1. INTRODUCTION
Reservoirs are subject to substantial water level fluctuations caused by water release operations and, as such, to shorter water residence times than lakes. Evaporation (latent heat flux, LE ) is a major component of the mass and energy budgets of reservoirs, which can compromise some typical key functions such as freshwater supply, irrigation, hydropower, navigation, and other related economic activities (Friedrich et al., 2018). In some arid regions of the world, structural measures are put in place to limit evaporation, such as floating balls or lattices (Assouline & Narkis, 2021). Evaporation is intangible and therefore a difficult hydrological flux to measure, making it difficult to fully understand its magnitude and controlling mechanisms.
Studies that have quantified reservoir evaporation are rare. Tanny et al. (2008) reported an average evaporation rate of 5.5 mm day–1 from July to September in the Eshkol reservoir in Israel (33°N), which has a hot and arid climate. Further north, in the Eastmain-1 reservoir in Canada (52°N), Strachan et al. (2016) found the evaporation rate to be 3.1 mm day–1 between August and October. For the Great Slave Lake in Canada (61°N), Rouse et al. (2003) reported 2 mm day−1 in summer and 5 mm day−1 in December. These studies indicate that even in cold regions, reservoir evaporation can be substantial.
In cold climates, reservoirs undergo two turnovers per year (dimictic). Their thermal regime typically evolves into three successive phases (Cole & Weihe, 2016). During the ice cover phase, ice acts as a lid over the water body, preventing direct interactions between the atmosphere and the water column. The water column then becomes stratified, with cold water (< 4°C) sitting on top of warmer water. Latent heat fluxes tend to remain low during this period (Wang et al., 2016). From ice breakup in spring to the middle or end of summer, the heat storage phase unfolds. Energy is first stored in the top several meters of the water column closer to the surface and eventually reaches deeper layers due to wind-induced mixing and internal hydrodynamics (Spence et al., 2003; Vincent, 2018). The result is a surface mixing layer (epilimnion) that is separated from the homogeneous deep layer (hypolimnion) by a zone with a steep temperature gradient (metalimnion). Latent heat fluxes remain low during this phase, with frequent and stable atmospheric stratification. The third and final phase corresponds to the heat release period. This is characterized by a decline in water temperature due to a substantial release of energy into the atmosphere through turbulent heat fluxes that are high and sustained day and night (Blanken et al., 2011). The epilimnion then slowly deepens until the fall turnover, during which the temperature of the entire water column becomes homogenous.
While evaporation varies seasonally in response to the three thermal phases, it also fluctuates on smaller time scales in response to meteorological forcing. For instance, incoming shortwave radiation causes latent heat fluxes to peak during the day, thereby increasing the rate of evaporation during peak times (Lensky et al., 2018). The atmospheric demand for water vapour, driven by wind speed and vapour pressure deficit, is also known to modulate evaporation in water bodies (Pérez et al., 2020). Evaporative demand can vary within a single day. For instance, changing wind direction can lead to a reduced or enhanced sheltering effect, increasing or decreasing evaporation rates (Markfort et al., 2010; Venäläinen et al., 1998). Evaporation can also vary over the course of a few days, due to passing synoptic systems that can generate sustained evaporation (Laird & Kristovich, 2002; Spence et al., 2013). Blanken et al. (2000) found that 50% of annual evaporation over the Great Slave Lake occurred over only 25% of the year through episodic evaporation water losses. Moreover, thermocline depth and intensity, which depends in part on the reservoir morphometry (Gorham, 1964), influence turbulent heat fluxes by limiting or enhancing the energy available in the upper water layers. Indeed, Piccolroaz et al. (2015) identified positive feedback between the lake surface temperature and the stratification dynamics of Lake Superior, Canada. Therefore, the timing of evaporation occurs at different scales and remains poorly documented or correlated to physical drivers (Beck et al., 2018).
Northeastern America is one of the densest regions of lakes and reservoirs around the world (Downing et al., 2006). These lakes and reservoirs are considered to be climate sentinels (Adrian et al., 2009; Williamson, Saros, & Schindler, 2009) as well as integrators and regulators of climate change (Williamson, Saros, Vincent, et al., 2009). Wang et al. (2018) showed that modifications in surface energy allocation under warmer climate conditions will accelerate global lake evaporation. In-situ evaporation observations are needed to develop and improve lake models (McJannet et al., 2017) for future climate estimates, particularly in remote areas.
There is a lack of direct in-situ measurements of turbulent heat fluxes over reservoirs in remote northern regions. The overarching goal of this study is to identify the characteristic time scales of evaporation from a deep subarctic hydropower reservoir. Using four years of eddy-covariance (EC) measurements, the specific objectives are to quantify turbulent heat fluxes at daily, monthly and annual time scales, and to identify the key processes and surface energy budget terms that govern LE at each time scale. The paper is organized as follows. We first introduce the study site and measurement methods. Then, we describe the meteorological conditions over the whole study period and the driving factors for each time scale. Finally, uncertainties in the flux data are discussed, given the energy budget of several water layers in the reservoir.