Water Resour Manage DOI 10.1007/s11269-013-0360-x

A Modified SEBAL Modeling Approach for Estimating Crop Evapotranspiration in Semi-arid Conditions Giorgos Papadavid & Diofantos G. Hadjimitsis & Leonidas Toulios & Silas Michaelides

Received: 5 October 2012 / Accepted: 21 April 2013 # Springer Science+Business Media Dordrecht 2013

Abstract Remote sensing methods are becoming attractive to estimate crop evapotranspiration, as they cover large areas and can provide accurate and reliable estimations; intensive field monitoring is also not required, although some ground-truth measurements can be helpful in interpreting satellite images. For the purposes of this paper, modeling and remote sensing techniques were integrated for estimating actual evapotranspiration of groundnuts (Arachishypogaea, L.) that is cultivated near Mandria Village in Paphos District of Cyprus. The Surface Energy Balance Algorithm for Land (SEBAL) was adopted for the first time in Cyprus, employing the essential adaptations for local soil and meteorological conditions. Landsat-5 TM and 7 ETM+ images were used to retrieve the needed spectral data. The SEBAL model is enhanced with empirical equations determined as part of the present study, regarding crop canopy factors, in order to increase its accuracy. Maps of ETa were created using the SEBAL modified model (CYSEBAL) for the area of interest. The results have been compared to the measurements from an evaporation pan (which was used as a reference) and those of the original SEBAL model. The statistical comparison has shown that the modified SEBAL yields results that are comparable to those of the evaporation pan. T-test application has revealed that the statistical difference between SEBAL and CYSEBAL is significant and quite crucial, especially in a place with limited surface and underground water resources. Keywords SEBAL model . Evapotranspiration . Crop canopy factors . Remote sensing G. Papadavid (*) Agricultural Research Institute, Athalassa 1516, Nicosia, Cyprus e-mail: [email protected] D. G. Hadjimitsis Department of Civil Engineering & Geomatics, Cyprus University of Technology, 31 Archbishop Kyprianos, P.O. Box 50329, 3603 Lemesos, Cyprus L. Toulios National Agricultural Research Foundation (NAGREF), 1, Theofrastou str., 413 35 GR Larissa, Greece S. Michaelides Cyprus Meteorological Service, Nicosia, Cyprus

G. Papadavid et al.

1 Introduction The actual evapotranspiration of crops (ETa) is one of the most useful indicators for optimizing crop production. The variability of ETa, from a spatiotemporal viewpoint and for different land use classes, is thought to be highly indicative for the adequacy, reliability and equity in water use. Evapotranspiration estimation is important for hydrologic modeling and irrigation scheduling (Rogers et al. 1983; Souch et al. 1996; Pereira et al. 1999). Unfortunately, ETa estimation under actual field conditions is still a very challenging task for scientists and water managers (Petra et al. 2010). The Surface Energy Balance Algorithm for Land (SEBAL; see Bastiaanssen 2000; Bastiaanssen et al. 1998, 2005) is a model used in several countries; however, the effectiveness of this model over different geographical areas is still an open question. Within this framework, the present study explores the importance of appropriately revising SEBAL over the island of Cyprus for estimating crop evapotranspiration of groundnuts, under semi-arid conditions. Currently, remote sensing based agro-meteorological models are the most suitable for estimating crop water use at both field and regional scales (Bastiaanssen et al. 2005; Papastergiadou et al. 2008). Numerous evapotranspiration algorithms have been developed making use of remote sensing data acquired from sensors on airborne and satellite platforms (D’Urso and Menenti 1995; Roerink et al. 1997; Hadjimitsis et al. 2008). This study demonstrates the application of SEBAL for estimating actual evapotranspiration of groundnuts (Arachis hypogaea, L.), by employing the necessary modifications and adaptations regarding crop canopy parameters, such as Leaf Area Index (LAI) and Crop Height (CH). The SEBAL model has been used in several studies in many countries around the world and it was validated with high accuracy (e.g., Spiliotopoulos et al. 2008; Bandara 2006; Bastiaanssen et al. 2005; Alexandridis and Chemin 2001; Bastiaanssen et al. 2000). SEBAL was originally applied in Egypt (Bastiaanssen et al. 1998) and subsequently in Turkey (Bastiaanssen 2000) and Greece (Alexandridis 2003). Cyprus is located in the crossroads of these countries and it would be very challenging to test its reliability in a country with similar meteorological conditions. Eighteen Landsat 5 TM & 7 ETM+ images were transformed into ETa maps. It is worth clarifying here that the use of different satellite images does not create any problem, because a radiometric calibration is performed and the respective calibration factors are available with each satellite image; therefore, different satellite images provide the same remotely sensed data in terms of reflectance. The images were acquired for specific dates in the irrigation period of the crop in study, which starts in mid May and ends in mid September. Classic SEBAL and modified SEBAL have been applied for the same satellite image; these estimations are subsequently compared to a direct evaporation measurement, namely that from an evaporation pan, Epan, which is hereby used as a reference; it was further examined to determine whether there is room for increasing the algorithm’s accuracy (Kite and Droogers 2000; D’Urso and Menenti 1995). The results of this paper refer to specific dates in the four-year period 2008–2011. A brief description of the SEBAL modeling approach is given in Section 2. Section 3 presents the resources used in the present study, such as data, methodologies and software. The findings are discussed in Section 4 and, finally, concluding remarks are given in Section 5.

2 Overview of the Surface Energy Balance Algorithm (SEBAL) SEBAL computes a complete radiation and energy balance along with the resistances for momentum, heat and water vapor transport for each pixel (Bastiaanssen et al. 1998;

A Modified SEBAL Modeling Approach for Estimating Crop

Bastiaanssen 2000). The key input data for SEBAL consists of spectral radiance in the visible, near-infrared and thermal infrared part of the spectrum. The model can be applied using satellite sensors having a thermal band; Landsat 5 and 7 images were used in this study. In addition to satellite images, the SEBAL approach requires weather parameters (wind speed, humidity, solar radiation, air temperature). These meteorological parameters were used as input for the algorithm and they were provided from a meteorological station very close to the study area. The latent heat flux is calculated from the relationship: lE ¼ Rn  ðG0 þ HÞ

ð1Þ

where, lE is the latent heat flux (W m−2), Rn is the net radiation (W m−2), G0 is the soil heat flux (W m−2) and H is the sensible heat flux (W m−2). In the stages for applying the SEBAL model, a crucial point is the selection of the two ‘anchor’ pixels (the ‘hot’ and the ‘cold’ pixels), over the area of interest. These two pixels are used to find the difference of the temperature between surface temperature (Ts) and air temperature (dT). A linear relationship is assumed between Ts and dT, in the form of: dT ¼ aTs þ b

ð2Þ

where, a and b are the linear relationship constants. To determine these constants, SEBAL uses the two “anchor” pixels for which a value for H can be reliably estimated. Ts is estimated from the thermal band of Landsat 5TM/7ETM+ for each pixel, while dT, for either the ‘hot’or ‘cold’ pixel, is calculated using the relationship .
 ð3Þ ρ cold=hot cp dTcold=hot ¼ Hcold=hot rah cold=hot where, H is sensible heat flux, which can be calculated for the anchor pixels using meteorological data only (temperature, relative humidity and wind speed), ρ is air density (kg/m3), cp is air specific heat (1,004 J/kgK), dT (K) is the temperature difference between two heights, and rah is the aerodynamic resistance to heat transport (s/m) for each ‘cold’ and ‘hot’ pixel. The above linear relationship between dT and Ts is a major presumption in SEBAL. Several studies indicate that this presumption appears to fit a large range of conditions (see Bastiaanssen et al. 1998, 2005 and Bastiaanssen 2000; Tasumi et al. 2000; Bandara 2006). Equation 3 is developed by using the dT values for the cold and hot pixels and surface temperature. The cold pixel is used to define the amount of evapotranspiration, through H, occurring from the most vegetated and well-watered areas of the image. Usually, an alfa-alfa cultivation or water body is used to identify cold pixels in the area of interest. The ‘cold’ pixel was selected from vegetated areas in the image as one where albedo ranges between 0.22 and 0.24 and LAI >3. The ‘hot’ pixel is one where evapotranspiration should be zero; this pixel is usually located in dry bare agricultural fields.

3 Resources and Methodology 3.1 Study Area The study area is located near Mandria village, in the vicinity of Paphos International Airport in Cyprus (see Fig. 1). The area is characterized by mild climate which provides the opportunity for early production of leafy and annual crops. The area is flat and almost at

G. Papadavid et al.

Fig. 1 Landsat 5 satellite image (26 March 2009)

sea level, while the surface can be considered homogenous with only annual leafy vegetables being cultivated. Weather can briefly be described as hot, humid and cloud-free during May to beginning of October (Papadavid and Hadjimitsis 2009). Groundnut is a traditional crop cultivated in Cyprus and especially in the Paphos District, since it requires mild weather conditions and certain type of soils (well-drained, loose, friable medium textured soils). Its growing period varies from 90 to 115 days for the sequential, branched varieties and from 120 to 140 days for the alternately branched varieties (Markou and Papadavid 2007). 3.2 Resources This sub-section outlines the instrumentation, data and software that were used for carrying out the research. Field Spectroradiometer The GER (Geophysical Environmental Research) 1500 field spectroradiometer is a light-weight, high performance, single-beam field spectroradiometer. It is a field portable spectroradiometer covering the ultraviolet, visible and near-infrared wavelengths from 350 nm to 1,050 nm. SunScan Canopy Analyzer for LAI Estimation LAI is commonly used for monitoring crop growth. Instead of the traditional, direct and labor-consuming method of physically measuring the plant with a ruler (direct method), an optical instrument, namely, the SunScan canopy analysis system (Delta-T Devices Ltd., UK) is used (indirect method). The instrument indirectly calculates LAI by measuring the ratio of transmitted radiation through canopy to incident radiation (Lang et al. 1991; Welles and Norman 1991). Time Series of Satellite Images For studies dealing with crop water requirements, spatial, spectral and temporal resolution of satellite images is very important. Landsat 5 TM and 7 ETM+ have been widely used for hydrological studies due to their relatively good temporal resolution (16 days) which is important for providing regular snapshots during the crop growth season (Oetter et al. 2000). Alexandridis (2003) has indicated that the resolution of Landsat images is sufficient when they are to be used for hydrological purposes. An advantage of Landsat images is that they provide a complete coverage of the island in a single image (Fig. 1).

A Modified SEBAL Modeling Approach for Estimating Crop

Meteorological Records Meteorological data for Paphos Airport and for the period of the project (2008–2011) were provided by the Cyprus Meteorological Service and used as input parameters in the SEBAL algorithm (temperature, relative humidity and wind speed). Image Processing Software ERDAS V.10 Imagine image processing software was used for the analysis and interpretation of multi-spectral digital satellite images 3.3 Methodology SEBAL is an algorithm that takes into account crop canopy factors in order to estimate ETa. This fact provides the opportunity to feed the model with local data in an attempt to increase its effectiveness. Empirical equations (using VI’s) describing LAI and CH could be ingested in the algorithm. The methodology adopted in this study is based on the hypothesis that SEBAL’s accuracy could be enhanced if the algorithm is supported with ground data, regarding crop canopy factors and meteorological conditions of the area. In the following, the methodology adopted is analyzed. Spectroradiometric Measurements A two-year (2009–2010 from April to July) field campaign was undertaken in order to collect spectral signatures of each crop included in the study. The aim was to have the reflectance of each crop during their phenological stages after the data was passed through the Relative Spectral Response filters. Field spectroradiometric measurements were made using a GER 1500 field spectroradiometer with reflectance spectrum from 350 nm to 1,050 nm. The task was to estimate the surface reflectance values equivalent to the Landsat TM/ETM+ bands 1, 2, 3 and 4 using in situ the GER1500 spectroradiometer. To filter the data through the Relative Spectral Response (RSR) values of Landsat TM/ETM+, the GER1500 reflectance values were interpolated to obtain the reflectance values at the incremental wavelength of the RSR. It has to be mentioned that the initial spectroradiometric measurement was taken after each crop had enough foliage (27 days) in order to avoid soil effects on spectroradiometric data, while the final measurement was taken just after the crop started drying and became yellowish. LAI and CH measurements were also taken simultaneously to spectroradiometric measurements, following the same phenological cycle of each crop for the corresponding cultivating periods. The purpose was to create time series of these two parameters and correlate them to VI (Baret et al. 2007). Vegetation Indices Time series of VI have been created based on the spectroradiometric reflectance of each crop, in each phenological stage. VI’s have been widely used for assessing vegetation condition, cover, phenology and various processes, such as evapotranspiration, climate- and land-use-change detection and drought monitoring (Haboudane et al. 2004; Glenn et al. 2008). NDVI (Normalized Difference Vegetation Index), SAVI (Soil-adjusted Vegetation Index) and WDVI (Weighted Difference Vegetation Index) are the spectral VI’s that were selected in order to be correlated to LAI and CH. Such indices are found to be widely used in various evapotranspiration algorithms and models (Bannari et al. 1995). Modeling LAI and CH in Terms of VI’s Different models were tested in order to identify the best possible model which better describes LAI and CH in terms of VI’s. The fact that there is no universal equation relating LAI to a VI (Qi et al. 2000), forces remote sensing users to develop equations for each crop under the current relevant conditions, using a substantial

G. Papadavid et al.

amount of true measurements of LAI and remote sensing data, such as spectroradiometric measurements. As a result, each image must be transformed into a LAI map. Factors such as soil, atmospheric effects and topography, which affect VI’s, must be considered and their impact must be minimized in this procedure And thus iImproved VI’s such as SAVI and WDVI, were used for this purpose (Huete 1988; Clevers 1989; Qi et al. 1994). In this present modeling procedure, an attempt has been made to model LAI and CH in terms of one of the three VI’s mentioned earlier (i.e., NDVI, SAVI, WDVI). The models for LAI used in the present study are mathematically expressed as:

& & & &

Linear: LAI ¼ aðVIÞ þ b Exponential: LAI ¼ a ebðVIÞ Logarithmic: LAI ¼ alnðVIÞ þ b Polynomial: LAI ¼ aðVIÞ2 þ bðVIÞ þ c

Similar mathematical relationships were developed for CH modeling. The coefficients a, b and c are determined from the statistical analysis and the most “sound” model, determined on the basis of the respective coefficient of determination (R2) will be used in the algorithms for estimating ETa. Mapping LAI, CH The three crop canopy parameters were mapped using the ERDAS Imagine software. The satellite images were transformed into maps in order to, firstly, test in practice the models and, secondly, to be inserted as inputs to the evapotranspiration algorithms. In order to find out the soundest relationship it is needed to create time series of the needed parameters and by using statistical methods to create empirical models to characterize LAI and CH. Application of SEBAL Algorithm Both the original SEBAL and the one modified by empirical equations (hereafter called as CYSEBAL) were applied in order to test the hypothesis that the latter can provide more accurate results. SEBAL is essentially a single source model that solves the Energy Balance equation and provides maps of ETa on a pixel basis. CYSEBAL — Modifying SEBAL The effectiveness of the proposed CYSEBAL method is assessed by using Epan data for the same dates as the satellite images. T-test is employed to evaluate results and cross-check if there is significant statistical difference between the CYSEBAL and SEBAL results.

4 Results and Discussion It is noticeable that in SEBAL, empirical equations are used to describe parameters that need to be directly measured. The intended purpose is to ingest into the algorithm the needed local empirical equations and check if the accuracy can be enhanced. In this respect, the authors have managed to develop empirical equations, adapted to the Cypriot conditions, and embed them in the algorithm, in an attempt to have as accurate results as possible. The results from both SEBAL and CYSEBAL are compared to those of the reference method measurements derived from the same plot. 4.1 Modeling LAI and CH with VI’s The use of VI’s for statistically describing LAI and CH is common in international literature and many empirical models are available depending on the conditions and the place (Running and Coughlan 1988; Tiktak and Van Grinsven 1995; Clevers 1989). Time series

A Modified SEBAL Modeling Approach for Estimating Crop

of LAI, CH and VI’s (i.e., NDVI, SAVI and WDVI) were created in order to proceed with correlations (Table 1). LAI is used in SEBAL algorithm to estimate the two surface emissivities which are used in turn to infer the surface temperature from the thermal band (band 6) of Landsat satellite in order to calculate the net radiation (Rn). LAI, defined as the ratio of the total area of all leaves of a plant to the ground area of the plant (see Watson 1947), is related to SAVI (Huete 1988, 1989) in the classic SEBAL as shown below:   ln 0:69SAVI 0:59 ð4Þ LAI ¼  0:91 The whole procedure of statistically describing LAI in terms of VI’s in this current effort is presented in Table 2: linear, exponential, logarithmic and polynomial models were used to infer the best fitted model under the current conditions. Using empirical modeling techniques and field spectroradiometric data, it was found that the coefficient of determination obtains the highest value (R2 =0.88) when LAI is correlated to SAVI. In order to be as accurate as possible, the measurements for both LAI and CH were taken following the phenological stages of groundnuts (Papadavid and Hadjimitsis 2009). The mathematical model used to describe LAI in terms of SAVI is the following: LAI ¼ 7:99ðSAVIÞ  3:29

ð5Þ

Table 1 Spectroradiometric data, VI and Crop canopy factors (the phenological stages are: 1 = vegetative, 2 = flowering, 3 = yield formation, 4 = ripening) Phenological stage

Band 1

Band 2

Band 3

Band 4

Vegetation indices

Canopy factors

(450– 520 nm)

(520– 600 nm)

(630– 670 nm)

(760– 900 nm)

NDVI

SAVI

WDVI

LAI

CH

1

0.03

0.14

0.08

0.42

0.68

0.54

0.31

0.90

0.19

1

0.02

0.14

0.08

0.40

0.66

0.52

0.29

0.90

0.20

1

0.03

0.13

0.09

0.48

0.69

0.59

0.36

0.90

0.21

1 2

0.03 0.03

0.15 0.14

0.10 0.10

0.50 0.50

0.66 0.67

0.58 0.58

0.36 0.36

1.00 1.10

0.22 0.25

2

0.02

0.12

0.11

0.51

0.65

0.57

0.36

1.40

0.27

2

0.02

0.13

0.10

0.52

0.68

0.60

0.38

1.40

0.28

2

0.02

0.13

0.09

0.53

0.71

0.63

0.41

1.60

0.29

2

0.02

0.12

0.08

0.53

0.74

0.65

0.42

2.00

0.35

3

0.05

0.15

0.09

0.56

0.72

0.65

0.44

2.40

0.37

3

0.05

0.16

0.08

0.57

0.75

0.68

0.46

2.50

0.41

3 3

0.03 0.03

0.13 0.13

0.07 0.08

0.66 0.69

0.81 0.79

0.77 0.77

0.57 0.58

2.70 2.90

0.43 0.44

3

0.04

0.13

0.08

0.68

0.79

0.76

0.57

2.80

0.44

3

0.03

0.09

0.07

0.63

0.80

0.74

0.53

2.70

0.43

4

0.03

0.09

0.05

0.62

0.86

0.78

0.55

2.50

0.43

4

0.01

0.08

0.05

0.53

0.81

0.70

0.45

2.50

0.42

4

0.02

0.05

0.06

0.43

0.76

0.60

0.35

2.20

0.41

y=7.99x−3.29

y=6.56x−1.08

LAI-SAVI

LAI-WDVI

y=1.32x−0.64

y=0.89x−0.28 y=0.807x−0.03

CH to NDVI

CH to SAVI CH to WDV

CH (y) to VI’s (x)

y=0.07x+0.58

LAI-NDVI

LAI(y) to VI’s (x)

Model

Linear

0.86 0.77

0.81

0.81

0.88

0.78

R

2

0.76 0.85 0.76

0.78

y=0.31e3.80x

y=0.04e2.90x y=0.098e2.6x

0.86

y=0.01e4.17x

0.79

y=0.09e4.19x

R

2

y=0.60e0.10x

Model

Exponential

Table 2 Results of the correlation analysis for LAI and CH to VI

y=0.65 ln(x)+0.62 y=0.36 ln(x)+0.63

y=0.98 ln(x)+0.64

y=2.96 ln(x)+4.31

y=4.81 ln(x)+3.75

y=0.12 ln(x)+0.66

Model

Logarithmic

0.87 0.79

0.83

0.82

0.87

0.74

R

2

0.85 0.86 0.8

y ¼ 1:02x2 þ 2:30x  0:75 y ¼ 1:61x2 þ 2:3243x  0:37

0.83

y ¼ 7:84x2 þ 13:91x  2:73 y ¼ 5:17x2 þ 9:02x  3:47

0.79 0.87

y ¼ 2:36x2 þ 10:39x  4:10

R2

y ¼ 0:01x2 þ 0:01x þ 0:63

Model

Polynomial

G. Papadavid et al.

A Modified SEBAL Modeling Approach for Estimating Crop

CH is used to infer the momentum roughness (Zom) in order to estimate the friction velocity and then the aerodynamic resistance to heat transport that is a basic element for Sensible Heat Flux (Santos et al. 2012). In classic SEBAL, Zom is found by using the empirical relationship: Zom ¼ 0:12 CH

ð6Þ

where, CH is set to a standard height, usually between 0.25 and 0.30 cm, or by utilizing a LAI map. In the present research, the above equation was modified by using a VI to estimate CH and create a map of Zom, which was subsequently used as input in the algorithm describing the whole area. Following the same procedure as for LAI, CH was described in the present research by using SAVI as follows: CH ¼ 0:65 lnðSAVIÞ þ 0:62

ð7Þ 2

This is the relationship in Table 2, having the highest coefficient of determination (R =0.87). By adopting the above logarithmic model, Eq. 6 was subsequently transformed to: Zom ¼ 0:07 lnðSAVIÞ þ 0:34

ð8Þ

Equations 5 and 7 are used to calculate LAI and CH, respectively, with the use of satellite data. These equations have been fed to ERDAS Imagine software (Modeler module), in order to create LAI and CH maps (Fig. 2). These maps are the satellite images transformed to pixel based maps, where users can retrievethe values of the two specific parameters, LAI and CH. The specific rasters are used as inputs in the SEBAL algorithm. One shoulc curry in mind

Fig. 2 Generation of LAI (b) and CH (c) maps (in pseudo color) using Landsat image (a)

G. Papadavid et al.

that since the equations refer to groundnuts, they can exclusively be used for this specific crop and not for any other. Since SAVI is the vegetation index that was found to be best fitted for both models, it is useful to address very briefly this particular vegetation index. SAVI is described by the following equation: SAVI ¼ ½ðNIR  RÞ=ðNIR þ R þ LÞð1 þ LÞ

ð9Þ

where NIR is the reflectance value of the near infrared band, R is reflectance of the red band, and L is the soil brightness correction factor. SAVI was adapted to the local conditions with the custom soil-adjustment constant L set, using more than 250 ground spectroradiometric measurements in the years 2009–2010 (Fig. 3); L is the sum of l1 and l2 (the co-ordinates of the point, where Early Vegetation Line and Soil Line are intersected), as described by Huete (1988). Indeed, L is basically the offset of the NIR/IR (Near Infrared/Infrared Reflectance), and it is set to the value of 0.41, as it is determined from Fig. 3. When semi-empirical models are retrieved from the statistical analysis, then LAI and CH maps are created using the available satellite images. 4.2 Employing the Modified SEBAL Model (CYSEBAL) Maps of actual evapotranspiration were finally created for each satellite image, using the empirical Eqs. 5 and 7. These maps were employed to infer the ETa of groundnuts for all the available images at from Landsat 5 and Landsat 7 satellites (Table 3). The value of ETa refers to the mean value of the four plots of groundnuts in the area of interest that follow the same phenological cycle and are subject to the same meteorological conditions. ETa values of groundnuts were compared to the Epan values provided by the Agricultural Research Institute (ARI) of Cyprus. The ARI has the resources for estimating ETa by using Epan data that were collected since 1978 (see Metochis 1997). As it can be inferred from Table 3, the results of the three methods applied are very close. It is noteworthy that that results from both SEBAL and CYSEBAL follow the same variations as the results of the direct Epan method, implying that when the Epan results

50

NIR reflectance (%)

40

-30

30 20

10 0 -20

-10

0 -10

l2

-20

l1

-30 -40 IR reflectance (%)

Fig. 3 Adapting SAVI to ground conditions

10

20

30

A Modified SEBAL Modeling Approach for Estimating Crop Table 3 SEBAL, CYSEBAL and Epan measurements. Numbers in parenthesis represent the percentage error (%) with respect to the reference method (Epan) Satellite image

Satellite Sensor

SEBAL(mm/day)

CYSEBAL(mm/day)

Epan (mm/day)

12 July 2008

Landsat ETM+

5.6 (2)

5.4 (−2)

5.5

28 July 2008

Landsat ETM+

5.6 (−3)

5.8 (0)

5.8

13 August 2008

Landsat ETM+

5.1 (21)

4.6 (10)

4.2

29 August 2008

Landsat ETM+

4.9 (20)

4.6 (12)

4.1

29 June 2009

Landsat ETM+

4.6 (7)

4.2 (−2)

4.3

7 July 2009

Landsat TM

4.8 (−13)

5.1 (−7)

5.5

15 July 2009

Landsat ETM+

6.1 (7)

5.9 (4)

5.7

23 July 2009 16 August 2009

Landsat TM Landsat ETM+

6.1 (5) 5.4 (20)

5.9 (2) 5.0 (11)

5.8 4.5

25 September2009

Landsat TM

4.8 (26)

3.9 (3)

3.8

31 May 2010

Landsat ETM+

4.3 (23)

3.9 (11)

3.5

27 August 2010

Landsat TM

5.8 (4)

5.2 (−7)

5.6

2 May 2011

Landsat ETM+

4.2 (8)

3.9 (0)

3.9

19 June 2011

Landsat ETM+

5.0 (19)

4.3 (2)

4.2

5 July 2011

Landsat ETM+

5.1 (4)

4.8 (−2)

4.9

21 July 2011 29 July 2011

Landsat ETM+ Landsat TM

5.6 (6) 5.9 (9)

5.0 (−6) 5.6 (4)

5.3 5.4

30 August 2011

Landsat TM

AVERAGE

5.2 (2)

5.2 (2)

5.1

5.2 (9.2)

4.9 (1.9)

4.8

decrease or increase, there is a respective decrease or increase of the results of SEBAL and CYSEBAL. In can generally be concluded that CYSEBAL yields values that fall between those estimated with the Epan method and the SEBAL (see Fig. 4); also, CYSEBAL’s values are, in general, closer to those of the Epan method. The Epan method was used here as a preliminary tool to assess the effectiveness of the revised proposed CYSEBAL method, despite the fact that Epan refers to point measurements of ETa. What is important to mention is that when SEBAL has its highest deviations for the dates 13 August 2008 (21 %), 25 September 2009 (26 % deviation) and 31 May 2010 (23 % deviation), CYSEBAL produces much better results with only 10 %, 3 % and 11 % deviation, respectively. The average deviation of SEBAL with respect to Epan reaches 9.2 %, while CYSEBAL yields an average deviation of only 1.9 %. Obviously, although the difference of the SEBAL to the Epan method is less than 1 mm/day, the cumulative error will be very important when it comes to monthly irrigation planning; however, by employing the empirical equations, as described above, CYSEBAL yields more accurate results. Figure 4 depicts graphically the results of using the two methods (against Epan method) for calculating evapotranspiration. Although it could clearly be inferred from this figure that CYSEBAL yields superior results compared to SEBAL, it is considered important to substantiate this superiority by using appropriate statistical testing. To this end, T-test was performed. The T-test results are shown in Table 4, revealing that SEBAL and CYSEBAL produce significantly different results. Comparing each of these two models to the Epan measurements, it was found that SEBAL has a significant difference, while CYSEBAL has no significant difference, at the 5 % significance level. It was also found that there is a statistically significant difference between the results produced by CYSEBAL and SEBAL.

G. Papadavid et al.

Fig. 4 Comparing SEBAL and CYSEBAL graphically to Epan method

5 Concluding Remarks The existing literature supports the use of SEBAL as the most promising algorithm that requires minimum input data of ground based variables; it has been widely applied in several countries due to its accurate estimation of actual evapotranspiration. However, further research is required to apply such algorithm and assess its effectiveness in different geographical areas with different meteorological and soil conditions. Indeed, the application of SEBAL algorithm in Cyprus, as performed in this study, has shown that the algorithm could provide better results if it is fed with local empirical equations regarding crop canopy factors. It is the first time that the specific algorithm is employed for estimating ETa in Cyprus. The modified SEBAL algorithm, named as CYSEBAL, has been found to yield closer results to the reference E-pan method. This is an important finding, since it stresses the need to apply SEBAL taking into consideration local soil, geomorphological and meteorological conditions. The results of the T-test revealed that CYSEBAL has no significant difference from Epan method and that the former can be used for estimating ETa in a systematic way. On the one hand, in its classic form, SEBAL could create anomalies in terms of irrigation planning, if it is not supported by appropriate empirical equations, regarding crop canopy factors. On the other hand, CYSEBAL utilizes empirical equations and produces better results, especially at times when SEBAL deviates highly from the Epan measurements. The hypothesis that if SEBAL algorithm is supported with ground data could work better was verified.

Table 4 T-test analysis (5 % significance level)

Analysis

SEBAL

CYSEBAL

T-test statistic

T-test to Epan

3.79

1.02

2.11

T-test to SEBAL

–

4.62

2.11

A Modified SEBAL Modeling Approach for Estimating Crop

The present study explores also the potential of using remotely sensed measurements for obtaining quite accurate estimates of evapotranspiration which can in turn provide irrigation managers and farmers with information that was not previously available and that can enhance irrigation performance for sustainable management of limited water resources. With its semi-arid climatic characteristics, Cyprus is characterized by long lasting very good weather conditions (more than 180 days of cloud free days in a year) that allow the use of satellite remote sensing on a systematic basis. Despite the fact that Landsat-5 TM is an old sensor and Landsat-7 ETM+ faces several acquisition problems, Landsat covers the area of Cyprus twice every month and is still considered as one of the useful tools for determining evapotranspiration. Future research consists of further applying the modified SEBAL algorithm using ASTER or other better resolution satellite imagery and examining more empirical relationships regarding crop canopy parameters of other crops too.

References Alexandridis T (2003) Scale effect on determination of hydrological and vegetation parameters using remote sensing techniques and GIS. PhD Thesis. Aristotle University of Thessaloniki, Greece Alexandridis T, Chemin Y (2001) Irrigation water consumption through remote sensing. Comparison at different scales in Zhanghe irrigation system, China. Presented in First International Conference on Water Resources Management, Halkidiki, Greece, 10pp Bandara KMPS (2006)Assessing irrigation performance by using remote sensing. Doctoral thesis, Wageningen University, Wageningen, The Netherlands Bannari A, Morin D, Huette AR, Bonn F (1995) A review of vegetation indices. Remote Sens Rev 13:95–120 Baret F, Hagolle O, Geiger B, Bicheron P, Miras B, Huc M, Berthelot B, Nino F, Weiss M, Samain O, Roujean JL, Leroy M (2007) LAI, fAPAR and fCover CYCLOPES global products derived from VEGETATION. Part 1: principles of the algorithm. Remote Sens Environ 110:275–286 Bastiaanssen WGM (2000) SEBAL-based sensible and latent heat fluxes in the irrigated Gediz Basin, Turkey. J Hydrol 229:87–100 Bastiaanssen WGM, Menenti M, Feddes RA, Holtslag AAM (1998) A remote sensing surface energy balance algorithm for land (SEBAL), part 1: formulation. J Hydrol 212-213:198–212 Bastiaanssen WGM, Noordman EJM, Pelgrum H, David G, Thoreson BP, Allen RG (2005) SEBAL model with remotely sensed data to improve water resources management under actual field conditions. J Irrig Drain Eng 131:85–93 Clevers JGPW (1989) The application of a weighted infrared-red vegetation index for estimating leaf area index by correcting for soil moisture. Remote Sens Envir 29:25–37 D’Urso G, Menenti M (1995) Mapping crop coefficients in irrigated areas from Landsat TM images. In: Proceedings European Symposium on Satellite Remote Sensing II, Europto, Paris, September 1995,SPIE (International Society of Optical Engineering) 2585, 41–47 Glenn EP, Huete AR, Nagler PL, Nelson SG (2008) Relationship between remotely-sensed vegetation indices, canopy attributes and plant physiological processes: what vegetation indices can and cannot tell us about the landscape. Sensors 8:2136–2160 Haboudane D, Miller JR, Pattey E, Zarco-Tejada PJ, Strachan IB (2004) Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop canopies: modeling and validation in the context of precision agricultural. Remote Sens Envir 90:337–352 Hadjimitsis DG, Papadavid G, Themistocleous K, Kounoudes A, Toulios L (2008) Estimating irrigation demand using satellite remote sensing: a case study of Paphos District area in Cyprus. Proc SPIE Eur Remote Sens, 15–18 September 2008 University of Wales Institute, Cardiff, UK,Proc. SPIE 7104, 71040I (2008);doi:10.1117/12.800366 Huete AR (1988) A soil-adjusted vegetation index (SAVI). Remote Sens Envir 25:295–309 Huete AR (1989) Soil influences in remotely sensed vegetation-canopy spectra. In: Asrar G (ed) Theory and application of optical remote sensing. Wiley, New York, pp 107–141 Kite GW, Droogers P (2000) Comparing evapotranspiration estimates from satellites, hydrological models and field data. J Hydrol 229:3–18 Lang A, McMutrie R, Benson M (1991) Validity of surface area indices of PinusRadiata estimated from transmittance of the sun’s beam. Agric Forest Meteorol 57:157–170

G. Papadavid et al. Markou M, Papadavid G (2007) Norm input output data for the main crop and livestock enterprises of Cyprus. Agricultural Economics Report 46, ISSN 0379-0827, p 196–199 Metochis C (1997) Assessment of irrigation water needs of main crops of Cyprus, Cyprus Agricultural Research Institute Series. Ministry of Agriculture, Natural Resources and Environment, Cyprus, Nicosia Oetter DR, Warren BC, Berterretche M, Maiersperger TK, Kennedy RE (2000) Land cover mapping in an agricultural setting using multiseasonal thematic mapper data. Remote Sens Envir 76:139–155 Papadavid G, Hadjimitsis DG (2009)Spectral signature measurements during the whole life cycle of annual crops and sustainable irrigation management over Cyprus using remote sensing and spectroradiometric data: the cases of spring potatoes and peas. Proc. of SPIE, Remote Sensing for Agriculture, Ecosystems, and Hydrology XI, Vol 7472, 747215. doi:10.1117/12.830552 Papastergiadou ES, Retalis A, Apostolakis A, Georgiadis T (2008) Environmental monitoring of spatiotemporal changes using remote sensing and GIS in a Mediterranean Wetland of Northern Greece. Water Resour Manag 22:579–594 Pereira LS, Perrier A, Allen RG, Alves I (1999) Evapotranspiration: concepts and future trends. J Irrig Drain Eng 125:45–51 Petra JG, Hellegers J, Soppe R, Perry CJ, Bastiaanssen WGM (2010) Remote Sensing and Economic Indicators for Supporting Water Resources Management Decisions. Water Resour Manag 24:2419–2436 Qi J, Kerr Y, Chehbouni A (1994) External factor consideration in vegetation index development. Proc. Phys Meas Signature Remote Sens ISPRS, pp 723–730 Qi J, Kerr YH, Moran MS, Weltz M, Huete AR, Sorooshian S, Bryant R (2000) Leaf area index estimates using remotely sensed data and BRDF models in a semiarid region. Remote Sens Envir: 73:18–30 Roerink G, Bastiaanssen WGM, Chambouleyron J, Menenti M (1997) Relating crop water consumption to irrigation water supply by remote sensing. Water Resour Manag 11:445–465 Rogers JS, Allen LH Jr, Calvert DV (1983) Evapotranspiration from a humid-region developing citrus grove with a grass cover. Trans ASAE 26:1778–1792 Running SW, Coughlan JC (1988) A general model of forest ecosystem processes for regional applications. I. Hydrological balance, canopy gas exchange and primary production processes. Ecol Model 42:125–154 Santos C, Lorite IJ, Allen RG, Tasumi M (2012) Aerodynamic parameterization of the satellite-based energy balance (METRIC) model for ET estimation in rainfed olive orchards of Andalusia, Spain. Water Resour Manag 26:3267–3283 Souch C, Wolfe CP, Grimmond CSB (1996) Wetland evaporation and energy partitioning: Indiana Dunes National Lakeshore. J Hydrol 184:189–208 Spiliotopoulos M, Loukas A, Vasiliades L (2008)Actual evapotranspiration estimation from satellite-based surface energy balance model in Thessaly, Greece. EGU General Assembly, 13-18 April 2008, Vienna, Austria, Geophysical Research Abstracts: Vol 10 Tasumi M, Bastiaanssen WGM, Allen RG (2000) Application of the SEBAL methodology for estimating consumptive use of water and stream flow depletion in the Bear River Basin of Idaho through Remote Sensing. EOSDIS Project Report, Raytheon Systems Company and the University of Idaho, USA Tiktak A, Van Grinsven HJM (1995) Review of sixteen forest–soil–atmosphere models. Ecol Model 83:35–53 Watson DJ (1947) Comparative physiological studies on growth of field crops: I. Variation in net assimilation rate and leaf area between species and varieties, and within and between years. Ann Bot-London 11:41– 76 Welles JM, Norman JM (1991) Instrument for indirect measurement of canopy architecture. Agron J 83:818– 825