Research Article Reliability of MODIS Evapotranspiration ... · Caatinga 10 FLUX TOWER N km 9 20 S...

Research ArticleReliability of MODIS Evapotranspiration Products forHeterogeneous Dry Forest A Study Case of Caatinga

Rodrigo de Queiroga Miranda1 Josiclecircda Domiciano Galviacutencio1

Magna Soelma Beserra de Moura2 Charles Allan Jones3 and Raghavan Srinivasan3

1Laboratorio de Sensoriamento Remoto e Geoprocessamento Universidade Federal de Pernambuco50670901 Recife PE Brazil2Empresa Brasileira de Pesquisa Agropecuaria Centro de Pesquisa Agropecuaria do Tropico Semi-Arido56302970 Petrolina PE Brazil3Spatial Sciences Laboratory Texas AampM University College Station TX 77845 USA

Correspondence should be addressed to Rodrigo de Queiroga Miranda rodrigoqmirandagmailcom

Received 23 September 2016 Accepted 29 November 2016 Published 23 January 2017

Academic Editor Minha Choi

Copyright copy 2017 Rodrigo de Queiroga Miranda et al This is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

Evapotranspiration (ET) is normally considered as the sum of all water that evaporates from the soil and transpires from plantsHowever accurately estimating ET from complex landscapes can be difficult because of its high spatial heterogeneity and diversityof driver factors which make extrapolation of data from a point to a larger area quite inaccurate In this paper we hypothesize thatMODIS products can be of use to estimate ET in areas of Caatinga vegetation the hydrology of which has not been adequatelystudied The experiment was conducted in a preserved level area of Caatinga in which meteorological and water flux measureswere taken throughout 2012 by eddy covariance Evapotranspiration estimates from eddy covariance were compared with remotelysensed evapotranspiration estimates from MOD16A2 and SAFER products Correlations were performed at monthly 8-day anddaily scales and produced 1199032 values of monthly scale = 092 8-day scale = 088 and daily scale = 085 for the SAFER algorithmMonthly MOD16A2 data produced a value of 1199032 = 082 and they may be useful because they are free downloadable and easy touse by researchers and governments

1 Introduction

The high equipment and maintenance costs involved in mea-suring water fluxes in agrosystems and natural ecosystemsthrough field experiments make remote sensing an attractivealternative [1] Remote sensing has been used worldwideas a low cost fast and practical methodology to measurephysical and biological parameters at multiple scales fromland surfaces [2] It has been applied in different climaticregions to determine and to map the spatial and temporalvariation of components of water balance [3 4] It has beencoupled with hydrological models such as the Soil andWaterAssessment Tool (SWAT) providing input data normallyobtained from agrometeorological stations [5]

Evapotranspiration is normally considered as the sumof all water that evaporates from the soil and transpires

from plants [6 7] It can be used to estimate the amount ofwater used by crops and the amount of irrigation requiredfor optimum crop production [8] It can also be used as acomponent of models used to estimate other components ofthe water balance including surface runoff lateral flow baseflow and percolation to aquifers [5 9] Such estimates areuseful for management of water in soils reservoirs and evenhydroelectric plants [2] As human population increases andclimate changes water management is becoming a greaterconcern for researchers [7] farmers and other decisionmakers The state of Pernambuco has developed policies forclimate change (state law number 14090 of June 17 2010) andcoastal management (state law number 14258 of December23 2010) and to combat desertification and mitigate theeffects of drought (state law number 14091 of June 17 2010)These policies are designed to help manage the water in the

Hindawi Publishing CorporationAdvances in MeteorologyVolume 2017 Article ID 9314801 14 pageshttpsdoiorg10115520179314801

2 Advances in Meteorology

WaterAgricultureCaatinga

10

FLUX TOWER

N

km

9∘20998400 S

9∘S

8∘40998400 S

40∘40998400W 40∘20998400W41∘W

Figure 1 Location of the Flux tower installed in an area of Caatinga at Embrapa Semiarido which is a research station of the state ofPernambuco Brazil Land use map source PROBIO

state and accurate information about temporal and spatialvariations in ET could assist in their implementation

Spatial heterogeneity of ET can be great due to spatialvariations in vegetation weather soils and topography As aresult extrapolation of ET estimates fromflux towers to largerareas can be misleading [1 10] Also ET is spatially auto-correlated so changes at a point can affect its surroundings[11] Direct measurement of ET in the field is expensive anddifficult therefore satellite-based products that can be usedto remotely estimate ET have become very popular amongresearchers and government agencies Several models oflandscape energy balance based on remotely sensed data havebeen developed for example SEBAL (Surface Energy BalanceAlgorithm for Land [12 13]) MOD16 [7 14] and SAFER(Simple Algorithm For Evapotranspiration Retrieving [8])Each of these models has advantages and disadvantagesdepending on where they are being applied [15] Becausemodels are often based on a global scale [7] or on a model-species approach [16 17] calibration and validation may berequired to parameterize a model to a given set of localconditions [18]

Validation is used to analyze the uncertainties of modeloutputs ensuring the accuracy of remotely sensed ET andenhancing its applications Awide range ofmethods has beenused but in most cases deviance measures and statisticaltests are recommended ones The first one evaluates thedifferences between the simulated and observed values [19]With squared deviations Root Mean Square Errors (RMSE)can be useful to derive statistical properties The second oneis better for bias analysis It is important to use both methodsbecause a dataset could be extremely close in absolute valuesbut spatially removed from the perfect covariance fit (ieregression curve) and vice versa especially when it is anonlinear relationship like an exponential fit For that thecoefficient of determination (1199032) is usually the measurementto evaluate the proportion of output variation that can beexplained by the fit curve

In this paper we hypothesize that MODIS products canbe used to accurately estimate ET in areas of Caatinga vege-tation in northeastern Brazil which is a neglected Brazilianvegetation type in terms of research and investments [20] Inour laboratory Machado et al [21] have previously estimatedET using Landsat imagery but we believe the MODIS sensormay be better suited for monitoring ET in Caatinga becauseof its high temporal resolution In this study we usededdy covariance data to calibrate MODIS estimates of ETfor Caatinga vegetation on monthly 8-day and daily timescales

2 Methodology

21 Study Area The experiment was conducted in a pre-served flat area of Caatinga (asymp600 ha) located at EmbrapaSemiarido research station of the state of Pernambuco Brazil(40∘1910158401610158401015840W 9∘0210158403310158401015840S 350m) (Figure 1) Caatinga is ahyperxerophilic vegetation type that consists of deciduousxerophytic shrubs and trees with an average height of 5mand over 1000 vascular plant species [20] In the studyarea dominant plant species include Poincianella microphylla(Mart ex G Don) Croton conduplicatus (Kunth) Bauhiniacheilantha (Bong) Manihot pseudoglaziovii (Pax amp Hoff-man) and Commiphora leptophloeos (Mart) The climatetype is BSwh (Koopen) or semiarid with rainy season fromJanuary to April Average annual rainfall is 510mm andit is temporally and spatially very heterogeneous Mean airtemperature is 262∘C [22]

22 Eddy Covariance Flux Tower Meteorological and waterflux measurements were taken throughout 2012 with sensorsinstalled in a 16m tower located in the study area Anet radiometer (model CNR-1 Kipp amp Zonen B V DelftNetherlands) installed at 133m above the soil surface wasused to determine incoming solar radiation (119877119878) Air tem-perature and relative humidity (HMP45C Vaisala Helsinki

Advances in Meteorology 3

Table 1 List of all MODIS imaged days of the year of 2012 used in the SAFER analysis All these images were selected by having clear andhigh quality data for the pixel regarding an area of Caatinga at Embrapa Semiarido which is a research station of the state of PernambucoBrazil (9∘051015840S 40∘191015840W 350m)

01072012 04192012 05312012 07262012 1020201201102012 04262012 06022012 07312012 1023201201282012 04282012 06062012 08072012 1026201203012012 04292012 06072012 08082012 1028201203022012 05012012 06092012 08102012 1029201203032012 05032012 06102012 08112012 1030201203042012 05052012 06152012 08292012 1214201203072012 05072012 06182012 09042012 1218201203092012 05122012 06202012 09082012 1220201203142012 05142012 06212012 09142012 1222201203172012 05152012 06222012 09212012 1225201203232012 05162012 06262012 09252012 mdash03252012 05172012 07082012 09262012 mdash03282012 05292012 07132012 10032012 mdash04062012 05302012 07232012 10102012 mdash

Finland) and rainfall (CS700-L Hydrological Services RainGauge Liverpool Australia) were measured at 157m and163m height respectively Wind speed values were obtainedwith an ultrasonic anemometer 3D (WindMaster Pro GillInstruments Ltd Lymington UK) at 169m height andET was directly measured by an Open Path H2O analyzer(IRGA model LI-7500 LI-COR Inc Lincoln NE USA)All sensors were connected to a data logger (model CR1000Campbell Scientific Inc Logan Utah USA) set up to takemeasurements every 10 seconds More information on thestudy site and monitoring system can be found in a previouspaper [22]

23 MOD16A2 Products MODIS MOD16A2 products weredownloaded for all 8-day and month periods of the yearof 2012 from httpwwwntsgumteduprojectmod16 The58 images were rescaled from 01mm 8-dayminus1 or 01mmmonthminus1 to correct units (mm 8-dayminus1 or mm monthminus1) bymultiplication of all pixels by the 01 using the GDAL library(Geospatial Data Abstraction Library) The ET MOD16A2dataset is composed of two components (i) meteorologicaland (ii) remote sensing based data and is computed usingan algorithm (1) [7 14] a modification of the equationdescribed by Cleugh et al [23] which in turn is a Penman-Monteith approach to estimate ET [24]

ET = Δ (119877119899 minus 119866) + 120588119862119901 (119890119904 minus 119890119886) 119903119886Δ + 120574 (1 + 119903119904119903119886) (1)

where 119877119899 is radiation budget (Jmminus2 dayminus1) 119866 is soil heatflux (asymp0 Jmminus2 dayminus1) 120588 is the air density (kgmminus3) 119862119901 is thespecific heat of air at constant pressure (1013 J kgminus1Kminus1)119890119904 is saturation vapor pressure (Pa) 119890119886 is current vaporpressure (Pa) Δ is the slope of the curve pressure versusair temperature (PaKminus1) and 120574 is the psychometric constant(kPaKminus1) The meteorological input data for that equation is

always provided by the Global Modelling and AssimilationOffice (GMAO) and includes daily total downward radiation(119877119878 MJmminus2 dayminus1) daily average air temperature (T ∘C)daytime and nighttime air temperatures (119879119863 119879119873 ∘C) dailyminimum air temperature (119879min

∘C) and vapor pressure(119890119904 119890119886 kPa) all at 10∘ times 125∘ spatial resolution The landsurface inputs are acquired from threeMODIS products witha spatial resolution from 500 to 1000m2 These products areMOD12Q1 [25] MOD15A2 [26] and MCD43B2B3 Collec-tion 5 (albedo) [27] Further details of MOD16 algorithm aregiven in Mu et al [14]

24 SAFER Products To create SAFER products we usedimages of level 1B MOD021KM products from sensorTerraMODIS for the year of 2012 Firstly all 366images were downloaded through the Level 1 andAtmosphere Archive and Distribution System (LAADShttpsladswebnascomnasagovdatasearchhtml) and sec-ondly 71 imageswere selected as having clear and high qualitydata for the pixel regarding Embrapa Semiarido (Table 1)The MOD021KM product has 36 spectral bands from0405 120583m to 14385 120583m with different spatial resolutionsfrom 250 to 1000m2 The SAFER algorithm is basicallya simplified and calibrated version of SEBAL [12 13] forBrazilian semiarid conditions [15] It was first proposed byTeixeira [8] using Landsat imagery but MODIS applicationdetails can be found in Teixeira et al [3] It utilizes only fourbands in a five-step process to acquire ET The first step wasto convert the digital numbers (DN) of two bands from thevisible spectrum (bands 1 and 2) into reflectance (119901119887) andDN of two thermal bands (bands 31 and 32) into radiance(119871119887 Wmminus2 srminus1 120583mminus1) For that we used following equation

119901119887 = 119886 (DN119887 minus 119887) (2)

where DN is the digital number and 119886 and 119887 are calibratedconstants retrieved from the metadata of each product


The second step is used to calculate values of albedo asrecommended by Valiente et al [28]

120572119904 = 119886 + 1198871199011 + 1198881199012 (3)

where 119886 119887 and 119888 are 008 041 and 014 which are valuescalibrated to Brazilian semiarid by Teixeira et al [29] Thethird step is to process the Normalized Difference VegetationIndex (NDVI) which can be acquired as follows [30]

NDVI = (1199012 minus 11990111199012 + 1199011) (4)

where 1199011 and 1199012 are the reflectance of band 1 and 2respectively The fourth step is used to calculate surface andbrightness temperatures

119879119887 = ( 1198702ln (1198701119871119887 + 1))

1198790 = 11988611987931 + 11988711987932(5)

where 11987931 and 11987932 are brightness temperatures from band 31and 32 respectively 119886 and 119887 are both coefficients with valueof 05 [29] 119879119887 is brightness temperature of each band 119887 119871119887 isspectral radiance and 1198701 and 1198702 are conversion coefficientsin Wmminus2 srminus1 120583mminus1 1198701 and 1198702 can be calculated with thefollowing equations [31]

1198701 = 2ℎ1198882120582minus510minus61198702 = ℎ119888

120590120582 (6)

where ℎ is the Planck constant (662606896 sdot 10minus34 J sminus1)119888 is the speed of light (299792458 sdot 108msminus1) 120590 is theBoltzmann constant (13806504 sdot 10minus23 J Kminus1) and 120582 is themean wavelength for each thermal band in meters (110186 sdot10minus6 and 120325 sdot 10minus6 for bands 31 and 32 resp) Finallyfor the ET ratio (ET ET0) inputs are NDVI 120572119904 and 1198790 (7)The ET ratio is related with soil moisture and vegetationchlorophyll content It can be used to estimate componentsof the water balance at the scale of the remote sensing pixelswhich may include different agrosystems and natural plantcommunities

ETET0

= exp [119886 + 119887( 1198790120572119904NDVI)] (7)

where ET0 is the reference evapotranspiration and 119886 and 119887are regression coefficients of values 190119890minus0008 respectivelyfor Brazilian semiarid conditions [8] All calculations wereperformed with the GDAL library

25 Reference Evapotranspiration To estimate the referenceevapotranspiration (ET0) required by the last step of SAFERalgorithm we used the following equations with daily resam-pled data from the eddy covariance Flux tower

ET0

= 0408Δ (119877119899 minus 119866) + 120574 (900 (119879 minus 273)) 1199062 (119890119904 minus 119890119886)Δ + 120574 (1 + 0341199062) (8)

where 119877119899 is radiation budget (MJmminus2 dayminus1) 119866 is soil heatflux (asymp0MJmminus2 dayminus1) 119879 is air temperature at 2m (∘C) 1199062is wind speed at 2m (m sminus1) 119890119904 is saturation vapor pressure(kPa) 119890119886 is current vapor pressure (kPa) Δ is the slope ofthe curve pressure versus air temperature (kPa ∘Cminus1) and 120574is the psychometric constant (kPa ∘Cminus1) 119877119899 was calculatedfollowing this series of equations

119877ns = (1 minus 120572) 119877119878120593 = 120587

180 119897119889119903 = 1 + 0033 cos( 2120587

365119895) 120575 = 0409 sin( 2120587

365119895 minus 139) 120596119904 = arccos [minus tan (120593) tan (120575)] 119877119886 = 24 (60)

120587sdot 119866SC119889119903 [120596119904 sin (120593) sin (120575) cos (120593) cos (120575) sin (120596119904)]

1198771198780 = (075 + 2 sdot 10minus5119911) 119877119886119877nl = 1205901198794 (034 minus 014radic119890119886) times (135 1198771198781198771198780 minus 035) 119877119899 = 119877ns minus 119877nl

(9)

where 119877ns is net solar or net shortwave radiation (MJmminus2dayminus1) 120572 is albedo or canopy reflection coefficient whichwas considered 0175 as indicated by de Souza et al [22] forthe study area 119877119878 is measured incoming solar radiation frommeteorological tower (MJmminus2 dayminus1) 119897 is latitude in degrees120593 is latitude in radians 119889119903 is the inverse relative Earth-Sundistance 119895 is the number of the day in the year 120575 is thesolar declination 120596119904 is the sunset hour angle (radians) 119877119886is daily extraterrestrial radiation (MJmminus2 dayminus1) 119866SC is thesolar constant (00820MJmminus2minminus1) 1198771198780 is clear-sky solarradiation (MJmminus2 dayminus1) 119911 is elevation above sea level inmeters 119877nl is net longwave radiation (MJmminus2 dayminus1) and 120590is the Boltzmann constant (4903 sdot 10minus9MJKminus4mminus2 dayminus1)The parameters 1199062 119890119904 119890119886 Δ and 120574 were acquired with theequations below respectively where 119906119911 is wind speed atelevation 119911 (m sminus1) and RH is relative air humidity ()

1199062 = 119906119911 487ln (678119911 minus 542) (10)

119890119904 = 06108 exp( 1727119879119879 + 273) (11)

119890119886 = RH100119890119904 (12)


0

20

40

60

80

100

Relat

ive h

umid

ity (

)

2 3 4 5 6 7 8 9 10 11 121Month

0

5

10

15

20

25

30

35G

loba

l rad

iatio

n (M

Jmminus2)

1 2 3 4 5 6 7 8 9 10 11 12Month

Month1 2 3 4 5 6 7 8 9 10 11 12

0

10

20

30

40

50

Rain

fall

(mm

)

1 2 3 4 5 6 7 8 9 10 11 12Month

0

5

10

15

20

25

30

35

Air

tem

pera

ture

(∘C)

Figure 2 Meteorological scenario for the year of 2012 recorded by the Flux tower installed in an area of Caatinga at Embrapa Semiaridowhich is a research station of the state of Pernambuco Brazil Global radiation is themonthlymean of accumulated daily incoming radiationsair temperature and relative humidity are presented as monthly mean of the daily mean and rainfall is the monthly sum of all precipitations

Δ = 4098 times 06108119890119904(119879 + 273)2 (13)

120574 = 0665 sdot 10minus3 (1013 (293 minus 00065119911293 )526) (14)

26 Statistical Analysis Covariance between SAFER andMOD16A2 and tower estimates of ET were analyzedusing linear and nonlinear regressions Normality andhomoscedasticity of all ET data were testedwith ShapiroWilktest and Brown Forsyth test respectively [32]The differencesbetween the simulated and observed absolute values wereevaluated using RMSE [19 32] All statistical analyses wereperformed with the package R (v323 [33]) and results wereconsidered to be significant when 119901 le 005 Graphs wereplotted with the software Veusz (v124 [34])

3 Results and Discussion

31 Climatology and Surface Properties A first analysis of theweather data showed that in general global radiation (119877119878)air temperature (119879) and relative humidity (RH) exhibitednormal behavior in 2012 with low values of119877119878 and119879 and highvalues of RH in July and August (Figure 2) Exceptions in thepatterns were observed only for February April and Novem-ber when rainfall had a strong direct or indirect influenceAnnual precipitation was 9042mm which was only 17 ofhistorical averages for the area (510mm) Temporal variationin rainfall was large with peaks in February April andNovember February was a cloudy month with 119877119878 reachingvalues similar to July and 119879 decreasing as RH increased Theeffects of the February rainfall peak can also be observedin ET0 which decreased to 766 of Januaryrsquos (Figure 3)Global radiation measured in meteorological stations can be


1 2 3 4 5 6 7 8 9 10 11 12Month

0

01

02

03

04

05N

DV

I1 2 3 4 5 6 7 8 9 10 11 12

Month

0

005

01

015

02

Alb

edo

()

0

20

40

60

80

100

Evap

otra

nspi

ratio

n (E

T 0 m

m)

2 3 4 5 6 7 8 9 10 11 121Month

2 3 4 5 6 7 8 9 10 11 121Month

30

325

35

375

40

425

45

Surfa

ce te

mp

(∘C)

Figure 3 Monthly variation of all input parameters used in the SAFER algorithm for the year of 2012 in an area of Caatinga at EmbrapaSemiarido which is a research station of the state of Pernambuco Brazil NDVI albedo and Surface Temperature are presented as monthlymean of all available data and evapotranspiration is the monthly mean of all available data multiplied by the number of days in that month

drastically decreased during cloudy days and that mightsignificantly affect the amount of energy available to the pro-cess of evapotranspiration Also ET0 is a modelled variableand varies positively in function of 119877119878 For these groundmeasurements cloudy days still produce reliable data sincesurface data is acquired beneath the clouds which is not truefor satellite-based data that attempts to acquire surface datafrom above it The values of NDVI 120572119904 and 1198790 could notbe processed to February due to these interferences Aprilpresented a pattern inverse to February with high 119877119878 and 119879and low values of RH Although it was a dry month NDVIwere still high because of the existing lag between rainfalland vegetation cover dynamics and that kept 120572119904 1198790 and ET0stable November was again a rainy month that followed thesame pattern as February but since vegetation cover is lowerin November than in February 119879 was proportionally higherthan in February

32 Monthly Evaluation The ET estimates for the MODISpixel of Caatinga in which the tower was located were com-pared with the observations in loco performed by equipmentinstalled on that tower The relationship between observeddata and MOD16A2 products presented coefficients of deter-mination (1199032) of 077 for the linear fit and 082 for theexponential one (Figure 4) These values are comparable tothat found by Ruhoff et al [9] (1199032 = 089) when comparingMOD16A2 ET against eddy covariance measurements in anarea of Cerrado which is a Brazilian ecosystem that consistsof a dense vegetation dominated by shrubs and trees with 5ndash10m height When using SAFER data instead of MOD16A2we obtained 1199032 of 092 for a linear relation and 079 for anexponential relation

The lower 1199032 obtained with MOD16A2 compared withSAFER suggest that the limitations cited by Mu et al [7]may be operative Those limitations include (i) inaccuracy


y = 07341x minus 567

r2 = 077

p lt 005

0

10

20

30

40

50

60To

wer

evap

otra

nspi

ratio

n (m

m m

onth

minus1)

0 10 20 30 40 50 60MOD16A2 evapotranspiration (mm monthminus1)

y = 30195e00477x

r2 = 082

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

y = 51181e00469x

r2 = 079

p lt 0001

0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

10 20 30 40 50 600SAFER evapotranspiration (mm monthminus1)

y = 06376x + 30758

r2 = 092

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

Figure 4 Monthly linear and exponential regression of MOD16A2 evapotranspiration (superior part) and SAFER evapotranspiration(inferior part) versus evapotranspiration from a Flux tower installed in an area of Caatinga at Embrapa Semiarido which is a research stationof the state of Pernambuco Brazil for the year of 2012

of algorithm inputs such as MODIS LAI (Leaf Area Index)that tends to be overestimated [35] andmay result in overesti-mates of ET inaccuracy of MODIS EVI that may lead to mis-calculation of vegetation cover fraction andmisclassificationof land cover (MOD12Q1) that may result in incorrect VPDand minimum air temperature values in equations [7 25]and (ii) average parameter values for stomatal dynamics thatmay not represent well all species within that biome [36 37]On the other hand the SAFER algorithm was created andvalidated for the Brazilian semiarid Its greatest advantage isthat both the algorithm itself [8] and its inputs (120572119904 and 1198790in Teixeira et al [15]) have been calibrated for the Caatingaconditions

The exponential model performed better than the linearone forMOD16A2 indicating that the productrsquos ET estimatessaturate and lose sensitivity at high ET rates However thatis not a disadvantage of the SAFER algorithm Since SAFER

uses an exponential regression This saturation pattern isbased on an intrinsic behavior of the vegetation indexes usedin both MOD16A2 and SAFER equations that are often over-or underestimated at low and high values for the Caatingarespectively and also on their relationships with LAI thathas proven to be always exponential For example Costa etal [38] (15) and Domiciano Galvıncio et al [39] (16) havefound exponential relationships between LAI and NDVI forCaatinga vegetation In addition vegetation indexes producenonlinear relationships with LAI due to their mathematicalstructure [40]

LAI = 06401119890(26929timesNDVI) (15)

LAI = exp [1426 + (minus0542NDVI

)] (16)


Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3 minus2 minus1 0 1 2 3Regression standardized predicted value

minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

10 2 3minus2 minus1minus3Regression standardized predicted value

minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)

Figure 5 Monthly regression standardized residuals versus regression standardized predicted values of all models (a) and (b) are linear andexponential models for MOD16A2 respectively and (c) and (d) are linear and exponential models for SAFER algorithm respectively

Linear estimation of ET using MOD16A2 produced agreater average residual (398mmmonthminus1) than exponentialone (282mmmonthminus1)These results corroborate theNDVI-LAI relation that is better for nonlinear than linear equationsAlso the linear model violates assumptions of linearity andhomoscedasticity We can observe curvilinearity in the dataand residuals are greater for greater values of predicted ET(Figure 5(a)) Note that themodel should only be used withinthe range of data used to develop the model which inour case is from 431 to 4015mm of ET The exponentialmodel for MOD16A2 was similar to the linear model but itviolates only the assumption of homoscedasticity presentingthe same increasing relation of residualrsquos variances andregression predicted values (Figure 5(b)) That is estimatesof small values of ET are more precise than large values ForSAFER neither linear nor exponential models present anyspecific pattern of errors suggesting that the assumptionswere met (Figures 5(c) and 5(d)) However the linear modelis preferred because its 1199032 is greater and its residuals aresmaller (16mm monthminus1) when compared with exponentialestimates (243mmmonthminus1)

33 Eight-Day Evaluation The relationship between ob-served ET and MOD16A2 estimates presented 1199032 values of062 and 069 for a linear and exponential fit respectively(Figure 6) These values are again close to that found byRuhoff et al [9] in the Cerrado (1199032 = 078) when comparing8-day MOD16A2 ET with tower flux measurements of ETLimitations regardingMOD16A2 products and linearmodelsare the same as described for monthly data Residualsfrom the linear model using MOD16A2 presented a greateraverage residual (156mm 8-dayminus1 Figure 7(a)) than theexponential model (12mm 8-dayminus1 Figure 7(b)) and wecan observe the same pattern as in the monthly analysiswhere both show signs of heteroscedasticity and only thelinear model violates the linearity assumption However the8-day exponential model overestimates ET more than themonthly model The mean positive residuals were 907greater than the mean negative residuals in the 8-day modelwhile in the monthly model mean positive residuals were124 larger That corroborates results of Ruhoff et al [9]in which MOD16A2 overestimated ET for another biomein Brazil We obtained 1199032 values of 088 and 065 for linear


y = 07933e01679x

r2 = 069

p lt 005

0 5 10 15 20MOD16A2 evapotranspiration (mm 8-dayminus1)

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

5 10 15 200SAFER evapotranspiration (mm 8-dayminus1)

y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

Figure 6 8-day linear and exponential regression of MOD16A2 evapotranspiration (superior part) and SAFER evapotranspiration (inferiorpart) versus evapotranspiration from a Flux tower installed in an area of Caatinga at Embrapa Semiarido which is a research station of thestate of Pernambuco Brazil for the year of 2012

and exponential SAFERmodels respectively (Figure 6) withaverage residuals of 081mm 8-dayminus1 and 115mm 8-dayminus1for linear and exponential models respectively (Figures 7(c)and 7(d)) For the linear model symmetry of residuals wasquite large with small differences (62 for 8-day) but againthe monthly model was better with only 001 of differencebetween average negative and positive residuals

34 Daily Evaluation As monthly and 8-day SAFER linearmodels gave better results than the exponential ones wedecided to develop a linear model for daily ET This modelgave 1199032 of 085 with residuals varying from minus04 to 06mm(Figure 8) In comparison Teixeira [1] found 1199032 of 089 in alinear relation between SAFER data and field measurementsof irrigated crops and Caatinga In our study the patterns ofresiduals suggested heteroscedastic behavior This result was

inconsistent with residuals of the linear monthly and 8-daySAFER models possibly the result of temporal downscalingof weather data However daily ET data is essential for aprecise environment monitoring especially in the case of theBrazilian semiarid where there is a high space-time variationof rainfall [41] for example 20 of the yearly rainfall mayoccur in a single day or 60 in a month [42] Becauseof limitations in quantity and quality of weather stationdata in Brazil use of remotely sensed data by researchersand government agencies has increased Among the varioussensors with freely available data that have been used toestimate ETMODISmay be themost widely used It providesdata on a daily basis (250 to 1000m) which not only allowsprecise monitoring of ET at daily monthly and yearly scalesbut also can provide daily inputs to hydrological models likeSWAT [5]


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)

Figure 7 8-day regression standardized residuals versus regression standardized predicted values of all models (a) and (b) are linear andexponential models for MOD16A2 respectively and (c) and (d) are linear and exponential models for SAFER algorithm respectively

35 Temporal Evaluation ET estimates of the SAFER algo-rithm were closer to Flux tower estimates than MOD16A2estimates (Figure 9) especially from April to Decemberwhere the mean monthly difference between the tower andremotely sensed estimateswasminus026mmmonthminus1 for SAFERin comparison to 1108mm monthminus1 for MOD16A2 andthe mean 8-day differences were 0019 and 289mm 8-dayminus1for SAFER and MOD16A2 respectively However for theperiod between January and March the differences werenot as great minus0012mm monthminus1 and 017mm 8-dayminus1 forSAFER compared to 1636mmmonthminus1 and 431mm8-dayminus1for MOD16A2 For the daily scale the differences betweenSAFERandFlux tower ET estimateswere 0017mmdayminus1 and021mm dayminus1 for the first and second periods respectivelyThe SAFER model gave better results than the MOD16A2model at all evaluated temporal scales MOD16A2 tendedto overestimate ET due we suggest to the meteorologicalinput data used in MOD16A2 algorithm It is derived froma 100∘ times 125∘ grid while meteorological input data for theSAFER algorithm was acquired from the Flux tower itself

Mu et al [14] have reported that GMAO data producesbiases if compared to measures frommeteorological stationsThe period when all products and the Flux tower wereclosest was from September to October As we showed inthe residual analysis low values of ET tend to have smallervariances than high values In thosemonths both SAFER andMOD16A2 presented the smallest values of ET because thiswas the only period that a dry month (asymp0mm of rainfall) wasfollowed by another dry month and no soil water rechargeoccurred However even during these months when norainfall occurred we observed levels of ET varying from asymp3to 10mm This ET may result from plant transpiration ofwater from deep soil layers and soil evapotranspiration ofdew In northern Israel during the dry season Agam andBerliner [43] found that even when no dew is deposited onthe soil surface soil evaporation can be observed suggestingthat humidity is absorbed by the soil during the night andevaporates during the day

SAFER linear ET models more closely matched Fluxtower estimates of ET than all other models (Table 2) with


Table 2 Brief statistical summary and Root Mean Square Error (RMSE) of all nine evapotranspiration models

Model Product Min and max obs values (mm) Min and max pred values (mm) RMSE (mm)119910 = 07341119909 minus 567 Monthly MOD16A2

431 to 4015

083 to 324 491119910 = 3019511989000477119909 46 to 358 368119910 = 06376119909 + 30758 Monthly SAFER 45 to 2894 197119910 = 5118111989000469119909 568 to 3432 286119910 = 07699119909 minus 17182 8-day MOD16A2

083 to 1895

033 to 1217 218119910 = 0793311989001679119909 107 to 1647 199119910 = 06856119909 + 08387 8-day SAFER 106 to 1302 113119910 = 1440111989001566119909 149 to 233 227119910 = 05351119909 + 01067 Daily SAFER 007 to 191 012 to 145 015

Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005

05 1 15 2 25 30SAFER evapotranspiration (mm dayminus1)

0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)

0 2 4minus2minus4Regr std pred value

minus4

minus2

0

2

4

Figure 8 Daily regression of SAFER evapotranspiration versusevapotranspiration fromaFlux tower installed in an area ofCaatingaat Embrapa Semiarido which is a research station of the state ofPernambuco Brazil for the year of 2012 In the insets dispersionpattern of residuals is relative to the model

Root Mean Square Errors (RMSE) of 197 and 113mm formonthly and 8-day scales respectively while MOD16A2rsquoserrors ranges were 199 and 491mm respectively In theCerrado Ruhoff et al [9] found RMSEs of 19 and 078mmwhen comparing monthly and 8-day ET estimates fromMOD16A2 to tower flux measurements respectively withmonthly differences between predicted and measured dataranging from 0 to 40mm We also compared the min-imum and maximum predicted values from the SAFERand MOD16A2 estimates to the Flux tower dataset TheMOD16A2 linear model underestimated the minimum Fluxtower value by 81 for the monthly scale and by 60smaller for the 8-day scale In contrast SAFER linear andMOD16A2 exponential models showed minimum values of

only 4 and 7 greater than the observed minimum valuesfrom Flux tower dataset for monthly scale and both were28 greater for the 8-day scale For the maximum valuesthe 8-day SAFER exponential model overestimated Fluxtower estimates by 23 Daily SAFER linear model showed015mm RMSE over- and underestimating the minimumand maximum values respectively compared to the RMSEof 034mm found by Teixeira [1] when analyzing ET fieldmeasurements and SAFER data The SAFER algorithm wascompared with another algorithm by its developers in theirreport [44] and they argued that SAFER outperformedbecause it estimates the ratio ET ET0 instead of ET whichallows spatial extrapolation of ET [44] However since theMOD16A2 model estimates the same ratio we suggest thatthe regionally calibrated inputs [8] are the reason that SAFERestimates may be more accurate over other methods

4 Conclusion

In this study we compared ground-based ET with remotelysensed ET from MOD16A2 and SAFER products and itproduced 1199032 values of monthly scale = 092 8-day scale= 088 and daily scale = 085 for the SAFER algorithmMonthly MOD16A2 data produced a value of 1199032 = 082and 8-day value = 069 Although dataset variance increasedin temporal downscaling ET data we showed that MODISderived products can be of use to model ET for the Caatingaecosystemwith acceptable 1199032 values for the SAFER algorithmat all temporal scales Moreover we also recommend theuse of MOD16A2 monthly products to monitor the Caatingawhen if locally observedmeteorological data are not availableto produce SAFER estimates MODIS data produced satis-factory estimates of Flux tower observed data and are freelydownloadable at httpwwwntsgumteduprojectmod16We believe this study contributes to the assessment evap-otranspiration data via remote sensing techniques whichmay provide better understanding of the evapotranspirationdynamics of the Caatinga ecosystem which might be a worldreference in the future for ecological disturbances due toclimate changes or simply key information to help federal andmunicipal governments plan land use changes with rationalcriteria


MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year

Figure 9 Monthly 8-day and daily variations of MOD16A2 and SAFER evapotranspiration and evapotranspiration from a Flux towerinstalled in an area of Caatinga at Embrapa Semiarido which is a research station of the state of Pernambuco Brazil for the year of 2012

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

The authors thank CAPES (Coordenacao de Aperfeicoa-mento de Pessoal de Nıvel Superior - Brazilian Coordinationfor the Improvement of Higher Level Personnel) for fundingthis study through the international project PVE A1032013

References

[1] A H D C Teixeira ldquoDetermining regional actual evapo-transpiration of irrigated crops and natural vegetation in theSao Francisco river basin (Brazil) using remote sensing andPenman-Monteith equationrdquo Remote Sensing vol 2 no 5 pp1287ndash1319 2010

[2] M Govender K Chetty and H Bulcock ldquoA review of hyper-spectral remote sensing and its application in vegetation andwater resource studiesrdquo Water SA vol 33 no 2 pp 145ndash1512007


[3] A H D C Teixeira M Sherer-Warren H L Lopes F B THernandez R G Andrade and C M U Neale ldquoApplication ofMODIS images for modelling the energy balance componentsin the semi-arid conditions of Brazilrdquo in Proceedings of theRemote Sensing for Agriculture Ecosystems and Hydrology XVvol 8887 of Proceedings of SPIE 2013

[4] A H de C Teixeira D C Victoria R G Andrade J FLeivas E L Bolfe and C R Cruz ldquoCoupling MODIS imagesand agrometeorological data for agricultural water productivityanalyses in the Mato Grosso State Brazilrdquo in Proceedings of theRemote Sensing for Agriculture Ecosystems and Hydrology XVIvol 9239 of Proceedings of SPIE September 2014

[5] M Strauch and M Volk ldquoSWAT plant growth modificationfor improved modeling of perennial vegetation in the tropicsrdquoEcological Modelling vol 269 pp 98ndash112 2013

[6] R G Allen L S Pereira D Raes and M Smith CropEvapotranspiration-Guidelines for Computing Crop WaterRequirements-FAO Irrigation and Drainage Paper 56 vol 300no 9 FAO Rome Italy 1998

[7] Q Mu F A Heinsch M Zhao and S W Running ldquoDevelop-ment of a global evapotranspiration algorithmbased onMODISand global meteorology datardquo Remote Sensing of Environmentvol 111 no 4 pp 519ndash536 2007

[8] A H C Teixeira ldquoModelling evapotranspiration by remotesensing parameters and agro-meteorological stationsrdquo inRemote Sensing and Hydrology C M U Neale and M HCosh Eds no 352 pp 154ndash157 International Association ofHydrological Sciences Jackson Hole wyo USA 2012

[9] A L Ruhoff A R Paz L E O C Aragao et al ldquoAssessmentof the MODIS global evapotranspiration algorithm using eddycovariance measurements and hydrological modelling in theRio Grande basinrdquo Hydrological Sciences Journal vol 58 no 8pp 1658ndash1676 2013

[10] A H De C Teixeira M Scherer-Warren F B T HernandezR G Andrade and J F Leivas ldquoLarge-scale water productivityassessments with MODIS images in a changing semi-aridenvironment a Brazilian case studyrdquo Remote Sensing vol 5 no11 pp 5783ndash5804 2013

[11] P Segurado M B Araujo and W E Kunin ldquoConsequencesof spatial autocorrelation for niche-based modelsrdquo Journal ofApplied Ecology vol 43 no 3 pp 433ndash444 2006

[12] W G M Bastiaanssen M Menenti R A Feddes and A A MHoltslag ldquoA remote sensing surface energy balance algorithmfor land (SEBAL) 1 Formulationrdquo Journal of Hydrology vol212-213 no 1ndash4 pp 198ndash212 1998

[13] W G M Bastiaanssen H Pelgrum J Wang et al ldquoA remotesensing surface energy balance algorithm for land (SEBAL) 2Validationrdquo Journal of Hydrology vol 212-213 no 1ndash4 pp 213ndash229 1998

[14] Q Mu M Zhao and S W Running ldquoImprovements to aMODIS global terrestrial evapotranspiration algorithmrdquo Re-mote Sensing of Environment vol 115 no 8 pp 1781ndash1800 2011

[15] A H D C Teixeira W G M Bastiaanssen M D Ahmad andM G Bos ldquoReviewing SEBAL input parameters for assessingevapotranspiration and water productivity for the Low-MiddleSao Francisco River basin Brazil part A calibration andvalidationrdquo Agricultural and Forest Meteorology vol 149 no 3-4 pp 462ndash476 2009

[16] R C Beeson Jr ldquoModeling actual evapotranspiration of Vibur-num odoratissimumduring production from rooted cuttings tomarket size plants in 114-L containersrdquoHortScience vol 45 no8 pp 1260ndash1264 2010

[17] J C B Hoedjes A Chehbouni F Jacob J Ezzahar andG Boulet ldquoDeriving daily evapotranspiration from remotelysensed instantaneous evaporative fraction over olive orchard insemi-arid Moroccordquo Journal of Hydrology vol 354 no 1-4 pp53ndash64 2008

[18] J G Arnold D NMoriasi PW Gassman et al ldquoSWATModeluse calibration and validationrdquo Transactions of the ASABE vol55 no 4 pp 1491ndash1508 2012

[19] D G Mayer and D G Butler ldquoStatistical validationrdquo EcologicalModelling vol 68 no 1-2 pp 21ndash32 1993

[20] J C Santos I R Leal J S Almeida-Cortez G W Fernandesand M Tabarelli ldquoCaatinga the scientific negligence experi-enced by a dry tropical forestrdquo Tropical Conservation Sciencevol 4 no 3 pp 276ndash286 2011

[21] C C Machado B B Silva M B de Albuquerque and JD Galvıncio ldquoEstimativa do balanco de energia utilizandoimagens TMmdashLandsat 5 e o algoritmo SEBAL no litoral sul dePernambucordquo Revista Brasileira de Meteorologia vol 29 no 1pp 55ndash67 2014

[22] L S B de Souza M S B de Moura G C Sediyama andT G F da Silva ldquoRadiation balance in Caatinga ecosystempreserved for a year drought in semiarid PernambucanordquoRevista Brasileira de Geografia Fısica vol 8 no 1 pp 41ndash552015

[23] H A Cleugh R Leuning QMu and SW Running ldquoRegionalevaporation estimates from flux tower and MODIS satellitedatardquo Remote Sensing of Environment vol 106 no 3 pp 285ndash304 2007

[24] J L Monteith ldquoEvaporation and environmentrdquo inThe State andMovement ofWater in LivingOrganisms pp 205ndash234AcademicPress New York NY USA 1965

[25] M A Friedl D K McIver J C F Hodges et al ldquoGlobal landcover mapping from MODIS algorithms and early resultsrdquoRemote Sensing of Environment vol 83 no 1-2 pp 287ndash3022002

[26] R B Myneni C D Keeling C J Tucker G Asrar and R RNemani ldquoIncreased plant growth in the northern high latitudesfrom 1981 to 1991rdquoNature vol 386 no 6626 pp 698ndash702 1997

[27] J G Salomon C B Schaaf A H Strahler F Gao and Y JinldquoValidation of theMODIS bidirectional reflectance distributionfunction and albedo retrievals using combined observationsfrom the aqua and terra platformsrdquo IEEE Transactions onGeoscience and Remote Sensing vol 44 no 6 pp 1555ndash15642006

[28] J A Valiente M Nunez E Lopez-Baeza and J F MorenoldquoNarrow-band to broad-band conversion for Meteosat-visiiblechannel and broad-band albedo using both AVHRR-1 and -2channelsrdquo International Journal of Remote Sensing vol 16 no6 pp 1147ndash1166 1995

[29] A H D C Teixeira W G M Bastiaanssen M D AhmadM S B Moura and M G Bos ldquoAnalysis of energy fluxesand vegetation-atmosphere parameters in irrigated and naturalecosystems of semi-arid Brazilrdquo Journal of Hydrology vol 362no 1-2 pp 110ndash127 2008

[30] J W Rouse R H Haas J A Schell and D W DeeringldquoMonitoring vegetation systems in the Great Plains with ERTSrdquoin Proceedings of the 3rd Earth Resources Technology Satellite-1Symposium NASA SP-351 pp 309ndash317 1973

[31] Y Oguro S Ito and K Tsuchiya ldquoComparisons of brightnesstemperatures of landsat-7ETM+ and TerraMODIS aroundHotien Oasis in the Taklimakan Desertrdquo Applied and Environ-mental Soil Science vol 2011 Article ID 948135 11 pages 2011


[32] J H Zar Biostatistical Analysis Prentice Hall Upper SaddleRiver NJ USA 3rd edition 1996

[33] J Fox ldquoThe R commander a basic-statistics graphical userinterface to Rrdquo Journal of Statistical Software vol 14 no 9 pp1ndash42 2005

[34] J Sanders VeuszmdashA Scientific Plotting Package Max PlanckInstitute Garching bei Munchen Germany 2015

[35] F A Heinsch M Zhao S W Running et al ldquoEvaluation ofremote sensing based terrestrial productivity from MODISusing regional tower eddy flux network observationsrdquo IEEETransactions on Geoscience and Remote Sensing vol 44 no 7pp 1908ndash1923 2006

[36] D P Turner S Urbanski D Bremer et al ldquoA cross-biomecomparison of daily light use efficiency for gross primaryproductionrdquo Global Change Biology vol 9 no 3 pp 383ndash3952003

[37] D P Turner W D Ritts W B Cohen et al ldquoScaling GrossPrimary Production (GPP) over boreal and deciduous forestlandscapes in support of MODIS GPP product validationrdquoRemote Sensing of Environment vol 88 no 3 pp 256ndash270 2003

[38] T C Costa L J Accioly M A Oliveira N Burgos and F HSilva ldquoPhytomass mapping of the ldquoserido caatingardquo vegetationby the plant area and the normalized difference vegetationindecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash715 2002

[39] J Domiciano Galvıncio M S Beserra de Moura T G Freireda Silva B B da Silva and C R Naue ldquoLAI improved to dryforest in semiarid of the Brazilrdquo International Journal of RemoteSensing Application vol 3 no 4 p 193 2013

[40] J R Jensen Sensoriamento Remoto do Ambiente Uma Perspec-tiva em Recursos Terrestres Parentese Sao Jose dos CamposBrazil 1st edition 2009

[41] S J Reddy ldquoClimatic classification the semi-arid tropics and itsenvironmentmdasha reviewrdquo Pesquisa Agropecuaria Brasileira vol18 pp 823ndash847 1983

[42] S H Bullock H A Mooney and E Medina Seasonally DryTropical Forests Cambridge University Press 1995

[43] N Agam and P R Berliner ldquoDiurnal water content changes inthe bare soil of a coastal desertrdquo Journal of Hydrometeorologyvol 5 no 5 pp 922ndash933 2004

[44] R G Allen M Tasumi A Morse et al ldquoSatellite-based energybalance for mapping evapotranspiration with internalized cal-ibration (METRIC)mdashapplicationsrdquo Journal of Irrigation andDrainage Engineering vol 133 no 4 pp 395ndash406 2007

Submit your manuscripts athttpswwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of


EarthquakesJournal of


Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining


Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of


Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering


GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of


OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in


MineralogyInternational Journal of


MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Geological ResearchJournal of


Geology Advances in


WaterAgricultureCaatinga

10

FLUX TOWER

N

km

9∘20998400 S

9∘S

8∘40998400 S

40∘40998400W 40∘20998400W41∘W

Figure 1 Location of the Flux tower installed in an area of Caatinga at Embrapa Semiarido which is a research station of the state ofPernambuco Brazil Land use map source PROBIO

state and accurate information about temporal and spatialvariations in ET could assist in their implementation

Spatial heterogeneity of ET can be great due to spatialvariations in vegetation weather soils and topography As aresult extrapolation of ET estimates fromflux towers to largerareas can be misleading [1 10] Also ET is spatially auto-correlated so changes at a point can affect its surroundings[11] Direct measurement of ET in the field is expensive anddifficult therefore satellite-based products that can be usedto remotely estimate ET have become very popular amongresearchers and government agencies Several models oflandscape energy balance based on remotely sensed data havebeen developed for example SEBAL (Surface Energy BalanceAlgorithm for Land [12 13]) MOD16 [7 14] and SAFER(Simple Algorithm For Evapotranspiration Retrieving [8])Each of these models has advantages and disadvantagesdepending on where they are being applied [15] Becausemodels are often based on a global scale [7] or on a model-species approach [16 17] calibration and validation may berequired to parameterize a model to a given set of localconditions [18]

Validation is used to analyze the uncertainties of modeloutputs ensuring the accuracy of remotely sensed ET andenhancing its applications Awide range ofmethods has beenused but in most cases deviance measures and statisticaltests are recommended ones The first one evaluates thedifferences between the simulated and observed values [19]With squared deviations Root Mean Square Errors (RMSE)can be useful to derive statistical properties The second oneis better for bias analysis It is important to use both methodsbecause a dataset could be extremely close in absolute valuesbut spatially removed from the perfect covariance fit (ieregression curve) and vice versa especially when it is anonlinear relationship like an exponential fit For that thecoefficient of determination (1199032) is usually the measurementto evaluate the proportion of output variation that can beexplained by the fit curve

In this paper we hypothesize that MODIS products canbe used to accurately estimate ET in areas of Caatinga vege-tation in northeastern Brazil which is a neglected Brazilianvegetation type in terms of research and investments [20] Inour laboratory Machado et al [21] have previously estimatedET using Landsat imagery but we believe the MODIS sensormay be better suited for monitoring ET in Caatinga becauseof its high temporal resolution In this study we usededdy covariance data to calibrate MODIS estimates of ETfor Caatinga vegetation on monthly 8-day and daily timescales

2 Methodology

21 Study Area The experiment was conducted in a pre-served flat area of Caatinga (asymp600 ha) located at EmbrapaSemiarido research station of the state of Pernambuco Brazil(40∘1910158401610158401015840W 9∘0210158403310158401015840S 350m) (Figure 1) Caatinga is ahyperxerophilic vegetation type that consists of deciduousxerophytic shrubs and trees with an average height of 5mand over 1000 vascular plant species [20] In the studyarea dominant plant species include Poincianella microphylla(Mart ex G Don) Croton conduplicatus (Kunth) Bauhiniacheilantha (Bong) Manihot pseudoglaziovii (Pax amp Hoff-man) and Commiphora leptophloeos (Mart) The climatetype is BSwh (Koopen) or semiarid with rainy season fromJanuary to April Average annual rainfall is 510mm andit is temporally and spatially very heterogeneous Mean airtemperature is 262∘C [22]

22 Eddy Covariance Flux Tower Meteorological and waterflux measurements were taken throughout 2012 with sensorsinstalled in a 16m tower located in the study area Anet radiometer (model CNR-1 Kipp amp Zonen B V DelftNetherlands) installed at 133m above the soil surface wasused to determine incoming solar radiation (119877119878) Air tem-perature and relative humidity (HMP45C Vaisala Helsinki






ET = Δ (119877119899 minus 119866) + 120588119862119901 (119890119904 minus 119890119886) 119903119886Δ + 120574 (1 + 119903119904119903119886) (1)





119901119887 = 119886 (DN119887 minus 119887) (2)




120572119904 = 119886 + 1198871199011 + 1198881199012 (3)


NDVI = (1199012 minus 11990111199012 + 1199011) (4)


119879119887 = ( 1198702ln (1198701119871119887 + 1))

1198790 = 11988611987931 + 11988711987932(5)


1198701 = 2ℎ1198882120582minus510minus61198702 = ℎ119888

120590120582 (6)


ETET0

= exp [119886 + 119887( 1198790120572119904NDVI)] (7)



ET0



119877ns = (1 minus 120572) 119877119878120593 = 120587

180 119897119889119903 = 1 + 0033 cos( 2120587

365119895) 120575 = 0409 sin( 2120587




(9)


1199062 = 119906119911 487ln (678119911 minus 542) (10)

119890119904 = 06108 exp( 1727119879119879 + 273) (11)

119890119886 = RH100119890119904 (12)


0

20

40

60

80

100

Relat

ive h

umid

ity (

)

2 3 4 5 6 7 8 9 10 11 121Month

0

5

10

15

20

25

30

35G

loba

l rad

iatio

n (M

Jmminus2)

1 2 3 4 5 6 7 8 9 10 11 12Month

Month1 2 3 4 5 6 7 8 9 10 11 12

0

10

20

30

40

50

Rain

fall

(mm

)

1 2 3 4 5 6 7 8 9 10 11 12Month

0

5

10

15

20

25

30

35

Air

tem

pera

ture

(∘C)


Δ = 4098 times 06108119890119904(119879 + 273)2 (13)

120574 = 0665 sdot 10minus3 (1013 (293 minus 00065119911293 )526) (14)





1 2 3 4 5 6 7 8 9 10 11 12Month

0

01

02

03

04

05N

DV

I1 2 3 4 5 6 7 8 9 10 11 12

Month

0

005

01

015

02

Alb

edo

()

0

20

40

60

80

100

Evap

otra

nspi

ratio

n (E

T 0 m

m)

2 3 4 5 6 7 8 9 10 11 121Month

2 3 4 5 6 7 8 9 10 11 121Month

30

325

35

375

40

425

45

Surfa

ce te

mp

(∘C)






y = 07341x minus 567

r2 = 077

p lt 005

0

10

20

30

40

50

60To

wer

evap

otra

nspi

ratio

n (m

m m

onth

minus1)


y = 30195e00477x

r2 = 082

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

y = 51181e00469x

r2 = 079

p lt 0001

0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)


y = 06376x + 30758

r2 = 092

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)





LAI = 06401119890(26929timesNDVI) (15)


)] (16)


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)





y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in






ET = Δ (119877119899 minus 119866) + 120588119862119901 (119890119904 minus 119890119886) 119903119886Δ + 120574 (1 + 119903119904119903119886) (1)





119901119887 = 119886 (DN119887 minus 119887) (2)




120572119904 = 119886 + 1198871199011 + 1198881199012 (3)


NDVI = (1199012 minus 11990111199012 + 1199011) (4)


119879119887 = ( 1198702ln (1198701119871119887 + 1))

1198790 = 11988611987931 + 11988711987932(5)


1198701 = 2ℎ1198882120582minus510minus61198702 = ℎ119888

120590120582 (6)


ETET0

= exp [119886 + 119887( 1198790120572119904NDVI)] (7)



ET0



119877ns = (1 minus 120572) 119877119878120593 = 120587

180 119897119889119903 = 1 + 0033 cos( 2120587

365119895) 120575 = 0409 sin( 2120587




(9)


1199062 = 119906119911 487ln (678119911 minus 542) (10)

119890119904 = 06108 exp( 1727119879119879 + 273) (11)

119890119886 = RH100119890119904 (12)


0

20

40

60

80

100

Relat

ive h

umid

ity (

)

2 3 4 5 6 7 8 9 10 11 121Month

0

5

10

15

20

25

30

35G

loba

l rad

iatio

n (M

Jmminus2)

1 2 3 4 5 6 7 8 9 10 11 12Month

Month1 2 3 4 5 6 7 8 9 10 11 12

0

10

20

30

40

50

Rain

fall

(mm

)

1 2 3 4 5 6 7 8 9 10 11 12Month

0

5

10

15

20

25

30

35

Air

tem

pera

ture

(∘C)


Δ = 4098 times 06108119890119904(119879 + 273)2 (13)

120574 = 0665 sdot 10minus3 (1013 (293 minus 00065119911293 )526) (14)





1 2 3 4 5 6 7 8 9 10 11 12Month

0

01

02

03

04

05N

DV

I1 2 3 4 5 6 7 8 9 10 11 12

Month

0

005

01

015

02

Alb

edo

()

0

20

40

60

80

100

Evap

otra

nspi

ratio

n (E

T 0 m

m)

2 3 4 5 6 7 8 9 10 11 121Month

2 3 4 5 6 7 8 9 10 11 121Month

30

325

35

375

40

425

45

Surfa

ce te

mp

(∘C)






y = 07341x minus 567

r2 = 077

p lt 005

0

10

20

30

40

50

60To

wer

evap

otra

nspi

ratio

n (m

m m

onth

minus1)


y = 30195e00477x

r2 = 082

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

y = 51181e00469x

r2 = 079

p lt 0001

0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)


y = 06376x + 30758

r2 = 092

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)





LAI = 06401119890(26929timesNDVI) (15)


)] (16)


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)





y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in



120572119904 = 119886 + 1198871199011 + 1198881199012 (3)


NDVI = (1199012 minus 11990111199012 + 1199011) (4)


119879119887 = ( 1198702ln (1198701119871119887 + 1))

1198790 = 11988611987931 + 11988711987932(5)


1198701 = 2ℎ1198882120582minus510minus61198702 = ℎ119888

120590120582 (6)


ETET0

= exp [119886 + 119887( 1198790120572119904NDVI)] (7)



ET0



119877ns = (1 minus 120572) 119877119878120593 = 120587

180 119897119889119903 = 1 + 0033 cos( 2120587

365119895) 120575 = 0409 sin( 2120587




(9)


1199062 = 119906119911 487ln (678119911 minus 542) (10)

119890119904 = 06108 exp( 1727119879119879 + 273) (11)

119890119886 = RH100119890119904 (12)


0

20

40

60

80

100

Relat

ive h

umid

ity (

)

2 3 4 5 6 7 8 9 10 11 121Month

0

5

10

15

20

25

30

35G

loba

l rad

iatio

n (M

Jmminus2)

1 2 3 4 5 6 7 8 9 10 11 12Month

Month1 2 3 4 5 6 7 8 9 10 11 12

0

10

20

30

40

50

Rain

fall

(mm

)

1 2 3 4 5 6 7 8 9 10 11 12Month

0

5

10

15

20

25

30

35

Air

tem

pera

ture

(∘C)


Δ = 4098 times 06108119890119904(119879 + 273)2 (13)

120574 = 0665 sdot 10minus3 (1013 (293 minus 00065119911293 )526) (14)





1 2 3 4 5 6 7 8 9 10 11 12Month

0

01

02

03

04

05N

DV

I1 2 3 4 5 6 7 8 9 10 11 12

Month

0

005

01

015

02

Alb

edo

()

0

20

40

60

80

100

Evap

otra

nspi

ratio

n (E

T 0 m

m)

2 3 4 5 6 7 8 9 10 11 121Month

2 3 4 5 6 7 8 9 10 11 121Month

30

325

35

375

40

425

45

Surfa

ce te

mp

(∘C)






y = 07341x minus 567

r2 = 077

p lt 005

0

10

20

30

40

50

60To

wer

evap

otra

nspi

ratio

n (m

m m

onth

minus1)


y = 30195e00477x

r2 = 082

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

y = 51181e00469x

r2 = 079

p lt 0001

0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)


y = 06376x + 30758

r2 = 092

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)





LAI = 06401119890(26929timesNDVI) (15)


)] (16)


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)





y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


0

20

40

60

80

100

Relat

ive h

umid

ity (

)

2 3 4 5 6 7 8 9 10 11 121Month

0

5

10

15

20

25

30

35G

loba

l rad

iatio

n (M

Jmminus2)

1 2 3 4 5 6 7 8 9 10 11 12Month

Month1 2 3 4 5 6 7 8 9 10 11 12

0

10

20

30

40

50

Rain

fall

(mm

)

1 2 3 4 5 6 7 8 9 10 11 12Month

0

5

10

15

20

25

30

35

Air

tem

pera

ture

(∘C)


Δ = 4098 times 06108119890119904(119879 + 273)2 (13)

120574 = 0665 sdot 10minus3 (1013 (293 minus 00065119911293 )526) (14)





1 2 3 4 5 6 7 8 9 10 11 12Month

0

01

02

03

04

05N

DV

I1 2 3 4 5 6 7 8 9 10 11 12

Month

0

005

01

015

02

Alb

edo

()

0

20

40

60

80

100

Evap

otra

nspi

ratio

n (E

T 0 m

m)

2 3 4 5 6 7 8 9 10 11 121Month

2 3 4 5 6 7 8 9 10 11 121Month

30

325

35

375

40

425

45

Surfa

ce te

mp

(∘C)






y = 07341x minus 567

r2 = 077

p lt 005

0

10

20

30

40

50

60To

wer

evap

otra

nspi

ratio

n (m

m m

onth

minus1)


y = 30195e00477x

r2 = 082

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

y = 51181e00469x

r2 = 079

p lt 0001

0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)


y = 06376x + 30758

r2 = 092

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)





LAI = 06401119890(26929timesNDVI) (15)


)] (16)


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)





y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


1 2 3 4 5 6 7 8 9 10 11 12Month

0

01

02

03

04

05N

DV

I1 2 3 4 5 6 7 8 9 10 11 12

Month

0

005

01

015

02

Alb

edo

()

0

20

40

60

80

100

Evap

otra

nspi

ratio

n (E

T 0 m

m)

2 3 4 5 6 7 8 9 10 11 121Month

2 3 4 5 6 7 8 9 10 11 121Month

30

325

35

375

40

425

45

Surfa

ce te

mp

(∘C)






y = 07341x minus 567

r2 = 077

p lt 005

0

10

20

30

40

50

60To

wer

evap

otra

nspi

ratio

n (m

m m

onth

minus1)


y = 30195e00477x

r2 = 082

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

y = 51181e00469x

r2 = 079

p lt 0001

0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)


y = 06376x + 30758

r2 = 092

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)





LAI = 06401119890(26929timesNDVI) (15)


)] (16)


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)





y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


y = 07341x minus 567

r2 = 077

p lt 005

0

10

20

30

40

50

60To

wer

evap

otra

nspi

ratio

n (m

m m

onth

minus1)


y = 30195e00477x

r2 = 082

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)

y = 51181e00469x

r2 = 079

p lt 0001

0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)


y = 06376x + 30758

r2 = 092

p lt 0001


0

10

20

30

40

50

60

Tow

er ev

apot

rans

pira

tion

(mm

mon

thminus1)





LAI = 06401119890(26929timesNDVI) (15)


)] (16)


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)





y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus3

minus2

minus1

0

1

2

3


(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus3

minus2

minus1

0

1

2

3

(d)





y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


y = 07933e01679x

r2 = 069

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)

y = 07699x minus 17182

r2 = 062

p lt 005


0

5

10

15

20To

wer

evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

y = 14401e01566x

r2 = 065

p lt 005

0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)


y = 06856x + 08387

r2 = 088

p lt 005


0

5

10

15

20

Tow

er ev

apot

rans

pira

tion

(mm

8-d

ayminus1)






Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(a)

Regr

essio

n st

anda

rdiz

ed re

sidua

l


minus6

minus4

minus2

0

2

4

6

(b)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(c)

Regr

essio

n st

anda

rdiz

ed re

sidua

l

minus6

minus4

minus2

0

2

4

6


(d)








431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in




431 to 4015


083 to 1895


Regr

std

res

idua

l

y = 05351x + 01067

r2 = 085

p lt 005


0

05

1

15

2

25

3

Tow

er ev

apot

rans

pira

tion

(mm

day

minus1)


minus4

minus2

0

2

4




4 Conclusion



MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


MOD16A2SAFER

Tower

1 5 10 15 20 25 30 35 40 45Period of 8 days (∘)

0

5

10

15

20

Evap

otra

nspi

ratio

n (m

m 8

-day

minus1)

MOD16A2SAFER

Tower

1 3 5 7 9 11Month

0

10

20

30

40

50

60Ev

apot

rans

pira

tion

(mm

mon

thminus1)

SAFERTower

0

05

1

15

2

25

3

456

Evap

otra

nspi

ratio

n (m

m d

ayminus1)

62 123 184 245 306 3661Day of the year


Competing Interests


Acknowledgments


References
























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in






















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in

Research Article Reliability of MODIS Evapotranspiration ... · Caatinga 10 FLUX TOWER N km 9 20 S...

Documents

Transcript of Research Article Reliability of MODIS Evapotranspiration ... · Caatinga 10 FLUX TOWER N km 9 20 S...