Soil water and salinity dynamics under sprinkler irrigated almond exposed to a varied salinity

14 Pages·2017·1.54 MB·English

by  V. Phogat

UC Riverside
UC Riverside Previously Published Works
Title
Soil water and salinity dynamics under sprinkler irrigated almond exposed to a varied
salinity stress at different growth stages
Permalink
https://escholarship.org/uc/item/3r29q8tr
Journal
Agricultural Water Management, 201(C)
ISSN
0378-3774
Authors
Phogat, V
Pitt, T
Cox, JW
et al.
Publication Date
2018-03-31
DOI
10.1016/j.agwat.2018.01.018

Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
Agricultural Water Management 201 (2018) 70–82
ContentslistsavailableatScienceDirect
Agricultural Water Management
journalhomepage: www.elsevier.com/locate/agwat
Soil water and salinity dynamics under sprinkler irrigated almond exposed
ff T
to a varied salinity stress at di erent growth stages
V. Phogata,b,⁎, T. Pitta, J.W. Coxa,c, J. Šimůnekd, M.A. Skewesa
aSouthAustralianResearchandDevelopmentInstitute,GPOBox397,AdelaideSA5001,Australia
bCCSHaryanaAgriculturalUniversity,Hisar125004,India
cTheUniversityofAdelaide,PMB1GlenOsmondSA5064,Australia
dDepartmentofEnvironmentalSciences,UniversityofCalifornia,Riverside,CA92521,USA
A R T I C L E I N F O A B S T R A C T
Keywords: Wateruseandsalinitydynamicsinthesoilsarethecrucialmanagementfactorsinfluencingtheproductivityand
Almond long-termsustainabilityofalmondandassociatedenvironment.Inthisstudy,HYDRUS-2Dwascalibratedand
HYDRUS-2D validatedonmeasuredspatialandtemporalwatercontentsandsoilsalinities(EC)distributionsunderalmond
e
Salinity irrigatedwithdifferentwaterqualities(EC )atdifferentphysiologicalstages.Duringtwoirrigationseasons
iw
Leaching (2014–15 and 2015–16), less saline irrigation water (average EC 0.78dS/m) was substituted for recycled
WWaatteerrbuaselaenfficeciency irrigation water (average ECiw 1.9dS/m) in three phenologicallyidwifferent growth stages; pre-pit hardening,
kernelgrowth,andpost-harvest,alongwithnoandfullsubstitutionduringtheentireseason.Graphicaland
statistical comparisons (RMSE,MAE, ME,theNashandSutcliffemodelefficiency,andthecoefficient ofde-
termination) between measured and simulated values of water contents and EC in the soil showed a close
e
agreementinalltreatments.Thewaterbalancedatarevealedthattheseasonalcropevapotranspirationofal-
mond(ET)variedfrom850to955mmindifferenttreatmentsoverthetwoseasonswhichrepresented68–79%
c
of the water application. Trees irrigated with only less saline water through the two seasons (average EC
iw
0.78dS/m)showed10%higherplantwateruptakeascomparedtothoseirrigatedwithrecycledwateronly
(averageEC 1.9dS/m).Substitutinglesssalineirrigationduringthekernelgrowthphase,betweenpit-hard-
iw
eningandharvest,showedgreaterwateruptakebyalmondandlowersalinitybuildupinthesoilascomparedto
treatmentsthatsubstitutedlesssalineirrigationearlyorlateintheseason.Foralltreatments,theaveragedaily
rootzoneEC (2.4–3.7dS/m)remainedabovethelevelofthealmondsalinitytolerancethreshold(EC =1.5dS/
e e
m) throughout the period of investigation. Water use efficiency of almonds varied in a narrow range
(0.21–0.25kgm−3)fordifferenttreatments.Deepdrainagebelowtherootzone(2m)variedfrom22.4–31.1%
ofthetotalwaterapplication(Rainfall+Irrigation),whichwasepisodicandinsufficienttocontainthesalinity
belowthealmondthreshold.Thisstudyprovidedagreaterunderstandingofsoilwaterandsalinitydynamics
underalmondirrigatedwithwatersofvaryingqualities.
1. Introduction 1982;Maas,1990;PitmanandLaüchli,2002;Assoulineetal.,2015).
Impactsofsoilsalinityarevariedandhighlyinfluencedbythesoiltype,
Water is becoming increasingly scarce worldwide, andsustainable inherentlayeringandhydraulicheterogeneity,qualityandquantityof
useoftheavailablewaterresourcesisthemajorwaterpolicychallenge irrigation, the method of irrigation, rainfall amounts and distribution
forthefuture.Inwater-stressedregions,strongconstraintsduetonat- patterns,andsalttoleranceofcrops.Atthesametime,theinstallation
uralclimaticvariabilityandincreasedusebyotherusersmaketheal- ofhighlyefficient,partialcover,irrigationsystemsinhigh-valuecrops
location of water the core challenge for water resource management including almond has anarrow wettedarea thattends to concentrate
(Alconetal.,2013).Inadditiontoissuesrelatedtowaterquantity,the saltsintherootzone.Duringperiodsofwaterscarcity,whenirrigators
qualityofwaterplaysanimportantroleinthesustainabilityofirrigated areforced to limit their leaching fractions orforced to apply ground-
lands,especiallyinthecontextofsalinitybuildupthatcouldadversely water/recycledwater,saltscanrapidlyaccumulatetodangerouslevels.
impact the agricultural/horticultural crop productivity (Bresler et al., Almondproduction inthearidandsemi-arid regions oftheworld
⁎Correspondingauthor.
E-mailaddress:[email protected](V.Phogat).
https://doi.org/10.1016/j.agwat.2018.01.018
Received24October2017;Receivedinrevisedform9January2018;Accepted19January2018
0378-3774/ © 2018 Elsevier B.V. All rights reserved.
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
remains vulnerable to water availability and soil salinity. Although recycledwater,(2)tosimulatewatermovementandsalinitydynamics
almond is a drought tolerant tree (Torrecillas et al., 1996), its pro- forotherexperimentalirrigationtreatmentsincludingsubstitutingless
ductionandprofitabilityarehighlydependentonthesupplyofirriga- salineirrigationwaterfortheresidentrecycledwaterirrigationatthree
tionwater.Theadoptionofefficientirrigationtechniquescanincrease phenologically different growth stages, and (3) to evaluate the water
theproductionbyasmuchasten-foldcomparedtonon-irrigatedlands useefficiencyofalmondsunderthedifferentirrigationtreatments.
(Egeaetal.,2010).Ontheotherhand,almondissensitivetosalinity,
having a production threshold with respect to the electrical con-
2. Materialsandmethods
ductivityofthesoilsaturationextract(EC)of1.5dS/mandareduction
e
in the growth rate of 19% for a unit increase in salinity beyond the
2.1. Experimentaldetails
threshold (Maas, 1990). Thus, the salinity levels higher than the
threshold anytime during the cropping season can have a serious im-
The experimental site was established at amature almond planta-
pactongrowthandnutproduction.Therefore,properirrigationman-
tion located in the Northern Adelaide Plains irrigation district, ap-
agement in the orchard is very important to obtain sustainable yield,
proximately 35km north of Adelaide, South Australia (34.628°S and
leachharmfulchemicalsfromtherootzone,savepreciouswater,and
138.683°E).Theorchardwasplantedin1998anddesignedtohavetwo
increasetheshelflifeofthefruits.Preciseinformationonthesalinity
adjacent rowsof acommercial variety, Nonpareil, bordered oneither
impactatdifferentgrowthstagescanhelpdevelopmorerobustorchard
side by pollinators, Price and Keane. All trees were grafted to Bright
salinitymanagementguidelines.
peach hybrid rootstock with rows planted in a north-south direction.
Differentdeficitirrigationoptions(regulateddeficitirrigation,RDI;
Trees were spaced at a distance of 5.5m within the rows and 7.5m
sustained deficit irrigation, SDI; and partial root zone drying, PRD),
betweenrows.
including their impact at different growth stages of almond (Girona
Thetrialwasdesignedasarandomizedunblockeddesign,withfour
et al., 2005; Goldhamer and Viveros, 2000; Romero et al., 2004;
treatmentsreplicatedfourtimes,plusanadditionaldemonstrationplot.
Goldhameretal.,2006;Egeaetal.,2010;Puertoetal.,2013)havebeen Each treatment plot consisted of micro-sprinklers (five trees) along a
extensively studied over the almond growing season (Nortes et al.,
treelineandwasthreerowswide.ItcoveredadoublerowofNonpareil
2009; Egea et al., 2013; Monks et al., 2017). These investigations
treesplusasinglerowofapollinatorvariety.Allsoilandplantmea-
highlighted theimpact ofreduced water applications on various phy-
surementswerecollectedfromthethreecentraltreesinthemiddlerow.
siological parameters, kernel size, and kernel yield. Other studies fo- The treatment infrastructure was installed before the 2013–14 irriga-
cussedonthealmondwaterrequirementandcropcoefficients(Stevens
tionseasonandwasdesignedtosubstituterecycled(moresaline)water
etal., 2012;Espadafor etal.,2015;García-Tejero etal.,2015), anir-
irrigation(2seasonsaverageEC =1.9dS/m)withlesssalinewater(2
rigationsystemdesign(Phogatetal.,2012),andwaterproductivityof iw
seasons average EC =0.78dS/m) at one of the three phenological
almondsunderreducedwaterapplications(García-Tejeroetal.,2011; iw
growthstages.Thetreatmentwhenrecycledwaterirrigationwasused
Phogatetal.,2013).Ontheotherhand,thestudiesontheimplications
fortheentireirrigationseasonrepresentsthecontroltreatment(A).In
of the application of saline/recycled water irrigation on the water
treatmentB,lesssalinewaterwasappliedbetweenthebudsburstand
availability to almonds and salinity dynamics in the soil are sparse.
pithardening(BB-PH)stages.IntreatmentC,lesssalineirrigationwas
Notably, the salinity related studies are specifically focused on the
appliedbetweenthepithardeningandharvest(PH-H)stages,whereas
identification of tolerant root stocks (Gradziel and Kester, 1998;
intreatmentD,lesssalineirrigationwasapplied betweentheharvest
Camposeo et al., 2011) and genotypes (Rouhi et al., 2007; Sorkheh
andleaffall(H-LF)stages.Thenumberofdaysduringthethreephy-
etal.,2012;Rajabpooretal.,2014;Bahramietal.,2015).Francoetal.
siologicalstageswhenlesssalinewaterwasappliedintreatmentsB,C,
(2000)reporteda46%reductioninthealmondkernelswhenseason- andDwas84,108,and73during2014–15and88,101,and88during
ally irrigated with high salinity water (4.6dS/m) compared to when and2015–16,respectively.Inaddition,anon-replicatedplot(treatment
less saline water (0.8dS/m) was used. Nightingale et al. (1991) ob-
E)oftreeswasirrigatedwithlesssalinewaterduringtheentireseason
served the greatest accumulation of salinity (5.7dS/m) beneath the
for demonstration purposes. Table 1 gives details about the four re-
tricklelinesourcefor50%ET irrigationandlaterallyawayfromthe
C plicated irrigation treatments and the non-replicated demonstration
tricklelinefor100and150%ET irrigation.Therefore,theinformation
C treatment. Further details about the trial design are described in Pitt
on the impact of saline/recycled water irrigation at different growth
etal.(2015).
stagesofalmondonthewaterbalanceandsalinitydynamicsinthesoil
Irrigations were scheduled to replace estimated tree evapo-
could improve the understanding of the use of such water for the al- transpiration, which was evaluated based upon a modified version of
mondproductionandensurethelong-termsustainabilityofhigh-value
theprotocoldevelopedbytheAlmondBoardofAustralia(2011).The
horticulturecrops.
Almond Board of Australia (ABA) provides a spreadsheet based on a
Predictive science and numerical models such as HYDRUS-2D
(Šimůnek et al., 2016) present an excellent opportunity to gauge the
impactsofirrigationpractices(KandelousandŠimůnek,2010;Ramos TThabelteim1ingofexposuretonon-salineirrigationwaterintreatmentsA–D(replicated)and
etal.,2012;Phogatetal.,2012;Gonzálezetal.,2015),waterquality treatmentE(anon-replicateddemonstrationplot).
(Hassan et al., 2005; Ramos et al., 2011; Hassanli et al., 2016) and
climate change (Austin et al., 2010) on potential water and salinity Treatments Irrigation GrowthStages
hazardsandtocontroloffsitemovementofcostlyinputsintothesur-
1 2 3
faceandsubsurfacewaterbodies(Phogatetal.,2014).Moreover,the
partial wetting pattern and irregular root water uptake in a drip-irri- BB*toPH#P# HtoHλ HtoLDη
gated orchard make it difficult to apply and interpret standard water
A-Control Salineallyear Saline
balancetechniquesandrequirealargenumberofmeasurements(Ben-
B Non-salineatBB-PH Non-saline Saline
Asher,1979).Thus,anumericalmodel(HYDRUS-2D)wasemployedto C Non-salineatPH-H Saline Non-saline Saline
evaluate the impact of different qualities of water applications to al- D Non-salineatH-LD Saline Non-saline
monds at different growth stages on the water balance and salinity E Non-salineallyear Non-saline
dynamicsinthesoil.
*Budburst.
The objectives of this study are (1) to calibrate and validate
#Pithardening.
HYDRUS-2Dtodescribewatercontentdistributionsandspatiotemporal λHarvest.
salinity dynamics in the soils under sprinkler-irrigated almond with ηLeafdrop.
71
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
waterbudgetingtooltofacilitatethisprocess.Thespreadsheetrequires Table2
theusertoenterdailyestimatesofpanevaporationfromaclassApan OptimizedsoilhydraulicparametersattheexperimentalsiteusedintheHYDRUS-2D
installedontheirpropertyanduses“in-house”cropfactors(C)within simulations.
f
the spreadsheet to calculate the crop water requirement for the day Soildepth(cm) Texture θr θs α n Ks l
(AlmondBoardofAustralia,2011).Thepotentialcropwateruse(ET )
C
is the result of a product of measured pan evaporation (ET ) and a cm3cm−3 cm−1 – cm/day –
pan
given crop factor on any particular day during the season. Therefore,
1 0–15 Sandyloam 0.06 0.42 0.075 1.89 120 0.5
dataonirrigationdepthsfortheexperimentalsiteweresourcedfrom
2 15–25 Loam 0.12 0.37 0.059 1.48 80 0.5
thecollaboratinggrowerandcross-checkedagainstflowmeterslocated 3 25–35 Sandyclay 0.18 0.40 0.041 1.40 65 0.5
in each treatment plot. Applied depths of irrigation were recorded loam
weeklythroughtheirrigationseasontoensurethesamedepthofirri- 4 35–55 Clayloam 0.23 0.45 0.027 1.23 40 0.5
gationwasappliedtoalltreatmentsateachgrowthstage.Theamount 5 55–110 Clayloam 0.22 0.40 0.045 1.40 70 0.5
6 110–130 Siltyclayloam 0.19 0.37 0.02 1.20 30 0.5
of seasonal irrigation to almonds was slightly higher (8%) during 7 130–150 Siltyclayloam 0.20 0.38 0.032 1.32 40 0.5
2015–16ascomparedto2014–15.TheorchardwasirrigatedwithEin- 8 150–200 Siltyclayloam 0.19 0.37 0.02 1.20 30 0.5
Dor (70L/h) micro-sprinklers spaced every 5.5m, halfway between
trees,alongthelengthoftreerowsspacedat7.5m.Thisresultedinan
applicationrateof1.7mm/h.Thespatialextentofsprinklerswasovera parameters (α, n, and Ks) were further optimized during calibration.
280-cm radial area and there was a small overlap from neighboring Theoptimizedsoilhydraulicparametersfordifferentsoildepthsused
sprinklers. Irrigation water was distributed more orlessuniformly on inthepresentmodelingstudyaregiveninTable2.
thesoilsurfacealongatreerowtoadistanceof280cmonbothsidesof
the tree line. While sprinkler irrigation with saline water can cause 2.2. Meteorologicaldata
direct salt injuryto thecrop foliage, itremained an understory inter-
ventionforalmondtreesattheexperimentalsite. HYDRUS-2Drequiresdailyrainfall,potentialevaporation(Es),and
The recycled water was drawn from the Bolivar Wastewater transpiration (Tp) as inputs, which define the dynamic climatic varia-
TreatmentPlant.Thewastewaterusedforirrigationinthisstudywas bility experienced by a plant. Data on rainfall and daily reference
classArecycledwaterandaspertheguidelinesClassAwaterhasre- evapotranspiration (ET0) during the modeling period (2014–15 and
ceivedaleveloftreatmentgreaterthanclassesBtoDandissuitablefor 2015–16) fortheexperimentalsite weregenerated byrunningadata
unrestricted irrigation to all crops and fodder types (Department of drill(Jeffreyetal.,2001)whichgeneratethedataforasitedepending
Human Services and South Australian Environmental Protection on the available climate data records in the adjoining Bureau of Me-
Agency, 1999). The less saline irrigation water was supplied with a teorology (BOM) observatories. The observatories closest to the ex-
potablewaterconnection(SAWater),installedin2013adjacenttothe perimental site are Edinburg Raaf (6.9km), Roseworthy (16km), and
trial site. Water samples were continuously collected throughout the Parafield(16km).Cropcoefficients(Kc),whichwereobtainedfromthe
growing season via sprinklers, and the electrical conductivity of the IRES irrigation scheduling software developed by Rural Solutions SA
irrigation water (ECiw) was assessed. In treatment A (recycled water), (Rural Solutions SA, 2011), were used to estimate daily ETC for al-
theECiwvariedfrom1.01to2.54and1.2–2.18dS/mduring2014–15 monds. Daily ETC was divided into potential Tp and Es following
and2015–16seasons,respectively,withcorrespondingaverageEC of Belmansetal.(1983)usingmeasuredLAIattheexperimentalsite.LAI
iw
1.94dS/mand1.81dS/m.Similarly,intreatmentE(lesssalinewater), was determined using the method described in Fuentes et al. (2014)
the EC varied from 0.72 to 0.78dS/m (average 0.76dS/m) and using an SLR Camera (Leica Digilux 2, New Jersey, USA). LAI was
iw
0.83–0.86dS/m(average0.85dS/m),respectivelyduring2014–15and measured three times during both seasons: just after pit-hardening
2015–16. (November),beforeharvest(February),andintheweeksapproaching
The experimental site was managed as per standard commercial senescence(April).Fourundercanopyimageswerecollectedfromeach
orchard practices including a full nutrient, fungicide, and herbicide of three central trees using a sensor located at the ground level and
spray program plus mechanical harvest operations. A mid-row cover 1.5mawayfromthetrunk.Fourimagescapturedthenorth,south,east,
crop of various volunteer weeds and grasses (both annual and per- and west portions of the canopy. Images were processed using an al-
ennial) were controlled with occasional slashing and under tree her- gorithmdevelopedbyFuentesetal.(2014). DailypotentialTpandEs
bicide operations. The experimental site was fertilized via foliar, values were then used as time-variable atmospheric boundary condi-
broadcastandfertigationmethodsasreportedinPittetal.(2015).The tionsinthemodel,alongwithprecipitationdataforthesiteduringthe
variationinEC duetofertigationwasnotconsideredinthemodeling simulationperiod.Seasonalrainfallattheexperimentalsiteamounted
iw
simulationsastheamountandtimingoffertilizationapplicationissi- to327and373mmduring2014–15and2015–16seasons,respectively.
milaracrossallthetreatments. Daily ET0, rainfall, and the amount of irrigation applied to almonds
Soilsatthesiteconsistofasandyloamtoloaminthesurface35-cm duringthe2014–15and2015–16seasonsareshowninFig.1.
horizon,followedbyhardandcalcareousclay(35–110cm),andvari-
able clay soil zoning below 110cm, as characterized by Dowley and 2.3. Watercontentandsoilsalinitymeasurements
Fitzpatrick(2001).Theynotedthatthetreerootdevelopmentisgoodin
theupper0.6moftheprofileandreportedgoodwatermovementinthe Soil water contents were measured on a monthly basis during the
surfacelayer,with24-hirrigation(atarateof2.5mm/h)wettingthe 2015and2016irrigationseasonsusinga503DRHydroprobeneutron
soiltoadepthof45–50cm.Themoisturecharacteristics,bulkdensity, moisturemeter(CPNInternational,California,USA).Twelvealuminum
soiltextures, particlesizedistribution,andchemical propertiesofthe accesstubeswereinstalledinMay2014forthemonitoringofspatio-
soils of this region have also been documented in theAPSIM Soil da- temporal moisture contents distribution in the soil in different treat-
tabase (2016) and the ASRIS (2011) database. The particle size dis- ments.ThetubeswereinstalledinthreereplicatesintreatmentsA,C,
tributionandthebulkdensityofvariousdepthswereusedtoestimate andD,tworeplicatesintreatmentB,andasinglereplicateinthenon-
thesoilhydraulicparametersusingROSETTA(Schaapetal.,2001),a replicated treatment E. Access tubes were installed 90cm from the
pedo-transfer function software package that uses neural networks to sprinkleremittersacrossthetreeline.Theprobewassetuptomeasure
predict soil hydraulic parameters from soil texture and related data. at 10cm depth increments between 20 and 60cm and 20cm depth
However,valuesofresidual(θr)andsaturated(θs)watercontentswere incrementsbetween60and200cm.Atthesametime,undisturbedsoil
takenfromthedatabasesmentionedabove.Thevaluesofsoilhydraulic sampleswerecollectedfromthreeseparatecoresforthedetermination
72
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
Feddesmacroscopicapproach(Feddesetal.,1978)implementedinthe
HYDRUS software. The following critical values of pressure heads for
almond in the Feddes et al. (1978) model were used: h =−10,
1
h =−25, h =−500 to −800, and h =−15000cm. These values
2 3 4
were taken from previous investigations in South Australia (Phogat
etal.,2012,2013).Thereductionofrootwateruptakeduetosalinity
stress, α2(hϕ), was described by adopting the salinity threshold and
slopefunction(Maas,1990).Thesalinitythreshold(EC )foralmonds
T
hasavalueofEC of1.5dSm−1andaslope(s)of19%.Asrequiredby
e
HYDRUS-2D, these values were converted into the electrical con-
ductivity at the actual soil water content (EC ), utilizing the linear
sw
relationship(Pittetal.,2017)andinitialwatercontentvalues.
Fig.1.Dailyreferencecropevapotranspiration(ET0,linegraph),rainfall(bluebars),and
irrigation (green bars) applied to almonds during 2014–15 and 2015–16. (For inter-
pretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtotheweb 2.5. Soilsalinitysimulation
versionofthisarticle.)
Thedistribution ofsoilsolution salinity(EC )wassimulatedasa
sw
non-reactive solute (Ramos et al., 2011; Phogat et al., 2012; Phogat
ofbulkdensity,fortheconversionofgravimetrictovolumetricwater
et al., 2014; Wang et al., 2014). These studies demonstrated a good
contents,andforthecalibrationoftheNeutronProbe.
predictionofsalinityunderintensiveirrigationandfertigation,similar
Soilsampleswerecollectedfromeveryplotofalltreatmentsatthe
beginningof2014–15andattheendofeachgrowthstage(budburst, topracticesintheregionunderstudy.Thelongitudinaldispersivitywas
pit hardening, harvest) during the 2014–15 and 2015–16 seasons. assumed as one/tenth of the modeling domain (with the transverse
dispersivity being one-tenth of the longitudinal dispersivity) (Beven
Samples were collected 100cm into the mid-row from the sprinkler
etal.,1993;Coteetal.,2003)andmoleculardiffusioninwaterequalto
emitterandin10cmincrementstoadepthof160cmusingahydraulic
1.66cm2/day(Phogatetal.,2014).Measuredvaluesofirrigationwater
soil sampling rig with a 50mm diameter collection tube (Christies
salinity(EC )wereusedasatimevariableinputfortherootzonesoil
Engineering,HorsleyPark,Australia).Soilsalinitywasmeasuredasthe iw
salinityprediction.TherainfallchemistryanalyzedbyCresswelletal.
electrical conductivity of 1:5 soil:water extracts (EC1:5) on duplicate
(2010)fordifferentlocationsinAustraliaprovidedreliableinformation
samples using a temperature compensated conductivity meter (model
aboutrainfallsalinity(EC )fortheAdelaideregion,whichiscloseto
CON510,Eutech,Singapore).However,soilsalinitywasreportedasthe rw
the experimental site. The average salinity of rainfall was 0.12dS/m,
electricalconductivityoftheextractfromasaturatedpaste(EC)using
e andthisvaluewasusedforallsalinitysimulations.
aconversionfactor(EC =5.89×EC ,R2=0.72)thatwasgenerated
e 1:5
by analyzing paired data from 35 soil samples covering all depths,
2.6. Modelingdomainandinitialandboundaryconditions
whichhadbeensplitsothatbothEC andEC couldbemeasured.All
1:5 e
salinities were determined following the method of Rayment and
Atwo-dimensionalmodelingdomainwasconstructedwithawidth
Higginson(1992).Moredetailsaboutthesamplingandtheanalysisare
equal to the half distance between tree lines (375cm) and a vertical
describedinPittetal.(2017).
depth of 200cm (Fig. 2). The domain represents a right half cross-
section ofthealmond treeplant spacing. Thedomain wasdiscretized
2.4. Rootdistribution into 23,505 elements, with a finer mesh size on the upper boundary
where the time-variable flux and atmospheric boundary conditions
Most almond roots at the experimental site are distributed in were imposed. Measured values of water contents and soil salinities,
shallow depths (0–25cm). Very few roots were encountered in the recordedon10/6/2014,wereusedastheinitialconditions.Thewater
25–50cmdepthduetothepresenceofhardclayattheshallowdepth, contents were measured using a Neutron Probe, and soil salinity was
which doesn’t allow the roots to penetrate deeper as the penetration measured on disturbed soil samples collected during the probe cali-
resistance exceeds 2MPa (Dowley and Fitzpatrick, 2001). The root bration. The top boundary consisted of a time-variable flux boundary
distributionconsideredinthemodelwasbasedonaroughestimateof conditionalongthe280-cmwidthandanatmosphericboundarycon-
thevolumeofrootsrecoveredfromvariousdepthranges(inincrements ditionontherestoftheupperboundary(100cm).A15%rainfallin-
of 10cm from the surface to a depth of 160cm) from soil samples terception by the tree was assumed, which is comparable to the re-
collectedusingahydraulicrig.Astherootswereonlyobservedabove ported value for similar deciduous trees (Xiao et al., 2000). Uniform
the 50cm soil depth, their distribution was assumed in the model distributions of net rainfall and irrigation were imposed as a time-
withinthe0–50cmregionwithamaximumrootdensityatthe20-cm variable flux boundary condition. However, the understory rainfall
soildepth.Rootwateruptakefromthisregionwascomputedusingthe distribution and interception by trees could vary depending on the
Fig.2.Amodelingdomain(375×200cmintheX–Zplane)representing
therighthalfspacingofthemid-rowalongthealmondtreeline,showing
imposedboundaryconditionsandthedistributionofsoiltexturallayersin
thesoilprofileattheexperimentalsite.
73
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
canopy characteristics. The side boundaries were no flowboundaries, N
∑ (M −S)2
andfreedrainagewasimposedatthelowerboundary.Themodelwas i i
calibrated using water content and salinity distributions during the E=1− i=1
N
entireseason(July10,2014,toJune30,2015).Themodelsimulated ∑ (M −M)2
i
soilsolutionsalinityvalues(ECsw)wereconvertedtothesoilsalinityat i=1 (5)
saturation (ECe) utilizing the corresponding water content data and a TherangeofEliesbetween−∞and1.0(perfectfit).Anefficiency
relation developed for the experimental soil samples. Validation was
value between 0 and 1 is generally viewed as an acceptable level of
performedusingdatafromJuly1,2015toJune30,2016inthecontrol performance. Additionally, the coefficient of determination (R2) be-
(recycledwaterirrigation)(treatmentA)bycomparingmeasuredand
tween the simulated and measured water contents and soil salinities
simulatedvaluesofthewatercontentsandsalinitiesatvariousdepths
wasalsoestimatedas:
(20, 30, 40, 50, 60, 80, 100, 120, 140, and 160cm). In treatment A,
irrigationswerecomposedofrecycledwaterhavinganaverageECiwof ⎛ N ⎞2
1.94dS/m (varied from 1.01 to 2.54dS/m) and 1.81dS/m (ranged ⎜ ∑ (Mi−M)(Si−S) ⎟
from1.2to2.18dS/m)duringthe2014–15and2015–16seasons,re- R2=⎜ i=1 ⎟
spectively. ⎜ N N ⎟
∑ (M −M)2 ∑ (S −S)2
Thecalibratedandvalidatedmodelwasthenusedtosimulatewater ⎜ i i ⎟
⎝ i=1 i=1 ⎠ (6)
andsalinitydynamicsunderothertreatments(B,C,andD)wherere-
cycledwaterwassubstitutedwithlesssalinewater(EC of<0.8dS/
iw
m) between different phenological stages of the almond growth. 3. Resultsanddiscussion
Simulations were also performed for treatment E where less saline
water was used throughout the almond season. Simulated spatio- 3.1. Modelcalibrationandvalidationforwatercontentdistributionsinthe
temporaldynamicsofwatercontentsandsoilsalinitieswerecompared soils
withthecorrespondingmeasuredvaluesinalltreatments.
Model calibration and validation were performed using measured
valuesofwatercontentsinthesoilatdifferentdepthsandatdifferent
2.7. Yieldcomponentsandwaterproductivity growth stages of almond in the saline water treatment (A; control).
Measuredandsimulatedwatercontentsatdifferenttimesduringmodel
Theyieldwasmeasuredonthethreecentraltreesofeachtreatment calibration(2014–15)arecomparedinFig.3a–d.Differentfiguresre-
plot. Prior to commercial harvest operations, the orchard floor was present different growth stages of almond (bud burst, pit hardening,
rakedclearunderthetargettrees.Treesweremechanicallyshakenon harvest, and leaf drop). The simulated distribution (S) of the water
16February2015and10February2016.Nutswerelefttodryonthe contentinthesoilprofilematchedwellwiththemeasured(M)valuesat
groundforfourdaysbeforetheexperimentalpickupoperation. alltimesduringthegrowingseason.However,therewasasmallover-
Water productivity of almonds for actual ET losses (WPET ) was estimationofwatercontentsattheharveststage(Fig.3c)betweenthe
C
estimated using the average seasonal ET values obtained from 60–120cm soil depth. Such occasional deviations in simulated values
C
HYDRUS-2D simulations for two seasons and measured average yield areunavoidableandareonthesameorderofmagnitudeasmeasure-
datafordifferenttreatmentsgivenas: ment errors of the sensors (e.g., Ganjegunte et al., 2012; Evett et al.,
2012).
WPET = Y Similarly, statistical errors (see Section 2.8)evaluating differences
C ETC (1) betweenmeasuredandsimulatedsoilwatercontentsindicatedaclose
fit (Table 3, the first row). The RMSE and MAE values obtained by
where Y is the average kernel yield (kg/tree) and ETC is the average comparing measured and simulated soil water contents at different
seasonal actual evapotranspiration (m3 of water/tree) estimated by depths and at different times varied from 0.020 to 0.036 and
HYDRUS-2Dfordifferenttreatments. 0.014–0.030cm3cm−3, respectively. Similarly, the mean error esti-
mated at different times also varied in a narrow range
(−0.030–0.015cm3cm−3)(Table3).ThenegativeMEvaluesindicated
2.8. Statisticalparameters
a slight under-estimation of water contents by the model. However,
small ME values indicate a good match between measured and simu-
The model’s performance was evaluated by comparing measured lated values. Also, the seasonal modeling efficiency of 0.79 indicates
(M)andHYDRUS-2Dsimulated(S)valuesofwatercontentsandelec-
that the model provided a good representation of water movement
tricalconductivities(ECe)inthesoilatdifferenttimes,andcalculating under experimental conditions. Similarly, the coefficient of determi-
themeanerror(ME),themeanabsoluteerror(MAE),andtherootmean
nation (R2; 0.85) for measured and simulated water contents also
squareerror(RMSE)asfollows:
showed a close correspondence. Hence, estimated values of soil hy-
draulicparametersusedinthemodelrepresentwelltheexperimental
ME= 1 ∑N (M −S) soil.
N i i (2) Duringvalidation,thecalibratedHYDRUS-2Dmodelwasrunwith
i=1
the weather and irrigation input (treatment A) for the second season
(2015–16). Overall, there was a good match between modeled and
MAE= 1 ∑N M −S measured water contents at different depths and times as shown in
N i=1 i i (3) Fig. 3e–h. Small deviations at some depths are common under field
conditions.Similarly,thestatisticalcomparisonbetweenmeasuredand
simulated soil water contents during validation (2015–16) showed a
RMSE= 1 ∑N (M −S)2 slightlywidererrorrange(Table3)ascomparedtocalibration.How-
N i=1 i i (4) ever,theRMSE(0.16–0.38cm3cm−3),MAE(0.11–0.34cm3cm−3),and
ME (−0.034–0.006cm3cm−3) values ranged within accepted limits
Model efficiency (E) was estimated as proposed by Nash and reportedinsimilarstudies(e.g.,Debetal.,2011;Phogatetal.,2012,
Sutcliffe(1970),whichisgivenas: 2013, 2014, 2017;Ramos et al., 2011, 2012). The seasonal modeling
74
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
Fig.3.Comparisonofmeasured(M)andHYDRUS-2Dsimulated(S)watercontentdistributionsinthesoilatindicatedtimesduring2014–15(a,b,candd)and2015–16(e,f,gandh)
underalmondirrigatedwithsalinewater(treatmentA).
efficiency(0.78)wasalsoslightlylowerthanthecorrespondingvalue water contents at the harvest stage in treatment D (see the fourth
obtained during calibration. However, the R2 value (0.86) between columninFig.4)atdepthsbetween40and140cmdepth.Nevertheless,
measuredandsimulatedseasonalwatercontentdistributionsindicated the statistical measures, such as error parameters (RMSE, MAE, and
aclosematch.Allstatisticalmeasurescomparingmeasuredandsimu- ME),modelefficiencyandR2betweenmeasuredandsimulatedwater
lated water contents during calibration and validation thus indicate contents,indicatedagoodmatchinalltreatments(Table3).
thatHYDRUS-2Disabletoreliablysimulatewaterdistributionpatterns Despitesomedeviations,simulatedresultshaveshownafairlygood
in the soil under evaluated experimental conditions during the entire agreement with measured water contents in all treatments, as sub-
almondgrowingseason. stantiatedbydifferentstatisticalerrorparameters,themodelefficiency,
and R2 values. Similar deviations between measured and simulated
3.2. Watercontentdistributionsinothertreatments watercontentshavebeenreportedinothersimilarinvestigations(e.g.,
Ramos et al., 2012; Phogat et al., 2012, 2013, 2017; González et al.,
Thecalibratedandvalidatedmodelwasthenusedtosimulatewater 2015). The extent of deviations can be attributed to various causes,
content distributions in other treatments (B, C, D, and E) where irri- possiblyincludingdifferentrootdistributioncharacteristicsduetosoil
gations with recycled water were substituted with less saline water heterogeneity, which was assumed constant in all treatments. Mod-
during different growth stages. Water content distributions in these ifying input parameters and calibrating the model for each treatment
treatments(B,C,D,andE)duringbothseasons(2014–15and2015–16) would probably improve HYDRUS-2D predictions for each treatment.
at selected times are compared in Fig. 4. Water content depth dis- Other causes for deviations between measured and simulated water
tributionsshowedagoodmatchbetweenmeasuredandsimulatedva- contents, which could also be involved in the current investigation,
lues at the bud burst growth stage in treatment B (see the second were explained by Ramos et al. (2012) and are related to field mea-
columninFig.4)andatthepithardeningstageintreatmentC(seethe surements,modelinputs,andmodelstructuralerrors.Additionally,the
thirdcolumninFig.4).However,therewasasmallover-estimationin measurementofsoilwatercontentsbysensorsisalsonotfreeoferrors,
Table3
Statisticalerrors(RMSE,MAE,ME),modelefficiency,andthecoefficientofdetermination(R2)forthecomparisonbetweenmeasuredandsimulatedsoilwatercontentsfordifferent
treatmentsduring2014–15and2015–16.
Treatment Year RMSE MAE ME Modelefficiency R2
(cm3cm−3)
A 2014–15a 0.020–0.036 0.014–0.030 −0.030to0.015 0.79 0.85
2015–16b 0.016–0.038 0.011–0.034 −0.034to0.006 0.78 0.86
B 2014–15b 0.018–0.046 0.013–0.40 −0.040to0.026 0.72 0.73
2015–16b 0.015–0.040 0.012–0.035 −0.032to0.010 0.74 0.80
C 2014–15b 0.015–0.62 0.011–0.058 −0.058to0.016 0.48 0.64
2015–16b 0.017–0.60 0.014–0.056 −0.56to−0.013 0.56 0.80
D 2014–15b 0.018–0.051 0.016–0.045 −0.038to0.020 0.46 0.57
2015–16b 0.014–0.048 0.012–0.039 −0.034to−0.002 0.60 0.74
E 2014–15b 0.017–0.040 0.015–0.036 −0.030to0.018 0.74 0.76
2015–16b 0.025–0.048 0.022–0.037 −0.032to0.019 0.66 0.73
aCalibrationperiod.
bValidationperiod.
75
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
Fig.4.Comparisonofmeasured(M)andHYDRUS-2Dsimulated(S)watercontentdistributionsinthesoilindifferenttreatments(B,C,D,E)atindicatedtimesduringthe2014–15(top
row)and2015–16(bottomrow)almondseason.
due to numerous assumptions and inherent complexities in the soil over-estimationofmeasuredEC valuesatthebudburststagebetween
e
(e.g., Rosenbaum et al., 2010; Ganjegunte et al., 2012; Evett et al., 40and80cmsoildepths.Thereafter,themodel-producedEC dynamics
e
2012),whichmaycontributetoasimilarextentofdeviationsinwater remained within the extent of deviations observed in measured EC
e
contentsasobtainedbymodellingpredictions.Finally,thesoilspatial replicates at all stages of the almond growth. The statistical error
variabilityindifferenttreatmentscouldleadtospatiallydynamicroot parameters,i.e.,RMSE,MAE,andMEestimatedatdifferenttimesbe-
growth,whichisnotconsideredinthepresentmodelinginvestigation. tweenaveragemeasuredandsimulatedEC valuesvariedfrom0.60to
e
0.86, 0.47–0.56, and −0.35–0.49dS/m (Table 4) and matched well
3.3. Modelcalibrationandvalidationforsalinitydynamicsinthesoils with other studies (e.g., Ramos et al., 2012). The model efficiency
(0.98) and R2 (0.73) also indicated a close match between measured
Simulated ECe (S) dynamics in the soil profile under almond is andsimulatedECevaluesduringtheentireseason(2014–15).
compared with measured values at different times at 4 locations (A1, Similarly, during the validation period (2015–16) simulated ECe
A2, A3 and A4) in treatment A during 2014–15 (Fig. 5a–d) and valuesatdifferentdepthsandtimesareincloseagreementwithmea-
2015–16(Fig.5e–h).Duringcalibration(2014–15),therewasaslight sured replicated values except for a small under-estimation in the
Fig.5.Comparisonofmeasured(atdifferentlocationsA1,A2,A3andA4)andHYDRUS-2Dsimulated(S)electricalconductivitiesinthesoilatsaturation(ECe)atindicatedtimesduring
2014–15(a,b,c,andd)and2015–16(e,f,g,andh)underalmondirrigatedwithsalinewater(treatmentA).
76
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
Table4
Statisticalerrors(RMSE,MAE,ME),modelefficiency,andthecoefficientofdetermination(R2)formeasuredandsimulatedECeinthesoilindifferenttreatmentsduring2014–15and
2015–16.
Treatment Year RMSE MAE ME Modelefficiency R2
(dSm−1)
A 2014–15a 0.60–0.86 0.47–0.56 −0.35to0.49 0.98 0.73
2015–16b 0.71–1.09 0.57–0.90 −0.03to0.76 0.49 0.64
B 2014–15b 0.63–1.05 0.53–0.86 −0.66to−0.29 0.44 0.69
2015–16b 0.82–1.27 0.66–1.02 −0.55to0.62 0.32 0.51
C 2014–15b 0.59–0.85 0.44–0.69 −0.25to0.18 0.52 0.57
2015–16b 0.64–1.20 0.46–0.82 −0.72to0.45 0.23 0.37
D 2014–15b 0.64–1.17 0.50–0.87 −0.67to0.12 0.43 0.54
2015–16b 1.13–1.49 0.78–1.22 −1.09to0.69 0.20 0.37
E 2014–15b 0.83–1.26 0.72–1.06 −0.33to0.38 0.38 0.40
2015–16b 0.50–1.60 0.49–1.32 −0.22to1.09 0.08 0.24
aCalibrationperiod.
bValidationperiod.
60–140cm soil depth at the pit hardening stage (Fig. 5e to h). The wider range (0.82–1.27dS/m) in the following season (2015–16). Si-
RMSE,MAE,andMEvaluesbetweenaveragemeasuredandsimulated milarly,MAEvariedfrom0.66-1.02dS/mduring2015–16.Apartfrom
EC valuesatdifferentsoildepthsandgrowthstagesvariedfrom0.71to thesedeviations,themodelefficiencyvalues(0.44and0.32)illustratea
e
1.09,0.57–0.90,and−0.3–0.76dS/m(Table4).However,themodel reasonablepredictionwhileR2values(0.69and0.51)indicatereduced
efficiency (0.49) and R2 (0.64) values were lower compared to the congruencyinthecomparison.Overall,similardeviationsinEC were
e
correspondingvaluedobservedduringcalibration(2014–15),whichis recordedintreatmentC(Table4).However,deviationsbetweenmea-
acommonoccurrence. sured and simulated EC in treatments D and E were comparatively
e
largerascomparedtoothertreatments(A,B,andC)butthevariability
remainedwithinareportedrange(Ramosetal.,2011,2012).
3.4. Soilsalinitydynamicsinothertreatments Deviationsbetweenfield-measuredandsimulatedvariablescanbe
attributed to different causes, including errors related to field mea-
The calibrated and validated model was then used to simulate surements, model inputs, and structural errors (Ramos et al., 2012).
salinity dynamics in the soil in other treatments (Fig. 6). Simulated Ramosetal.(2012)havediscussedextensivelyvariouscausesthatmay
salinity distributions in the soil in different treatments (B to E) were result in deviations between field measurements and model simula-
comparedwithcorrespondingmeasuredECereplicates(B1,B2,andB3 tions.InadditiontovariouscausesoferrorsmentionedinRamosetal.
fortreatmentB;C1,C2,C3,andC4fortreatmentC;andD1,D2,D3, (2012),anadditionalfactorcausingdeviationsinourstudyespecially
and D4 for treatment D) during both seasons. Simulated ECe values forECevalueswasthedestructivesampling.Whilethemodelpredicted
remainedwithinthemeasuredrangeofdifferentreplicatesduringboth thesalinitydynamicsatthesamespot,fieldmeasurementsweremade
seasons. However, small under- or over-estimations also occurred atdifferentlocationseachtime.Moreover,thesoilsattheexperimental
duringsometimeindifferenttreatments.Amongtheerrorestimatesin site have immense spatial heterogeneity (nine different textures at
treatmentB,RMSEvariedfrom0.63-1.05dS/mandMAErangedfrom differentdepths;Table2),whichhadastrongimpactonthemeasured
0.53-0.86dS/m during 2014–15 (Table 4). However, RMSEshowed a
Fig.6.ComparisonofmeasuredandHYDRUS-2Dsimulated(S)electricalconductivitiesinthesoilatsaturation(ECe)indifferenttreatments(B,C,D,E)atindicatedtimesduringthe
2014–15(toprow)and2015–16(bottomrow)almondseasons.
77
V.Phogatetal. Agricultural Water Management 201 (2018) 70–82
ECevaluesasshownindifferentreplicatesintreatmentsA,B,C,andD. Table5
IntreatmentE,EC measurementsweremadeonlyatonelocationand Seasonalwaterbalancecomponentsforalmondindifferenttreatments(A,B,C,D,andE)
e
hence,therewasawidervariationamongmeasuredandsimulatedEC during 2014–15 and 2015–16. Seasonal evapotranspiration, drainage and depletion/
e storage data was evolved from modelling simulations for different treatments. Water
values.TheassumptionofasimilarconversionfactorforEC andEC
e 1:5 balanceerrorrepresentstheextentoferrorbetweensource(irrigation,rainfall),sink
forbothyearshavealsocontributedtotheextentofvariabilityines- (Evapotranspiration,drainage),andsoildepletion/storagedata.Depletionisconsidered
timatedEC values. assourcetermwhenfinalwetnessinthesoilwaslowerthantheinitialwetnessandsink
e
The model input parameters, especially soil hydraulic parameters (storage)termifviceversa.
andsolutedispersivities,mayalsointroducehugeerrorsifnotproperly
Treatments Components 2014–15 2015–16
measured at the field site. In our study, these parameters were taken
from other local studies or from the literature, which may have in- (mm) (%)# (mm) (%)#
troducederrorsinsimulatedvalueseventhoughthemodelwasprop-
erlycalibratedandvalidatedforthefieldsite.Calibrationandvalida- A Irrigation 889.4 71.3 960.4 76.8
Rainfall 326.9 26.2 372.7 29.8
tionofthemodelonlyforonetreatmentcouldalsoresultinerrorsin Soildepletion(+)/storage(−) 30.4 2.4 −81.8 −6.5
simulated values. In spite of these deviations between measured and Evapotranspiration(ETC) 849.9 68.2 868.2 69.4
simulated EC values, the error estimates are lower than reported in Drainage 387.8 31.1 376.9 30.1
e
otherstudies(Ramosetal.,2011,2012). Waterbalanceerror* 9 0.7 6.2 0.5
B Irrigation 889.4 72.2 960.4 76.5
Rainfall 326.9 26.5 372.7 29.7
Soildepletion(+)/storage(−) 15.3 1.2 −78.2 −6.2
3.5. Seasonalwaterbalance
Evapotranspiration(ETC) 878.6 71.3 892.8 71.1
Drainage 359.3 29.2 352.1 28.1
Thesimulatedwaterbalancerevealedthatevapotranspirationlosses Waterbalanceerror* −6.3 −0.5 10 0.8
(ET )byalmondsunderrecycledwaterirrigations(treatmentA)during C Irrigation 889.4 74.2 960.4 75.6
201C4–15amountedto849.9mm,whichwas68.2%ofthetotalapplied RSoaiilndfaelplletion(+)/storage(−) −32168.9.29 −271.3.5 −37622.7.1 −294.3.9
water (rainfall+irrigation). Similarly, during 2015–16, although the Evapotranspiration(ETC) 880.9 73.5 923.5 72.7
seasonalETC(868.2mm)wasslightlyhigher,thedrainagewasreduced Drainage 303.8 25.4 321.1 25.3
by1%comparedtothepreviousseason.Theerrorinthewaterbalance Waterbalanceerror* 13.31 1.1 26.4 2.1
simulationsinAwas0.7and0.5%during2014–15and2015–16sea- D Irrigation 889.4 73.2 960.4 76.7
Rainfall 326.9 26.9 372.7 29.8
sons, respectively, which is not uncommon in such situations. Deep Soildepletion(+)/storage(−) −2.1 −0.2 −80.9 −6.5
drainageintreatmentArepresented31.1%ofthetotalappliedwater, Evapotranspiration(ETC) 859.8 70.8 877.2 70.1
whichisanimportantamountofwaterthathelpsintransportingsalts Drainage 361.8 29.8 367.7 29.4
outoftherootzone. Waterbalanceerror* −7.4 −0.6 7.3 0.6
Monthlydrainagefluxesbelowthe2-msoildepthunderthealmond E Irrigation 889.4 74.9 960.4 76.2
Rainfall 326.9 27.5 372.7 29.6
tree irrigated with saline water (treatment A) during calibration Soildepletion(+)/storage(−) −28.1 −2.4 −72.6 −5.8
(2014–15) and validation (2015–16) is shown in Fig. 7. Large deep Evapotranspiration(ETC) 941.1 79.2 954.7 75.7
drainage (80mm) during July 2014 is the result of a low water re- Drainage 265.6 22.4 289.2 22.9
Waterbalanceerror* −18.5 −1.6 16.6 1.3
quirement by the tree due to its dormancy and the early bud sprout
stage coupled with high late winter rains. Deep drainage in other
*is(∑source−∑sink).
months during the first season (2014–15) varied between #represents%ofsource(irrigation+rainfall±depletion/storage).
13.7–39.2mm.DeepdrainageduringSeptemberandOctober2014was
very low because of a high water requirement by the almond tree received during the same period of the 2014–15 season. This kind of
during this period of profuse vegetative growth. Similarly, during rainfallduringthedormantperiodcouldbehelpfulindrivingsaltsout
2015–16 more deep drainage (57.7mm) occurred during the early oftherootzone.Thissuggeststhatthepresenceofhigheramountsof
season (August2015) ascomparedto mid-andlate-period.However, saltsinirrigationwaternotonlyreducedwateruptakebyalmondtrees
enormousdeepdrainageoccurredduringMayandJunemonthsofthe butitmayhavealsoencouragedgreaterhydraulicflowthroughthesoil
second season as compared to 2014–15. This was ascribed to higher profile. A similar monthly drainage pattern was observed in other
rain during May-June 2016, which was almost double the amount treatmentaswell.
SimulatedseasonalET ofalmondintreatmentBwasincreasedby
C
3.4and2.8%,during2014–15and2015–16,respectively,ascompared
tothecorrespondingvaluesintreatmentA(Table5).Thissuggeststhat
switchingtoirrigationwithnon-salinewaterduringthebudburstand
pithardeningstagesexposedthealmondtreetolessosmoticpressurein
therootzoneandincreasedrootwateruptake,therebyincreasingthe
water consumption. As a consequence, this favorable impact also re-
duced the drainage component by 6–7% compared to treatment A
(Table5).IntreatmentC,wherelesssalineirrigationwasusedbetween
thepithardeningandharveststages,seasonalET ofalmondincreased
C
ascomparedto thecontrol treatment(treatment A).Theextent ofan
increaseinET wasalsohigherthanintreatmentB,especiallyduring
C
2015–16.SeasonalET increasedby3.6and6.4%duringthe2014–15
C
and2015–16seasons,respectively,ascomparedtotreatmentA.Onthe
otherhand,seasonaldrainageintreatmentCwasreducedby21.7and
14.8% during 2014–15 and 2015–16, respectively, as compared to
treatmentA,irrigatedwithsalinewaterthroughouttheentireseason.
This suggests that the impact of switching to less saline irrigation on
Fig.7.Simulatedmonthlydrainage(belowthe2-mdepth)underalmondirrigatedwith water uptake by almond was slightly more effective during the mid-
salinewater(treatmentA)during(a)2014–15and(b)2015–16. stage of almond growth (treatmentC) compared to the initial period
78