Source of information: http://www.ecy.wa.gov/biblio/0203052.html
The Water Quality Index (WQI) presented here is a unitless number ranging from 1 to 100. A higher number is indicative of better water quality. For temperature, pH, fecal coliform bacteria and dissolved oxygen, the index expresses results relative to levels required to maintain uses according to criteria specified in WAC 173-201A. For nutrient and sediment measures, where standards are not specific, results are expressed relative to expected conditions in a given Ecoregion. Multiple constituents are combined and results aggregated over time to produce a single score for each sample station. In general, stations scoring 80 and above met expectations for water quality and are of "lowest concern," scores 40 to 80 indicate "marginal concern," and water quality at stations with scores below 40 did not meet expectations and are of "highest concern." A spreadsheet-version for calculating the WQI is available from the author.
Monthly WQI scores are suitable for statistical trend analysis. Prior to adjusting for flow, statistically significant (p < 0.05) improving trends in overall (aggregated constituents) WQI scores were indicated at four stations and declining trends at one station out of 62 evaluated.
There are three parts to calculating the index:
1. Convert each result to an index score ranging from 1 to 100.
Every result in the selected date range is converted to an index score by a quadratic equation (coefficients are listed in Appendix A). The particular formulas used for a particular station depended on the stream class or ecoregion for that station. For temperature, oxygen, pH, and fecal bacteria, formulas were scaled to yield a score of 80 for results at the water quality criterion for that constituent. The geometric mean criterion was used for fecal coliform bacteria. For example, a temperature of 18 °C in a Class A stream would yield an index score of approximately 80. For nutrient and sediment constituents, formulas were designed so that about 20 percent of the data from long-term stations would convert to index scores below 80. (See “Converting Raw Data to WQI Scores,” below, for more detail.)
2. Aggregate index scores.
WQI analyses including multiple years can be aggregated into a single score. A score for each measured water quality constituent for each month is determined as the mean of all scores for that constituent and that month (e.g., all Januaries are averaged). However, I have chosen to present annual scores individually to avoid confusion when interpreting scores from stations where data were collected during different years. The WQIs for the different constituents are then aggregated for each month by calculating a simple average and subtracting a penalty factor for monthly scores less than 80. The penalty factor is (85-WQI Score)/2. (For example, if the average WQI score in January was 89 and pH, at 75, was the only constituent below 80 , the penalty factor for pH would be (85-75) / 2 = 5 and the overall average score for that month would be 89-5=84.) The penalty factor approach is used to weight low-scoring (poor water quality) constituents more heavily and thus reduce the likelihood of one low-scoring constituent—which could have severe affects on the ecosystem—being masked by the averaging process. (Oregon uses a square harmonic mean to weight low-scoring constituents (Cude, 2001).) The overall WQI for a station is the average of the three lowest-scoring months.
A WQI is also determined for each evaluated water quality constituent. For fecal coliform bacteria and sediment and nutrient measures, the constituent score is the average of the three lowest scores for that constituent. For temperature, pH, and dissolved oxygen, the constituent score is the minimum monthly score. Unlike other measures, the distribution of these last three constituents is not particularly patchy. A single high temperature measurement is better correlated with the average 7-day minimum than is the average of three monthly grab samples.
Note, however, that this procedure applies only to constituent scores, not to the overall score.
3. Apply weightings and other miscellaneous rules.
Some adjustments were made to moderate low scores that could be attributed to naturally occurring influences. The following rules are applied:
a) A harmonic mean is used to combine turbidity and suspended solids. This prevents doubleweighting these strongly correlated constituents.
b) The score for the limiting nutrient is used for total phosphorus and total nitrogen. This prevents double weighting of a nutrient index.
c) A maximum penalty (20) is set for nutrient and sediment scores below 80 because these scores are based on distribution of historical data and not on environmental impact or beneficial use support. Setting a maximum penalty helps prevent nutrient and sediment scores from overwhelming the overall index.
I considered an adjustment to reduce pH scores in eastern Washington, where pH is typically a half unit higher than in western Washington (Table 1), probably due, at least in part, to geological differences. However, the pH curves are not very restrictive anyway (a score of 60 requires a pH measurement of 9.1). Instead, I elected to discuss this and other potential natural influences on scores in a narrative accompaniment to the numerical WQI.