MWS-Report

3.0. Chapter 3 – Evaluation of the Ecofish Research Ltd. Cumulative Effect Indexing Tool

3.1. Summary

Ecofish Research Ltd. developed a water quality cumulative effect indexing tool for NWT CIMP. This tool was developed to help select sites where water quality monitoring would be most beneficial to understanding cumulative effects to water quality in a given region. For the next three years (2020- 2022), NWT CIMP will use the tool for a research project monitoring cumulative effects to water quality in the Upper Coppermine River Basin. The project first uses the tool to categorize lakes in the Upper Coppermine as low, medium or high cumulative effect risk by combing several remote sensing metrics. Secondly, the tool uses the Halton Iterative Partitioning (HIP) method to remotely select lakes greater than 1km (so that they can be accessed and sampled in a float or ski plane) in length throughout the basin or sub-basin to be used in the monitoring program. Eventually, data from this three year water quality sampling program of 30 lakes in the Daring Lake sub-basin will be used to fully validate the tool. The first monitoring campaign took place in July 2020. The ultimate goal is a validated cumulative effect tool that can be applied across the Northwest Territories to efficiently determine where monitoring should be directed.

Prior to the monitoring campaign and collection of new water quality data, a first evaluation of the tool can be undertaken using existing data, the same data analysed in Section 2.0 of this report. These data were collected by Environment and Natural Resources – Water Management and Monitoring Division (ENR-WMMD) for their long term water quality network. Sites used in this evaluation are Daring Lake, Desteffany Lake and Lac de Gras. Observed data from these sites can be compared against the tool’s prediction. Rocknest Lake was not used in the evaluation because it is outside of the cumulative effect indexing tool’s borders.

Results from the tool predict low cumulative effect risk for Desteffany and Daring Lake and high risk for Lac de Gras. The tool also predicts low risk in all natural disturbance metrics in all three lakes, except for a medium water area change at Lac de Gras. Most of the risk comes from anthropogenic disturbance. The observed results echo the tools predictions such that Lac de Gras shows the most consistent change in parameters over the last 20 years in parameters constant with anthropogenic disturbance. Desteffany Lake is rated low due to its high flushing rate, however, the observed data would suggest a medium risk based on its significant increasing trends. Future recommendations for the tool would be adding a permafrost slump metric to the natural disturbance category and looking into the application of the vulnerability category.

3.2. Development of the site selection tool

The cumulative effect indexing portion of the tool was developed by Ecofish using publicly available GIS data layers (Ecofish, 2020). It uses cumulative effect (CE) metrics from three categories. Natural disturbances include wetting and drying indices, greening and browning indices, and total absolute water surface area change. Anthropogenic disturbances include, distance downstream from disturbance, disturbance footprint density and road density. Finally, landscape vulnerability was determined using a flushing rate index, which is the ratio of lake surface area to contributing area.

The greening and browning index is a measure of long term rates of greening or browning using Landsat 5 imagery (Normalized Difference Vegetation Index (NDVI) trends) from 1985-2010. The wetting and drying uses the same methodology as greening and browning, however, Normalized Difference Water Index (NDWI) is used instead. Water change percentage uses Landsat data from 1985-2010 sourced from the Global Water Surface Explorer. It compares the area of the changed area against the entire area of the basin. Therefore, a high value indicates a larger disturbed area, most likely caused by a natural disturbance. The disturbance footprint measures the area of anthropogenic disturbance (ex. mine sites) in the upstream drainage area compared to the total basin area. The road density metric measures road length in the upstream drainage area and divides this by the total basin drainage area. Vulnerability was determined based on the lake size, compared to the upstream drainage area. It is based on the idea that smaller lakes will be flushed and larger lakes will take longer to expel contaminants.

**Table 8.** Cumulative effect metrics and overall risk predicted by the Ecofish CE indexing tool. The results are then rolled up into a 1 (low), 2 (medium) or 3 (high) ranking. Ranking thresholds are determined by the 33rd and 67th percentiles or where established thresholds already exist. The next step is to, roll up the metrics into one value that represents the CE risk for each lake. Three rules are used in rolling up the three categories. First, if a lake scored high in any natural or anthropogenic metric, it scores high over all. Second, if a lake scored medium in any natural or anthropogenic metric, it scores medium overall. Third, if a lake has a low vulnerability rating and is not high in both natural and anthropogenic disturbances, then the overall score is reduced by one. The predicted risk categories for Daring, Destefany and Lac de Gras from the tool are: Low (Daring), Low (Desteffany) and High (Lac de Gras) (Table 8). Natural disturbances are all low (1) for all three lakes, except for increased wetting (2) at Lac de Gras. Anthropogenic disturbances are low at Daring (1), and increased at Desteffany (2) and Lac de Gras (2). **3.3. Methodology** First, an analysis of the tool’s CE parameters was conducted for each lake. Second, the tool’s results were compared against the observed water quality data. The baseline analysis from section 2.0 was compared to the prediction of the CE tool for evaluation. The number of significant trends was used to categorize the lakes into low, medium or high cumulative effect risk. A similar statistical threshold designation to the tool was used: <33% parameters with significant trends is low CE, 33.3%-66.6% is medium CE, and >66.6% is high CE (Table 9). For example, if >66.6% of parameters showed long-term significant changes over the lake’s history, it is considered to have been impacted to a high degree from disturbances. The lake categories based on the observed water quality were then compared to the categories predicted by the tool. **3.4. Results** Predicted natural disturbances are low (1) for all three lakes, except for increased wetting (2) at Lac de Gras. It can therefore be inferred, that based on the tool’s parameters, the tool predicts there has been little natural disturbance owing to climate change at these three lakes. Further teasing out natural disturbance should involve more detailed metrics such as slumping and surficial geology, since they were outlined as important earlier in this report (section 1.0). Anthropogenic disturbances are low at Daring Lake (1), and increased at Desteffany Lake(2) and Lac de Gras (2). The difference is most likely owing to the two active diamond mines in the Lac de Gras sub-basin and the accompanying road networks (Figure 10). Currently, very few anthropogenic disturbances are present in the Yamba sub-basin, however, the proposed all season Slave Geological Province Road to Resources may increase Daring Lake’s anthropogenic metric in the future. The lake vulnerability is low at Daring (1) and Desteffany (1) and high at Lac de Gras (3). Differences in lake vulnerability are a result of lake size and lake contributing area. The large lake area of Lac de Gras (568 km2) and relatively small contributing area indicates that the storage capacity is likely large and flushing rate small, perhaps resulting in a storage of nutrients and other water constituents. The observed CE categories are similar to the tool’s predicted categories (Table 9). Assuming that significant changes over time in water quality parameters are indicators of cumulative effects, Daring is low and Desteffany and Lac de Gras are medium. It should be noted also that the majority of annual median trends at Daring were found to be insignificant when analyzed for linear regression and ultimately further supports a lack of cumulative effect evidence at Daring. Exceedance percentages are highest at Daring most likely because the baseline is ambient and includes temporal trends. Therefore, the IQR is smaller at Daring and exceedances are more likely. It is expected that future monitoring will begin to show higher exceedance percentages in Desteffany and Lac de Gras, and less at Daring. In the future, observed exceedances and trends can be used to calibrate the tool’s predictions. **Figure 10.** Observed vs. predicted cumulative effects to water quality at Daring Lake, Desteffany Lake and Lac de Gras. Observed symbols are reflective of the actual sampling location. Furthermore, similar to the tool, the most significant increasing trends in parameters reflect anthropogenic activity in the basin (strontium and conductivity) such as diamond mining and road building. Strontium increased by 223% at Lac de Gras and 103% at Desteffany. Conductivity increased by 94% at Lac de Gras and 45% at Desteffany. **Table 9.** Percent of significant trending parameters at Daring Lake, Desteffany Lake and Lac de Gras from section 2.0. **3.5. Discussion and Conclusion** The results from the cumulative effect index tool developed by Ecofish Research Ltd. produces similar results to the analysis of nearly 20 years of water quality data at Daring Lake, Desteffany Lake and Lac de Gras. Lac de Gras and Desteffany show higher cumulative effects, mainly contributed by anthropogenic disturbance in their sub-basin. The actual data supports the tool’s interpretation of Desteffany Lake and Lac de Gras with a high number significant temporal trends and significant increases in anthropogenic related parameters, such as strontium and conductance. Daring Lake is predicted to have a low cumulative effect risk and its baseline data supports this with few temporal trends and a lack of changing anthropogenic related parameters. As outlined in Ecofish Research Ltd’s report, future work with the tool should include layers that address permafrost thaw, slumping, and forest fires in the natural disturbance section and land cover and surficial geology in the landscape vulnerability section (Ecofish et al., 2020). Considering that section 1.0 of this report highlighted the significant effect geology, slumping, and forest fires can have on water quality in the north, the tool should especially consider these. For example, the water quality data showed significant increase in sodium (an indication of slumping) at Lac de Gras (419%) and Desteffany Lake (174%) but is not directly accounted for by any of the tool’s layers. A more robust surficial analysis by the tool may be helpful in teasing out sources of existing water quality disturbances. Overall, the preliminary analysis of the cumulative effect index tool based on existing water quality data shows agreement with observed data and potential as an effective predictive tool. Moving forward, the three year sampling program in the Daring Lake sub-basin will help perfect this tool further. [Index](/MWS-Report/)