In response to suggestions by EPA and DEP, MWRA has, on an interim basis, modified some of the thresholds in the Contingency Plan. MWRA made these interim modifications on July 1, 2001, and they became effective thirty days later as provided in the NPDES permit (Part I.8.d). MWRA requests OMSAP review of these modifications prior to incorporating them in the annual request for Contingency Plan modifications, which will be submitted by November 15, 2001, pursuant to Part I.8.c of the NPDES permit.

Water column Nearfield Bottom and Stellwagen Basin Bottom Dissolved Oxygen

In December 1997, the Outfall Monitoring Task Force recommended that the threshold for bottom dissolved oxygen saturation be deleted from the Contingency Plan, because dissolved oxygen saturation frequently fell below the thresholds in the baseline period, both in the Nearfield and in Stellwagen Basin. (This recommendation was also discussed at the November 2000 OMSAP meeting.) EPA and DEP declined MWRA’s request to allow deletion of the threshold, but suggested in a letter dated April 3, 2001, that the phrase "unless background conditions are lower" be added to the threshold descriptions for both dissolved oxygen saturation and dissolved oxygen concentration. This makes the thresholds more consistent with the state water quality standard, which was the intended basis for the dissolved oxygen concentration and saturation thresholds. MWRA has calculated background conditions as suggested by EPA and DEP; the results and methodology for the calculations are described below.

Calculation of Dissolved Oxygen Background and Survey Means

This section summarizes how survey means are calculated, how background values are calculated, and how the threshold comparison is performed. Contingency plan threshold comparisons for both the nearfield and Stellwagen Basin for bottom-water dissolved oxygen (mg/L) and percent oxygen saturation (%) are performed for each survey occurring from June through October. The mean survey values are compared to the threshold and background values to determine if there is an exceedance. There is no threshold exceedance unless the survey mean is below both the threshold value (caution or warning) and the background value. The background values are calculated from data gathered during the baseline period (summers of 1992 through 1999). Table 1 below shows the caution, warning, and background values for each parameter and location.

Parameter	Location	Caution	Warning	Background
Dissolved Oxygen (mg/L)	Nearfield	6.5	6.0	5.75
	Stellwagen Basin	6.5	6.0	6.2
Percent Oxygen Saturation (%)	Nearfield	80	75	64.3
	Stellwagen Basin	80	75	66.3

The calculation uses dissolved oxygen and percent saturation sensor readings from the CTD upcast, bottom depths only. The sensor data are calibrated to bottle measurements taken during the same survey. Data qualified as suspect/invalid, under investigation, and missing are not used. Non-detects are treated as zeros if they occur. (There are currently no non-detects in the database.) For the Nearfield thresholds, data from all nearfield stations sampled on nearfield surveys are used. For Stellwagen Basin, data from stations F12, F17, F19, and F22 during all surveys are used.

Data are aggregated as follows:

1. Calculate the mean bottom dissolved oxygen and percent saturation for each station within a survey.
2. Calculate the survey mean by averaging station means during the survey.

1. Find the lowest survey mean for each of the eight years (1992 through 1999) for all months.
2. Fit a normal distribution to the eight annual minimum survey means. (The four sets of means, bottom dissolved oxygen and percent saturation at the nearfield and Stellwagen Basin, are all normally distributed.)
3. The "Background" value is the 5^th percentile of this fitted distribution, calculated as
   
   5^th percentile = mean - 1.645*Standard deviation.

Threshold testing:

Each survey mean is compared to the threshold value (caution or warning) and to the background value in Table 1. If it is below the background value and the threshold value then there is an exceedance. If the mean is not below both values then there is no exceedance.

The graphs below show the fluctuations of bottom dissolved oxygen concentration and percent saturation since the start of monitoring in 1992, along with the threshold and background values.

Figure 1. Dissolved oxygen concentration (mg/L) graphs are in the top row and DO saturation graphs are in the bottom row. (Click on charts below to view enlarged image).

Dissolved oxygen percent saturation fell below the caution threshold in October 2000, in the nearfield and in Stellwagen Basin. Neither percent saturation nor dissolved oxygen concentration have fallen below the background levels since outfall start-up in September 2000.

OMSAP recommended in a July, 2000 letter to EPA and DEP that the threshold for Alexandrium cell counts be deleted from the Contingency Plan, because the MWRA’s Outfall Monitoring sampling design is not optimized to sample the patchy, variable occurrence of Alexandrium, and because a better, more integrated measure of paralytic shellfish toxicity using long-term shellfish monitoring is under development. EPA and DEP declined to allow deletion of the threshold, but suggested in a letter dated April 3, 2001, that an alternative threshold of 100 cells/liter in any sample could be substituted, in recognition of the fact that the maximum observed pre-discharge cell count was 163 cells/liter (that value was measured in April 1993).

This information briefing describes the method used to compare Alexandrium cell counts to the threshold of 100 cells/l. It also describes how the seasonal mean would be compared to the 95^th percentile of the baseline seasonal means. EPA and DEP have requested that both approaches be presented for OMSAP review.

The 100 cells/liter has several advantages over the former, 95^th percentile approach. It has a much shorter turn-around – preliminary data can be available in a few weeks rather than waiting for the all the data for a season to be available. It is sufficiently protective – PSP toxicity in shellfish is generally not observed until ambient cell counts reach more than 300 cells/liter (Don Anderson, pers. comm.). Nevertheless, it is unlikely to be exceeded except in an Alexandrium bloom year.

This section summarizes the methods used to calculate the Alexandrium cell counts. For each post-discharge survey, the Alexandrium cell count for each sample is compared against the caution threshold of 100 cells/liter.

Cell counts from 20m screened water samples are used. Two Alexandrium taxa are included: Alexandrium tamarense and Alexandrium spp. Data from nearfield stations, all depths (usually two), during all surveys are used. Data qualified as suspect/invalid are not used.

Data are aggregated into sample average counts as follows:

1. For each taxon, subsample averages are generated by averaging analytical replicates.
2. Sample averages are generated by averaging the bottle averages.

Threshold Testing:

For each post-discharge survey, the Alexandrium cell count for each nearfield sample is compared against the caution threshold of 100 cells/liter. If any sample average exceeds the threshold, there is an exceedance for that survey.

The figure below shows the sample counts of Alexandrium compared to the 100 cells/liter threshold, for all stations and for the nearfield only.

ALL STATIONS	NEARFIELD
Final plankton data from MWRA's outfall discharge ambient monitoring are available through the end of June, 2001. Alexandrium was observed in a few samples during spring and early summer of 2001, with the highest counts (35 cells/liter) found in a June nearfield sample.

This section describes how a 95^th percentile seasonal mean for Alexandrium could be calculated. Because there are several seasons where Alexandrium was not detected, the usual method of calculating the 95^th percentile cannot be used. An alternative "frequency adjustment" method outlined by Andrew Solow (Solow, pers. comm. 2000) was used, as described below. The resulting 95^th percentile values are shown in the table below.

Cell counts from 20 micrometer screened water samples are used. Two Alexandrium taxa are included in the threshold: Alexandrium tamarense and Alexandrium spp. Data from all depths (usually two), during all surveys are used in calculating the baseline values. Data qualified as suspect/invalid are not used.

Data from nearfield and farfield stations, all depths, during all surveys are used in calculating the baseline values. Farfield stations are included because there is no difference between the nearfield and farfield, so all the available data are used. However, for comparing to thresholds, post-discharge data from only nearfield stations would be used, as changes from baseline in the nearfield are potentially outfall-related.

Data are aggregated into sample average counts as follows:

1. For each taxon, subsample averages are generated by averaging analytical replicates.
2. Sample averages are generated by averaging the subsample averages.
3. If more than one sample is taken at a depth on the same day, the sample averages for each station, day and depth are averaged.
4. Sample averages for the two taxa are added together, resulting in a total concentration for Alexandrium for each station, day and depth.
5. For each station and date, the average value over all depths (usually two) is computed.
6. Seasonal averages are calculated as the average of the daily station averages for the following periods:

Calculation of 95^th percentile:
There were several seasons during the eight or nine baseline years when Alexandrium was not detected. Therefore, the set of baseline means does not fit a normal or log-normal distribution, and the usual method of calculating the 95^th percentile (mean plus 1.645 times the standard deviation) can not be used. An alternative method that adjust for the frequency of zero years (Solow, personal communication 2000) was used, as follows:

To date we have received data for autumn 2000 and winter/spring 2001 post-discharge seasons. The autumn 2000 nearfield seasonal mean (0.060 cells/l) falls below the 95^th percentile of the baseline autumn means. The winter/spring 2001 nearfield seasonal mean (0.67308 cells/l) falls above the 95^th percentile of the baseline winter/spring means. The winter/spring seasonal mean is the average of one non-zero observation in April 2001 (17.5 cells/l) and several samples without Alexandrium.

MWRA. 2001. Massachusetts Water Resources Authority Contingency Plan Revision 1. Boston: Massachusetts Water Resources Authority. Report ENQUAD ms-071. 47 p.

Outfall Monitoring Science Advisory Panel. July 2000 letter to EPA and DEP

http://www.epa.gov/region01/omsap/omsapletter.html

Go to Contingency Plan page