EXAMPLE - Frequently asked questions about the Canada BC Water Quality Monitoring Program - 48826

Last updated on August 27, 2020

Program Overview

 

What is the purpose of the Federal-Provincial Water Quality Monitoring Network?

Rivers, streams and lakes in British Columbia provide habitat for aquatic life, drinking water, irrigation for crops and recreation for everyone. The purpose of the Network is to deliver, comparable data in an open and transparent format to assess current water quality conditions and determine trends, to identify emerging threats to freshwater aquatic ecosystems, and to track the results of remedial measures and regulatory decisions.

 

What water quality parameters are monitored?

Biweekly or monthly water samples are collected from each station and analysed for a range of water quality parameters. There is a core set of parameters for every station including metals, different forms of nitrogen and phosphorus, turbidity, specific conductance, hardness, suspended sediment, colour, organic carbon, calcium, magnesium, silicon, and potassium. Field measurements are also collected for air and water temperature. Additionally, specific parameters are added based on water quality threats at a particular station.

 

Is water quality data from the Network available to the public?

Raw data can be downloaded from either the Government of Canada Open Data Portal or from the Environment Canada and Climate Change website.

 

Does the data support other projects?

Data from the Network is used to calculate the national water quality indicator, it supports Provincial State of Environment reporting and reporting associated with bilateral agreements between BC and the Governments of Yukon, Alberta and Northwest Territories.

 

Water Quality Indicators

 

What is the Water Quality Indicator?

The national water quality indicator provides a measure of the ability of river water across Canada to support plants and animals. Specific water quality data from monitoring stations are compared to water quality guidelines to create a score for an individual station (ECCC, 2020).

 

How is the indicator calculated?

The indicator is calculated using the water quality index (CCME, 2001). For each station, concentration values over a three year period for 5 to 15 parameters are compared to their guideline value and a score (1-100) is calculated. Stations are assigned a category based on that score. The frequency and amplitude by which a specific parameter does not meet its guideline value drives the score down.

 

What are the ratings and what do they mean?

Water quality stations are rated from poor to excellent. For example, water quality is considered excellent when parameters in a river almost always meet their guidelines. Conversely, water quality is rated poor when parameters usually do not meet their guidelines, sometimes by a wide margin (ECCC, 2020; CCME, 2001).

  • Excellent: water quality is protected with a virtual absence of threat or impairment; conditions very close to natural or pristine levels.
  • Good: water quality is protected with only a minor degree of threat or impairment; conditions rarely depart from natural or desirable levels.
  • Fair: water quality is usually protected but occasionally threatened or impaired; conditions sometimes depart from natural or desirable levels.
  • Marginal: water quality is frequently threatened or impaired; conditions often depart from natural or desirable levels.
  • Poor: water quality is almost always threatened or impaired; conditions usually depart from natural or desirable levels.

 

 

Water Quality Trends

 

Why evaluate changes in water quality?

Pollution from urban, industrial and agricultural areas pose a threat to water quality. To see if the quality of BC’s rivers has become better, worse or stayed the same we have assessed water quality data from the network for trends.

 

What data were included in your analyses?

Publicly available water quality data were obtained from the Government of Canada Open Data Portal. Data with a status code of V (Validated) or P (Provisional) were included in the datasets. Replicate samples collected for quality assurance/quality control were not included.

 

What parameters were included in your analyses?

We assessed river water quality using 43 parameters that characterized the physical and chemical conditions in rivers and streams. Parameters with either Approved or Working BC Water Quality Guidelines were selected for evaluation, as well as total phosphorus and nitrogen, hardness and specific conductance.

  • Nutrients (6) – ammonia, nitrate, nitrate+nitrite, nitrite, total nitrogen and total phosphorus

  • Carbon (2) – dissolved organic carbon, total organic carbon

  • Major Ions & Other Inorganics (4) – chloride, fluoride, sulphate and hardness

  • General (7) – total alkalinity, dissolved oxygen, pH, non-filterable residue, specific conductance, turbidity and water temperature

  • Metals (4 dissolved, 19 total) – aluminum* , antimony, arsenic, beryllium, boron, cadmium* , chromium, cobalt, copper* , iron* , lead, manganese, molybdenum, nickel, selenium, silver, thallium, uranium, zinc

 * indicates dissolved and total metals, otherwise only total were included

 

What is a censored value?

Censored data is the term used for measurements that are reported as less than (or greater than) a specific value. In water quality data, censored values are usually a result of instrument limitations or laboratory analytical techniques, where a parameter cannot be accurately measured below a specific concentration. This is called the method detection limit (MDL). Concentrations that fall below the MDL are reported as ‘less than’ the MDL.

 

What time period did you assess in your trend analyses?

Trends were assessed for the ten year period for all stations from January 2005 to December 2014.  There were major laboratory changes in both 2003 and 2015. Further evaluation on how this change may impact the trend analysis is required before additional periods will be available.

 

Why do your trends not extend to most recent data?

There are many processes that happen between the time a sample is collected and when the data is ready for analysis - from laboratory analysis to data validation and review. Following this, additional time is required to develop the study, analyze, interpret and review the results before they are released.

 

How did you decide which stations were suitable for trend analysis?

There are a number of reasons for which a water quality dataset may not be suitable for trend analysis. A set of criteria were set up to ensure that each dataset we used met a minimum standard for trend analysis. See the Calculating Water Quality Trends Factsheet for details. Data had to meet the following criteria to be considered suitable:

  1. The dataset must begin in 2005 and end in 2014 (i.e. data over the entire 10 period is required).

  2. The first two years and the last two years of the trend period must have data in at least 75% of the seasons, while the remaining years must have 60% of the seasons.

  3. No more than 80% of the values are censored at the highest method detection limit.

Stations that did not meet these criteria were not analysed and are reported as ‘not assessed’.

 

What trend analysis methods did you use?

This study involved analysing a large number of water quality parameters collected from many stations. We needed a method that was easy to use and that could be applied consistently to a wide variety of data from all stations. For this reason, non-parametric methods were used in this trend analysis. Non-parametric methods are less affected by characteristics typically exhibited by water quality data (such as non-normal distribution of the data, presence of outliers, censored and missing values).

The Seasonal Kendall test (Hirsch et al., 1982) was used to identify statistically significant trends, and the seasonal Sen slope estimate (Sen, 1968; Hirsch and Slack, 1984) was used to assess the magnitude of the trend. The Seasonal Kendall test is an extension of the Mann Kendall (MK) test for monotonic trends that accounts for the presence of seasonality in the data by combining the MK test computed on each of the seasons separately (Hirsch et al, 1982). This test was selected because it does not make assumptions about the distribution of the data, and allows missing values and censored data without biasing the analysis (Hirsch et al., 1982).

More details on the trend analysis method are in the Factsheet on Calculating Water Quality Trends.

 

What do the trend categories mean?

Trend results are classified as increasing, decreasing or indeterminate. This classification is based on the statistical significance and the direction of a detected trend:

  • Increasing Trend: the trend was significantly different from zero with an increasing trend

  • Indeterminate Trend: there is insufficient evidence to determine if the trend is increasing or decreasing

  • Decreasing Trend: the trend was significantly different from zero with a declining trend

A trend is classified as increasing or decreasing only when the Seasonal Kendall test for trend was found to be statistically significant (with a p-value < 0.1 considered significant). When the trend was ‘not significant’ (p >= 0.1) it was classified as indeterminate. This means that there is insufficient evidence to confidently determine if the trend is increasing or decreasing; it does not mean there is “no trend”.

 

Why is the slope not always reported?

For this analysis, a slope value was only reported when fewer than five percent of the values were above the highest method detection limit and the trend was statistically significant.
In order to calculate a slope between two values when a censored value is present, the censored data were replaced by a value of one-half of the highest method detection limit. The slope values associated with a censored value are therefore imprecise and affected by the value chosen to represent the censored observations, making the Thiel-Sen method invalid for censored data unless there are very few observations (Helsel, 2012).

 

Why can I get a significant estimated slope of zero?

The Seasonal Kendall test is a non-parametric test based on ranks rather than actual values (i.e. comparison of values are made to determine which is larger), and is therefore unaffected by censored data. This is not the case for the Sen slope estimate which relies on values in its calculation and is dependent on the value selected to represent the censored values (Helsel, 2012).

Since the calculations for significance and slope are based on two separate procedures, it is possible for the analysis to report a ‘significant trend’ (classified as increasing/decreasing) but have a slope estimate of zero. This typically occurs when many of the values in the dataset are the same. 

 

Why are results for all constituents not available for every station?

In short, because monitoring varies among stations. The Canada-BC monitoring program has a core set of variables it typically monitors for at each station; however, some stations are monitored for additional variables based on specific water quality threats.. The characteristics and completeness of the data also affect whether the data can be included in the analysis. For example, a data record was required to have at least 20 percent of values above the method detection limit for trend analysis. The concentrations of some water quality parameters are very low and do not meet this criteria. See ‘FAQ - How did you decide which stations were suitable for trend analysis?’ for more details on how the data was screened.

 

What is shown on the trend graphs for each station?

The graphs display the concentrations for a single water quality parameter over time for a ten year period (2005-2014).

The solid circles ( ● ) on the graph are concentration values above the highest laboratory detection limit. Values below the highest laboratory detection limit are open circles ( ○ ) in the graph and referred to as “value too low to detect”. If a trend is detected (i.e. statistically significant), the estimated trend over the period is shown as a straight blue line on the graph. This trend slope estimate becomes unreliable if there are many values that are below the detection limit in the dataset. Therefore, the graph will only have a trend line if a trend is statistically significant and the dataset has less than 5% of the values below the highest detection limit.

The box above the graph indicates if the trend is increasing or decreasing and shows the slope of the trend in concentration units per year. The slope of the trend is estimated using a separate analysis so it is possible for a trend to be classified as increasing/decreasing and have a slope estimate of zero. This typically occurs when many of the values in the dataset are the same. This box may also show “indeterminate.” This means that there is insufficient evidence to determine if there is an increasing or decreasing trend; it does not mean there is “no trend.”

BC ENV has developed guidelines to protect water quality for various water uses. The BC Guideline for the protection of aquatic life is presented beside the box. Although a trend is statistically significant, it may not be of environmental concern. Guidelines are used to assess potential risks to water quality and can provide a reference to help determine if an increasing trend is of environmental concern. Guideline values have been converted to match the units on the graph.

 

FAQ References

Canadian Council of Ministers of the Environment (2001) CCME Water Quality Index 1.0 User’s Manual (PDF; 84.3 kB). Retrieved on April 23, 2020. http://www.ccme.ca/files/ceqg/en/138.pdf

Environment and Climate Change Canada (ECCC), 2020. Canadian Environmental Sustainability Indicators: Water quality in Canadian Rivers. Retrieved on April 22, 2020. Available at:  https://www.canada.ca/en/environment-climate-change/services/environmental-indicators/water-quality-canadian-rivers.html

Helsel, D. R. (2012). Statistics for censored environmental data using Minitab and R: . New York: John Wiley & Sons.

Hirsch, R. M., Slack, J. R., & Smith, R. A. (1982). Techniques of trend analysis for monthly water quality data. Water Resources Research, 18(1), 107-121.doi:10.1029/WR018i001p00107

Hirsch, R. M., & Slack, J. R. (1984). A Nonparametric Trend Test for Seasonal Data With Serial Dependence. Water Resources Research, 20(6), 727-732. doi:10.1029/WR020i006p00727

Sen, P. K. (1968). Estimates of the Regression Coefficient Based on Kendall’s Tau. Journal of the American Statistical Association, 63(324), 1379-1389. doi:10.1080/01621459.1968.10480934

Contact information

For enquiries related to the data, monitoring program or study methods, please contact the water quality scientists at
ec.msqeoperationswqmoperations.ec@canada.ca