Estimating reservoir evaporation losses for the United States: Fusing remote sensing and modeling approaches
Introduction
It has been projected that 5.3 billion people will live under water stress and water scarcity globally by 2030 (Organisation for Economic Co-operation and Development, 2008). Most of the affected population relies on surface water—especially the water impounded by reservoirs, which can be easily accessed and managed (United Nations Environment Programme, 2013). In addition to supplying water for agricultural, municipal, and industrial uses, reservoirs can also be used for flood control and hydropower generation. From 1950 to 2007, the cumulative volume of water impounded by global reservoirs rose from about 1000 km3 to 11,000 km3, reducing the global sea level rise by 30 mm (Chao et al., 2008). According to the Global Reservoir and Dam Database (GRanD; Lehner et al., 2011), the United States is the country with the largest number of reservoirs. These reservoirs are capable of storing 1300 km3 of water, which is almost equivalent to the region's annual mean runoff (Graf, 1999).
Globally there are about 16.7 million reservoirs that have a surface area of 100 m2 or greater (Lehner et al., 2011). These reservoirs have increased the global terrestrial water surface area by about 305,000 km2. With the large amount of surface area that is produced by these artificial impoundments, the evaporative loss is significant—especially in semi-arid and arid regions (Ali et al., 2008; Morton, 1979). For example, the annual evaporation loss from 200 reservoirs in Texas, USA, equals to 20% of their active storage (Zhang et al., 2017). The long-term evaporation from Lake Tahoe, which is located in the arid western United States, accounts for 40%–60% of the total reservoir output (Friedrich et al., 2018). The reservoir evaporation of Lake Mead (~1800 mm/year; Moreo, 2015) is much larger than the surrounding evapotranspiration (~50 mm/year; Mu et al., 2011), which can be regarded as pre-reservoir ET. From a global perspective, Shiklomanov (1999) estimated a total of ~270 km3/year of reservoir evaporation, which is larger than the combined domestic and industrial water use in the year 2010 (~250 km3). Therefore, to better support efficient water resources management, it is essential to incorporate accurate reservoir evaporation information into current reservoir operation rules.
Despite the critical need for reservoir evaporation information, no continentally consistent and locally practical evaporation dataset has been produced that can be used in the policy making process at a regional scale. To precisely quantify the evaporation losses from a given reservoir, water surface area and evaporation rate data are needed. However, both high quality reservoir surface area and evaporation rate data can be difficult to gather.
Reservoir surface area is usually inferred from in-situ measurements, or estimated from remote sensing images. By applying in-situ measured reservoir elevation values to a known elevation-area relationship, a reservoir's area can be calculated. The elevation-area relationship is typically derived from bathymetry investigations (either before or after the reservoir is constructed) using sonar/laser and GIS technologies. However, this method is limited by its large expense and the changes of reservoir bathymetry due to sedimentation. Remote sensing has the advantage of estimating water surface area from satellite images at low cost (McFeeters, 1996; Sawaya et al., 2003; Gao, 2015). Even though there is always a compromise between spatial and temporal resolution with remote sensing technologies, usually high quality images with acceptable time intervals can be obtained. Compared with other remote sensing data, Landsat has the advantages of long temporal coverage and high spatial resolution, which makes it suitable for water surface area change studies (Pekel et al., 2016; Donchyts et al., 2016; Khandelwal et al., 2017). However, a major limitation of using Landsat images for continuous water area monitoring is the frequent contamination from multiple sources, such as clouds, cloud shadows, terrain shadows, and the Scan Line Corrector (SLC) failure (for Landsat 7). As a result, direct water extraction from the contaminated satellite images can lead to significant underestimation. To generate reliable water area estimates, most studies have simply removed the contaminated images. For instance, Busker et al. (2018) removed all of the images which are >5% contaminated to get the surface area time series for 135 global lakes. However, this has led to many missing values in the time series, especially for the regions that have frequent cloud coverage. To bridge this gap, Zhao and Gao (2018) developed an image enhancement algorithm to automatically repair the contaminated reservoir images extracted from the Global Surface Water Dataset (GSWD; Pekel et al., 2016). The new algorithm resulted in a Global Reservoir Surface Area Dataset (GRSAD), which significantly improved the continuity of the reservoir area time series (i.e., 81% improvement on a global scale).
The evaporation rate of open water has been studied for decades. Comprehensive reviews of evaporation rate estimation methods can be found in Morton (1994) and the more recently Friedrich et al. (2018). The pan evaporation method has been employed by the National Weather Service (NWS) for estimating the point evaporation rate operationally for many decades. In addition to the primary purpose of assessing the spatial variability of atmospheric evaporative demand (AED) for irrigation scheduling, the pan evaporation data are also used for other applications such as investigating climate change and estimating reservoir evaporation (Stanhill, 2002; Ohmura and Wild, 2002; Rotstayn et al., 2006). Although there are about 950 pan evaporation stations across the CONUS, only a very small portion of these are located close enough to dams to estimate reservoir evaporation. Furthermore, lake evaporation estimation based on pan evaporation is subject to large errors due to multiple factors. These include ignoring the microclimate difference between the reservoir and the pan, not accounting for heat storage effects, extra heat absorption from the pan's sides, water splashing, overflow due to intensive rainfall, freezing conditions, human error, and others (McMahon et al., 2013; Friedrich et al., 2018). Thus, it is regarded as one of the least accurate evaporation estimation methods and is not suitable for precise water management practices (Tanny et al., 2008; Harwell, 2012; McJannet et al., 2017). In addition to the pan evaporation method, eddy covariance (EC), scintillometer, mass balance, Bowen ratio energy budget (BREB), and combination equation methods are frequently used. In general, EC is considered the most accurate approach—but it has been primarily used for evapotranspiration related research. Constrained by the expensive cost and the sensitivity to wind direction (relative to both sensor and reservoir location), very few lake evaporation data have been collected using this approach. By measuring the sensible heat flux, a scintillometer can estimate the latent heat flux if other energy terms are known, even though it has multiple limitations (such as signal saturation) (Moene et al., 2009). The mass balance and BREB methods are both data intensive. Mass balance requires inputs of inflow (tributary + lateral), outflow (outlet + lateral), storage change, and water use data. The BREB method requires measuring heat storage changes and water surface temperatures, in addition to other meteorological forcings (Morton, 1986; Morton, 1994). Both methods can result in considerable error introduced by the complex inputs (Stannard et al., 2013; Friedrich et al., 2018).
Among the various approaches for estimating the evaporation rate on a large scale, the most practical one involves using a physically based combination equation such as the Penman equation (Penman, 1948). Some variants include PenPan (Rotstayn et al., 2006; for estimation of pan evaporation), Penman-Monteith (Monteith, 1965; commonly used for potential evapotranspiration estimation), and the Priestley-Taylor equations (Priestley and Taylor, 1972; for quantifying wet-surface evaporation in advection-free conditions). The PenPan equation has the same form as the Penman equation but includes different parameterizations. The Penman-Monteith equation introduced physically based aerodynamic resistance to replace the empirical wind function. For the Priestley-Taylor equation, an empirical coefficient (αPT = 1.26) was used to approximate the aerodynamic term in the Penman equation. The value 1.26 was found to be appropriate for non-advective conditions but may change under advective conditions (Assouline et al., 2016; Eichinger et al., 1996; Flint and Childs, 1991). Detailed comparisons of these methods can be found in McMahon et al. (2013) and Wang and Dickinson (2012). These physically based models are proven to be reliable for applications over shallow water reservoirs (typically <3 m in depth) where heat storage is insignificant (Abtew, 2001; Linacre, 1993; Zhao et al., 2016). However, lakes and reservoirs usually have a considerable heat storage effect, causing combination equations to be biased with regard to seasonal evaporation rate estimation (Finch and Hall, 2001; McMahon et al., 2013). For example, in Lake Tahoe (California, US), the air temperature is highest in July, but the largest evaporation rate occurs in September (Tahoe Environmental Research Center, 2015). To address this issue, Edinger et al. (1968) introduced the equilibrium water temperature, and de Bruin (1982) incorporated it into evaporation rate estimation. The equilibrium temperature is the water temperature at which there is no heat exchange between the air and water under constant forcings. It can help calculate the water column temperature and (then) the heat storage changes. This concept has been used in several studies, and has proven to be appropriate for open-water evaporation estimation (Finch, 2001; Finch and Hall, 2001; Bogan et al., 2003; Caissie et al., 2005; McJannet et al., 2008; Mekonnen and Hoekstra, 2012). Although the derivations of the equilibrium temperature in these studies were all based on the energy balance of the water body, different studies have adopted different simplified energy terms. For instance, the most generic form of the equilibrium temperature was from Mohseni and Stefan (1999), which used a simplified latent heat flux formulation. Therefore, there is a lack of a generalized formulation of the equilibrium temperature to improve upon the open-water evaporation estimation using the Penman equation.
Therefore, this study focuses on breaking the above key barriers in reservoir evaporation quantification to better support more precise water resources management at both local (individual reservoir) and regional (multiple reservoirs) scales. A total of 721 reservoirs, which account for 90.2% of the large reservoir storage capacity in the CONUS (Fig. 1), were chosen as our study sites (Lehner et al., 2011). Specifically, our three objectives were to: 1) adopt continuous reservoir surface area time series generated from a Landsat-based water classification dataset that is free of image contamination; 2) quantify heat storage changes in the Penman equation to better simulate the monthly reservoir evaporation rate; 3) generate the long-term monthly evaporation volume dataset for the 721 reservoirs and analyze the long-term trends of reservoir evaporation. Even though this study focuses on reservoirs in the CONUS, the retrieval algorithms and the data analysis approaches are transferable to other regions or to a global scale. These objectives provide the structural sub-headings used in the following Methods, Results and Discussions sections.
Section snippets
Estimation of reservoir surface area
The surface area time series data for the 721 reservoirs were extracted from the Landsat based GRSAD by Zhao and Gao (2018). The dataset was built upon the GSWD, which includes the global water areas for each month from March 1984 to October 2015 (Pekel et al., 2016). By applying a complex expert system classification method (on ~3 million Landsat images from 1984 to 2015), the GSWD generated monthly global water coverage maps at 30-meter resolution. Each 30 m by 30 m pixel was classified as
Validation of reservoir surface area
Detailed validations of the GSWD and the GRSAD can be found in Pekel et al. (2016), and Zhao and Gao (2018), respectively. In addition, two reservoirs are used as examples here to demonstrate the robustness of the remotely sensed surface area results. The Amistad Reservoir is located between the USA and Mexico, and it has a maximum area of about 150 km2. Lake Mead is located between the states of Nevada and Arizona, USA and has a maximum area of about 600 km2. The remotely sensed surface area
Reservoir surface area
This long term reservoir evaporation dataset directly benefits from the high quality remotely sensed surface area estimations. Satellite images captured by VIS/NIR sensors usually suffer from multi-source contaminations including clouds, cloud shadows, and terrain shadows. For Landsat 7, the collected images also suffer from gaps due to the SLC failure. As a result, direct extraction of water area from the satellite images can lead to notable underestimations. Compared with the GSWD raw water
Conclusion
This study presents an advanced algorithm framework for reservoir evaporation quantification at a large scale. By applying the algorithm to 721 reservoirs in the CONUS, a first of its kind continentally consistent and locally practical evaporation data product was generated, providing significant benefits to the water resources management, hydrology, and remote sensing communities. The major conclusions are as follows:
- 1.
For the 721 reservoirs in the CONUS, long term average evaporation was found
Acknowledgement
The dataset containing the evaporation time series of the 721 reservoirs will be publicly available online at https://ceprofs.civil.tamu.edu/hgao/ and at https://dataverse.tdl.org/dataverse/tamu.
This research was supported by the NASA Science of Terra, Aqua, and Suomi NPP (TASNPP) Program (80NSSC18K0939), and the Earth and Space Science Fellowship (NESSF) Program (80NSSC17K0358). It was also partially supported by the U.S. Department of Energy Water Power Technologies Office as a part of the
References (123)
- et al.
Evaluating anemometer drift: a statistical approach to correct biases in wind speed measurement
Atmos. Res.
(2018) - et al.
Trends in evaporation for the Canadian prairies
J. Hydrol.
(2007) Temperature and energy balance of a water reservoir determined from standard weather data of a land station
J. Hydrol.
(1982)- et al.
Multi-season lake evaporation: energy-budget estimates and CRLE model assessment with limited meteorological observations
J. Hydrol.
(1998) - et al.
Use of the Priestley-Taylor evaporation equation for soil water limited conditions in a small forest clearcut
Agric. For. Meteorol.
(1991) - et al.
Evaporation and energy budget in Lake Vegoritis, Greece
J. Hydrol.
(2007) - et al.
Google earth engine: planetary-scale geospatial analysis for everyone
Remote Sens. Environ.
(2017) - et al.
An approach for global monitoring of surface water extent variations in reservoirs using MODIS data
Remote Sens. Environ.
(2017) Data-sparse estimation of lake evaporation, using a simplified Penman equation
Agric. For. Meteorol.
(1993)- et al.
Evaluation of potential impacts on Great Lakes water resources based on climate scenarios of two GCMs
J. Great Lakes Res.
(2002)
Evaporation variability of Nam Co Lake in the Tibetan Plateau and its role in recent rapid lake expansion
J. Hydrol.
Water on an urban planet: urbanization and the reach of urban water infrastructure
Glob. Environ. Chang.
An area-dependent wind function for estimating open water evaporation using land-based meteorological data
Environ. Model Softw.
Measurements of evaporation from a mine void lake and testing of modelling approaches
J. Hydrol.
Developing a decision support tool for China's re-vegetation program: simulating regional impacts of afforestation on average annual streamflow in the Loess Plateau
For. Ecol. Manag.
Global review and synthesis of trends in observed terrestrial near-surface wind speeds: implications for evaporation
J. Hydrol.
Stream temperature/air temperature relationship: a physical interpretation
J. Hydrol.
Operational estimates of areal evapotranspiration and their significance to the science and practice of hydrology
J. Hydrol.
Improvements to a MODIS global terrestrial evapotranspiration algorithm
Remote Sens. Environ.
A comprehensive study across methods and time scales to estimate surface fluxes from Lake Kinneret, Israel
J. Hydrol.
Comparison of 15 evaporation methods applied to a small mountain lake in the northeastern USA
J. Hydrol.
Comparison of energy-budget evaporation losses from two morphometrically different Florida seepage lakes
J. Hydrol.
Extending satellite remote sensing to local scales: land and water resource monitoring using high-resolution imagery
Remote Sens. Environ.
Is the Class A evaporation pan still the most practical and accurate meteorological method for determining irrigation water requirements?
Agric. For. Meteorol.
New findings about the complementary relationship-based evaporation estimation methods
J. Hydrol.
TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015
Scientific Data
Evaporation estimation for Lake Okeechobee in south Florida
J. Irrig. Drain. Eng.
Evaluating best evaporation estimate model for water surface evaporation in semi-arid region, India
Hydrol. Process.
Crop evapotranspiration-guidelines for computing crop water requirements-FAO irrigation and drainage paper 56, FAO
Rome
Regional assessment of evaporation from agricultural irrigation reservoirs in a semiarid climate
Agric. Water Manag.
On the variability of the Priestley-Taylor coefficient over water bodies
Water Resour. Res.
When will Lake Mead go dry?
Water Resour. Res.
Stream temperature-equilibrium temperature relationship
Water Resour. Res.
Evapotranspiration réelle et potentielle, signification climatique
IAHS Publ.
Evaporation into the Atmosphere: Theory, History and Applications
Indications of increasing land surface evaporation during the second half of the 20th century
Geophys. Res. Lett.
Hydrologic cycle explains the evaporation paradox
Nature
A global lake and reservoir volume analysis using a surface water dataset and satellite altimetry
Hydrol. Earth Syst. Sci. Discuss.
Predicting river water temperatures using the equilibrium temperature concept with application on Miramichi River catchments (New Brunswick, Canada)
Hydrol. Process.
El Niño 1997–1998: The Climate Event of the Century
Impact of artificial reservoir water impoundment on global sea level
Science
Assessment of clear and cloudy sky parameterizations for daily downwelling longwave radiation over different land surfaces in Florida, USA
Geophys. Res. Lett.
The effects of climate change on the hydrology and water resources of the Colorado River Basin
Clim. Chang.
Earth's surface water change over the past 30 years
Nat. Clim. Chang.
Toward improved validation of satellite sea surface skin temperature measurements for climate research
J. Clim.
The response of water temperatures to meteorological conditions
Water Resour. Res.
On the concept of equilibrium evaporation and the value of the Priestley-Taylor coefficient
Water Resour. Res.
Bulk parameterization of air-sea fluxes for tropical ocean-global atmosphere coupled-ocean atmosphere response experiment
Journal of Geophysical Research: Oceans
A comparison between measured and modelled open water evaporation from a reservoir in south-East England
Hydrol. Process.
Estimation of Open Water Evaporation: A Review of Methods
Cited by (99)
Lake evaporation in arid zones: Leveraging Landsat 8's water temperature retrieval and key meteorological drivers
2024, Journal of Environmental ManagementEnsemble modeling of global lake evaporation under climate change
2024, Journal of HydrologyEffect of continuous and modular floating covers on evaporation losses and microalgal growth
2024, Journal of King Saud University - Engineering Sciences