Hydrokinetic resource assessment for the Canadian Arctic for turbine-based power generation Open Access

With the increasing energy demand (WHO 2009) and the continued warming of the Arctic regions at double the rate of mean global warming (Lee et al. 2021), there has been an increasing focus on harnessing renewable energy sources to reduce greenhouse gas emissions. Hydrokinetic energy is a renewable form of energy, where the kinetic energy of the flowing water is converted first into mechanical energy and then into electrical energy. Unlike traditional hydropower systems, which require the construction of dams, reservoirs, or diversion channels, hydrokinetic systems use turbines, generators, and other devices, which are placed directly into the flowing water in order to extract energy. As a result, it can be installed even in the most remote and extreme locations depending on the necessity and ease of installation. For identifying potential locations for hydrokinetic power extraction, it is important first to develop detailed datasets and estimates of available resources based on observed streamflow records, targeted hydrologic and hydraulic simulation experiments, or analysis of remotely sensed datasets. Such information, particularly for high latitudes, is generally not readily available.

Selected global to regional scale studies that have investigated the hydrokinetic potential utilizing observed streamflow, numerical modeling, and simulation techniques, and characteristics of various hydrokinetic energy systems and their feasibility for the Arctic regions are discussed below. At the global scale, Ridgill et al. (2021) assessed available riverine hydrokinetic resources using GRADES, a modeling system that incorporates the variable infiltration capacity land surface model (Liang et al. 1994) at 0.25° resolution and the routing application for parallel computation of discharge (David et al. 2011) with the MERIT-Hydro version 0.0 dataset (multi-error-removed-improved-terrain global hydrography data set; Yamazaki et al. 2019). They estimated the global resources to be of the order of 58,400 ± 109 TWh/year, which also demonstrated the high hydrokinetic potential of the North American continent. Although this study provides useful insights, the 25 km resolution, which is suitable for global-scale assessment, is too coarse for detailed site-specific analysis required for turbine installation.

At a regional level, the Underwood & McLellan group (1980) conducted one of the earliest Canada-wide assessments of hydropower potential based on observed streamflow data. As this study used data available prior to the 1980s, the analysis focused mainly on large rivers but suggested potential for hydrokinetic power in river reaches with flow greater than 450 m³/s. Building up on the methodology used by the Underwood and McLellan (UMA) group, a similar study was conducted in the USA by Miller et al. (1986) for the US Department of Energy, focusing on the river reaches with mean flows greater than 113 m³/s and flow velocity greater than 1.3 m/s. The methodology considered provided a conservative assessment of the hydrokinetic power, suggesting an estimated total mean annual power of 12.5 GW (i.e., 110 TWh/year of hydrokinetic energy), with the highest contributions being associated with western, northwestern, and Alaskan regions. Later, Jacobson (2012) expanded this study to reaches with mean flows greater than 28 m³/s. They reported technically recoverable hydrokinetic energy of up to 119.9 TWh/year, indicating an approximate 9% increase compared to the estimate provided in Miller et al. (1986). However, they could not validate their results due to the lack of empirical data from existing hydrokinetic energy conversion systems. Duvoy & Toniolo (2012) and Toniolo et al. (2010) studied the hydrokinetic potential of two river reaches in Alaska to assess the suitability of an in-stream river turbine installation. Velocity outputs generated by an existing depth-integrated two-dimensional hydrodynamic numerical model CCHE2D (Jia & Wang 2001; Zhang 2018), coupled with HYDROKAL (hydrokinetic calculator), were used to compute the instantaneous power density and identify potential installation sites for hydrokinetic power extraction. However, the employed model had difficulty in representing the helical flows, which developed on the outer slopes of river bends. For the Canadian high-latitude regions, although many studies have looked at streamflow characteristics (e.g., Su et al. 2005; Poitras et al. 2011; Clavet-Gaumont et al. 2013; Huziy et al. 2013; Teufel & Sushama 2021), very few studies have focused on hydrokinetic potential. Recently, Teufel & Sushama (2022b) did a preliminary analysis of hydrokinetic potential over Canada, including the high-latitude regions, for the current climate, by considering outputs from a coarse resolution regional climate modeling system. This study highlighted the need for high-resolution climate model outputs for better estimation of hydrokinetic potential.

Irrespective of the approach adopted for assessing the hydrokinetic potential globally or regionally for the historical periods, it is also important to consider the impact of future climate change, especially on streamflow regimes, magnitudes, and velocities, as these are the most important factors that determine hydropower potential. Teufel & Sushama (2021) studied the Canadian rivers using an ensemble of regional climate model (RCM) simulations and found mostly increases in the median annual streamflow for the Canadian Arctic region for the Representative Concentration Pathway 8.5 scenario. A previous RCM-based study by Poitras et al. (2011), for the Special Report on Emissions Scenarios (SRES) A2 scenario, also reported future increases in mean annual streamflow for the Arctic basins Mackenzie and Yukon. Climate change impacts on streamflow were considered by Zhang et al. (2023) for four major pan-Arctic river basins, namely Mackenzie, Ob, Lena, and Yenisei basins, using machine learning models such as support vector regression, artificial neural network, and multi-variable regression under climate change scenarios SSP2-4.5 and SSP5-8.5, using data from five CMIP6 global climate models (GCMs) for the 2020–2100 period. Their results suggest increases in mean annual streamflow for the studied basins.

Over the years, various hydrokinetic energy conversion systems have been studied and tested to harness the hydrokinetic energy of the flowing water bodies. In some cases, the conversion of hydrokinetic energy to electrical energy through a turbine can operate at almost zero head (Balat 2006). However, the amount of energy that can be harnessed from the flow depends on the turbine's efficiency and the flow velocity necessary for the operation of a hydrokinetic device, which typically ranges from 1 to 2 m/s, but in some cases, it can be as low as 0.5 m/s, depending on the additional methods adopted (Johnson & Pride 2010). Considering the orientation of the rotor with respect to the direction of the flow, hydrokinetic turbines are classified broadly into horizontal axis turbines or axial flow turbines and vertical axis turbines (Khan et al. 2009). Several deployments over the past decade, including the EVG-005H in Ruby, Alaska; the EVG-025H in British Columbia, Canada; and the Riverlution turbine in Alberta, Canada, adopted the vertical axis turbines, such as the H-Darrieus or Squirrel cage Darrieus (straight-bladed) turbines for river-based applications over the horizontal axis turbines, as they do not require a yaw mechanism, are quieter in operation, and are simple in design with the possibility to design a turbine with a diameter larger than the depth of the river in shallow rivers to generate higher power (Saini & Saini 2019; Khan et al. 2022). However, in cold regions, many cold weather operational adjustments are required, such as heating of the gearbox oil and ice removal, which are important for the functioning of the device, as any component that pierces the water interface will accumulate ice through direct splashing and freezing or buildups through frazil ice processes (Woods 2017). Several studies have been conducted to improve the efficiency of turbines in different flow conditions. Guney (2011) and Khan et al. (2008) exemplified that emphasis was on enhancing the performance coefficient and tip speed ratio by incorporating duct augmentation, variable pitch blade structures for vertical turbines, a flip wing mechanism, fixed-wing rotor, amongst other strategies. The power from a single hydrokinetic turbine is usually small, and thus the energy density can be increased by the installation of the turbines in arrays or farms (Balat 2006).

Due to limited and sporadic observational data in the higher latitudes, hydrological models can be employed to estimate streamflow. The conventional approach to studying climate change impacts on streamflow and power generation involves the use of hydrological model simulations driven by climate model outputs corresponding to selected emission scenarios. Interactive modeling of streamflow at several kilometers to sub-kilometers scale using a climate-hydrology integrated modeling approach is also starting to emerge with recent advances in the areas of high-resolution modelling (e.g., Prein et al. 2015; Diro & Sushama 2019; Teufel & Sushama 2022a; b) and high-performance computing. Such fine resolutions help further improve the simulated streamflow due to better representation of surface characteristics and soil and landscape heterogeneities in the climate model. The objective of this study is to assess the hydrokinetic power potential in the Canadian Arctic using continuous streamflow sequences derived from ultra-high-resolution climate model simulations performed using a state-of-the-art regional climate model for both current and future climates.

The rest of the paper is structured as follows: Section 2 comprises a detailed description of the streamflow data and related climate and routing models, along with the methodology. Section 3 focuses on streamflow validation and discusses characteristics of estimated flows, flow velocities, and hydrokinetic power for the current 2001–2020 period. Projected changes to hydrokinetic power for the near-future 2021–2040 period are presented in Section 4. Section 5 discusses potential areas for hydropower extraction. Discussion and conclusions are provided in Sections 6 and 7, respectively.

The state-of-the-art regional climate model, GEM, solves non-hydrostatic, deep atmosphere dynamics with an implicit, two-time-level semi-Lagrangian numerical scheme. In the horizontal, the model uses a regular latitude–longitude grid with Arakawa C staggering and a rotated pole configuration such that the domain is approximately centered on the equator to minimize changes in grid spacing across the domain. In the vertical coordinate, following Girard et al. (2014), Charney–Phillips staggering is used. The radiation scheme is represented by the correlated K solar and terrestrial radiation of Li & Barker (2005) and the planetary boundary layer scheme follows Benoit et al. (1989) and Delage (1997). The double-moment microphysics scheme of Milbrandt & Yau (2005) is used for condensation processes. In addition to the large-scale precipitation schemes, the model includes the deep convection scheme of Kain & Fritsch (1990) and the shallow convection based on Bélair et al. (2005). As discussed in Teufel & Sushama (2022a), the use of convection parameterization for ∼3–8 km grid spacing is still a topic of debate and considered a gray zone, as convection is neither fully resolved nor can it be assumed to be smaller than the grid box spacing (Gerard et al. 2009).

The land processes in GEM are represented by the Canadian Land Surface Scheme (Verseghy 2011). Given the importance of permafrost in the study domain, the scheme uses a 60 m deep soil layer configuration, consisting of 26 layers of varying thicknesses: 0.1, 0.2, 0.3, 0.4, 0.5 (× 10), 1.0, 3.0, and 5.0 (× 10) m.

WATROUTE is based on the routing algorithm of the WATflood-distributed hydrological model (Kouwen et al. 1993; Soulis et al. 2000). In the original version of WATROUTE, routing through the digital river network is performed using a single surface reservoir. The modified version of WATROUTE used in this study includes a groundwater reservoir modeled as a linear reservoir as proposed in Sushama et al. (2004). Detailed information on the model formulation can be found in Teufel & Sushama (2022b).

The flow directions of rivers, channel length, and slopes that are required by WATROUTE are obtained from the Hydrological Data and Maps based on Shuttle Elevation Derivatives at Multiple Scales (HydroSHEDS) database (Lehner et al. 2008). The primary sources of data used in the development of HydroSHEDS are the digital elevation model from National Aeronautics and Space Administration (NASA)'s Shuttle Radar Topography Mission (SRTM), with ancillary data sources such as the SRTM Water Body Data, the river networks of the Digital Chart of the World (ESRI 1993), and Arc World (ESRI 1992), and the Global Lakes and Wetlands Database (Lehner & Döll 2004). The processing steps of generating HydroSHEDS are available in the technical documentation by Lehner et al. (2013). As the SRTM elevation data are not available beyond 60° N latitude, DEM from the HYDRO1k database (USGS 2000) is used to complete the hydrographic data and thus to obtain full global coverage of the drainage networks and sub-basin delineations. The flow direction from HydroSHEDS is upscaled to 4 km resolution following Huziy et al. (2013) and Döll & Lehner (2002).

Two streamflow simulations are considered in this study. The first spans the period from 2001 to 2020 and is derived from runoff simulated by the GEM model at 4 km resolution using the WATROUTE routing scheme. Here, the GEM simulation is driven by ‘perfect boundary conditions’, namely, the European Centre for Medium Range Weather Forecasts ERA5 reanalysis (Hersbach et al. 2020) over the domain shown in Figure 1(a). The initial soil conditions for this simulation are based on a pan-Arctic GEM simulation driven by ERA-Interim (Teufel et al. 2019), while the atmospheric initial conditions are specified from ERA5. This simulation, hereafter referred to as GEM_ERA5, is used for validation purposes. The second streamflow simulation, spanning from 2001 to 2040, is derived from a transient climate change GEM simulation driven by the Canadian Earth System Model (CanESM2) under the Representative Concentration Pathway (RCP) 8.5 emission scenario (Riahi et al. 2011); this simulation is denoted as GEM_CanESM2, referencing the specific period of interest, i.e., ‘current’ for 2001–2020 and ‘future’ for 2021–2040. When driven by CanESM2, GEM simulations are initially downscaled to 10 km resolution (the outer domain shown in Figure 1(a)); the outputs (temperature, wind, humidity, surface pressure, and geopotential height fields) from which are used as the lateral boundary conditions for the ultra-high-resolution GEM simulation over the inner domain at 4 km resolution. The initial conditions for the 4 km GEM simulation are obtained from the 10 km GEM simulation driven by CanESM2.

The available hydrokinetic power P_k is estimated from the simulated streamflow at 4 km resolution as P_k = (ρ/2)AV³, where ρ is the fluid density (considered as 1,000 kg/m³), and A and V are the cross-sectional area and flow velocity, respectively. The power captured by a hydrokinetic turbine is a fraction of that estimated using this equation, as it also depends on the swept area of the rotor and the performance coefficient of the turbine, which vary with the turbine and modifications considered.

To identify river reaches suitable for hydropower extraction, a velocity threshold of 1.5 m/s is considered, following the guidelines for the practical operation of hydrokinetic turbines (Johnson & Pride 2010). The number of days with velocity greater than this threshold is estimated, which could be useful in decision-making. Although other criteria, including site characteristics, can be considered, this study focuses only on the velocity threshold.

Projected changes to streamflow, flow velocities, and hydrokinetic power are obtained by comparing the future 2021–2040 period with the current 2001–2020 period of GEM_CanESM2-based WATROUTE simulations.

For regions where the current annual mean hydrokinetic power potential is in the 10–10³ kW range, the projected changes are in the 0–4 kW range. Similarly, where the current power potential is in the 10³–10⁶ kW range, a change of up to 10 kW is noted. However, the highest change in the 100–500 kW range is noted for the regions where the current power potential is in the 10⁶–10⁹ kW range, which is at the outlets of the basins. Table 1 provides the list of the top 10 Nunavut basins with hydrokinetic power potential and their projected changes. The relatively modest changes noted are an expected scenario, as more prominent changes are anticipated toward the end of the 21st century under the RCP 8.5 scenario.

Table 1

The basin ID and names of Nunavut basins with the highest hydrokinetic power potential in descending order are shown in columns 1 and 2

Potential basins .		Hydrokinetic power (kW) .		.
Basin ID .	Names .	2001–2020 .	2021–2040 .	.
8	Baker Lake	794,328	794,128
5	Thelon	100,000	99,654
1	Seal	100,000	99,600
31	Back	79,433	79,203
7	Kazan	10,000	9,800
32	Back-Hayes	3,163	3,255
2	Thlewaiza	3,163	3,150
4	Than-anne	1,000	880
9	Quoich	759	769
30	Queen Maud Gulf	501	531

Potential basins .		Hydrokinetic power (kW) .		.
Basin ID .	Names .	2001–2020 .	2021–2040 .	.
8	Baker Lake	794,328	794,128
5	Thelon	100,000	99,654
1	Seal	100,000	99,600
31	Back	79,433	79,203
7	Kazan	10,000	9,800
32	Back-Hayes	3,163	3,255
2	Thlewaiza	3,163	3,150
4	Than-anne	1,000	880
9	Quoich	759	769
30	Queen Maud Gulf	501	531

Note. The third column shows the hydrokinetic power for the grid cells with median values for respective basins for the current 2001–2020 period and the fourth column shows their future values for the 2021–2040 period.

View Large

In the future 2021–2040 period, the number of days shows increases of up to 10 days for the northern basins, while decreases of the same order are noted for the southern basins (Figure 11(b)). These are consistent with the velocity patterns shown in Figures 10 and S4.

In this paper, hydrokinetic power potential is estimated for Nunavut and adjoining regions using an integrated climate-hydrology modeling approach at 4 km resolution for the 2001–2020 period, which is considered a surrogate for limited historical records. This physically sensible approach is adopted due to the severe lack of hydrometric station-based observations and to produce estimates of hydrokinetic power potential at reasonably fine spatial resolutions throughout the region to support optimal siting of hydrokinetic power extraction systems. A triangular cross-section of the river channel is assumed when using WATROUTE. This choice simplifies the calculations required for estimating hydraulic properties, such as flow areas and hydraulic radius, and reduces the need for detailed cross-sectional and profile data. This approach is particularly beneficial when dealing with large basins, such as those considered in this research, where data collection is both challenging and resource-intensive. It is important to note that other cross-sectional shapes may also be utilized depending on specific hydraulic conditions and data availability (Kouwen 2023).

Considering that the extraction of hydrokinetic power can only be realized in future periods, climate warming consequences on the hydrokinetic power potential are also analyzed for the 2021–2040 near-future period for the RCP 8.5 scenario for informed decision-making. It must be noted that this type of analysis for the region is accomplished for the very first time using ultra-high-resolution regional climate model simulations. Although the ability of GEM to simulate various climatic parameters and land features such as soil temperature, air temperature, depth to bedrock, and precipitation characteristics that influence surface runoff and drainage has already been established in the literature (Teufel & Sushama 2022a), some focused evaluations are also accomplished in this study, which seems satisfactory.

Additional research is needed to better understand streamflow dynamics during ice cover conditions. While ice cover typically increases the flow resistance and reduces surface flow velocity, variations in ice thickness or the formation of ice dams can result in temporary increases in water velocity by elevating the upstream water head (Rokaya et al. 2022). Thus, observations of streamflow velocities during river ice conditions are required for the northern regions for in-depth river reach-specific studies, which are very scarce at the moment. This, coupled with hydrodynamic model simulations of flow velocities for river reaches with observed data and machine learning approaches, can contribute to the development of spatio-temporal velocity reduction factors for large-scale applications such as that considered in this study.

From the various analyses and assessments presented and discussed in this paper, the following conclusions can be drawn. Shortcomings of the study and future extensions and research avenues are also highlighted.

• 1. Comparisons of observed flows from HYDAT with those derived from ERA5 and CanESM2-driven GEM simulations, one way coupled with the WATROUTE scheme, confirm the ability of the climate-hydrology integrated modeling approach in simulating variables that are required for the estimation of hydrokinetic power for the Canadian Arctic River network.

• 2. Projected changes for the 2021–2040 period with respect to the 2001–2020 period under the RCP 8.5 scenario are minimal for streamflow, flow velocity, and hydrokinetic power over the entire study region. Projected changes could be different in the far future, which is not considered in this study. Increases in the active layer thickness and degrading permafrost conditions in the higher latitudes, projected toward the end of the century, can significantly influence flow patterns and velocities and therefore the hydrokinetic potential of Arctic rivers.

• 3. As zero-head hydrokinetic turbines can only capture a certain percentage of the kinetic energy of flow, it is important to consider river reaches with flow velocities greater than 1.5 m/s and also where a higher potential of hydrokinetic power exists throughout the year or during a significant part of the year. Further granular analysis of the hydrokinetic power estimates will help in understanding the behavior of individual river reaches and in recommending a suitable hydrokinetic conversion system.

• 4. The Back (Basin ID: 31) and Back-Hayes (Basin ID: 32) basins along the Arctic coast in central Nunavut and the basins draining into Hudson Bay in the east, which include the Thelon (Basin ID: 5) and Baker Lake (Basin ID: 8) basins, show the highest potential for hydrokinetic power. The Seal (Basin ID 1), Than-anne (Basin ID: 4), and the Thlewaisa (Basin ID: 2) basins in the southeast of the domain also show potential for hydrokinetic power generation owing to their consistent flow velocity maintained throughout the year.

One of the main limitations of this study is that the projected changes are analyzed based on a single transient climate change GEM simulation for the RCP 8.5 scenario extending until 2040. This was unavoidable and is due to the high computational cost involved in undertaking the ultra-high-resolution simulations at 4 km resolution. To better quantify the uncertainty, it is also useful to consider a multi-model ensemble of simulations at 4 km or even finer resolution. Additionally, the representation of glaciers is considered static and not dynamic, which hinders the accurate representation of streamflow dynamics for glaciated basins. Moreover, due to limited data specific to river channels, the channel modifications due to the installation of turbines, such as the changes in flow velocities and depth of the channel, are not considered in this study. This omission highlights the need for more comprehensive data collection and analysis. Detailed granular studies of the identified reach, employing hydrodynamic models and considering the impact of river ice on the flow volumes and velocities, will contribute to more precise identification of optimal locations and site-specific research for the installation of hydrokinetic conversion systems.

A lack of available information on resource assessments is one of the main limitations in the development of appropriate hydrokinetic energy conversion technology. The ultra-high-resolution climate model coupled with the routing scheme used in the study to estimate flows, flow velocities, and hydrokinetic power for the high-latitude region of Canada contributes toward addressing this deficiency by introducing a science-based approach for hydrokinetic resource assessment and thus reduces reliance on highly uncertain gross regional estimates. This study offers a more reliable assessment, providing valuable and new insights for the development of hydrokinetic energy technologies, and lays the foundation for advanced modeling to support hydrokinetic technology installations.

This research was funded by the Ocean Coastal and River Engineering Research Center of the National Research Council Canada and the Trottier Institute for Sustainability in Engineering and Design. The simulations considered in this study were performed on the supercomputer managed by Digital Research Alliance and Calcul Québec.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.