Evaluation of evapotranspiration of wheat using SEBAL algorithm (Case study: Agricultural Research Station of Haji Abad)

Introduction
Evapotranspiration (ET), which includes water evaporation from soil surface and vegetation transpiration, represents a fundamental process of the hydrological cycle. For water resource management, especially in arid and semi-arid regions, this is a key element.  
A lot of empirical methods have been developed in order to estimate ET from meteorological data since 50 years ago. The major problem of this method is that it can be used for evaluation of uniform regions near the station. This could not be extracted for other regions. Nowadays, remote sensing based methods often used for calculation of different parameters of ET, are suitable to extract different parameters of ET at proper temporal and spatial scales.
 
Materials and methods
One of the most famous algorithms for estimation of actual evapotranspiration is Surface Energy Balance Algorithm (SEBAL). This algorithm calculates all fluxes of the energy balance at the earth's surface including net radiation (R_n), soil heat flux (G), and sensible heat flux (H) from satellite images. Finally, actual ET is computed based on the energy balance at the earth surface. The aim of the current study is to evaluate the spatial and temporal variation of actual evapotranspiration of wheat in Agricultural Research Station of Haji Abad using SEBAL algorithm. The area has geographic coordinates 55° 54 'N, 28° 18' E, with an elevation of 900 m above mean sea level. For this purpose, 4 cloud free Landsat 7 / ETM+ images are used. Landsat 7 Enhanced Thematic Mapper Plus (ETM+) in cloud free satellite images were downloaded from the USGS Earth Explorer site [(http://edcsns17.cr.usgs.gov/NewEarthExplorer/)].
 
Table 1. Applied LANDSAT data to estimate actual evapotranspiration




Satellite / sensor


Date


Pass No




LANDSAT 7 /  ETM⁺


2005-01-25


p160r40




LANDSAT 7 /  ETM⁺


2005-02-10


p160r40




LANDSAT 7 /  ETM⁺


2005-03-30


p160r40




LANDSAT 7 /  ETM⁺


2005-04-15


p160r40




 
Results and discussion
In order to evaluate SEBAL estimation of evapotranspiration, recorded data of lysimeter (located in Agricultural Research Station of Haji Abad) is used for grass. The data is converted into ET of wheat based on a field study by Moradi, 2002 (Table 2). 
 
Table 2: converting ET of grass to wheat (Moradi, 2002)




time


Evapotranspiration of grass (lysimeter)


Correction Coefficient


Evapotranspiration of wheat (for 10 days)


Daily evapotranspiration of wheat




20 to 29 January 2005


22.8


0.92


20.97


2.09




8 to 18 February 2005


27.3


1.1


30.03


3.003




21 to 30 march 2005


46.5


1.1


51.15


5.11




9 to 19 April 2005


69


1.1


75.9


7.59




 
We have compared the results of the SEBAL algorithm with lysimeter data. In the comparison, the Mean Absolute Error between the results was 0.7 mm per day and the coefficient of determination of R²=0.77. Statistical analysis by T test does not show a significant difference between the results of the SEBAL algorithm and lysimeter. Therefore, the results show that the SEBAL algorithm has reasonable potential to estimate evapotranspiration in the study area. 
The difference between SEBAL and Lysimeter data show:

Cold and warm pixels; this is a very sensitive phase of SEBAL algorithm based on surface temperature and on leaf area index. Alen et al., 2002, used a well irrigated Alfalfa field as cold pixels, while we used wheat field.  
SEBAL algorithm used hourly meteorological data to estimate ET to estimate ET of Reference crop (grass) for daily average of these data. Evaporation is a function of temperature and wind speed. The effects of these parameters are simultaneous with the passing time of satellite. Thus, this plays an important role in estimation of ET.
SEBAL parameters are not calibrated for our study area.

 
These results confirm Hongjun et al. 2008; Rahimiyan et al. 2012; Teixeira et al. 2009 and Karimi et al. 2013.
 
Conclusion
Results of our study is different from Ayenew (2003). He believes that using SEBAL algorithm is not appropriate for a short period. Comparison between evapotranspiration from SEBAL algorithm and from Lysimeter shows a small difference in growing season, because SEBAL algorithm uses hourly weather data (at the passing moment of satellite) whereas mean daily data is used for Lysimerer. The evapotranspiration is a function of temperature and wind speed, so the effects of these parameters on the passing moment of satellite play an important role in estimation of the evapotranspiration.