PFG 2014 / 5, 0369–0381 Article Stuttgart, October 2014

CropWaterRequirementsonRegionalLevelusingRemote Sensing Data – ACaseStudyintheMarchfeldRegion

NIKOLAUS NEUGEBAUER &FRANCESCO VUOLO, ,

Keywords: irrigation, crop water requirements, evapotranspiration, satellite image, remote sensing, DEIMOS-1

Summary: This research presents an operative ap- Zusammenfassung: Berechnung des Pflanzen- plication of satellite–basedtechnologiestoesti- wasserbedarfs für Sommerfeldfrüchte mittels Fer- mate crop water requirements (CWR) on regional nerkundungsdaten. Eine Fallstudie in der March- level. The study is located in the agricultural area feld-Region. Diese Arbeit präsentiert eine Berech- ofMarchfeld(LowerAustria)andfocusesonthe nung des Pflanzenwasserbedarfes auf regionaler year2012.Agriculturalproductioninthisregion Ebene.DieStudiewirdinnerhalbderlandwirt- requires irrigation which draws its resources from schaftlichen Region Marchfeld (Österreich) fürdas thegroundwater.Duetoitsimpactonhydrology Jahr 2012 durchgeführt. Die landwirtschaftliche and environment this practice necessitates close Produktion dieser Region ist definiert durch Feld- monitoring and management. In this process the bewässerung,dieihreRessourcenausdemGrund- CWRisanimportantpieceofinformation,which wasser bezieht. Der Einfluss auf die hydrologi- can be derived from satellite data. We used multi- schen- und Umweltbedingungen macht eine genaue temporal (6 acquisitions, with an average revisit Beobachtung dieser Anwendung unumgänglich. timeof16days)andmulti-spectral(visibleand Der Pflanzenwasserbedarf ist dabei eine wichtige near-infrared) imagery of the DEIMOS-1 satellite Information, sowohl für die Beobachtung als auch sensor data in combination with local agro-meteo- für operationelle Entscheidungen. Die Analyse ba- rologicalmeasurements.Inafirststep,theleafarea siert auf multi-temporalen (6 Bildaufnahmen, 16 index(LAI)andalbedomapswerederivedfrom Tage Wiederholzyklus), multispektralen (sichtba- atmospherically corrected satellite data. Those res Licht und nahes Infrarot) Satellitenbildaufnah- were used to generate maps of potential evapotran- men des DEIMOS-1 Sensors und lokalen agro-me- spiration (ETp)bydirectapplicationofthePenman- teorologischen Daten. Zuerst wurde der Blattflä- Monteith equation. Then, we produced a crop mask chenindex (Leaf Area Index – LAI) und die Albedo of the areas potentially irrigated. This was achieved aus den atmosphärisch korrigierten Satellitenbil- byusinganunsupervisedimageclassificationap- dernberechnet.MitdiesenDatenwurdendurch proachincombinationwithapost-classification eine direkte Anwendung der Penman-Monteith analysis of the time-series of the crop development. Gleichung Karten der potentiellen Evapotranspira- Rain gauges data from a number of weather sta- tion erstellt. Mit einer unüberwachten Klassifikati- tions were spatially interpolated to produce a map on der zeitlichen Pflanzenentwicklung und einer of precipitation (P). Finally, the crop water require- nachfolgenden Analyse wurde die räumliche Ver- ments were calculated as the difference between teilung von Sommerfeldfrüchten ermittelt. Weiters

ETp andtheprecipitationandaggregatedover10- wurden Niederschlagsdaten von elf Wetterstatio- day interval periods. Reference evapotranspiration nen interpoliert, um eine räumliche Abschätzung

(ETo)forthegrowingperiodresultedin391mm. der Niederschlagsverteilung zu erhalten. Zuletzt The extent of irrigated crops covered an area of wurde der Wasserbedarf der Sommerfeldfrüchte

21,278 ha with a maximum ETp of 5.6 mm/day at als Differenz der potentiellen Evapotranspiration theendofJune.Thetotalcropwaterrequirement und des Niederschlages berechnet und in 10-Tages- resulted in 34.26 million m³ for the year 2012. Intervallen aggregiert. Das Untersuchungsgebiet wies während der Vegetationsperiode eine totale

ET0 von391mmauf.DieProduktionvonSommer- feldfrüchten nahm eine Fläche von 21.278 ha in

Anspruch und erreichte eine maximale ETp von 5,6 mm/Tag. Der gesamte Pflanzenwasserbedarf für Sommerfeldfrüchte innerhalb der Marchfeld- Region im Jahr 2012 war 34,26 Mio. m3.

© 2014 E. Schweizerbart'sche Verlagsbuchhandlung, Stuttgart, Germany www.schweizerbart.de DOI: 10.1127/1432-8364/2014/0230 1432-8364/14/0230 $ 3.25 370 Photogrammetrie • Fernerkundung • Geoinformation 5/2014

1Introduction In the Marchfeld region the withdrawal of waterforirrigationiscurrentlyapproximated Irrigated agricultural land has a high produc- by measurements of the groundwater level. tive capacity producing between 30% to 50% Field to catchment scale estimates of water re- of the world food crop on 17% of all arable quirements are important in the assessment of land (SECKLER etal.1999).Efficientmanage- regional water consumption and groundwater mentofwaterresourcesisofparticularimpor- depletionandcouldhelptoimproveestimates tance considering that 70% of global freshwa- of irrigation withdrawal. This information can ter withdraws are done by agriculture. The last be used to coordinate allocation of resources decadeshaveseenagrowthinuseofwater in irrigation, like volumes of water and invest- forirrigation,whichispredictedtofurtherin- mentsininfrastructure,andinthecompila- crease (COLLINS et al. 2009). Adjacent to nutri- tion of water management plans to achieve an ent management, efficient water management increasedefficiencyinthewateruse(PEREIRA is considered to be the main component that et al. 2002). cancontributetoproductionincreases(45% Crop development is one of the driving fac- to 70% for most crops) thus providing future tors determining water needs and it highly foodsecurityinasustainableway(MUELLER dependsonspatialandtemporalfactors.A et al. 2012). temporal variation is a result of vegetation dy- The need for irrigation arises when pre- namics over the growing season. Spatial vari- cipitation does not compensate the water loss ation is related to climatic conditions on large caused by evapotranspiration (ET) of the plant scale,andonsmallerscalerelatedtoother surface. Usually, farmers and irrigation man- factorslikesoiltypeandfertility,cropvari- agers will aim to meet crop water require- ety and management. Estimations of water re- ment (CWR), defined as the deficit between quirements based on regional statistics, static ET loss and precipitation, by applying irriga- crop maps and standard crop coefficients are tion to achieve optimal crop development and not capable of accounting for this variability yield (ALLEN et al. 1998). Water is taken from (FRENKEN &GILLET 2012). Therefore, precise availablesources,e.g.riversandgroundwa- estimations of water requirements need to ter, and extensive infrastructure, such as ca- takeintoaccountspatialandtemporalinfor- nals, damns, water reservoirs, pumps, is need- mation on the actual development of crops. ed to make water accessible for irrigation. An Crop development can be monitored using example is the Marchfeld region in Austria multi-spectral and multi-temporal observa- whereirrigationisdeployedsincethe1970s. tions from satellite platforms (OZDOGAN et al. Betweentheyears1986and2004the“March- 2010).Currentplatformsareacquiringmulti- feldkanal-Projekt” wasrealizedinresponseto spectralimageryinaspatialresolutionpre- dropping groundwater levels. This canal sys- cise enough for analysis of agricultural veg- tem diverts water from the Danube River and etationwithsuitablerevisitintervals.Forin- conveysittotheMarchfeldregiontomakeit stance, the commercial Spanish satellite of the accessible for irrigation practice. The method DMC (disaster monitoring constellation) DEI- canresultinconsiderableeffectsonthesur- MOS-1providesdatain3spectralbandswith rounding environment like biodiversity loss- agroundsamplingdistance(GSD)of22m es due to lowered groundwater tables or pol- and 2 – 3daysrevisittime(DEIMOS-IMAGING lution of rivers by agricultural inputs (DAV I D 2014). The freely available Landsat-8 satellite etal.2000).Furtherproblemsarelimitedwa- featuring 8 spectral bands provides 16 days tersupplyfordownstreamusers,soilerosion revisittimeandaGSDof30m.Sentinel-2, and nutrient elution (SHANAN 1987, DAV I D et al. an upcoming European Space Agency (ESA) 2000). In the Marchfeld the main issues are mission, will provide 13 spectral bands with theregulationofgroundwaterlevelsandthe aresolutionofupto10minarevisitinterval dilution of nitrate into the groundwater. These of 2 – 3 days. These satellite sensors offer the problems deserve special recognition since possibilitytomonitorlargecropareasininter- theresourcesarealsousedasindustrial,do- vals (7 – 10days)suitableforthecalculationof mestic and drinking water. CWRinavarietyofagriculturalconditions. N. Neugebauer & F. Vuolo, Crop Water Requirements 371

OtherpossibleapplicationsofEOdatainag- al. 2008, EITZINGER et al. 2009, TRNKA et al. riculturearetheimplementationincropmod- 2009). The high average wind speed of about els (CASA etal.2012),monitoringofthenitro- 3.5m/shasanamplifyingeffectofthedrycli- gen status (SCHARF et al. 2002) and to predict mate considering plant transpiration. The ag- yields (JÉGO et al. 2012). riculturalareahasanextentof60,000haof

This paper focuses on the calculation of ETp arable land of which more than 21,000 ha are and an estimation of the seasonal crop water regularly irrigated each year. The soil consists requirements in the area of Marchfeld, an im- mainly of fertile aeolian slit deposit which, to- portant agricultural region located in Lower gether with the pannonian climate, high solar Austria, where irrigation is regularly applied radiation and flat terrain forms a well-quali- during summer months. Our approach com- fied region for agricultural production. The bines daily agro-meteorological data from lo- main crops cultivated in summer are vegeta- calstationsandtheprocessingofmulti-spec- bles(11%ofthecroparea),sugarbeet(10%) tralimagesbasedonthemethodologyde- and potatoes (7%). Cereal crop types are cul- scribed by D’URSO et al. (2001, 2010) using tivatedonabouttwo-thirdsofthecroparea. thePenman-Monteithapproach(ALLEN et al. Limitations to agricultural production are low 1998). We used rain gauges data from eleven precipitation and a predominantly low field weather stations located in the region and 6 watercapacityof70mmandless.Thismeans images with a spatial resolution of 22 m ac- thatthesoilcanholdonlyalowamountofwa- quiredwithanaveragerevisittimeof16days. ter thus limiting availability to plants in dry We performed an image classification using periods. the satellite-based crop development curves as AlthoughMarchfeldisoneofthedriestre- input feature to differentiate different types of gions in the country, it is still one of the pri- crops. This was done to identify areas used for mary producers of agricultural goods in Aus- productionofsummercrops,whicharemost tria.Thisispossibleduetotheuseofirriga- likelytobeirrigatedduringthetimeofinter- tion that is used to compensate scarce pre- est. cipitation. The infrastructure of the “March- Thepaperpresentstheresultsofacase feldkanal Projekt” consists of 100 km of ca- study and discusses the findings in the context naland8weirswhichcontrolsurfacewater of operative tools to support irrigation water flow and height. To control groundwater- lev management at regional scale. els 22 pumping stations and 7 infiltration basinswereimplemented.Thecostofcon- struction was 207.8 million Euro (NEUDOR- 2Material FER &WEYERMAYER 2007). Water for applica- tion on field is pumped from the groundwater 2.1 Study Area throughwells.Inthiscasedecentralized,mo- bilepumpsarerunondieselorlessoftenby Marchfeld is located in , north electricity. In some cases centralized pumping east of Vienna (Fig. 1). The average annual stations provide pressurized water via a distri- precipitation is 500 mm – 550 mm making it butionnetwork.Themajorityofcropsareir- thedriestregionofAustria.Annualprecipita- rigated from May until end of September. The tion during the vegetation period (April – Sep- only crop regularly irrigated afterwards (until tember) is 200 mm – 440 mm. Dry periods mid-October) is spinach. (timewithnodailyprecipitationhigherthan 5 mm) of three weeks can occur averagely 5 times per year, longer dry periods of 30 – 2.2 Satellite Data 34daysareayearlyoccurrence(NEUDORFER 2002). Modelling of future climatic develop- In this study, we used satellite images acquired mentforMarchfeldanditssurroundingre- by DEIMOS-1, a satellite of the DMC (dis- gions suggests a decrease of precipitation vol- aster monitoring constellation). DEIMOS-1 umesandanexpectedincreaseoflikelihood features a multi-spectral sensor providing for drought events (PAL 2004, DUBROVSKY et 22 m resolution in 3 bands (NIR: 0.77 μm – 372 Photogrammetrie • Fernerkundung • Geoinformation 5/2014

0.90 μm, Red: 0.63 μm – 0.69 μm, Green: Theimagerywasprovidedorthorectifiedto 0.52 μm – 0.60 μm) with a revisit interval of a sub-pixel accuracy of about 10 m. The cor- 2 – 3 days. DEIMOS-1 was chosen because rectionwasperformedbythedatasupplierus- it offers a cost-effective solution to monitor inggroundcontrolpointsandtheshuttleradar crop development with suitable temporal, spa- topographymission(SRTMv3)dataasdigital tialandspectralresolutiontoworkatfieldand elevation model. An atmospheric correction districtscales.Imagerywasacquiredonsix using ATCOR (RICHTER 1998) was performed different dates in 2012 to cover the growing by the authors before further processing. season:June17th,July30th,August1st and 20th, September 5th and 18th.

Fig. 1: Study area of Marchfeld (Lower Austria) and location of the weather stations used for the spatial interpolation of precipitation. N. Neugebauer & F. Vuolo, Crop Water Requirements 373

2.3 Agro-Meteorological Data Precipitation(P)canbemeasuredfromlo- calweatherstations.Acriticalfactorinthe Precipitationwasrecordedatelevenweath- precise estimation of CWR is the potential er stations in the region (Fig. 1). Fig. 2 re- evapotranspiration of crops. A review of dif- ports the mean cumulated precipitation vol- ferent procedures for the determination of ET umesofallelevenstationsandthedatafor is given by COURAULT et al. (2005) and VER- thestationswhichrecordedtheseasonalmini- STRAETEN et al. (2008). The approach for map- mum (148 mm at Zwerndorf) and maximum pingETusedinthisstudyisbasedonthe (249 mm at Wien-Unterlaa) values. For the methodology proposed by D’URSO et al. (2001, calculation of the reference evapotranspira- 2010)usingtheFAO-56Penman-Monteith equation (ALLEN etal.1998).Itconsistsoftwo tion (ET0), we considered solar radiation, air temperature, windspeed and air humidity re- maincomponents.Firstweestimatetheref- cordedattheZwerndorfstationonly. erence evapotranspiration (ET0)basedonthe FAO-56 standards. ET 0 indicates the amount ofwaterremovedfromastandardvegetation 3Methods surface (grass) through the process of evap- otranspiration. In this case, ET0 isafunc- 3.1 CalculationofCropWater tion of meteorological factors (solar radia- Requirements tion,windspeed,temperature,airhumidity), which can be measured by local weather sta- Crop water requirements are defined as the po- tions, and crop specific factors (LAI, albedo andcropheight– h ), which are kept constant tential evapotranspiration (ETp)ofacropped c surfacearea,netoftheprecipitation(P): in relation to the theoretical reference surface (LAI=2.88,albedo=0.23,h =0.12m). – P c CWR = ETp (1)

Fig. 2: Cumulated precipitation. The black line represents the mean value from all eleven weather stations, the shaded area is the standard deviation. Wien-Unterlaa recorded the total maximum of precipitation volumes (249 mm), Zwerndorf the minimum (148 mm). 374 Photogrammetrie • Fernerkundung • Geoinformation 5/2014

RICHTER &SCHLÄPFER 900 tion ( 2014). This calcu- 0.408Δ−+ (RG )γ uee ( − ) nsa+ 2 lation was performed using ATCOR module ET = T 273 0 Δ+γ + u implemented in ERDAS Imagine. Crop height (1 0.34) 2 (2) (hc),whichcontrolstheresistanceincanopy aerodynamic properties, was set to a constant Rn =netradiationatthecropsurface value of 0.3 m, representing an average height [MJ m-2 day-1], for crops in the region. This was done because G=soilheatfluxdensity[MJm-2 day-1], measurements of crop heights are not feasible T=meandailyairtemperatureat2m forlargeareasoveratimeperiodofseveral height [°C], monthsandafixedvalueisconsideredtobea -1 u2 = wind speed at 2m height [m s ], satisfactory compromise for the estimation of e =saturationvapourpressure[kPa], D’URSO &CALERA BELMONTE s ETp ( 2006). ea =actualvapourpressure[kPa], To produce daily estimates of ET ,K maps – p c es ea =vapourpressuredeficit[kPa], werederivedforeachacquisitiondate(ac- Δ = slope vapour pressure curve [kPa °C-1], cordingto(2))andtheyweretemporallyin- γ=psychrometricconstant[kPa°C-1]. terpolated to obtain daily Kc values from June 17th to September 18th.Forthispurpose,we ” WHITTAKER Subsequentlythecropcoefficient(Kc)is used the “Whittaker Smoother ( EILERS ATKINSON calculated. Kc is defined as: 1922, 2003, et al. 2012). The smoothingparameter(λ)wassetto15.Sub- ETp sequently, a daily estimation of ET was per- K = (3) p c formed by multiplying daily K with corre- ET0 c sponding daily ET0 estimates. In a final step Traditionally, i.e. in non-satellite-based ap- the total water requirements were aggregated proaches, Kc is calculated from measurements in 10-day time steps. of actual evapotranspiration and contempora- neous ET0 estimates. This way Kc has been cal- culated for different conditions and is report- 3.2 Spatial Interpolation of ed in tables for several crops. Tabulated data Precipitation represents a fixed value suitable for compar- ingcroptypesundergeneralconditions.Be- To derive the variability of precipitation causecropdevelopmentissubjecttochange events over the region of interest, rain gauges in regard to management and environmental data were spatially interpolated on a daily ba- factors, the Kc valueshouldbedynamically sis using the inverse distance weighted (IDW) adjusted to the actual conditions in order to method.IDWinterpolationisawellperform- achieve precise estimation of water require- ing method for interpolating precipitation and mentonfieldandregionalscale. less complex to apply than the similar per- Inthisstudy,weusedtheapproachpro- formingregression-krigingmethod(WAGNER posed by D’URSO et al. (2001, 2010) to estimate etal.2012).TheIDWalgorithmwasapplied FATICHI ETp, where crop parameters used in the FAO- in MATLAB using a code provided by 56 Penman-Monteith equation are replaced (2012). The interpolation was set to take into by LAI and albedo maps. LAI was calculated accountthe3nearestneighboursofeachpoint from the weighted difference vegetation in- of measurements to produce a 22 m × 22 m CLEV- dex(WDVI)usingtheCLAIRmodel( grid corresponding to the Kc raster dataset. ERS 1989).Asitespecificmodelwascalibrat- ed and the satellite-based estimation of LAI was validated using ground measurements. 3.3 Crop Category Classification The calibration procedure and results are pre- sented in VUOLO et al. (2013). The surface al- The aim of the crop category classifica- bedo is substituted by the wavelength-inte- tionwastoidentifythecropareasthatare grated ground reflectance where wavelength potentiallyirrigated.Forthispurpose,wede- gap regions are supplemented with interpola- fined two crop categories (summer and win- N. Neugebauer & F. Vuolo, Crop Water Requirements 375 ter crops) and used the summer crop areas to to September 18th 2012 is referred to as the ” computethetotalwaterrequirements.TheKc “growing season . estimation from the satellite time series pro- vided a quantification of the crop development overtimewhichwasusedasaninputfeature 4.1 Crop Area Estimation in the image classification. In our case, the timeseriesstartingonJune17th covered the Resultsofthemulti-temporalclassificationin- development of winter crops in their late or se- dicatedthatthetotalareausedforcroppro- nescencestages.Likewisesummercropswere ductioninMarchfeldwas60,691ha.Summer recorded during their full development stage. crop cultivation accounted for 35% (21,278 ha) Inafirststep,non-agriculturalareassuchas ofthetotalcroparea.Cultivationofsecond- urban, forest and water were identified using ary and winter crops accounted for 17,149 ha CORINE land cover (CLC) with a pixel size of and22,248ha,respectively.Anaccuracyas- 100 m (EUROPEAN ENVIRONMENT AGENCY 2006) sessment of the crop map was performed us- andwereremovedfromtheKc dataset. ing 140 random sampling points, which were In a second step, an unsupervised image visuallyinterpretedusingthecropdevelop- classificationofthetimeserieswithanini- mentandcolourcomposites.Theclass“sum- tial number of 12 clusters was performed. The mer crop” reachedanaccuracyof90.5%.An 12 clusters were then aggregated to summer errormatrixforthisclassisgiveninTab.1. and winter crop classes based on a visual in- Thecropareasandtheintensityoflanduse terpretation of the temporal crop coefficient forsummercropswerefurtheranalyzedbased developmentofeachcluster.Weintroduceda ontheboundariesofadministrativedistricts thirdclass(namelysecondarycrop)toaccount (Fig.3).TheregionAndlersdorfdedicated fordoublecroppingsystems,i.e.twoormore most of its area to summer crop production cropsinthesamefieldduringasinglegrow- (43%ofthedistrictarea,231ha).Themini- ing season. mumof5%(61ha)wassituatedinStrasshof, adistrictconsistingmainlyofurbanareas. ThedistrictofGroß-Enzersdorfhadthelarg- 4 Results est total area of summer crop production with 2,245 ha (29% of the area). The per- First we present the results of the summer crop centageoflanduseforsummercropcultiva- area classification. This is followed by the re- tion within Marchfeld for each administrative sultsofanalysisoftheagro-meteorological district is given in Tab. 2. data.TheCWRispresentedattheendofthis section. ETp and CWR estimates were calcu- lated considering summer crops only, because 4.2 Analysis of Agro-Meteorological of the overlap between their development and Data theirrigationseason.Thesewerecalculated onadailybasisandtemporallyaggregatedto The average daily temperature recorded 10-day intervals. For ease of discussion, the at Zwerndorf weather station ranged from considered reference period from June 17th 12.6 °Cto29.8°C during the growing season.

Tab. 1: Error matrix of the crop mask classification.

Reference data Summer Other User’s accuracy Summer 38 4 90.5 % Classification result Other 410196.2% Producer’s accuracy 90.5 % 96.2 % Overall accuracy 94.6 % 376 Photogrammetrie • Fernerkundung • Geoinformation 5/2014

Tab. 2: Percentages of summer crop area to The average wind speed at 2 m height was district area. 2.7m/s(σ=±0.8m/s). The reference ET reached a maximum val- Area of summer crop ueof6.1mm/dayattheendofJune.Themini- Municipality cultivation (%) mumof3.2mm/dayoccurredatthebeginning of September. The total ET0 over the growing 27.2 seasonwas391mm(94days).Fig.3reports 43.1 theweeklyaveragesofET0 in 2012 in relation an der March 16.1 totheaveragesoftheyears2009to2011. From the eleven stations, the highest amount 21.8 of precipitation during the growing season was Bockfließ 19.2 recordedinWien-Unterlaawith249mm.The Deutsch-Wagram 20.5 lowest volume was 148 mm at Zwerndorf. The interpolated measures of precipitations were 23.0 aggregated to 10-day intervals and are present- 27.3 ed in Tab. 3. Records of precipitation showed Gänserndorf 21.5 an expected eastward downtrend. Stations in ä 37.6 theeasternpartofthestudyarea,i.e.G nsern- dorf, Zwerndorf, Bad Deutsch-Altenburg and Großebersdorf 24.5 Bruck-Neudorf, recorded lower amounts of Groß-Engersdorf 30.3 precipitation with a mean (μ) of 163 mm and Groß-Enzersdorf 29.4 astandarddeviation(σ)of±16mmcompared tostationsinthewesternpart(μ=222mm Großhofen 29.0 andσ=±16mm).Acomparisonofaverage Groß-Schweinbarth 12.8 weekly values between 2012 and the years 33.3 from 2009 to 2011 is given in Fig. 4. 28.3 Leopoldsdorf im Marchfelde 33.4 4.3 Crop Water Requirements 31.5 Themaximumwaterrequirementsforthe 28.5 growingseasonconsideringatimeframeof94 15.0 days resulted in 34.26 million m3,whichcor- -Raggendorf 12.9 responds to an average requirement of 3.8 mil- lion m3 per 10-days (σ = ± 4.4 million m3). The 35.2 aggregationto10-dayintervalspresenteda 22.9 maximum water requirement of 7.9 million m3 36.7 in mid-August. Precipitation exceeded poten- Pillichsdorf 25.8 tial evapotranspiration during two of the 10- day periods. In this case, the surplus of water 16.5 resulted in 4.7 million m3 and 0.6 million m3 23.1 attheendofJulyandinmid-September,re- Schönkirchen-Reyersdorf 21.9 spectively. Strasshof an der Nordbahn 5.7 Water requirements per district 25.6 23.7 The satellite-based calculation of water re- quirements was spatially aggregated to dis- 24.0 trictlevelandnormalizedtotheextentofthe Wolkersdorf im 16.0 summer crop area per district to represent the spatial distribution of CWR. On average, the seasonal water requirement per district N. Neugebauer & F. Vuolo, Crop Water Requirements 377

Fig. 3: Comparison between year 2012 and the average values measured in the last three years – (2009 2011) for ET0.Theerrorbarsindicatethestandarddeviation(1time)withinthefouryears (Zwerndorf).

Fig. 4: Comparison between year 2012 and the average values measured in the last three years (2009 – 2011)fortotalrainfall.Theerrorbarsindicatethestandarddeviation(1time)withinthefour years (Zwerndorf).

Tab. 3: Calculated values of reference evapotranspiration (ET0), potential evapotranspiration of th summer crops (ETp),precipitationdataandwaterrequirementsfortheyear2012from17 of June to 18th of September; dates in the first column indicate starting points for 10-day intervals. Mean values (μ) and standard deviations (σ) refer to the spatial variation within the region of interest.

Water requirements of summer ET ET Precipitation 0 p crops (mm) (mm) (mm) (million m3) μσ μ σ June 17th 52.1 50.6 8.9 14.6 2.8 7.66 June 28th 61.4 56.0 13.6 19.4 5.0 7.87 July 9th 41.8 36.8 9.4 30.5 5.9 1.38 July 20th 36.6 32.0 7.7 55.0 14.1 -4.77 July 31st 47.7 42.8 10.0 13.4 4.8 6.08 Aug. 11th 42.9 39.8 10.1 2.3 1.2 7.95 Aug. 21st 43.1 38.4 11.5 2.1 1.2 7.75 Sep. 1st 32.2 23.7 9.2 19.0 4.5 0.98 Sep. 11th 33.2 20.4 9.9 24.0 2.2 -0.64 Overall 391 34.26 378 Photogrammetrie • Fernerkundung • Geoinformation 5/2014 resulted in 1,523 m3/ha(σ=±288.4m3/ha). rigation period in the agricultural region of ThemaximumCWRwasobservedinthedis- Marchfeld (Austria) for the year 2012. Refer- trictofWeidenanderMarchwithaCWRof ence evapotranspiration was calculated using 1943 m3/ha. The minimum CWR of 968 m3/ha theFAO-56Penman-Monteithequation(AL- was observed in Pillichsdorf. Fig. 5 shows the LEN etal.1998)andmeteorologicaldatare- spatial distribution of CWR. corded on-site. For a spatial estimation of pre- cipitation events data from eleven rain gauges in the region of interest were interpolated. A 5 Discussion and Conclusion time series of DEIMOS-1 imagery was ac- quired to follow the crop development during This study described an application of satel- thegrowingseason.DerivedLAIandalbedo lite-based technologies combined with agro- maps were used to calculate the crop coeffi- meteorological data and spatial modelling to cient maps based on an approach proposed estimate water requirements during the ir- by D’URSO etal.(2001,2010).Thecropco-

Fig. 5: Crop water requirements (m3/ha) of summer crops per district in Marchfeld for the growing season 2012. N. Neugebauer & F. Vuolo, Crop Water Requirements 379 efficient maps were temporally interpolated vide an important base for further decision to derive daily estimations of crop develop- making processes. Closer examination of the ment. Subsequently reference evapotranspi- water deficit at field level can help to develop ration and satellite-based Kc data were used amorepreciseframeworkfortheimplementa- to calculate potential evapotranspiration on a tion and regulation of water management prac- pixel-basis. The satellite time series was fur- tices. Current assessments of this variable in ther used to classify the agricultural surface in theMarchfeldregionarebasedonlocalpoint winter or summer crops. The land cover map measurements and are thus fixed to small scale and the pixel-based potential evapotranspira- calculations (CEPUDER &NOLZ 2007). tion were used to calculate the total water re- In his publication about “Accounting for quirements of summer crops. Water Use and Productivity” MOLDEN (2006) Results indicated that the total area extent identifies three levels of water use: “[…] a use of summer crop production in the year 2012 level such as an irrigated field or household, was21,278hawithatotalpotentialevapo- aservicelevelsuchasanirrigationorwater transpirationof340mm.Thetotalprecipita- supply system, and a water basin level that tion for the observed time period was 180 mm may include several uses”. EO data can help to calculated as the average within the study area bridge the gaps between these levels since wa- resulting in total water requirements for sum- ter deficits can be assessed in detail on large mer crops of 34.26 million m³ (in 94 days; scaleinrelativelyshorttimesteps. 368,429 m3/dayonaverage).Thisrepresents Estimations of regional water require- themaximumlevelofwaterneedsinthere- ments similar to the presented research were gion for the year 2012. performed in a recent study by AKDIM et al. Contemporaneous measurements of the ac- (2014) for the Doukkala area in western Mo- tual water withdraws were not available for the rocco.Inadditiontotheregionalcalculation same reference period. A coarse comparison is of water deficit this publication further sur- only possible considering the aggregated data veys the water withdrawal for irrigation by perseasonandtheestimationsofgroundwa- measurements at central pumps. In doing so, ter withdrawal from the “Marchfeldkanal Be- the authors identified times of water deficits, triebsgesellschaft”, the company responsible theappliedirrigationandtheoccurrenceof forthecanalinfrastructure. Theestimationis mismatchesbetweenthetwo.ForMarchfeld basedonmeasurementsofthefluctuationsof thevolumeofwaterwithdrawalforirrigation groundwater levels for representative meas- on regional level is difficult to obtain due to urements points. The Marchfeldkanal Compa- the decentralized network of pumping sta- ny estimates the withdrawal for irrigation pur- tions,theirindividualmanagementanddata posesbylimitingthetimeframetosummer privacyissues.SimilartothestudyofAKDIM periods where distinct drops in the ground- etal.(2014)itwouldallowquantifyingthead- water levels can be observed. The timeframe equacy of water allocation in certain times or ofthemeasurementsvariesbetween55days areas, an important topic worldwide as em- (1995) and 162 days (2012). The resulting esti- phasized by WALLACE (2000). The estimation mation for agricultural irrigation for the years of regional water deficit in Marchfeld is a first from 1991 to 2008 ranges between 7.5 mil- steptowardsthisdirection.Thelackofcom- lion m3 (year 2005) and 45.1 million m3 (year parabledataemphasizestheneedforexpand- 2000).Fortheyear2012,theestimationofwa- ed research activities on this topic. ter withdrawal was 35 million m3 considering atimeframeof162daysstartingattheendof April (NEUDORFER 2013). Acknowledgements For the Marchfeld region water regula- tion is a major management and resource is- The study was supported by the Austrian sue which is closely monitored at groundwa- SpaceApplicationProgramme(ASAP)inthe ter level (NEUDORFER &WEYERMAYER 2007). contextoftheproject‘EO4Water– Applica- Knowledge of the regional water requirement tion of Earth observation technologies for ru- and its spatial and temporal distribution pro- ral water management’ (www.eo4water.com). 380 Photogrammetrie • Fernerkundung • Geoinformation 5/2014

References project. – Agricultural Water Management 98 (2): 271–282. AKDIM,N.,ALFIERI,S.,HABIB,A.,CHOUKRI,A., DAV I D,B.,ELEN,C.,JANER,D.,SILKE,E.,ERIK,P.J., CHERUIYOT,E.,LABBASSI,K.&MENENTI,M., JOSE,S.-V.&CONSUELO,V.-O.,2000: The envi- 2014: Monitoring of Irrigation Schemes by Re- ronmental impacts of irrigation in the European – DWYER,J. mote Sensing: Phenology versus Retrieval of Union. (ed.): A report to the Envi- Biophysical Variables. – Remote Sensing 6 (6): ronmentDirectorateoftheEuropeanComission. DEIMOS-IMAGING,2014: – 5815–5851. Our satellite: DEIMOS-1. ALLEN,R.G.,PEREIRA,L.L.,RAES,D.,SMITH,M.& http://www.deimos-imaging.com/technology/ AB,W.,1998: Crop evapotranspiration – Guide- our-satellite-deimos-1 (14.2.2014). DUBROVSKY,M.,SVOBODA,M.D.,TRNKA,M.,HAYES, linesforcomputingcropwaterrequirements.– M.J., WILHITE,D.A.,ZALUD,Z.&HLAVINKA,P., http://www.fao.org/docrep/x0490e/x0490e00. htm (9.7.2014). 2008: Application of relative drought indices in ATKINSON,P.M.,JEGANATHAN,C.,DASH,J.&ATZ- assessingclimate-changeimpactsondrought – BERGER,C.,2012:Inter-comparisonoffourmod- conditions in Czechia. Theoretical and Ap- – – els for smoothing satellite sensor time-series plied Climatology 96 (1 2): 155 171. EILERS,P.H.C., – data to estimate vegetation phenology. – Remote 2003:APerfectSmoother. Ana- – Sensing of Environment 123: 400–417. lytical Chemistry 75 (14): 3631 3636. CASA,R.,VARELLA,H.,BUIS,S.,GUÉRIF,M.,DE SO- EITZINGER,J.,KERSEBAUM,K.C.&FORMAYER,H., LAN,B.&BARET,F.,2012:Forcingawheatcrop 2009:LandwirtschaftimKlimawandel:Aus- ü model with LAI data to access agronomic vari- wirkungen und Anpassungsstrategien f rdie ables: Evaluation of the impact of model and land- und forstwirtschaftlichen Betriebe in Mit- LAIuncertaintiesandcomparisonwithanem- teleuropa.–1.Auflage,376S.,Agrimedia, ü pirical approach. – European Journal of Agrono- M ncheberg. EUROPEAN ENVIRONMENT AGENCY my 37 (1): 1–10. 2006: CORINE – CEPUDER,P.&NOLZ,R.,2007: Irrigation manage- Land Cover. http://www.eea.europa.eu/ mentbymeansofsoilmoisturesensortechnolo- publications/COR0-landcover (16.12.2013). FATICHI,S., gies. – JournalofWaterandLandDevelopment 2012: Inverse Distance Weight for – 11 (1): 79–90. MATLAB. http://www.mathworks.com/ CLEVERS,J.G.P.W.,1989:Applicationofaweighted matlabcentral/fileexchange/24477-inverse- infrared-red vegetation index for estimating leaf distance-weight (30.6.2014). FRENKEN,K.&GILLET,V., AreaIndexbyCorrectingforSoilMoisture.– 2012: Irrigation water Remote Sensing of Environment 29: 25–37. requirement and water withdrawal by country. – COLLINS,R.,KRISTENSEN,P.&THYSSEN,N.,2009: http://www.fao.org/nr/water/aquastat/water_ Water resources across Europe-confronting wa- use_agr/index.stm (1.7.2014). JÉGO,G.,PATTEY,E.&LIU,J., ter scarcity and drought. – European Environ- 2012: Using Leaf ment Agency. Area Index, retrieved from optical imagery, in COURAULT,D.,SEGUIN,B.&OLIOSO,A.,2005: Re- theSTICScropmodelforpredictingyieldand view on estimation of evapotranspiration from biomass of field crops. – Field Crops Research – remote sensing data: From empirical to numeri- 131: 63 74. MOLDEN,D.J., calmodelingapproaches.– Irrigation and Drain- 2006:AccountingforWaterUseand – age Systems 19 (3–4): 223–249. Productivity. International Irrigation Manage- – D’URSO,G.,2001: Simulation and management of ment Institute (May): 1 16. MUELLER,N.,GERBER,J.,JOHNSTON,M.,RAY,D., on-demand irrigation systems: a combined agro- RAMANKUTTY,N.&FOLEY,J., hydrological and remote sensing approach. – 2012: Closing yield – Wageningen University Press 174, Netherlands. gaps through nutrient and water management. – D’URSO,G.&CALERA BELMONTE,A.,2006: Opera- Nature 490 (7419): 254 257. NEUDORFER,W., tiveApproachesToDetermineCropWaterRe- 2002: Wasserschutz und Lebens- quirements From Earth Observation Data: aderMarchfeldkanal.–1.Auflage,174S., Ö Methodologies And Applications. – AIP Confer- Deutsch-Wagram, sterreich. NEUDORFER,W., ence Proceedings: 14–25. 2013:personalcommunication. NEUDORFER,W.&WEYERMAYER,H., D’Urso,G.,Richter,K.,Calera,A.,Osann,M.A., 2007: Securing Escadafal, R., Garatuza-Pajan, J., Hanich, L., groundwater use and reestablishing the water Perdigao, A., Tapia, J.B. & Vuolo, F., 2010: Earth balance by artificial recharge of groundwater in – Observation products for operational irrigation theregionofMarchfeld,Austria. Water Prac- – management in the context of the PLEIADeS tice&Technology2 (3): 1 7. N. Neugebauer & F. Vuolo, Crop Water Requirements 381

OZDOGAN,M.,YANG,Y.,ALLEZ,G.&CERVANTES, Soil Moisture Content Across Different Scales C., 2010:RemoteSensingofIrrigatedAgricul- of Observation. – Sensors 8 (1): 70–117. ture: Opportunities and Challenges. – Remote VUOLO,F.,NEUGEBAUER,N.,BOLOGNESI,S.,ATZ- Sensing 2 (9): 2274–2304. BERGER,C.&D’URSO,G.,2013: Estimation of PAL,J.S.,2004: Consistency of recent European Leaf Area Index Using DEIMOS-1 Data: Appli- summer precipitation trends and extremes with cationandTransferabilityofaSemi-Empirical future regional climate projections. – Geophysi- Relationship between two Agricultural Areas. – calResearchLetters31 (13): 1–4. Remote Sensing 5 (3): 1274–1291. PEREIRA,L.S.,OWEIS,T.&ZAIRI,A.,2002: Irriga- WAGNER,P.D.,FIENER,P.,WILKEN,F.,KUMAR,S.& tion management under water scarcity. – Agri- SCHNEIDER,K.,2012: Comparison and evaluation cultural Water Management 57 (3): 175–206. of spatial interpolation schemes for daily rainfall RICHTER,R.,1998: Correction of satellite imagery in data scarce regions. – Journal of Hydrology over mountainous terrain. – Applied Optics 37 464: 388–400. (18): 4004–4015. WALLACE,J.,2000:Increasingagriculturalwater RICHTER,R.&SCHLÄPFER,D.,2014: Atmospheric / use efficiency to meet future food production. – TopographicCorrectionforAirborneImagery. Agriculture, Ecosystems & Environment 82 (1- – ATCOR-4UserGuide,Version6.3.1. 3): 105–119. SCHARF,P.C.,SCHMIDT,J.P.,KITCHEN,N.R.,SUD- WHITTAKER,E.T.,1922: On a New Method of Grad- DUTH,K.A.,HONG,S.Y.,LORY,J.A.&DAV IS,J.G., uation. – Edinburgh Mathematical Society 41: 2002:Remotesensingfornitrogenmanagement. 63–75. – Journal of Soil and Water Conservation 57 (6): 518–524. SECKLER,D.,AMARASINGHE,U.,MOLDEN,D.,DE SIL- Address of the Authors: VA,R.&BARKER,R.,1999: World Water De- mandandSupply,1990to2025:Scenariosand NIKOLAUS NEUGEBAUER &FRANCESCO VUOLO, Uni- Issues. – International Water Management Insti- versityofNaturalResourcesandLifeSciences, tute, Colombo, Sri Lanka. Vienna (BOKU), Institute of Surveying, Remote SHANAN,L.,1987: The impact of irrigation. – Sensing and Land Information, Peter-Jordan- SCOPE 32: 115–132. Straße82,A-1190Wien,Österreich, Tel.: +43-1 TRNKA,M.,DUBROVSKÝ,M.,SVOBODA,M.,SEMERÁ- 47654 5138, Fax.: +43-1 47654 5142, e-mail: dová,d.,Hayes,M.,Žalud,Z.&WilHite,d., {nikolaus.neugebauer} {francesco.vuolo}@ 2009: Developing a regional drought climatolo- boku.ac.at gy for the Czech Republic. – International Jour- nal of Climatology 29 (6): 863–883. VERSTRAETEN,W.W.,VEROUSTRAETE,F.&FEYEN,J., Manuskript eingereicht: März 2014 2008:AssessmentofEvapotranspirationand Angenommen: Juli 2014