100

39

13.CalibrationOutputs

95ppuplot

Thiscommandshowsthe95ppuofallvariables.Alsoshownareobservationsandbestsimulationofthe

currentiteration.PleaseNotethatthebestsimulationisonlyshownforhistoricreason.Thesolutionto

thecalibrationatthisstageisthe95PPUgraphandtheparameterrangesthatwereusedtogenerateit.

Notethefeatureswitharrowwhereyoucanchangethevariablesaswellaszoomingthehydrograph.


40/100

40

Also,pleasenotetheoptionsgivenbyChartLayoutandPrintPreview

95ppuNo_Observedplot

Thiscommandshowsthe95ppuofallvariableswithnoobservations.Hereyouonlyseetheuncertainty

insimulationofSoilMoisture,avariableforwhichwehavenoobservation.

DottyPlots

Thiscommandshowsthedottyplotsofallparameters.Theseareplotsofparametervaluesorrelative

changesversusobjectivefunction.Themainpurposeofthesegraphsaretoshowthedistributionofthe

samplingpointsaswellastogiveanideaofparametersensitivity.Inthefollowingfigureyouseeanice

trend for CN2 as it increases. Objective function is NashSutcliffe (NS). Clearly CN2 is a sensitive

parameteranditsbestfittingvaluesarelessthan0.1inrelativechange(r__).ButALPHA_BFdoesnot

appeartobesensitiveasthevalueofobjectivefunctiondoesntreallychange.GW_DELAYalso isnot

verysensitive,but itsvalueshouldprobablynotbeabove300,GWQMNalsonotverysensitive,butit

shouldprobablybesomewhereabove0.6.Moreaboutsensitivitylater.


41/100

41

Best_Par.txt

This fileshowsthebestparametervaluesaswellas theirranges.Theseare theparameters,which

gavethebestobjectivefunctionvalueinthecurrent iteration.Again,I liketoemphasizethatthebest

parameterreallydoesnotmeanverymuchasthenextobjectivefunctionvaluemaynotbestatistically

nottoodifferentfromthebestone.Theparameterrangesarethesolutionforthisiteration.


42/100

42

Best_Sim.txt

This file shows the best simulated values for all the variables used in the objective function. Both

observedandsimulatedvaluesaregivensothattheycouldeasilybeplottedwithothersoftwaresas

desired.

Goal.txt

Thisfileshowsthevalueofallparametersetsforthesimulationsperformedaswellasthevalueofthegoalfunctioninthelastcolumn.Thisfileisusedlatetocalculatethesocalledglobalsensitivity.

New_Pars.txt

This file shows the suggested valuesof thenewparameters tobeused in thenext iteration.These

values can be copied and pasted in the Par_inf.txt file for the next iteration, or alternatively, the

Import New Parameters could be used to copy new parameters into par_inf.txt file. The new

parametersshouldbecheckedforpossibleunreasonablevalues(e.g.,negativehydraulicconductivity,


43/100

43

etc.). These suggested parameter ranges should bemanually corrected and if desired directed to a

certainrangebytheuserincaseofavailableinformationorknowledgeofthesystem.

Summary_Stat

Thisfilehasthestatisticsofcomparingobserveddatawiththesimulationbandthroughpfactorandr

factorandthebestsimulationofthecurrentiterationbyusingR2,NS,bR2,MSE,SSQR,PBIAS,KGE,RSR,

andVOL_FR.Themeanandstandarddeviationoftheobservedandsimulatedvariablesarealsogiven

attheend.Fordefinitionofthesefunctionsseethesectiononobjectivefunctions.Alsoshown isthe

goalfunctiontype,bestsimulationnumberofthecurrentiteration,andthebestvalueoftheobjective

functionforthecurrentrunonthetop.

Ifbehavioral

solutions

exist,

then

the

p

factor

and

rfactor

for

these

solutions

are

also

calculated.

As

showninthefollowingTabletheeffectofusingbehavioralsolutionsistoobtainsmallerpfactorandr

factor,orasmallerpredictionuncertainty.


44/100

44

14.Sensitivityanalysis

Thismoduleoftheprogramperformssensitivityanalysis.Twotypesofsensitivityanalysisareallowed.

GlobalSensitivityandOneatatimesensitivityanalysis.

Globalsensitivityanalysiscanbeperformedafteraniteration.Oneatatimesensitivityisperformedfor

oneparameteratatimeonly.Theprocedureisexplainedinthenextsection.

15.Maps

TheMapsmoduleenablesvisualizationoftheoutlets.TheBingmap isusedtoprojectthe locationof

outlets,rivers,climatestations,andsubbasinsontheactualmapoftheworld.


45/100

45

Whenyou invoke theOutletMap, theBingmap isactivatedand theprogramasks for theArcSWAT

projectShapesfolder.

Upon locatingtheShapefilefolder intheArcSWATproject, ...\Watershed\Shapes,severaloptionsare

providedforvisualizationoftheoutletlocation:


46/100

46

Thesubbasinboundaries,andthelocationofraingauges.Thismodule isextremelyuseful inanalyzing

weather theoutlets are in their correct locationornot,whether the rivers areproperlydigitizedby

SWATornot,iftheoutletisundertheinfluenceofsnowandglacier,intensiveagriculture,etc.

SomeexamplesfromtheEuropeanprojectshow:

(from http://www.sciencedirect.com/science/article/pii/S0022169415001985)

a) Wrong positioning of an outlet on Viar river instead of on the main Rio Guadalquivir in Spain.

The red-green symbol indicates location of the outlet.


47/100

47

b) The position of an outlet downstream of a dam on Inn River near Munich, Germany. SWAT

cannot calibrate the flow in this outlet unless the reservoir is modeled.

c) a complex river geometry on Pechora River near Golubovo in Russia. SWAT cannot beexpected to simulate the flow in this outlet with high accuracy.


48/100

48

d) The flow in the outlet below is governed by glacier melt near Martigny in Switzerland. These

features could explain some of the discrepancies between simulation and observed results in

SWAT calibration.

16.Utilityprograms(C:\SWAT\SWATCUP5.1.6.2\ExternalData\utilityprograms)

Thismodule

currently

has

two

program

in

it:

Make_ELEV_BAND,

not

shown

in

the

interface,

and

Upstreamsubbasins,whichisshownintheinterface.

Make_ELEV_BAND,thisprogramcancalculate theelevationband foraSWATprojectanduseSWAT

CUPtoputtheinformationinSWAT*.subfiles.Thereisanexplanatoryfilecalledelev_band.doc,which

explainshowtodothis.


49/100

49

Upstreamsubbasins,thisprogramcandeterminetheupstreamsubbasins.This isauseful information

for parameterization. A READ_ME.txt file explains how to use this program. The file

upstream_sorted.out,which is not shown in the interface shows, in a sortedmanner, all the above

subbasinsofanyobserved subbasin.Theupstream.outcontainsavisualizationoptionthatshowswhich

outletsareconnectedtoeachother.

Theabovetextandvisualaidshowthatsubbasin(oroutlet)number1hasnoupstreamsubbasin,while

outlets3,7,18(whichareheremeasuredoutlets)haveupstreamsubbasins.Ifyourightclickonthenode

connectingoutlets3and18above,thenyouwillfindallthenonintersectingsubbasinsbetween3and

18.Thatmeansallthesubbasinsbetween3and18.Usingthisinformation,onecancalibrateforoutlet

18byparameterizingsubbasin19and20first,whichcontributetooutlet18.Thenkeeptherangesfixed

fortheparametersofsubbasins18and19,andparameterizesubbasins inbetween18and3(seethe

picturebelow,thearrowshowsalistoftheseparametersinthetextform).Usingthisprocedure,outlet

3canbecalibrated.


50/100

50

17.IterationsHistory

Alliterations

can

be

saved

in

the

iteration

history.

This

allows

studying

the

progress

to

convergence.

Afteracomplete iteration, review the suggestednewparameters in thenew_pars.txt,copy them to

par_inf.txtandeditthemasexplainedbefore,andmakeanewiteration.Therearenohardrulesasto

whenacalibrationprocesscanbeterminated.Buttheprocesscanstopwhensatisfactorystatisticsare

achievedandtherearenofurtherimprovementsintheobjectivefunctionvalue.


51/100

51

ParameterizationinSWATCUP

Thefollowingschemecanbeusedtoparameterize,orregionalizeparametersofawatershed.InSWAT,

theHRU is the smallestunitof spatialdisaggregation.Asawatershed isdivided intoHRUsbasedon

elevation,soil,andlanduse,adistributedspatialparametersuchashydraulicconductivity,bulkdensity,

orCN2canpotentiallybedefinedforeachHRU.Ananalystis,hence,confrontedwiththedifficulttask

of collecting or estimating a large number of input parameters, which are usually not available. An

alternativeapproach fortheestimationofdistributedparameters isto lumpthembasedonsoiltype,

landuse type, location, slope,or a combination of these. They can then be calibratedusing a single

globalmodificationtermthatcanscaletheinitialestimatesbyamultiplicative,oranadditiveterm.This

leadstothefollowingproposedparameteridentifiersexplainedbelow.

x__.__________

Where

x__ =Identifiercodetoindicatethetypeofchangetobeappliedtotheparameter:

v__meanstheexistingparametervalueistobereplacedbyagivenvalue,

a__meansagivenvalueisaddedtotheexistingparametervalue,and

r__meansanexistingparametervalueismultipliedby(1+agivenvalue).

Note:thattherearealwaystwounderscores__aftertheidentifier

= SWAT parameter name as it appears in the SWAT I/O manual or in the

Absolute_SWAT_Values.txtfile.

= SWAT file extension code for the file containing the parameter(e.g.,.sol,.hru,.rte,etc.)

=(optional)soilhydrologicalgroup(A,B,CorD)

=(optional)soiltextureasitappearsintheheaderlineofSWATinputfiles

=(optional)nameofthelandusecategoryasitappearsintheheaderlineofSWAT

inputfiles

= (optional) subbasinnumber(s)as itappears in theheader lineof SWAT input

files

= (optional)slopeasitappearsintheheaderlineofSWATinputfiles

Anycombinationoftheabovefactorscanbeusedtodescribeaparameteridentifier.Iftheparameters

are used globally, the identifiers , , , , and can be

omitted.

Note: thetwounderscoresaftereverypreviousspecificationsmustbeused,i.e.,tospecifyonlythe

subbasinwemustwriteeightunderscoresafter.crp v__USLE_C.crp________2


52/100

52

The presented encoding scheme allows the user to make distributed parameters dependent on

importantinfluentialfactorssuchas:hydrologicalgroup,soiltexture,landuse,elevation,andslope.The

parameters canbe assigned and calibrated regionally,orbe changed globally. This gives the analyst

largerfreedominselectingthecomplexityofadistributedparameterscheme.Byusingthisflexibility,a

calibrationprocesscanbestartedwithasmallnumberofparametersthatonlymodifyagivenspatial

pattern, with more complexity and regional resolution added in a stepwise learning process. Some

examplesoftheparameterizationschemeisasfollows:

SpecificationofSoilParameters

Parameteridentifiers Description

r__SOL_K(1).sol KofLayer1ofallHRUs

r__SOL_K(1,2,46).sol KofLayer1,2,4,5,and6ofallHRUs

r__SOL_K().sol KofAlllayersandallHRUs

r__SOL_K(1).sol__D Koflayer1ofHRUswithhydrologicgroupD

r__SOL_K(1).sol____FSL Koflayer1ofHRUswithsoiltextureFSL

r__SOL_K(1).sol____FSL__PAST Koflayer1ofHRUswithsoiltextureFSLand

landusePAST

r__SOL_K(1).sol____FSL__PAST__13 Koflayer1ofsubbasin1,2,and3withHRUs

containingsoiltextureFSLandlandusePAST

SpecificationofManagementParameters


v__HEAT_UNITS{rotationno,operationno}.mgt Managementparametersthatare

subjecttooperation/rotationmust

havebothspecified

v__CNOP{[],1}.mgt Thischangesanoperation's

parametersinallrotations

v__CNOP{2,1,plant_id=33}.mgt ChangesCNOPforrotation2,

operation1,andplant33only

v__CNOP{[],1,plant_id=33}.mgt Similartoabove,butforallrotations

v__CNOP{[],1,plant_id=33}.000010001.mgt Withthiscommandyoucanonly

modifyonefile


53/100

53

r__FRT_KG{9,1}.mgt Inthesethreeexamples,rotation9,

operation1,andtherestarefilters

where,meansANDr__FRT_KG{9,1,PLANT_ID=12}.mgt

r__FRT_KG{9,1,PLANT_ID=12,HUSC=0.15}.mgt

SpecificationofCropParameters


v__T_OPT{30}.CROP.DAT ParameterT_OPTforcropnumber30inthe

crop.datfile

v__PLTNFR(1){3}.CROP.DAT Nitrogenuptakeparameter#1forcrop

number3incrop.datfile

SpecificationofPesticideParameters


v__WSOL{1}.pest.dat ThischangesparameterWSOLforpesticide

number1inpest.datfile

SpecificationofPrecipitationandTemperatureParameters


v__precipitation(1){1977300}.pcp1.pcp (1)meanscolumnnumber1inthepcpfile

{1977300}specifiesyearandday

v__precipitation(13){1977300}.pcp1.pcp (13)meanscolumn1,2,and3


v__precipitation(){1977300,1977301}.pcp ()meansallcolumns(allstations)

{1977300,1977301}means1977days300and

301

v__precipitation(){1977001

1977361,19780011978365,1979003}.pcp

()meansallcolumns

fromday1today361of1977,andfromday1

today365of1978,andday3of1979

v__MAXTEMP(1){1977001}.tmp1.tmp (1)meanscolumn1inthetmp1.tmpfile


54/100

54


v__MAXTEMP(2){1977002

1977007}.tmp1.tmp

(2)meanscolumn2inthetmp1.tmpfile

fromday2today7in1977

v__MINTEMP(){19770021977007}.tmp1.tmp ()meansallcolumnsintmp1.tmpfile

SpecificationofslopeParameters


v__SOL_K(1).sol______________010 Koflayer1forHRUswithslope010

Pleasenotethatbrackets()areusedtodistinguishlayersinparametersthathavemanylayers.

Also,pleasenotethatprecipitationandtemperaturearealsoallowedtobeusedasfittingparameters.

Thisoptionmustbeusedwithcautionbecausefittingrainfallcanmakecalibrationofparameters

irrelevantasrainfallisthesinglemostimportantdrivingvariablecontrollingthebehaviorofflow.


55/100

55

ObjectiveFunctionDefinition

Intheobserved.txtfile,10differentobjectivefunctionsarecurrentlyallowed.Theseinclude:

1=mult Minimize:

....***

222

N

i

ism

S

i

ism

Q

i

ism

n

NN

n

SS

n

QQ

g

This isamultiplicativeformofthesquareerrorwhereQ,S,andNstandforvariables(e.g.,discharge,

sediment,andnitrate),nisthenumberofobservations,andmandsstandformeasuredandsimulated.

Sometimesthedenominatorisdividedby1000tokeepgsmall.

2=sum Minimize: .....1 2 3

1 1 1

2

3

2

2

2

1

n

i

n

i

n

iismismism

NNwSSwQQwg

This isthesummationformofthesquareerrorwhereQ,S,andNstandforvariables(e.g.,discharge,

sediment,andnitrate),mandsstandformeasuredandsimulated,nisthenumberofdatapoints,and

weightswscouldbecalculatedas:

i) 21

jj

jn

w

where2

j isthevarianceofthe

j

thmeasuredvariable(seeAbbaspour,etal.,2001),or

ii)m

m

m

m

N

Qw

S

Qww 321 ,,1

wherebars indicateaverages (seeAbbaspouretal.,1999).Note thatchoiceofweighscanaffect the

outcomeofanoptimizationexercise(seeAbbaspour,etal.,1997).

3=R2 Maximize:

i

sis

i

mim

sis

i

mim

QQQQ

QQQQ

R2

,

2

,

2

,,

2

CoefficientofdeterminationR2whereQisavariable(e.g.,discharge),andmandsstandformeasured

and simulated, i is the ithmeasuredor simulateddata. If therearemore thanonevariable, then the

objectivefunctionisdefinedas:


56/100

56

j

jjRwg2

Wherewjistheweightofjthvariable.

4=Chi2 Minimize:

2

2

2

m

i

ism QQ

whereQisavariable(e.g.,discharge),andmandsstandformeasuredandsimulated,respectively,and2

m isthevarianceofmeasureddata.Iftherearemorethanonevariable,thentheobjectivefunctionis

calculateas:

j

jjwg2


5=NS Maximize:

imim

iism

QQ

QQ

NS2

,

2

1

NashSutcliffe (1970), where Q is a variable (e.g., discharge), and m and s stand for measured and

simulated, respectively,and thebar stands foraverage. If there ismore thanone variable, then the

objectivefunctionisdefinedas:

j

jjNSwg


6=bR2 Maximize:

1

1

21

2

bifRb

bifRb

WhereCoefficientofdeterminationR2 ismultipliedby the coefficientof the regression linebetween

measuredandsimulateddata,b.Thisfunctionallowsaccountingforthediscrepancy inthemagnitude

oftwosignals(depictedbyb)aswellastheirdynamics(depictedbyR2).Ifmorethanonevariable,the

objectivefunctionisexpressedas(Krauseetal.,2005):


57/100

57

incaseofmultiplevariables,gisdefinedas:

j

jjwg


7=SSQR Minimize:

, ,

whereQisavariable(e.g.,discharge),andmandsstandformeasuredandsimulated,respectively.Here

irepresentstherank.TheSSQRmethodaimsatfittingthefrequencydistributionsoftheobservedand

the simulated series. After independent ranking of the measured and the simulated values (van

GriensvenandBauwens,2003):


j jjSSQRwg


8.PBIAS Minimize:

n

i

im

n

iism

Q

QQ

PBIAS

1

,

1*100

whereQ isavariable (e.g.,discharge),andmand s stand formeasuredand simulated, respectively.

Percentbiasmeasures the average tendencyof the simulateddata tobe largeror smaller than the

observations. The optimum value is zero, where low magnitude values indicate better simulations.

Positive values indicate model underestimation and negative values indicate model over estimation

(Guptaetal.,1999).


j

jjPBIASwg



58/100

58

9.KGE Maximize: 1 1 1 1

Where

,and

,andristhelinearregressioncoefficientbetweensimulatedandmeasured

variable, s and m are means od simulated and measured data, and s and m are the standard

deviationofsimulatedandmeasureddata.KlingGuptaefficiency(Guptaetal.,2009).


j

jjKGEwg


10.RSR Minimize:

n

i

mim

n

iism

QQ

QQ

RSR

1

2

,

1

2

whereQisavariable(e.g.,discharge),andmandsstandformeasuredandsimulated,respectively.RSR

isthestandardizestheRMSEusingtheobservationstandarddeviation.RSRisquitesimilartoChiin4.It

variesfrom0tolargepositivevalues.ThelowertheRSRthebetterthemodelfit(Moriasietal.,2007).


j

jjRSRwg


11.MNS Maximize:

i

p

imim

i

p

ism

QQ

QQ

NS

,

1

Modified NaschSutcliffe efficiency factor. Ifp=2, then this is simply NS as in 5 above. Ifp=1, the

overestimationofapeakisreducedsignificantly.Themodifiedformisreportedtobemoresensitivetosignificantover orunderpredictionthanthesquareform.Increasingthevalueofpbeyond2resultsin

an increase inthesensitivitytohighflowsandcouldbeusedwhenonlythehighflowsareof interest,

e.g.forfloodprediction(Krauseetal.,2005)

NOTE:Afteraniteration,trychangingthetypeofobjectivefunctionandrunSUFI2Post.bataloneto

see the effect of different objective functions, without having to run SWAT again. This is quite

informativeasitshowshowthechoiceofobjectivefunctionaffectsthecalibrationsolution.


59/100

59

SensitivityAnalysis

1 GlobalSensitivityanalysis

Parametersensitivitiesaredeterminedby calculating the followingmultiple regressionsystem,which

regresses the Latin hypercube generated parameters against the objective function values (in file

goal.txt):

m

i

iibg1

A ttest is thenused to identify the relative significanceofeachparameterbi.The sensitivitiesgiven

aboveareestimatesof theaveragechanges in theobjective function resulting from changes ineach

parameter, while all other parameters are changing. This gives relative sensitivities based on linear

approximations and, hence, only provides partial information about the sensitivity of the objective

functiontomodelparameters.Inthisanalysis,thelarger,inabsolutevalue,thevalueoftstat,andthe

smallerthepvalue,themoresensitivetheparameter. Intheexamplebelow,CN2,ESCO, followedby

GE_DELAY,CH_N2,andALPHA_BFarethefivemostsensitiveparameters.


60/100

60

tstatandpvalue

Amultipleregressionanalysisisusedtogetthestatisticsofparametersensitivity.Thetstatis

thecoefficientofaparameterdividedbyitsstandarderror.Itisameasureoftheprecisionwith

whichtheregressioncoefficientismeasured.Ifacoefficientislargecomparedtoitsstandard

error,thenitisprobablydifferentfrom0andtheparameterissensitive.Whatislarge?

YoucouldcomparethetstatofaparameterwiththevaluesintheStudent'stdistributiontable

todeterminethepvalue,whichisthenumberthatyoureallyneedtobelookingat.The

Student'stdistribution(youfindattheendofmoststatisticsbook)describeshowthemeanof

asamplewithacertainnumberofobservationsisexpectedtobehave.Thepvalueforeach

termteststhenullhypothesisthatthecoefficientisequaltozero(noeffect).Alowpvalue(


61/100

61

bymuch.Therefore,thevaluesofthefixedparametersmakeadifferencetothesensitivityofa

changingparameter.

Toperformtheoneatatimesensitivityanalysis:

1 DoasshowninthefollowingFigure.SetthenumberofparametersinthePar_inf.txtfileto1,and

performaminimumof3simulations.

2 ThensetthevaluesoffileSUFI2_swEdit.defasfollows:

3 FinallyperformtheiterationbyrunningunderCalibration,SUFI2_pre.batandthenSUFI2_run.bat.

4 Now,thethreesimulationcanbevisualizedforeachvariablebyexecutingoneat

atimecommand

underSensitivityanalysisasshownbelow:


62/100

62

ThedashedlineistheobservationandthedischargesignalforFLOW_OUT_1isplottedforthreevalues

ofCN2withinthespecifiedrange.Clearly,CN2issensitiveandneedstohavelargervalues.

NOTE:TheusersmustbeawarethattheparametersintheSWATfilesinthemainSWATCUPproject

directoryarealwayschanging.Onceyouperformasensitivityiteration,thentheparametervaluesin

thosefilesarethevaluesofthelastrun(lastparameterset)ofthelastiteration.Toperformtheoneat

atimesensitivityanalysis,oneshouldsetthevaluesoftheparametersthatarekeptconstanttosome

reasonablevalues.Thesereasonablevaluescould,forexample,bethebestsimulation(simulationwith

thebestobjectivefunctionvalue)ofthelastiteration,ortheinitialmodelparametersthatresideintheBackupdirectory.


63/100

63

ParallelProcessing

Parallel processing is a licensed product. Its function is to speed up the calibration process by

parallelizingtherunsinSUFI2.Thespeedoftheparallelprocessingdependsonthecharacteristicsofthe

computer.Newlaptopsnowhaveatleast4CPUs.Theparallelprocessingmodulecanutilizeall4CPUs

sothata1000runiterationcanbedividedinto4simultaneousrunsof250eachperCPU.Thespeedup

willnotbe4timesbecauseofprogramandWindowsoverheads;but therunwithparallelprocessing

willbesubstantiallyfasterthanasingle1000runsubmission.

Nowadaysitispossibletobuildquiteinexpensivelyacomputerwith48to64CPUsandmorethan96

GBofRAM.MostSWATmodelsofanydetailcouldberunonsuchmachineswithouttheneedforcloud

orgridcomputing(seeRouholahnejad,etal.,2012formoredetail).

Currently,20 simulationsareallowed tobemadewithout theneed fora license.Toobtaina license

follow the direction under license and activation and send the hardware ID, for the time being, to

([email protected]).Afterobtainingalicensefilebyemail,followtheactivationprocess.

Torunparallelprocessing,simplyclicktheParallelProcessingbuttononthecommandbar.Anewsetof

command icons appear. Press parallel Processing to seehowmanyjobs canbe submitted to your

computer.UnderProcesscountyoucanchoosehowmanyparalleljobsyouwanttosubmitIfthesize

oftheprojectislargeandthereisnotenoughmemory,thensmallernumberofparallelprocessesthan

thenumberofCPUsmaybepossible.TheCalibrationiconworksasbefore.


64/100

64

ValidationinSUFI2

Forvalidation,youshouldusethecalibratedparameterrangeswithoutanyfurtherchangesandrun

aniteration(withthesamenumberofsimulationsasyouusedforcalibration).

Toperformvalidation inSUFI2,editthefilesobserved_rch.txt,observed_hru.txt,obsrved_sub.txt,and

observed.txtasnecessary forthevalidationperiod.Also,theextraction filesand the file.ciotoreflect

thevalidationperiod.Thensimplyusethecalibratedparameterrangestomakeonecompleteiteration

(usingthecalibrationbutton).The95PPUandtheSummary_statfileshouldreflectthevalidationresults.

Thevalidationkeybringsupthefollowingmenu,whichexplainsthevalidationsteps.


65/100

65

Thesequenceofprogramexecution

Thesequenceofprogramexecutionandinput/outputsareshowninbelow.Inthefollowing,eachinput

andoutputfileisdescribedindetail.

INPUTFILES OUTPUTFILES

SUFI2_Extract_*_No_obs.exe

- model.in

- Absolute_SWAT_Values.txt

- BACKUP file

-New SWAT parameter files

-Swat EditLo .txt

SUFI2_make_input.exe- SUFI2.IN\\trk.txt- SUFI2.IN\\par_inf.txt

- SUFI2.IN\\par_val.txt

-Echo\echo_make_par.txt

-model.in

SWAT_Edit.exe

SUFI2_LH_sample.exe- SUFI2.IN\\trk.txt

- SUFI2.IN\\par_inf.txt

-ECHO\\echo_LH_sample.txt

-SUFI2.IN\\par_val.txt

-SUFI2.IN\\str.txt

SWAT.exe-SWAT output files

SUFI2_Extract_*.exe

- SUFI2_Extract_*.def- output.*

- SUFI2.IN\var_file_*.txt

- SUFI2.IN\trk.txt

- SUFI2.IN\observed *.txt

-Echo\echo_extract_*.txt

-SUFI2.OUT\files listed in

var_file_*.txt

SUFI2_goal_fn.exe

SUFI2_95ppu.exe

SUFI2_new_pars.exe

-Echo\echo_goal_fn.txt-SUFI2.OUT\\goal.txt

-SUFI2.OUT\best_sim.txt-SUFI2.OUT\\best_par.txt

-SUFI2.OUT\\beh_pars.txt

-SUFI2.OUT\\no_beh_sims.txt

-SUFI2.OUT\best_sim_nr.txt

- SUFI2.IN\par_inf.txt

- SUFI2.IN\observed.txt

- SUFI2.IN\par_val.txt- SUFI2.IN\\var_file_name.txt

-Echo\echo_95ppu.txt

-SUFI2.OUT\95ppu.txt-SUFI2.OUT\\95ppu_g.txt

-SUFI2.OUT\\summary_stat.txt


- Files liste in var_file_name.txt

- SUFI2.IN\observed.txt

- SUFI2.IN\\best_sim.txt

Echo\new_pars_all.txt

SUFI2.OUT\new_pars.txt

SUFI2.IN\observed.txt

SUFI2.OUT\\goal.txt

SUFI2.OUT\\best_par.txt

SUFI2_Post.bat

SUFI2_Run.bat

- extract_*_No_Obs.def

- output.*

- SUFI2.IN\var_file_*_No_obs.txt

- SUFI2.IN\trk.txt

SUFI2.OUT\files listed in

var_file_*_No_obs.txt

95ppu_No_Obs.exe- 95ppu_No_Obs.def- SUFI2.IN\par_inf.txt

SUFI2.OUT\95ppu_No_Obs.txt

SUFI2.OUT\95ppu_g_No_Obs.txt

NO_Observation

SUFI2_95ppu_beh.exe

-Echo\echo_95ppu_beh.txt-SUFI2.OUT\\summary_stat.txt


-SUFI2.OUT\\no_beh_sims.txt

- Files liste in var_file_name.txt

- SUFI2.IN\observed.txt- SUFI2.IN\\best_sim.txt


66/100

66

Howtoseetheresultsofmyinitialmodel?

BeforestartingiterationsinSWATCUP,youshouldchecksimulationofyourinitialmodelsetup.Itis

assumedthatsomethoughtandinvestigationhasgoneintodatacollectionandthebestinformationis

usedtobuildtheSWATmodel.Tochecktheinitial(default)simulationofyourmodelinSWATCUPdo

thefollowing:

1InPar_inf,putthenumberofsimulationsandthenumberofparametersto1andsetupadummy

parameterchangesuchas

r__SFTMP.bsn 0 0 (thisdoesnotchangeanything)

2InSUFI2_swEditputthebeginningandtheendingsimulationalsoto1

3Thenexecute:Pre,Run,andPostprocessing.

Nowlookatthe95PPUresultofyourdefaultorinitialmodelrun.Ifsimulationsandobservationsaretoo

different,thenyouneedtotakeacloserlookatyourswatmodel,includingrainfall,etc.

Iftheyarenottoodifferent,thenforeachoutletadjustrelevantparametersintherelevantsubbasins

(referredtoasparameterization),anddoafewiterationsbasedonthat.

Foraprotocolseetheopenaccesspaper:

http://www.sciencedirect.com/science/article/pii/S0022169415001985

HowtosettheparametersinSWATtextfilestothebestparametervaluesof

thelast

iteration?

IfyouwanttheSWATTxtInOutfilesreflectthebestparametersyouobtainedinaniterationdothe

following:

1 NotethenumberofthebestsimulationintheSummary_Stat.txtfile

2 IntheSUFI2_swEdit.txtsetthestartingandendingsimulationvaluesbothtothenumberofthebest

simulationinstep1.

3 UnderCalibration,runSUFI2_run.bat.DonotrunSUFI2Pre.bat.

Thiscommandwillreplacetheparametervaluesandsetthemtothebestvaluesofthelastiteration.


67/100

67

HowtodoLatinHypercubeSampling

ThebatchfileSUFI2_pre.batrunstheSUFI2_LH_sample.exeprogram,whichgeneratesLatinhypercube

samples.Thesesamplesarestoredinpar_val.txtfile.

ThisprogramusesLatinhypercubesamplingtosamplefromtheparameterintervalsgiveninpar_inf.txt

file.The sampledparametersaregiven inpar_val.txt file,while the structureof the sampleddata is

writtentostr.txtjustforinformation.Ifthenumberofsimulationsis3,thenthefollowinghappens:

1)Parameters(say2)aredividedintotheindicatednumberofsimulations(say3)

2)Parametersegmentsarerandomized

3)Asampleistakenatthemiddleofeverysegment

Everyverticalcombinationisthenaparameterset.

2 31

1 2 3

2 1 3

3 2 1

2 1 3

3 2 1


68/100

68

HowtoCalibratemorethanoneVariable

Ifyouwanttocalibrateusing,forexampledischarge,nitrate,andphosphate,youshouldfirstcalibrate

fordischarge.Thisisbecauseflowisthemaincontrollingvariable.Aftercalibratingforflow,keepflow

parameterrangesasyouobtainedfromflowcalibrationandaddsedimentparameters.Therearetwo

typesofsedimentparameters,thosethataffectonlysediment,andthosethataffectflowandsediment.

SeetheTablebelowfromAbbaspouretal.,(2007).

Initially,add theparameters thatonlyaffectsedimentand runan iteration.Youshouldget the same

dischargeresultsasbefore,sotrycalibratingonlyforthesedimentparametersthatdontaffecttheflow


69/100

69

first.After,oneortwoiterations,ifsedimentresultsarenotsatisfactory,thenaddtheotherparameters

thataffectsedimentandflowanddoacoupleofiterationsbyallowingflowparameterstoalsochange

slightly.

ForNitraterepeatthesameprocedurewithnitrateparameters.Notethatforcalibratingphosphoryou

mustcalibrateforsedimentfirst,becausemuchofthephosphormoveswithsediment,butnitratecan

becalibratedwithoutsediment.

Itisimportanttoalsonotethatthesolutiontothecalibratedmodelisthe95PPUgeneratedby

theparameterranges.DONOTtrytoonlyusethebestparametersetforfurtheranalysis.By

doing this youare assuming that the calibratedmodelonlyhasone solution and this isnot

correct. It is never correct to assume that only one set of parameters can represent a

watershed, which was modeled by very uncertain information about soil, landuse, climate,

management, measured data used for calibration, etc. Always propagate the range of

parametersyouobtainedduringcalibrationforallpurposesofmodeluse.


70/100

70

PSO

ParticleSwarmOptimization


71/100

71

IntroductiontoPSO

Particleswarmoptimization(PSO)isapopulationbasedstochasticoptimizationtechniquedevelopedby

Dr.EberhartandDr.Kennedy in1995,inspiredbysocialbehaviorofbirdflockingorfishschooling.

PSOsharesmanysimilaritieswithevolutionarycomputationtechniquessuchasGeneticAlgorithms(GA).

The system is initializedwithapopulationof random solutionsand searches foroptimabyupdating

generations.However,unlikeGA,PSOhasnoevolutionoperatorssuchascrossoverandmutation. In

PSO, thepotential solutions, calledparticles, fly through theproblem spaceby following the current

optimumparticles.Thedetailedinformationwillbegiveninfollowingsections.

Compared to GA, the advantages of PSO are that PSO is easy to implement and there are few

parameterstoadjust.PSOhasbeensuccessfullyappliedinmanyareas:functionoptimization,artificial

neuralnetworktraining,fuzzysystemcontrol,andotherareaswhereGAcanbeapplied.

There are two popular swarm inspired methods in computational intelligence areas: Ant colony

optimization (ACO)andparticleswarmoptimization (PSO).ACOwas inspiredby thebehaviorsofants

and has many successful applications in discrete optimization problems.

(http://iridia.ulb.ac.be/~mdorigo/ACO/ACO.html)

Theparticleswarmconceptoriginatedasasimulationofsimplifiedsocialsystem.Theoriginalintentwas

tographicallysimulate thechoreographyofbirdofabirdblockor fishschool.However, itwas found

that particle swarm model can be used as an optimizer.

(http://www.engr.iupui.edu/~shi/Coference/psopap4.html)

Thealgorithm

Asstatedbefore,PSOsimulatesthebehaviorsofbirdflocking.Supposethefollowingscenario:agroup

of birds are randomly searching food in an area. There is only one piece of food in the area beingsearched. All the birds do not know where the food is. But they know how far the food is in each

iteration.Sowhat'sthebeststrategyto findthefood?Theeffectiveone istofollowthebirdwhich is

nearesttothefood.

PSOlearnsfromthescenarioandusesittosolvetheoptimizationproblems.InPSO,eachsinglesolution

is a "bird" in the search space. We call it "particle". All of particles have fitness values which are

evaluatedby the fitness function tobeoptimized, andhave velocitieswhichdirect the flyingof the

particles.Theparticlesflythroughtheproblemspacebyfollowingthecurrentoptimumparticles.

PSOisinitializedwithagroupofrandomparticles(solutions)andthensearchesforoptimabyupdating

generations.Inevery iteration,eachparticle isupdatedbyfollowingtwo"best"values.Thefirstoneis

thebestsolution (fitness) ithasachievedso far. (The fitnessvalue isalsostored.)Thisvalue iscalledpbest.Another"best"valuethatistrackedbytheparticleswarmoptimizeristhebestvalue,obtainedso

farbyanyparticle inthepopulation.Thisbestvalue isaglobalbestandcalledgbest.Whenaparticle

takespartofthepopulationasitstopologicalneighbors,thebestvalueisalocalbestandiscalledlbest.

After finding the two best values, the particle updates its velocity and positions with the following

equations(a)and(b).

v[]=v[]+c1*rand()*(pbest[] present[])+c2*rand()*(gbest[] present[]) (a)


72/100

72

present[]=persent[]+v[] (b)

v[]istheparticlevelocity,persent[]isthecurrentparticle(solution).pbest[]andgbest[]aredefinedas

statedbefore.rand()isarandomnumberbetween(0,1).c1,c2arelearningfactors.usuallyc1=c2=2.

Thepseudocodeoftheprocedureisasfollows:

Foreachparticle

Initializeparticle

END

Do

Foreachparticle

Calculatefitnessvalue

Ifthefitnessvalueisbetterthanthebestfitnessvalue(pBest)inhistory

setcurrentvalueasthenewpBest

End

ChoosetheparticlewiththebestfitnessvalueofalltheparticlesasthegBest

Foreachparticle

Calculateparticlevelocityaccordingequation(a)

Updateparticlepositionaccordingequation(b)

EndWhilemaximumiterationsorminimumerrorcriteriaisnotattained

Particles' velocities on each dimension are clamped to a maximum velocity Vmax. If the sum of

accelerations would cause the velocity on that dimension to exceed Vmax, which is a parameter

specifiedbytheuser,thenthevelocityonthatdimensionislimitedtoVmax.

ComparisonsbetweenGeneticAlgorithmandPSO

Mostofevolutionarytechniqueshavethefollowingprocedure:

1.Randomgenerationofaninitialpopulation

2.Reckoningofafitnessvalueforeachsubject.Itwilldirectlydependonthedistancetotheoptimum.

3.Reproductionofthepopulationbasedonfitnessvalues.

4.Ifrequirementsaremet,thenstop.Otherwisegobackto2.Fromtheprocedure,wecanlearnthatPSOsharesmanycommonpointswithGA.Bothalgorithmsstart

withagroupofarandomlygeneratedpopulation,bothhavefitnessvaluestoevaluatethepopulation.

Bothupdatethepopulationandsearchfortheoptimumwithrandomtechniques.Bothsystemsdonot

guaranteesuccess.

However,PSOdoesnothavegeneticoperatorslikecrossoverandmutation.Particlesupdatethemselves

withtheinternalvelocity.Theyalsohavememory,whichisimportanttothealgorithm.

Compared with genetic algorithms (GAs), the information sharing mechanism in PSO is significantly

different.InGAs,chromosomesshareinformationwitheachother.Sothewholepopulationmoveslike

aonegrouptowardsanoptimalarea.InPSO,onlygBest(orlBest)givesouttheinformationtoothers.It

isaonewayinformationsharingmechanism.Theevolutiononlylooksforthebestsolution.Compared

withGA,alltheparticlestendtoconvergetothebestsolutionquicklyeveninthelocalversioninmostcases.

7.OnlineResourcesofPSO

ThedevelopmentofPSOisstillongoing.AndtherearestillmanyunknownareasinPSOresearchsuchas

themathematicalvalidationofparticleswarmtheory.


73/100

73

Onecanfindmuchinformationfromtheinternet.Followingaresomeinformationyoucangetonline:

http://www.particleswarm.net lots of information about Particle Swarms and, particularly, Particle

SwarmOptimization.LotsofParticleSwarmLinks.

http://icdweb.cc.purdue.edu/~hux/PSO.shtml lists an updated bibliography of particle swarm

optimizationandsomeonlinepaperlinks

usermanual swat cup

Documents