Evapotranspiration (ET) is one of the most important components of the hydrological cycle, but it is often the most difficult variable to determine at basin scale. Traditionally, ET is estimated using point-based measurements collected at meteorological stations, but the non-spatial nature of these measurements often leads to significant errors when utilized at watershed scale. In this study, the METRIC (mapping evapotranspiration at high resolution with internalized calibration) approach was evaluated using remotely sensed observations from the moderate resolution imaging spectroradiometer sensor and data from meteorological stations in the lower catchment of the Buyuk Menderes Basin in western Turkey in the form of actual ET maps at daily and monthly intervals between 1st April and 30th September 2010. The energy fluxes and daily ET maps resulting from METRIC were compared with ET data estimated with the help of meteorological parameters. These results were found to be compatible with the ground-based estimations which suggest considerable potential for the METRIC model for estimating spatially distributed actual ET values with little ground-based weather data.

energy balance, ET, METRIC, MODIS, remote sensing

Evapotranspiration (ET) is an important component of the Earth's water and energy cycles. For example, average annual evapotranspiration from the global land surface is around 60 to 65% of precipitation (Baumgartner & Reichel 1975). Estimation of actual ET is also crucial for hydrologic modeling, water and irrigation planning, management, drought assessment, water rights, aquifer recharge assessment, and many other areas. One of the main parameters for the determination of water needs for irrigation is actual ET. For this reason, estimation of actual ET provides very valuable information for the effective management of irrigation. Evapotranspiration is also crucial for the energy cycle. More than half of the solar energy absorbed by land surfaces is currently used to evaporate water (Trenberth et al. 2009), significantly affecting the Earth's climate.

The methods for estimating basin-wide ET are generally grouped into three categories: the water budget approach, meteorological estimate, and remote sensing of the land–atmosphere interface. Hydrological methods such as Thornthwaite & Matter (1955) and Grindley (1969) utilize an accounting approach where inflow, outflow, and change in storage are used in the water budget. Meteorological methods are based mainly on temperature and radiation data. Formulations proposed by Turc (1961) and Thornthwaite (1948) are commonly used by hydrology practitioners. Agro-meteorological methods use potential ET data as a reference, and crop coefficients are multiplied according to crop type and crop stage (Penman 1948). Remote sensing-based ET methods utilize the energy budget concept and generally provide accurate and spatially explicit estimates of ET (Allen et al. 2005).

The first two methods can be identified as conventional techniques and use point-based hydrometeorological data. However, with proper adaptations to use remote sensing data, conventional methods such as Monteith (1965) and Priestley & Taylor (1972) are also widely applied in a spatially distributed manner (e.g., Vinukollu et al. (2011), Miralles et al. (2011), among others). The spatial variation of the parameters affecting the ET process is so high that uncertainty increases with the increasing scale of hydrological basins. To this end, remote sensing techniques provide large and continuous spatial coverage and are applicable especially for large areas. Moreover, remote sensing techniques are advantageous over point-based methods for estimating ET from crop coefficients or vegetation indices in that neither crop development stages nor specific crop types need to be known (Allen et al. 2011). The use of remote sensing data in ET models (both conventional or energy balance models) enables the spatial distribution of ET over large areas.

Remote sensing-based ET algorithms are based on the rationale where available energy in the environment is used for the vaporization process (Su 2002). Along with other surface parameters such as leaf area index (LAI), surface albedo, surface emissivity and radiometric surface temperature – also derived from remote sensing – and surface meteorological variables, remote sensing algorithms are capable of estimating ET on local, regional, and global scales. Generally known as the energy budget methods, these algorithms use physical and empirical relations together to estimate ET as the residual of the land surface energy budget equation. There are many models based on residual methods, such as SEBAL (Bastiaanssen et al. 1998a, 1998b), SEBI (Menenti & Choudhury 1993), SEBS (Su 2002), mapping evapotranspiration at high resolution with internalized calibration (METRIC) (Allen et al. 2007a, 2007b), and S-SEBI (Roerink et al. 2000).

Many studies have tested the performance of the METRIC model at various scales. Gowda et al. (2008) evaluated the performance of the METRIC model in Texas plains using Landsat 5 TM data and found good correlation with ET estimations based on the soil water budget model. Choi et al. (2009) conducted a comparison of the two-source energy balance (TSEB) model, METRIC, trapezoid interpolation model, METRIC, and TSEB models, and consequently, found relatively good agreement between models for Rn and G but greater disagreement for H and LE between models and also among the observation data. Gonzalez-Dugo et al. (2009) compared METRIC and TSM models and underlined the requirement of models for accurate surface temperature input. Singh & Senay (2016) found that estimated daily ET for both cropland and grassland had some degree of linearity with METRIC, SEBAL, and SEBS. Paul et al. (2013) showed the sensitivity of SEBAL to excess resistance to heat transfer parameter (kB-1). Paul et al. (2014) underlined the importance of hot and cold pixel selection, and proposed spatial and temporal adoption of kB-1 variable for better model performance. ET models have strengths and limitations for applications in water resource management. In this study, we used the METRIC algorithm to calculate ET using moderate resolution imaging spectroradiometer (MODIS) remotely sensed images and ground-based meteorological observations on cloud-free days. METRIC is a variant of SEBAL, which incorporates autocalibration procedures by ground-based reference ET and uses evaporative fraction to extrapolate ET in image time to daily scales in order to lessen regional advection differences (Allen et al. 2011). In the study hourly, daily, and monthly values of ET were calculated for the study area. Further, the effect of spatio-temporal change on the ET values was evaluated. Inter-comparison of the model results was made by ET values calculated by the Penman–Monteith equation ASCE version in an experimental area covered with olive orchards in the study area. The main objective of this study was to assess the capability of remote sensing-based ET estimation via the METRIC model in the lower part of the Buyuk Menderes River Basin in western Turkey. The ET maps produced as outputs are planned to be used in the assessment of irrigation planning studies.

The Buyuk Menderes River, which is the largest river of the Aegean region of Turkey, has a length of 584 km and a drainage area of 24,873 km2. The downstream part of the basin was chosen as the study area, which includes the Soke Plain in Aydın province. The study area lies between 37.36 °–38.06 ° N and 27.03 °–27.93 °E and covers an area of 3,560 km2 (Figure 1). The altitude increases from west to east. The study area is an important irrigation area for Turkey, and cotton is the primary irrigated crop. Therefore, the study period was chosen from the beginning of the cultivation (irrigation) to the end of the cotton harvest in the lower part of Buyuk Menderes Basin. The chosen study area only includes the lower plain of the Buyuk Menderes Basin so as to exclude the uncertainties resulting from high elevation and aspect effects. Annual average, monthly minimum, and maximum temperatures are 17.6 °C, 8.1 °C, and 30.7 °C, respectively. The long-term average annual total precipitation and pan evaporation values are 627 mm and 1,378 mm, respectively.
Figure 1
Figure 1. Location of study area and meteorological stations.
View largeDownload slide

Location of study area and meteorological stations.

Figure 1
Figure 1. Location of study area and meteorological stations.
View largeDownload slide

Location of study area and meteorological stations.

Close modal

Data

Meteorological data required by the surface energy balance model include minimum, maximum, and average temperatures, actual vapor pressure, pan evaporation (class A-pan), wind speed, air temperature, and cloudiness. The data are based on hourly and daily time scales. In this study, the observations from six meteorological stations located in and around the study area were used.

Remotely sensed data were obtained from MODIS products (see details in Table 1) because it provides surface observations for regional/basin water management. MODIS is the key instrument aboard the Aqua and Terra satellites, acquiring data in 36 spectral bands. The primary remotely sensed inputs to the METRIC algorithm are normalized difference vegetation index (NDVI), surface emissivity, broadband surface albedo, and land surface temperature. The MODIS albedo is calculated as (Liang 2001):
formula
1
where α is the broadband albedo and rn represents spectral reflectance in band n.
Table 1

MODIS products

MODIS standard productsParameterSpatial resolutionTemporal resolution
MOD13A2 Vegetation indices (NDVI) 1,000 m 16 days 
MOD15A2 Leaf area index (LAI) 1,000 m 8 days 
MOD11A1 Land surface temperature and emissivity 1,000 m Daily 
MOD09GA Surface reflectance 1,000 m Daily 
MOD35L2 Cloud mask 1,000 m Daily 
MODIS standard productsParameterSpatial resolutionTemporal resolution
MOD13A2 Vegetation indices (NDVI) 1,000 m 16 days 
MOD15A2 Leaf area index (LAI) 1,000 m 8 days 
MOD11A1 Land surface temperature and emissivity 1,000 m Daily 
MOD09GA Surface reflectance 1,000 m Daily 
MOD35L2 Cloud mask 1,000 m Daily 
View Large

Since clouds have negative effects on the detection of reflectivity from the Earth's surface, cloud-free days for the study area were determined by the evaluation of the cloudiness data obtained from meteorological stations for the satellite passing times as well as with the help of the MODIS cloud mask product.

The MODIS Reprojection Tool was used to reproject the coordinate system to UTM with WGS 84 Datum. COoRdination of INformation on the Environment (CORINE) land cover data set of the year 2006 was used to determine the surface roughness for momentum transport (Figure 2).
Figure 2
Figure 2. Inter-comparison points shown on CORINE map.
View largeDownload slide

Inter-comparison points shown on CORINE map.

Figure 2
Figure 2. Inter-comparison points shown on CORINE map.
View largeDownload slide

Inter-comparison points shown on CORINE map.

Close modal

Model description

The METRIC model is a satellite image processing model which estimates ET as a residual of energy balance for the time of sensor overpass (Choudhury 1994; Allen et al. 2007a, 2007b; 2011) considering the following equation:
formula
2
where Rn (W/m2) is the net radiation, G (W/m2) is the ground (soil) heat flux, H (W/m2) is the turbulent sensible heat flux, and LE (W/m2) is the turbulent latent heat flux (energy equivalent of ET). The net radiation flux is the difference between downward and upward radiation fluxes at the land surface both in shortwave and longwave parts of the spectrum, and calculated as:
formula
3
where RS ↓ is downward solar radiation (W/m2), α is land surface albedo, RL ↓ is downward longwave radiation (W/m2), RL ↑ is upward longwave radiation (W/m2), and εo is emissivity of the surface (−).
Rs↓ is calculated assuming clear-sky conditions as a constant for image time using Equation (4):
formula
4
where Gsc is the solar constant (1,367 W/m2), cos θ is the cosine of the solar incidence angle, dr is the inverse squared relative earth–sun distance, and τsw is the atmospheric transmissivity. τsw is obtained from Equation (5) (Allen et al. 1998):
formula
5
where z is the elevation above sea level (m).
RL↓ is computed using the Stefan–Boltzman equation:
formula
6
where ɛa is the atmospheric transmissivity (dimensionless), σ is the Stefan–Boltzman constant (5.67 × 10−8 W/m2/K4), and Ta is the near surface air temperature (°K). The following empirical equation for ɛa by Bastiaanssen (1995) was applied:
formula
7
RL↑ is also calculated by the Stefan–Boltzman equation:
formula
8
where ɛo is the broad-band surface emissivity (dimensionless) and Ts is the surface temperature (oK).
Soil heat flux is defined as the rate of heat flow into the soil due to conduction. Soil heat flux (G) was modeled as a function of Rn, vegetation index, surface temperature, and surface albedo near midday values (Bastiaansen 1995) as given by:
formula
9
where Ts is surface temperature (°K).
Calculation of the sensible heat flux (H) is performed by using the bulk aerodynamic resistance method:
formula
10
where ρa is the density of moist air (kg/m3), Cp is specific heat capacity of dry air (∼1,004 J/kg.K), T is the average air temperature (°K) at screen height (typically at 2 m), dT is the near surface temperature difference between two levels (°K K), which are 2 m and 0.1 m (Allen et al. 2011), and rah is the aerodynamic resistance to heat transfer (s/m).
In METRIC, the momentum roughness length (zom) is estimated for each pixel based on land use type. Brutsaert (1982) and Allen (1996) defined the general values for different land use types. In agricultural areas, a relationship proposed by Tasumi (2003) is used:
formula
11
formula
12
The key to METRIC is to determine the a and b constants using the hot pixel (Ts is high due to no evapotranspiration) and cold pixel (low surface temperature because of the evaporative cooling) which were identified with the help of visual image interpretation, and sensible heat flux values were calculated for hot and cold pixels. This value is corrected for atmospheric stability, which involves an iterative process using the Monin–Obhukov similarity theory. H values are converted to dThot and dTcold values, and then a line is fitted through the pair of Ts and dT values to determine a and b values.
The instantaneous LE is obtained using Equation (2), and then it is converted to Etinst:
formula
13
where ETinst is hourly actual evapotranspiration (mm/hr), and λ is latent heat flux (J/Kg).
METRIC uses the fraction of alfalfa reference ET (ETrF) to obtain the daily ET from instantaneous observations. This is the ratio of the actual crop ET to ETr at the time of overpass, and it is also assumed to be constant throughout the day:
formula
14
where ETr is alfalfa reference evapotranspiration (mm/hr).
Therefore, the daily ET is obtained by multiplying ETrF by ETr-24; the latter being ETr for the day, obtained by summing hourly reference ET for the entire day. This approach is capable of capturing most of the advection impacts and any other change in weather conditions during the day.
formula
15
The METRIC approach assumes that the ET for the entire area of interest changes in line with ETr calculated for the representative weather station (Allen et al. 2007a, 2007b). Estimation of ET for non-processed image days (whether cloudy or clear sky) by using clear-sky processed satellite image data is an effective approach (Tasumi et al. 2005). Generally, one satellite image per month is sufficient to construct an accurate ETrF curve for estimating seasonal ET. However, more frequent image intervals are needed in periods of rapid vegetation growth. In this study, the intervals of images were chosen as 10 to 20 days for better representation of all vegetation growth period in the study area (Table 2).
Table 2

Processed image dates

Date (year: 2010)Julian daysDateJulian days
23 April 113 14 July 195 
3 May 123 2 August 214 
14 May 134 17 August 229 
30 May 150 29 August 241 
15 June 166 11 September 254 
2 July 183   
Date (year: 2010)Julian daysDateJulian days
23 April 113 14 July 195 
3 May 123 2 August 214 
14 May 134 17 August 229 
30 May 150 29 August 241 
15 June 166 11 September 254 
2 July 183   
View Large
Monthly ET is calculated as:
formula
16
where ETm is cumulative ET for a month ending on day m, EtrFi is interpolated ETrF for day i, and ETr24i is 24-h ETr for day i. Units are in mm day−1.
In this research, the scheme for estimation of ET by METRIC is based on MODIS products and is given in Figure 3.
Figure 3
Figure 3. Flowchart of ET estimation by using METRIC.
View largeDownload slide

Flowchart of ET estimation by using METRIC.

Figure 3
Figure 3. Flowchart of ET estimation by using METRIC.
View largeDownload slide

Flowchart of ET estimation by using METRIC.

Close modal

As there is no direct measurement of ET over the study area, the method used by Jia et al. (2009) in the case of the Yellow River Basin in China was used for the inter-comparison process. Jia et al. (2009) proposed an alternative method to make comparisons with modeled and observed data. In their method, ET was estimated by multiplying the reference ET derived from the Penman–Monteith equation by the crop coefficients. In this study, reference evaporation estimation calculated by the Penman–Monteith ASCE version was used to make comparison with the METRIC ET outputs. Converting the reference ET to actual ET was made by the crop coefficient for olive trees. Pastor & Orgaz (1994) determined the crop coefficient (Kc) values of olive trees on a monthly basis for the area, 60% of which is covered with olive trees (Table 3). For inter-comparison purposes, 10 points covered with olive trees in the study area were chosen.

Table 3

Crop coefficients for olive orchard (Pastor & Orgaz 1994)

MonthsKc coefficientsMonthsKc coefficients
January 0.5 July 0.45 
February 0.5 August 0.45 
March 0.65 September 0.55 
April 0.60 October 0.60 
May 0.55 November 0.65 
June 0.50 December 0.50 
MonthsKc coefficientsMonthsKc coefficients
January 0.5 July 0.45 
February 0.5 August 0.45 
March 0.65 September 0.55 
April 0.60 October 0.60 
May 0.55 November 0.65 
June 0.50 December 0.50 
View Large

Bafa Lake, located in the study area, was selected for the estimation of evaporation from water surfaces such as lakes and reservoirs. Class A-pan evaporation measurements at Kusadasi Station were used to compare Bafa Lake evaporation estimations by METRIC. Class A-pan conversion coefficients ranged between 0.6 and 0.8 (WMO 2008). Class A-pan measurements multiplied with different pan coefficients ranged between 0.6 and 0.8, and METRIC and pan estimations are presented in Figure 4. When combined with an appropriate pan coefficient or with an adjustment for the energy exchange through the sides and bottom of the tank, evaporation data from a Class-A pan can be considered to be open-water evaporation (Dingman 1992).
Figure 4
Figure 4. Bafa Lake monthly total evaporation and pan evaporation (mm).
View largeDownload slide

Bafa Lake monthly total evaporation and pan evaporation (mm).

Figure 4
Figure 4. Bafa Lake monthly total evaporation and pan evaporation (mm).
View largeDownload slide

Bafa Lake monthly total evaporation and pan evaporation (mm).

Close modal

In this study, a remote sensing algorithm, METRIC, was applied to estimate basin-wide actual ET in the lower part of the Buyuk Menderes Basin in Western Turkey. The spatial distribution of actual evapotranspiration for the cotton growing season was mapped at daily and monthly time steps. The model results were validated by comparing remote sensing estimates to ground-based actual ET estimations calculated with the help of the Penman–Monteith equation adjusted by coefficients on olive orchards.

The average standard error in daily ET estimates for the processed 11 imagery is 16.1%, and standard deviation is 27.1% (Figure 5). When compared to results of a similar study conducted under similar climatic conditions, Allen (2014) found METRIC to have 1.6% minimum bias in an application for an agricultural area equipped with flux measurements. In our work, the maximum differences were found at points 3 and 4, which also correspond to locations with the maximum slope values (24.8% and 24.0%, respectively). The mean and standard deviation of slopes in all points are 11.5% and 7.5%, respectively. Therefore, the deviation between METRIC and ground-based ET estimates appears to be associated with the ground slope. However, the METRIC's slope-correction ability was not used in this study since the cotton fields are almost flat. In addition to high slope areas, another reason for the less than ideal bias in our work is related to the nature of limited data used in the Penman–Monteith equation.
Figure 5
Figure 5. Comparision between METRIC and ground-based estimations for inter-comparison points (1:1 scale).
View largeDownload slide

Comparision between METRIC and ground-based estimations for inter-comparison points (1:1 scale).

Figure 5
Figure 5. Comparision between METRIC and ground-based estimations for inter-comparison points (1:1 scale).
View largeDownload slide

Comparision between METRIC and ground-based estimations for inter-comparison points (1:1 scale).

Close modal
Average monthly actual ET during the cotton growing season in the lower part of the Buyuk Menderes Plain ranges from 99.6 mm in April to 194.3 mm in July, whereas maximum and minimum monthly actual ET were estimated as 240.7 mm in July and 37.2 mm in April for the entire study area. To this end, the temporal and spatial details of ET maps afforded by the METRIC model are significant especially when compared to estimation by point-based measurements that can cause substantial errors (Figure 6).
Figure 6
Figure 6. Monthly total ET (in mm/month) from April to September.
View largeDownload slide

Monthly total ET (in mm/month) from April to September.

Figure 6
Figure 6. Monthly total ET (in mm/month) from April to September.
View largeDownload slide

Monthly total ET (in mm/month) from April to September.

Close modal

The major land cover types in the study area are agriculture (50%), forest and semi-natural lands (46%), the remainder being water and wetlands (3%) and artificial areas (1%). It was found that open water bodies have the highest ET values followed by agricultural areas due to large-scale irrigation. The areas of the land cover categories and the corresponding amount of water consumption estimated in this study are given in Table 4. As shown, the water consumption in the study area is dominated by agricultural and forest/semi-natural lands which cover the majority of the landscape (94.8%), and the ET rate of these areas accounts for over 92% of total consumption during the study period. Agricultural lands (49.4%) have the highest consumption (50.4%) as regards to the integrated impact of land coverage on the ET rate.

Table 4

Land use and monthly ET

Land cover typeArea (km2)MonthsTotal ET (mm)Mean (mm)Standard deviation
Agricultural lands 1,757.7 April 176,064.0 95.9 26.8 
May 209,312.1 113.8 33.1 
June 232,404.6 126.0 32.2 
July 302,526.2 162.5 48.6 
August 196,126.0 104.8 32.0 
September 111,855.3 84.9 44.4 
Forest and semi-natural lands 1,615.8 April 148,655.0 90.4 35.5 
May 167,255.9 102.9 46.2 
June 188,299.9 114.2 49.1 
July 218,578.0 132.8 66.9 
August 198,516.7 114.6 42.8 
September 105,428.0 65.7 53.4 
Wetlands 3.1 April 229.6 57.5 37.3 
May 248.5 62.1 40.4 
June 414.4 103.6 20.0 
July 758.9 189.7 25.0 
August 663.8 166.0 21.1 
September 482.2 120.6 38.9 
Water bodies 75.7 April 12,108.4 126.1 32.1 
May 15,966.6 164.6 47.5 
June 17,782.5 183.3 40.0 
July 22,385.3 228.4 40.0 
August 18,510.4 210.3 38.8 
September 14,140.2 148.8 31.5 
Artificial areas 105.1 April 10,257.7 71.2 39.0 
May 13,441.4 87.3 50.1 
June 14.822.3 98.2 50.5 
July 18,145.3 121.8 58.5 
August 16,301.7 91.1 37.5 
September 6,844.1 54.8 45.7 
Land cover typeArea (km2)MonthsTotal ET (mm)Mean (mm)Standard deviation
Agricultural lands 1,757.7 April 176,064.0 95.9 26.8 
May 209,312.1 113.8 33.1 
June 232,404.6 126.0 32.2 
July 302,526.2 162.5 48.6 
August 196,126.0 104.8 32.0 
September 111,855.3 84.9 44.4 
Forest and semi-natural lands 1,615.8 April 148,655.0 90.4 35.5 
May 167,255.9 102.9 46.2 
June 188,299.9 114.2 49.1 
July 218,578.0 132.8 66.9 
August 198,516.7 114.6 42.8 
September 105,428.0 65.7 53.4 
Wetlands 3.1 April 229.6 57.5 37.3 
May 248.5 62.1 40.4 
June 414.4 103.6 20.0 
July 758.9 189.7 25.0 
August 663.8 166.0 21.1 
September 482.2 120.6 38.9 
Water bodies 75.7 April 12,108.4 126.1 32.1 
May 15,966.6 164.6 47.5 
June 17,782.5 183.3 40.0 
July 22,385.3 228.4 40.0 
August 18,510.4 210.3 38.8 
September 14,140.2 148.8 31.5 
Artificial areas 105.1 April 10,257.7 71.2 39.0 
May 13,441.4 87.3 50.1 
June 14.822.3 98.2 50.5 
July 18,145.3 121.8 58.5 
August 16,301.7 91.1 37.5 
September 6,844.1 54.8 45.7 
View Large

Bafa Lake daily evaporation rates were estimated with 16.4% average standard error and 15.0% standard deviation for the 11 processed images. These results are compatible with Class-A pan-based lake evaporation rates. Monthly evaporation on Bafa Lake ranged between 145.4 mm in April 2010 and 248.7 mm in July 2010.

In addition to the spatial and temporal distribution of ET, energy balance components were also determined (Figure 7). The produced data on energy balance components can be a tool for further analysis such as comparison of available energy for evapotranspiration and measured ET values.
Figure 7
Figure 7. Daily actual evapotranspiration and energy balance components (15 June 2010).
View largeDownload slide

Daily actual evapotranspiration and energy balance components (15 June 2010).

Figure 7
Figure 7. Daily actual evapotranspiration and energy balance components (15 June 2010).
View largeDownload slide

Daily actual evapotranspiration and energy balance components (15 June 2010).

Close modal

In this study, the use of remote sensing along with ground-based data to determine ET at a regional scale is demonstrated, and promising results are achieved. The spatial and temporal variation of ET was found to be significant due to the land use, meteorological conditions, and vegetation patterns. METRIC evaporations are in agreement with the Class A-pan data. As the METRIC ET estimations are closely aligned with the Class A-pan data, the method can also be used for estimating evaporation from lakes in data-scarce areas. Remote sensing-based pan coefficients can also be developed for monitoring of lakes.

The gridded ET maps can be used as input for hydrological models. They also have the potential to increase the effectiveness of water management strategies by determining irrigation requirements in spatially explicit ways. Moreover, spatially distributed evapotranspiration data on cultivated lands can be used to calculate crop water productivity (kg/m3), which is a powerful tool to assess the efficiency of different irrigation management systems including structural and non-structural applications such as irrigation schemes and legal and institutional frameworks.

This study forms a basis for estimation of ET for regional-scale applications particularly over agricultural areas in western Turkey. Future studies should include other ET models based on remote sensing data to determine the efficiency of respective models under the climate regime of western Turkey.

The authors thank their colleagues Dr S. Bayari and Dr A.U. Sorman for continuing support and discussions during this study.

Allen
R. G.
2014
Ecostress METRIC evapotranspiration algorithm
. In:
HyspIRI Science and Application Workshop
,
14–16 October
,
Pasadena, CA
,
USA
.
Google Scholar
 
Allen
R. G.
Pereira
L. S.
Raes
D.
Smith
M.
1998
Crop Evapotranspiration – Guidelines for Computing Crop Water Requirements
.
FAO Irrigation and Drainage Paper 56
,
FAO
,
Rome
,
Italy
.
Allen
R. G.
Walter
I. A.
Elliot
R. L.
Howell
T. A.
Itenfisu
D.
Jensen
M. E.
Snyder
R.
2005
The ASCE standardized reference evapotranspiration equation
.
American Society of Civil Engineers
,
Reston, VA
.
Google Scholar
 
Allen
R.
Tasumi
M.
Morse
A.
Trezza
R.
Wright
J. L.
Bastiaanssen
W.
Kramber
W.
Lorite
L.
Robison
C. W.
2007b
Satellite-based energy balance for mapping evapotranspiration with internalized calibration, METRIC (applications)
.
J. Irrig. Drain. Eng.
133
,
395
–
406
.
Google Scholar
Crossref
Search ADS
 
Allen
R.
Irmak
A.
Trezza
R.
Hendrickx
J. M. H.
Bastiaanssen
W. G. M.
Kjaersgaard
J.
2011
Satellite-based ET estimation in agricultural using SEBAL and METRIC
.
Hydrol. Process.
25
,
4011
–
4027
.
Google Scholar
Crossref
Search ADS
 
Bastiaanssen
W. G. M.
1995
Regionalization of surface flux densities and moisture indicators in composite terrain: a remote sensing approach under clear skies in Mediterrenean climates
.
PhD Dissertation
,
Koninklijke Bibliotheek
,
The Netherlands
.
Google Scholar
 
Bastiaanssen
W. G. M.
Menenti
M.
Feddes
R. A.
Holtslag
A. A. M.
1998a
A remote sensing surface energy balance algorithm for land (SEBAL): 1. Formulation
.
J. Hydrol.
212–213
,
198
–
212
.
Google Scholar
Crossref
Search ADS
 
Bastiaanssen
W. G. M.
Pelgrum
H.
Wang
J.
Ma
Y.
Moreno
J. F.
Roerink
G. J.
van der Wal
T.
1998b
A surface energy balance algorithm for land (SEBAL): 2. Validation
.
J. Hydrol.
212–213
,
213
–
229
.
Google Scholar
Crossref
Search ADS
 
Baumgartner
A.
Reichel
E.
1975
The World Water Balance
.
Elsevier
,
Amsterdam
,
The Netherlands
, p.
179
.
Google Scholar
 
Brutsaert
W. H.
1982
Evaporation into the Atmosphere: Theory, History and Applications
.
D. Reidel
,
Norwell, MA
,
USA
.
Google Scholar
Crossref
Search ADS
 
Dingman
S. L.
1992
Physical Hydrology
.
Prentice Hall
,
Upper Saddle River, NJ
,
USA
.
Google Scholar
 
Gonzalez-Dugo
M. P.
Neale
C. M. U.
Mateos
L.
Kustas
W. P.
Prueger
J.
Anderson
M. C.
Li
F.
2009
A comparison of operational remote-sensing-based models for estimating crop evapotranspiration
.
Agr. Forest Meteorol.
149
,
1843
–
1853
.
Google Scholar
Crossref
Search ADS
 
Gowda
P. H.
Chavez
J. L.
Howell
T. A.
Marek
T. H.
New
L. L.
2008
Surface energy balance based evapotranspiration mapping in the Texas high plains
.
Sensors
8
,
5186
–
5201
.
Google Scholar
Crossref
Search ADS
PubMed
 
Grindley
J.
1969
The calculation of actual evapotranspiration and soil moisture deficit over specified catchment areas
.
Hydrological Memorandum No. 38
,
Meteorological Office
,
Bracknell
,
UK
.
Menenti
M.
Choudhury
B.
1993
Parameterization of land surface evaporation by means of location dependent potential evaporation and surface temperature range
. In:
Proceedings of IAHS Conference on Land Surface Processes
,
IAHS, Wallingford
,
UK
,
212
, pp.
561
–
568
.
Google Scholar
 
Miralles
D. G.
Holmes
T. R. H.
De Jeu
R. A. M.
Gash
J. H.
Meesters
A.
Dolman
A. J.
2011
Global land-surface evaporation estimated from satellite-based observations
.
Hydrol. Earth Syst. Sci.
15
,
453
–
469
.
Google Scholar
Crossref
Search ADS
 
Monteith
J. L.
1965
Evaporation and environment
. In:
Proceedings of the 19th Symposium of the Society for Experimental Biology
.
Cambridge University Press
,
New York
, pp.
205
–
233
.
Google Scholar
 
Pastor
M.
Orgaz
F.
1994
Riego deficitario del olivar: los programas de recorte de riego en olivar
.
Agricultura
746
,
768
–
776
.
Google Scholar
 
Paul
G.
Gowda
P. H.
Varaprasad
P. V.
Howell
T. A.
Aiken
T. M.
Hutchinson
S. L.
Neale
C. M. U.
2013
Role of hot and cold pixel concept in remote sensing based single source energy balance algorithms, using 21st century technology to better manage irrigation water supplies
. In:
Seventh International Conference on Irrigation and Drainage
,
Phoenix, AZ
,
USA
.
Google Scholar
 
Penman
H. L.
1948
Natural evaporation from open water, bare soil and grass
.
Proc. Roy. Soc. London
A193
,
120
–
146
.
Google Scholar
Crossref
Search ADS
 
Priestley
C. H. B.
Taylor
R. J.
1972
On the assessment of surface heat flux and evaporation using large-scale parameters
.
Monthly Weather Review
100
,
81
–
92
.
Google Scholar
Crossref
Search ADS
 
Singh
R. K.
Senay
G. B.
2016
Comparison of four different energy balance models for estimating evapotranspiration in the Midwestern United States. Water 8, 9
.
Tasumi
M.
2003
Progress in operational estimation of regional evapotranspiration using satellite imagery
.
PhD Dissertation
,
University of Idaho
,
Moscow, ID
,
USA
.
Google Scholar
 
Tasumi
M.
Allen
R. G.
Trezza
R.
Wright
J. L.
2005
Satellite-based energy balance to assess within-population variance of crop coefficient curves
.
J. Irrig. Drain. Eng.
131
(
1
),
94
–
109
.
Google Scholar
Crossref
Search ADS
 
Thornthwaite
C. W.
Mather
J. R.
1955
The Water Balance. Publications in Climatology
,
Vol. VIII, No. 1
.
Drexel Institute of Technology
,
Centerton, NJ
,
USA
.
Google Scholar
 
Trenberth
K. E.
Fasullo
J. T.
Kiehl
J.
2009
Earth's global energy budget
.
Bull. Am. Meteor. Soc.
87
,
597
–
606
.
Google Scholar
 
Turc
L.
1961
Evaluation des besoins en eau d'irrigation, evapotranspiration potentielle, formule climatique simplifice et mise a jour [Estimation of irrigation water requirements, evapotranspiration potential – a simple climatic formula evolved up to date]
.
Ann. Agron.
12
,
13
–
49
.
Google Scholar
 
WMO
2008
Guide to Hydrological Practices (WMO No. 168)
, 8th edn.
World Meteorological Organization
,
Geneva
.
© IWA Publishing 2017