Alternative method for determining available winter water volumes from lakes to support small-scale projects

Water is required for a variety of activities during the development of industrial projects. These include domestic use to support remote camps, drilling, and process water and dust suppression. In northern regions, exploration continues in winter months and water may also be required to construct ice roads and overland winter (snow) roads to provide access to areas that cannot be accessed during the open water months (Cott et al. 2015). As a result, water is often removed from ice-covered water bodies for winter activities. Aquatic populations in lakes are at higher risk during the winter, as the removal of water will expose sensitive habitat in nearshore areas to freezing, the absolute loss of water volume reduces the volume of available overwintering habitat and reduces the available mass of dissolved oxygen available for aquatic life (Cott et al. 2008b; Leppi et al. 2016). Water level fluctuations, particularly in winter, can be detrimental to aquatic life including fish and their habitat by, for example, stranding (Nagrodski et al. 2012) and freezing eggs and larvae as water levels drop (Cott et al. 2008a). As a result, it is necessary to restrict the volumes of water that can be taken during winter. Although proponents may provide their own estimates for the licencing process, these are derived using unproven methods and can only be verified during operations, risking damage to aquatic habitats if the methods are not protective. There is therefore a need to develop and verify a method to determine appropriate limits for water withdrawal under ice that (a) can be easily used by proponents, (b) is acceptable to regulators, and (c) provides conservative estimates of available water to ensure that water withdrawal does not exceed 10% and aquatic life is protected.

Regulatory boards operate in the Northwest Territories (NT) of Canada under federal and territorial legislation and are responsible for issuing licences to regulate water use above certain thresholds. The Minister of Environment and Climate Change of the Government of the Northwest Territories (GNWT-ECC) is responsible for administering the Territorial Waters Act and approving Type A Water Licences¹ on territorial lands while the Land and Water Boards (LWBs) conduct the water licence processes and issue water licences. As a result, the parties often collaborate on the water licencing process.

As part of that process, the Boards are required to conduct a preliminary screening of proposed activities under the Mackenzie Valley Resource Management Act to assess potential environmental impacts related to development and ensure that appropriate mitigation measures are in place. Such screening includes the potential for water withdrawal to impair aquatic habitat. Mitigation may include specific volume thresholds, timing limitations, and other conditions that consider protection of the aquatic environment. Water licence applicants are required to provide information regarding proposed water uses, location of proposed water sources, timing and proposed volume of water to be used, and a comparison of the proposed water use volume to the available water for each proposed source.

Previous initiatives in the NT developed guidance on the amount of water that could be safely withdrawn without harming fish habitat, thus complying with the Fisheries Act. A withdrawal threshold of 10% of the under-ice volume was determined to be protective of fish habitat for lakes under maximum ice cover through a whole lake manipulation experiment (Cott et al. 2008b). Withdrawals beyond 10% resulted in decreases in dissolved oxygen below levels required to protect overwintering fish species (Cott et al. 2008b). The Department of Fisheries and Oceans Canada then established the Protocol for Winter Water Withdrawal from Ice Covered Waterbodies in the Northwest Territories and Nunavut (Fisheries and Oceans Canada (DFO) 2010, ‘the Protocol’) which outlined thresholds for the amount of water to be removed under ice in small lakes depending on location, lake depth, and ice cover. The Protocol also outlined the requirements to establish available water volumes by specifying minimum bathymetric information, depth criteria, and maximum expected ice cover for water sources. The Protocol has been considered best practice by several of the regulatory boards in the NT and in other jurisdictions including Alaska, Nunavut, and Saskatchewan (P.Cott, GNWT-ECC, pers. comm.). It is often included as a condition of water licences as a standard means to determine if appropriate information on water sources has been included with an application.

While this has become the standard, in recent years there has been some concern raised from industrial proponents, particularly smaller-scale mineral exploration companies, regarding the amount of effort required to complete the bathymetric field programs defined in the Protocol during a typical exploration project. The Protocol outlines specific field methods to gather the necessary information on the bathymetry of proposed water sources and requires each to be field verified either using a boat and depth-sounding equipment in the summer or ground-penetrating radar in the winter. Mineral exploration activities are mobile and transient in nature, however, with drill targets identified and revised depending on the results of geophysical exploration and exploratory drilling. As such, exploration companies are unable to confirm all potential water sources at the time of the Water Licence application, resulting in a need for operational flexibility in the field. In addition, exploration companies often have limited capital, expertise and resources to conduct extensive environmental field programs as required under the Protocol prior to commencing drilling activity. Therefore, the amount of time and financial commitment required to conduct these bathymetric surveys on all potential water sources in the field could delay the licencing process and limit the feasibility of conducting exploration projects.

The LWBs and the GNWT therefore initiated a project in 2019 to develop a simple method that could be applied to estimate available under-ice water volumes in small lakes in support of early-stage exploration or other small-scale activities. The intent was to determine if a simple method would provide a conservative and reliable volume estimate that could be used without requiring detailed bathymetry and which, if followed, would be supported by regulators as protective of fish habitat. The initial requirements were outlined in a request for proposal for an independent consultant, a report based on the methods outlined below (HESL 2021) and additional investigations and method verification by the authors.

Lake data sets from the NT and Nunavut were used because the guidance method would be used in the NT and, as such, data from the NT (and adjacent Nunavut) would include lakes from the area of application. The study lakes included a large range of lakes across the region and would therefore accurately reflect their range of characteristics. In addition, the project rationale was focused on the needs of project proponents, many of whom had readily accessible data sets from existing projects that had been prepared for the licencing process. Data sets used were the extent of those that were available to the team during the time of the review and were not selected as a subset of a larger data set or in any other specific manner. The following lake data sets were used:

• The DFO data set (Cott et al. 2005) was collected from 37 lakes in the Mackenzie River Delta region of the northwestern NT by Fisheries and Oceans Canada to assist with the development of the original protocol for the protection of fish habitat. Average depths ranged from 0.5 to 5.3 m and bathymetry was collected in the ice-free season.

• The Golder Back River data set of 55 lakes was tested for initial model development using measured total lake volumes from the Back River Gold Project and then as a test data set for under-ice volume estimation assuming 2 m ice cover (Golder Associates 2018). Average depths in the 55 lakes ranged from 1.7 to 33.6 m and maximum depths from 3.5 to 64.45 m.

• A data set for 17 lakes along the Tibbett to Contwoyto Ice Road which extends 400 km NE of Yellowknife for which ice cover was measured (EBA 2013) was used for testing the model. Average water depths ranged from 0.8 to 4.3 m and maximum depths were not provided.

• A Kennady Diamonds mineral exploration data set of 28 lakes ∼300 km NE of Yellowknife with an assumed ice depth of 2 m (Hunt 2020, pers. comm.) was used for testing the model. Eight lakes were <2 m maximum depth, assumed to freeze to the bottom and excluded from the analysis. Maximum depths of the remaining lakes ranged from 2.3 to 14.5 m and average depths from 0.4 to 4.7 m.

• A set of data for 59 lakes with an assumed ice depth of 2 m obtained by Wilfred Laurier University along the Inuvik – Tuktoyaktuk Highway (ITH) in the Mackenzie River Delta region was used for testing the model. Seven of these lakes were <1.8 m maximum depth and frozen to the bottom. Average depths of the remaining 52 lakes ranged from 0.9 to 10 m and maximum depths ranged from 2.2 to 19.9 m.

The volume estimation method was developed in four steps:

• – A review of primary and technical literature established that there were no published methods that could be readily adapted to the purposes (see ‘Literature Review’).

• – Candidate volume estimation methods were sought by canvassing government, industrial, and academic operators in the NT.

• – Submitted methods were tested for their ability to estimate water volumes for a data set of lakes for which measured bathymetry was available (DFO data set – Cott et al. 2005; Table 1).

• – The method that gave the best estimates of water volume for the DFO data set (Golder Nighthawk model) was then tested using four sets of lakes (Golder Back River, Tibbett to Contwoyto Ice Road, Kennady Diamonds, and Wilfred Laurier University, Table 1) for which under-ice water volumes had been established by measurement to provide confidence in the estimates generated by the preferred method and average depths for which the preferred method was protective.

Detailed analyses and complete data sets are provided in Hutchinson Environmental Sciences Ltd (HESL 2021).

HESL (2021) completed a literature review which identified several potential approaches based on remote sensing or the use of a regional bathymetric data set and existing topographic maps:

• The HAB-2 model (Walther et al. 2011) estimated depths from aerial, high-resolution, film-based, true-colour imagery in the McKenzie River, Oregon. Equivalent photography was not available for the NT, the regional topography differed such that direct application of the model was not feasible and the model was intended for estimation of stream flow and could not be applied to lakes in the NT.

• Hamilton et al. (1993) used an aircraft-mounted Airborne Visible Infrared Imaging Spectrometer (AVIRIS) to measure depths in Lake Tahoe. Although such remote sensing technology could be usefully applied over large areas to generate regional data sets of lake bathymetry, within certain limits of transparency and depth to estimate volumes of water bodies, it did not provide a simple means that could be easily applied by proponents and required costly development of a regional lake bathymetry set using remote sensing and interpretation. The development of an empirical approach from available regional data sets (Table 1) was therefore preferred.

• Emmerton et al. (2007) used a data set of 81 lakes in the Mackenzie River Delta and used depth measured in summer and estimated ice thicknesses to develop a regional data set of lake volumes. The method was not carried forward for analysis as the assumed average depth of 0.818 m meant that all lakes would freeze to the bottom in winter with no available water.

These methods were limited for application in northern Canada and so were not carried forward.

Water source volume estimation methods developed on the basis of operational experience were solicited from industrial proponents and regulatory bodies. In essence, all submitted methods were a variation on the same theme of attempting to make a coarse volume calculation based on known information, such as surface area, or by extrapolating relationships developed for a few lakes to apply to many. The project team used the submitted methods to estimate volumes and then compared these estimates to measured volumes from the DFO data set (Cott et al. 2005).

No under-ice volumes were included with the DFO data set so an initial screening of the alternative estimation methods was conducted by comparing the outputs of the various methods against 10% of the total lake volumes as a conservative first step. The two methods that provided accurate estimates of total water volume were then carried forward for the estimation of under-ice volumes. Additional methods submitted by two industrial operators were found to have an unacceptably high error rate and were not included in this discussion paper.

The Northwest Territories Power Corporation proposed two approaches:

• Total volume = (Area*Depth)/3, where depth = assumed maximum depth of 3 m and

• Total volume = (Area*Depth), where depth = assumed average depth of 1.5 m.

Both methods overestimated water volumes; in 10 and 4 of the 37 lakes, respectively, and the magnitude of the error exceeded that generated by other methods so these models were not carried forward.

Both models produced very large overestimates of lake volume such that the estimated takings of 10% exceeded the DFO threshold for all 37 lakes and ranged as high as 225 and 666% for the shallow and deep lake models, respectively, and so these models were not carried forward. Additional information on these can be found in HESL (2021). Models were also submitted that had been derived from the Golder Back River and Golder Nighthawk Projects and which provided more accurate estimates of volume and are discussed below.

The Back River Project included the requirement for an ice road to service a proposed gold mine located inland of the Arctic Ocean in the central Canadian Arctic (Figure 1). The ice road required water use of 675 m³/km. Golder (2018) developed bathymetric profiles and water volumes for 41 lakes from a digital elevation model (DEM) developed for 118 waterbodies and watersheds using photogrammetric interpretation of satellite imagery. The interpreted slope of the terrain surrounding each water body was used to derive the slopes entering the water bodies and extrapolated from there to estimate water depth from the blue and green satellite spectral bands which ‘allow the identification of detailed lakebed topography to a depth of 30 m’ (Golder 2018). No estimates of error or uncertainty of the bathymetry measurements estimates were provided. The error was derived by testing the lake volumes that were estimated by the model against volumes derived from the bathymetry (below).

Golder (2018) concluded that the withdrawal of 10% of the under-ice volume during winter resulted in a mean water level change of 0.183 ± 0.065 m for the 41 waterbodies that ranged in area from 2 to 93 ha (Table 1). They proposed a water level change of 12 cm or less (i.e., 0.183 − 1 SD) as a conservative threshold for protecting aquatic habitat during water withdrawals such that: ‘Allowable water taking (m³) = Lake area (m²) * 0.12 ’.

The model was applied to the DFO data set and the allowable volumes derived using the model were compared to the measured allowable volumes from the DFO data set. A target withdrawal of 10% of the total lake volume was used as a conservative test criterion of model accuracy to compare the two methods.

The Golder Nighthawk model was submitted for the water licencing process for the proposed Nighthawk Gold Indin Project located 200 km north of Yellowknife in the NT (Figure 1). The method assumed a conservative estimate of 1 m of available water depth under 2 m of ice, such that the assumed average depth was 3 m. The submission stated that, in practice, an average depth of at least 3 m would be verified in the field prior to water extraction by a minimum of three measurements taken within 500 m of the water intake. The Nighthawk model assumed that (a) a maximum of 2 m of ice would form and (b) 10% of a minimum of 1 m of available water under the ice was conservative enough to protect overwintering aquatic life. The model was therefore stated as: ‘Allowable water taking (m³) = Lake area (m²) * (10% of 1 m)’.

The Back River and Nighthawk models were variations of the same theme; the Back River model with an application of 12 cm of water level drop to the surface area (surface area × 0.12 m) or the Nighthawk model with an application of 10% of a conservative assessment of 1 m of lake volume (=surface area × 0.1 m) for lakes with an assumed ice depth of 2 m. The more protective threshold of a 10 cm water level decrease (Nighthawk model²) was chosen to increase confidence in the protection of fish habitat in the water sources. The 10 cm reduction increased the percentage of lakes protected from 81% (Back River model: 0.12 m) to 90% (Nighthawk model: 0.1 m).

Application of 10 cm of drawdown to the surface area (equivalent to 10% of 1 m of available water) was chosen as the preferred model as it maintained water losses at <10% for 90% of lakes in the DFO data set. The remaining 10% (4 of the 37 lakes) were shallow lakes with an average depth of 1.5 m or less in which the 0.1 m reduction resulted in the removal of 11–22% of the water. The Golder Nighthawk model was therefore chosen as the preferred means of estimating allowable water takings using the formula:

Allowable volume (m³) = (Surface area (m²) * 0.10 (m)).

The DFO (2010) protocol speaks to a requirement to take up to 10% of the water volume under ice. The Golder Nighthawk model was therefore tested using four independent bathymetric data sets from small lakes within the NT that included measured under-ice volumes to determine if the model derived for open water was protective under ice cover. Four data sets were obtained for which detailed bathymetry and ice cover were available (Figure 1; Table 1): the Golder Back River data set, recalculated with an assumed 2 m ice cover, the Tibbett to Contwoyto Winter Road data set, the Kennady Diamonds mineral exploration data set, and the Wilfred Laurier University data set for the Inuvik – Tuktoyaktuk Highway (ITH). Estimates of 10% of the under-ice volumes were calculated for each data set using the Golder Nighthawk model and compared to 10% of under-ice volumes from the measured bathymetry of each lake.

A means to estimate available water volumes in small ice-covered lakes is required to ensure that water withdrawal requirements for small projects will not exceed 10% of the under-ice volumes of water necessary to protect fish habitat. The estimation method is intended to be used in the water licencing process as regulatory guidance and in the field so that estimates can be made reliably with existing data. The Nighthawk model (Allowable volume (m³) = (Surface area (m²)*0.10(m)) was proposed to meet this need.

There will always be uncertainty associated with estimation methods when they are compared with measured bathymetric data but such detailed measurements are not generally feasible at the early stages of project development. We therefore recommend the alternative of an estimation along with measurements of surface area which can be easily obtained from topographic maps, satellite, or Google Earth© imagery.

Application of the estimation model (Allowable volume (m³) = (Surface area (m²)*0.10 ) based on measurement of surface area showed, however, that the protective threshold of 10% of under-ice volume could not be maintained when tested against measured bathymetry in 111 of the 159 (70%) lakes from four Arctic data sets. These 111 lakes were all less than 2.8 m in average depth and 15 were frozen to the bottom and would not support fish. There is therefore substantial risk of an overestimation of water volumes for shallow lakes in the absence of accurate estimates of average depth.

The proposed estimation model would therefore require field verification of depth prior to initiating water use from under the ice. Golder (2019) proposed that an average depth of at least 3 m would be verified in the field prior to water extraction by a minimum of three measurements taken within 500 m of the water intake. Our analysis showed, however, that fish habitat would be protected if the average depth exceeded 2.8 m. We recommend that uncertainty be reduced by confirming average depth through measurements made in at least five locations within 20 m of the proposed withdrawal site to ensure that the minimum lake depth is 2.8 m. Efforts should be made to increase the number of field measurements where feasible. Although the stipulation of at least five data points is arbitrary, we recognize the need to strike a balance between operational feasibility and habitat protection such that some level of field verification is required.

Some potential for bias is inherent in the decision of where to locate a water intake. Proximity to shore would be a consideration to minimize the amount of pipe required and the potential that it could freeze. In most cases, field crews would have little information on the characteristics of a water source although the terrain surrounding the water body would provide some indication of the bathymetric slope of the water body. If five measurements near the intake exceeded 2.8 m in depth, then there is some assurance that this was typical for the lake.

There is also some risk inherent in developing empirical predictions of water volume from current data in an environment that is changing. Zhang et al. (2019) reported that the average annual temperature in northern Canada increased by 2.3 °C between 1946 and 2016 and projected increases of 1.9 to 2.7 °C by 2050 under models of decreasing (RCP2.6) and increasing (RCP8.6) carbon emissions. These increases will increase the rate of snow melt and evapotranspiration, changing the hydrologic dynamics of northern regions. Zhang et al. (2019) also predicted that annual precipitation would increase by 8.2–11.3% under the same emissions scenarios, adding additional uncertainty to hydrologic predictions.

While the intent of this study was not to address changing hydrology under a changing climate, we recognize that the empirical relationships described herein may be altered in a warmer future. Changes in water volumes of northern lakes will alter both the depth and surface area of water bodies, and potentially the relationships between them that informed the development of the estimation model we present. It is therefore important to encourage the acquisition of more data sets of measured bathymetry to test model accuracy over time as the climate changes. Future industrial development will add to the database of bathymetry as more ice roads and mines are built and acquisition of these data sets can be required as part of the licencing process as more advanced projects are undertaken. Future data sets should be compared with, and added to, the data sets used here to confirm the model presented or revise it as warranted. Additional research is also recommended to confirm if the threshold of a 10% reduction in water volume proposed by Cott et al. (2008b) and DFO (2010) provides adequate protection of fish habitat.

Relationships between measured surface area and measured lake volumes were undertaken to develop a simple empirical model to predict how much water could be withdrawn for industrial purposes without threatening under-ice fish habitat. Guidance of ‘Allowable volume (m³) = (Surface area (m²)*0.10 (m))’ is therefore proposed to provide operators with the desired flexibility to make water use decisions in the field while being protective of fish habitat. The model was tested using 4 independent data sets of 159 lakes (Back River, Ice Road, Kennady Diamonds, and Wilfred Laurier University) and shown to be protective if the average lake depth exceeded 2.8 m. Verification of water depth by measurement at 5 points within 20 m of the water taking is therefore required to increase confidence in volume estimates and protection of aquatic habitat. The model will be implemented through the regulatory process of reviewing water licences for small-scale projects in northern Canada.

The warming climate will alter temperature, precipitation, and the hydrologic responses of the northern landscape. It is therefore important to verify and refine the proposed model through the acquisition of additional bathymetric measurements to test model accuracy as the climate changes.

The authors would like to thank staff of the Land and Water Boards of the Mackenzie Valley, specifically Ryan Fequet and Shannon Allerston, for their support and involvement in this process. Appreciation also goes out to the Government of the Northwest Territories for providing funding support. We would also like to thank Pete Cott and Bruce Hanna for being an invaluable resource on the ‘DFO Protocol for Winter Water Withdrawal from Ice Covered Water Bodies in the Northwest Territories and Nunavut’ and supporting research, as well being an integral part of the working group initiated to develop this method and for providing feedback on the manuscript. The authors would especially like to thank the NWT/Nunavut Chamber of Mines for their involvement and support in this process as well as all the organizations that assisted us with providing methods and/or data sets; Golder Associates, Northwest Territories Power Corporation, Terra X (now Gold Terra), NorEx Engineering, Aurora Geosciences and Laurier University. Finally, special thanks for staff at Hutchinson Environmental Sciences Ltd. for mapping support.

All relevant data are available from an online repository or repositories: https://mvlwb.com/resources/lwb-policies-and-guidelines.

The authors declare there is no conflict.