The economic, agricultural and social development of a region is affected by the proper management of its water resources. Planning and management of water consumption in the basin area require knowledge of the hydrological behavior of the system, including the temporal and spatial variation of important components like actual ET in that basin. ET is important in irrigation management [1
], drought monitoring [2
], water storage [4
], water balance [5
], water productivity [6
] and groundwater management [7
]. Therefore, considering methods that estimate actual ET accurately is required.
The estimation of ET is difficult due to its complex nature. There are various methods for estimating this component. Remote sensing-based methods are more popular than classic (in situ) ones because of their wide coverage and comparability with numerical climate models. Jackson et al., through the infrared thermometer in the wheat field in the state of Arizona, discussed the importance of using remote sensing to estimate ET [10
]. Different methods have been developed to combine satellite imagery with ground data [11
]. Subsequent studies in this field have led to the development of a Surface Energy Balance Algorithms for Land (SEBAL) in the 1990s as one of the new methods for estimating actual ET [12
The SEBAL algorithm is a model based on the empirical and physical relationships that were first used in Egypt and Spain to estimate the ET of agricultural areas by Landsat, and then used extensively in similar studies [13
]. Later, METRIC (Mapping Evapotranspiration at high Resolution with Internalized Calibration) was developed by the University of Idaho as a modified version of SEBAL, that used Landsat satellite data to compute and map ET [1
]. In another study by Zwart and Bastiaanssen using METRIC and MODIS images of wheat cultivars in Mexico, an error of about 9% was reported over a period of 110 days [14
]. Also, Du et al. estimated ET on Sanjiang plain in China using 12 MODIS images [15
]. The difference between the actual daily ET and the measurement performed by Eddy Covariance was reported as an average of 10.5%.
Urmia Lake, largest lake in the Middle East and the sixth-largest saltwater lake on Earth, has shrunk dramatically in the last two decades mainly because of the increasing agricultural water consumption in the basin (Figure A1
]. Hesami and Amini demonstrated that the total irrigated area in the basin increased by 20% between 1989 and 2000 and the water use change in the area was 35%. In arid and semi-arid regions, water used for irrigation is highly correlated with evapotranspiration [19
]. Various studies have been conducted on the estimation of actual ET in the Urmia Lake Basin. In a study by Iran Water Research Institute, the actual ET was estimated using the SEBAL algorithm and the NOAA-AVHRR images on an annual scale for a dry and wet periods and it was shown that SEBAL can perform well with different satellites [20
]. In another study, the SEBAL algorithm used for estimating water budget in the Lake Urmia basin [21
]. The output was that SEBAL algorithm works well for this purpose [21
]. Estimating agricultural water consumption in the Lake Urmia basin is an important issue. Taheri et al. estimated actual ET in the Lake Urmia basin from METRIC, in order to estimate the agricultural irrigation water requirement. They found that all climatic factors and human activities in the Lake Urmia basin led to increase in agricultural water consumption [22
Most of energy balance models such as METRIC require thermal infrared band taken from cloud-free and corrected images to produce land surface temperature maps [23
]. As a result, cloudy images reduce the accuracy in the results. The ETLook algorithm uses soil moisture extracted from the passive microwave sensor instead of the land surface temperature. Microwave data provide surface information even in cloudy days, because they are less affected by cloud cover [24
]. Few studies have been conducted on this algorithm. Bastiaanssen et al. introduced ETLook in the Indus basin [25
]. The results showed that the algorithm has a good accuracy in these areas and even shows up great on cloudy days.
The United Nations World Food and Agriculture Organization (FAO) developed the “Free Access Water Productivity System “, aims to cover countries that are facing water crisis in Africa and the Middle East. Actual ET from WaPOR is one of the most important products of this system, which provides an annual and 10-day spatial mapping with 250 m pixel size for the 2009–2016 period using the ETLook algorithm (http://www.fao.org/in-action/remote-sensing-for-water-productivity/wapor/
). Since the public availability of WaPOR product (i.e., 2017), we are not aware of any published evaluation study over the Middle East and Africa, although such studies are urgently needed given the large uncertainties of ET estimates over these regions and the need for alternative estimates.
Given that METRIC approach has been a popular method to estimate actual ET over Urmia Lake basin, comparison of METRIC with WaPOR can shed light onto differences that WaPOR might bring to actual ET analysis over the region. In this study, we first calculate the actual ET values using the METRIC algorithm applied to MODIS images for the Urmia Lake Basin in year 2010 (representing a dry year in the basin) and 2014 (representing a normal year), and then these values are compared with the actual ET of WaPOR product in the Lake Urmia basin and the differences between them are analyzed as a function of elevation, land cover, and sub-basin. We used the only available lysimeter over the study region and water consumption measurements in an irrigation network system to complement our analysis.
4. Concluding Remarks
Overconsumption of water for agriculture can cause serious problems in arid and semi-arid regions. A good example of such regions is Lake Urmia, which has significantly been shrunk in recent years due to increase in agricultural water consumption. Water consumption is well correlated with ET, thus improved estimate of ET can provide valuable information on water consumption planning and management. In this study, we compared the actual ET maps from two high resolution ET products (the WaPOR product, derived from the ETLook algorithm, and ET from the METRIC algorithm) over the Lake Urmia basin for year 2010 and 2014. METRIC algorithm has commonly been used over this region using MODIS images, it is important to quantify the differences between ET estimates from the newly available high resolution WaPOR product (i.e., 250 m spatial resolution) and METRIC, the results of which can add valuable insights onto hydrologic studies over the region. ETLook calculates ET using microwave images, so has no gap under cloudy conditions. This two-source algorithm has a high sensitivity to soil moisture data, while the METRIC one-source algorithm is more sensitive to land surface temperature (LST).
The maps produced by the WaPOR were found to be more realistic than the METRIC estimate (derived using MODIS images) in terms of temporal and spatial scales, potentially due to the ability of the ETLook algorithm to calculate ET in cloudy conditions using microwave images. Furthermore, we found that the ET rates from the WaPOR product are much smaller than the ET rates calculated from the METRIC algorithm in most part of the basin, especially over rangelands. By moving from low to high elevations, this difference tends to increase in both years. The temporal and spatial analysis of the results in the sub-basins of Lake Urmia basin indicates that the difference in the Miandoab plain (lowland sub-baisn) is lower than the other plains studied and the highest difference is in the Sarab plain that can be due to the elevation changes of the sub-basin and the sensitivity of production maps to changes in levels. Assessment of ET over precipitation ratio in rainfed agriculture areas indicates that METRIC produces unrealistically high ET (i.e., annual ET/P >1), but WaPOR produces a more realistic ET rates (i.e., annual ET/P <1).
Moreover, using the limited lysimeter data over the city of Tabriz, 80 km west of Lake Urmia, it was found that ET from METRIC algorithm matches lysimter measurements with no more than 19% difference, while in average WaPOR underestimates the ET rates by 71%. However, the outcomes of this comparison could be affected by large differences between footprint sizes of the METRIC estimates from MODIS (i.e., 1 km × 1 km), WaPOR product (i.e, 250 m ×250 m), and lysimeter (i.e., 70 m2). Lastly, by using reports of annual water consumption rates in an irrigation network South of Lake Urmia, we found that WaPOR and METRIC slightly under- and over-estimate ET rates. Knowing that water consumption reports do not account for unauthorized pumping, it was concluded that METRIC’s estimate is likely more realistic than that from WaPOR.
In a warming climate most lakes, especially those in arid and semi-arid regions are going through significant changes due to combination of natural responses (e.g., lower relative humidity, higher vapor pressure deficit, etc.) [32
] and anthropogenic activities (e.g., surface and ground water extraction for agriculture, etc.). This heightens the need for more accurate measurements of regional water and energy cycle components for planning, prediction, and mitigation of negative social and environmental aspect of such changes. ET has remained one of the mostly poor measured component of water and energy cycle in most regions of the world, a good example of which is Lake Urmia and Aral Sea [35
]. It is hoped that the new generation of instruments (e.g., ECOSTRESS [36
], among others) can provide more reliable estimate of ET regionally and globally.