Does Lake Whatcom Qualify as an Impaired Waterbody?

Tom Pratum is a Lake Whatcom resident who is trying to help preserve the lake and put his Ph.D. in Chemistry to beneficial use.

Last month we discussed the trophic status of Lake Whatcom, and what that might tell us about its overall health. For both water quality and aesthetic values, one would want the degree of eutrophication to be as close to zero as possible -- in other words we want to have an oligotrophic lake. Both federal and state regulations are aimed at this goal. While eutrophication does occur naturally, if accelerated eutrophication is observed, it should be reversed post haste.

Eutrophication can be measured from both its cause and its effects. The causes are high nutrient levels-nitrogen and phosphorus. The effects are high turbidity (leading to low light penetration), high chlorophyll absorbance, and low dissolved oxygen (DO).

A couple of other effects which have been noted in Lake Whatcom are also related to eutrophication: detection of hydrogen sulfide (from anaerobic degradation), and detection of higher iron levels (iron is normally sequestered in the sediment in its more oxidized ferric form, but is reduced under anoxic conditions to more soluble ferrous form).

While the debate over whether the lake is prematurely aging or not is something in which we all have a part, the major players have been the state Department of Ecology (Ecology) and Whatcom County Water District 10 (WD10).

Ecology is responsible for administering the federal Clean Water Act, and has produced the data analysis of DO levels which led to the lake being placed on the 303(d) list as an impaired water body.

WD10 has multiple responsibilities, including providing water and sewer service within its boundaries, along with protecting any water resources which exist within those same boundaries. WD10 has always had a conundrum to deal with. They are saddled with the provision of service to many platted lots in the Sudden Valley area for which sewer capacity does not currently exist.

Starting in 1991, with their initial proposal for a second sewer line to provide this additional service (see Whatcom Watch February and March 2000; note that this second sewer line is currently scheduled for completion in the fall of 2002), they have argued that additional development of their service area would not harm the lake.

These arguments have been presented by their consultants Adophson and Associates (Seattle), and Envirovision (Olympia). These two entities have worked together, with the bulk of the surface water quality analysis being produced by Envirovision’s Joy Michaud.

Here we are concentrating on dissolved oxygen (DO) data, and the question of whether these data imply an increase in the rate of eutrophication of the lake. Each of the players has analyzed the very same data, for there is only one data set to analyze.

These are the data which have been acquired by the Institute for Watershed Studies (IWS) at WWU; most recently by Professor Robin Matthew’s group. However, by using slightly different methods, differing conclusions have been reached.

Figure 1 shows the DO content of the water at various depths near the center of basin 1. This shows the general trend for a lake which undergoes thermal stratification (see Figure 2). The oxygen content of the water is generated by atmospheric mixing at the surface, as well as photosynthetic respiration by phytoplankton.

The surface DO concentration drops as the water warms due to the decreasing solubility of oxygen at higher temperatures. However, at depth there is a dramatically different effect.

In the late spring the lake stratifies into a warmer upper layer known as the epilimnion and a cooler bottom layer known as the hypolimnion. These are separated by a steep temperature gradient-the thermocline. As can be seen, in basin 1 this separation occurs at approximately 10 meters in depth.

In the hypolimnion, the water is well isolated from the surface and here the DO content drops due to the occurrence of oxidative degradation which was discussed in Part I of this article. As can be seen in the plot, the oxygen content dropped to zero for nearly a month in September 2000 below about 12 meters in depth.

Low oxygen conditions are not uncommon in stratified lakes during part of the year. The question is—are these low oxygen conditions occurring more rapidly or at shallower depths as a function of time? If the low oxygen conditions are degrading from year to year, this indicates that the Washington Administrative Code (WAC) criteria of “no decrease from natural conditions” is being violated and the lake must be listed under section 303(d) of the Federal Clean Water Act.

One way to obtain a reading of the trend in DO over time is to look at the hypolimnetic oxygen deficit rate (HODR). This is the rate at which the oxygen content of the hypolimnion decreases over time.¹

The situation using recent data was initially analyzed by Adolphson and Associates in the 1997 South Shore Sewage Alternatives FEIS2 for WD10. They estimated the oxygen deficit rates for 1983, 1984 and 1990-1996. While they were unable to discern a trend, their analysis did not properly treat the oxygen content of the hypolimnion and will not be presented here.

Analysis of the situation by Greg Pelletier of Ecology was undertaken in 1998.³ In this analysis, the water-volume-weighted DO content of the hypolimnion was calculated for sampling dates in the years 1983, 1984, and 1988-1997. The years 1985-1987 have been excluded from this and subsequent analyses due to questions about their availability and/or reliability.

Volume weighting the DO makes it reflective of the total oxygen content of the water rather than just its concentration. The data for the years 1983, 1984, and 1990 and 1991 are plotted in Figure 3. As expected, the downward trend which occurs throughout the summer is similar to what is observed in Figure 1.

Pelletier obtained the HODR by taking the slope of the best fitting line 4,5 through the volume-weighted DO between June 1 and August 15-—these being dates between which the lake is known to be well stratified.⁶ When the oxygen deficit rates thus determined were plotted as a function of year, a linear correlation with a positive slope was shown. This suggested that the yearly drop-off in summer oxygen levels in Basin 1 was occurring more rapidly than it had in the past.

Recently this analysis has been extended to include data from 1998-2000 and a positive correlation has been shown to persist through the more recent data.⁷ This indicates that the rate of depletion of oxygen in the lake is increasing and thus the lake itself has departed from “natural conditions.”

This analysis lead Ecology in 1998 to add Lake Whatcom to its list of impaired water bodies. This listing has triggered the total maximum daily load (TMDL) study which Ecology is currently gearing up for.

Not surprisingly, the other primary player in this drama did not accept this analysis of the data. Envirovision, consultant for WD10, immediately disputed this analysis, and the need for the TMDL study it implied. They independently analyzed the data, but used what they refer to as the “standard method,” referring to the method used by Ecology as a “modified method.” ⁸

The differences in the results of the two methods were presented in the 1999 Entranco Report,⁹ and the Envirovision analysis has recently been extended to include data from years 1998 and 1999.¹⁰ These data are plotted along with the Ecology data in Figure 4.

The differences in the analyses are most striking for the early years (1983 and 1984), and the credibility of the positive trend (increasing rate of DO reduction) in the Ecology data is considerably weakened by the removal of these two years from their dataset.¹¹ However, it must be pointed out that the Envirovision analysis, including the years 1998 and 1999, has a positive trend which is considerably strengthened by the removal of these two years of data.¹²

While the Entranco Report stated that no reason could be ascertained for the relatively large differences in the 1983 and 1984 data, an examination of these data along with those from two years in which the Envirovision and Ecology data agree (1990 and 1991) gives a possible clue.

Rather than a nice continuous downward trend, the earlier data contains considerable scatter. The reason for this scatter is unknown, but the methods used for the measurement at that time differ considerably from those used to acquire the later data.¹³

By taking only two of these values to estimate the downward trend, Envirovision would appear to have completely ignored the scatter in the data. The more data values which are used to determine the trend, the more this scatter is taken into account.¹⁴ This would tend to vindicate the method which Pelletier used in the Ecology report. It should be pointed out that no analysis of the HODR data has shown a negative slope and this strongly hints that the increase in HODR is indeed real.

WD10, feeling strongly that Ecology had listed the lake in error, appealed to Tom Fitzsimmons (director of the Washington State Department of Ecology), and later to Governor Gary Locke. These letters have been detailed in previous issues of Whatcom Watch (August 2001, and October/November 2001) and will not be covered here.

In Ecology’s latest response to the water district, they included another presentation of DO data which indicates this parameter is indeed worsening over time.¹⁵ We have reproduced this plot in Figure 5. Rather than attempt to calculate rates, Ecology presents the volume-weighted DO for the years 1988-2000 in the months June-September. All years prior to 1988 have been excluded to remove differences due to sampling methods.

It is shown here that the volume-weighted hypolimnetic oxygen concentrations have been dropping over this time. Similar data using the raw (unweighted) DO concentrations and temperatures are presented in the Lake Whatcom Monitoring Report for the year 2000.¹⁶ The data show no significant correlation between temperature and dissolved oxygen levels. Thus the increasing rates of oxygen deficit appear to be unrelated to temperature.¹⁷

Based on this review of the data, it seems likely that Lake Whatcom has been properly listed as an impaired waterway, and the TMDL study is needed. Even if one didn’t believe the data, a quote from WD10’s own 1997 FEIS2 is worth keeping in mind: “Trends indicating a decline in water quality may not be noticeable until many years after the fact, and it may be prudent to address these issues before a definitive trend appears. Lake Whatcom represents a large volume of water with a high retention time.This means the impact from development may not be discernible for a long time, but once impact has occurred it may be much more difficult to correct or reverse the trend.”

The trend is there, let’s hope it is not too late to address it.

1 In regard to the HODR, a standard textbook on limnology, Limnological Analyses (R.G. Wetzel and G.E. Likens, 1975, pg. 317) states that “…the basic method is to calculate the total quantity of oxygen in the hypolimnion on two dates. The difference in oxygen content is then expressed as a rate of oxygen change per cm2 of hypolimnetic surface.”
2 South Shore Sewage Disposal Alternatives Final Environmental Impact Statement, Adolphson and Associates, September 1997, pp. 3-23 et seq.
3 Dissolved Oxygen in Lake Whatcom-Trend in the Depletion of Hypolimnetic Oxygen in Basin I 1983-1997, G. Pelletier, Department of Ecology report 98-313, May 1998.
4 Line fitting is a procedure whereby the line which passes through the points with the least residual error is determined. In this case, the line fitting is merely an attempt to obtain a quantitative estimate of the trend being observed. Line fits are characterized by two fitted parameters-the slope and intercept-and one parameter which determines how well the data fit to a line.
This parameter is usually referred to as the correlation coefficient R, and its squared value R2 is often stated as the goodness of fit criteria. One interpretation of R2 is that it is the fraction of total squared error which is explained by a linear model. If you fit a group of data which lie on a line to a linear model the R2 value would be its maximum of one.
In general, R2 will be lower than that, e.g. values near 0.1 would indicate a poor fit to a linear model. Another way to express the goodness of fit would be to give it in terms of a standard error for the slope and intercept. Both of these approaches are used here to express the results obtained.
5 A more recent text, Restoration and Management of Lakes and Reservoirs (G. Dennis Cooke et al, 1993, pg. 408) states that “The ODR can be determined from a plot of mean hypolimnetic oxygen concentrations over time and estimating the slope from the regression line…..Although less accurate, the rate may also be estimated from two observations of widely differing DOs…”
6 The calculated HODR is greatly dependent on the DO content at the start of the measurement. If DO is already low at the start, then, because it can’t get much lower, the measured rate will be biased low. This can occur due to early stratification, and may have lead to the low value measured by Ecology in 1998.7 My own analysis of the data appears to confirm this: in 1998 the lake appeared to be thermally stratified in basin 1 on May 5, while in 2000 it was substantially in thermal equilibrium on May 2 (note that the monthly measurement date was May 5 in 1998 and May 2 in 2000).
7 Bob Cusimano, personal communication.
8 The method described in footnote 1 is referred to by Envirovision as the “standard method,” while that employing a regression line as in footnote 5 is referred to by them as a “modified method.” The difference is that of using two points to determine a line, or determining the line from the best fit to a number of points.
9 Technical Report, Water Quality Assessment/Conditions, Lake Whatcom Stormwater Program, Entranco, Inc. November 1999 (Draft- a final version could not be located and may not exist).
10 South Shore Sewage Disposal Alternatives Supplemental Environmental Impact Statement, Adolphson and Associates, April 2001 (Draft-unpublished) pp. 2-4.
11 Analysis of the Ecology results (MacCurveFit 1.4, Kevin Raner Software, Victoria, Australia) shows that with 1983/1984 data included the slope = 8.2±2.9 (R2=0.39); without 1983/1984 data the slope = 3.4±3.4 (R2=0.08).
12 Analysis of the Envirovision results as in (10) shows that with 1983/1984 data included the slope is 4.0±2.9 (R2=0.14); without 1983/1984 data the slope is 10.0±3.4 (R2=0.42).
13 R. Matthews, personal communication.
14 This is an averaging of sorts, and comes into play in many fields of data analysis. If each measurement has some randomness associated with it, then many such measurements must be made in order to determine the most likely value for the measurement. If each measurement is made under different conditions (e.g. different dates), then, assuming the functional form of the trend in the data is known, fitting the many measurements to that trend will have a similar effect.
15 Department of Ecology, letter to Whatcom County Water District 10, October 11, 2001.
16 Lake Whatcom Monitoring Project 1999/2000 Final Report, Institute for Watershed Studies, WWU, April 2001, pp. 85-87 (DO) and pg. 89 (temperature).
17 Because the solubility of oxygen drops with temperature increase, one explanation of the observed DO values would be an increase in temperature of the hypolimnion over the years. This has not been observed to occur.