Sampling Locations
The OSM program is a comprehensive program, with snowpack measurements and water quality and quantity assessments. High frequency (15 minute sampling) water quality sondes measuring electrical conductivity, pH, dissolved oxygen, and turbidity are deployed alongside nine Water Survey of Canada (WSC) gauging stations detailed in the map below. Grab samples are currently taken at the WSC stations as well as the other marked locations on the following schedule:
- 3/wk mid-April to mid-May
- 2/wk mid-May to mid-June
- 1/wk mid-June to mid-August
- 1/2wks mid-August to mid-September
- Other times pending sampling conditions and river safety (See Data page for sampling breakdown)
Figure 4. Sampling locations within the AOSR and associated WSC hydrometric stations. These stations are collocated with the grab samples collected for this study. Station numbers here match those in the results section.
Sampling Methodology
Grab samples were collected from the average depth of the thalweg in these wadable rivers. Samples were stored at 4 degrees celsius and processed within appropriate holdings times. Particulate PAH concentrations were recorded using these samples.
Figure 5. Decision tree to determine ongoing monitoring locations under the OSM Program.
Method of Analysis
Measurement of PAHs in water is difficult to capture accurately. PAHs are hydrophobic, and adsorb onto sediments and suspended solids entrained in the currents. The adsorbing nature of PAHs leads to errors in measuring the "true" PAH concentration in a given water sample. A shift in sampling from dissolved and sediment PAHs to only suspended sediment analysis occurred in 2018.
Classifying the concentration of PAHs adsorbed to sediments is also difficult. This is because PAHs must first be released from the sediments to be accurately measured. As some PAHs will remain on the sediments regardless of removal processes there is a natural error in each measurement. To classify this error a surrogate recovery assessment was performed. This involves spiking a sample with a known concentration and conducting the test to determine how much of the additive is extracted successfully by the process. This efficiency is reported as a percentage and ranges within our data set from 5-120% but can be as high as 140%. Low surrogate recoveries indicate that measured concentrations are conservative values. Due to the potential for added error, no correction was applied to the measured samples. There was no observed bias between surrogate recovery and sample concentration (e.g., low surrogate recovery did not bias low sample concentrations). Varying surrogate recovery should not bias statistical analysis however extremely low recovery samples were removed to avoid analysis impacts on fractionation ratios.
Classifying the concentration of PAHs adsorbed to sediments is also difficult. This is because PAHs must first be released from the sediments to be accurately measured. As some PAHs will remain on the sediments regardless of removal processes there is a natural error in each measurement. To classify this error a surrogate recovery assessment was performed. This involves spiking a sample with a known concentration and conducting the test to determine how much of the additive is extracted successfully by the process. This efficiency is reported as a percentage and ranges within our data set from 5-120% but can be as high as 140%. Low surrogate recoveries indicate that measured concentrations are conservative values. Due to the potential for added error, no correction was applied to the measured samples. There was no observed bias between surrogate recovery and sample concentration (e.g., low surrogate recovery did not bias low sample concentrations). Varying surrogate recovery should not bias statistical analysis however extremely low recovery samples were removed to avoid analysis impacts on fractionation ratios.
Statistical ANalysis
There was 76 PAH compounds measured as part of the routine grab samples between 2015-2021. Following the initial assessment a subset of the 13 PAH group sums was selected. This produced a similar ordination as the full dataset. Data was normalized to evaluate relative differences between PAH groupings. Euclidean distances were calculated for the dataset as Bray-Curtis and Mahalanobis distances failed to provide clearer insights. Principal coordinate analysis was completed on these values. This was done to reduce the chemical signature of each sampling point into relative fractions. This allowed us to view and understand the non-parametric data in simpler, Euclidean space. Using groupings we looked at large scale spatial differences, differences in reach chemistry on individual rivers, and temporal differences as a whole.
This analysis has been completed as part of the University of Alberta's Renewable Resources 690 - Multivariate Statistics For Environmental Sciences course. The raw data is publicly available from Alberta Environment and Parks. Analysis is preliminary and is not intended to inform management or policy directions. Neither the University of Alberta or Brandon Hill accept any risk of liability with the use or misuse of the analysis provided.