1SM = Standard methods for the examination of water and wastewater. Eighteenth edition (1992). American Public Health Association, American Water Works Association, and Water Environment Federation. Washington, D.C.

EPA = Methods for the chemical analysis of water and wastes. Environmental Monitoring Supply Laboratory. U.S. Environmental Protection Agency. Cincinnati, OH. EPA-600/4-74-020. 1983.

NCASI = A procedure for the estimation of ultimate oxygen demand (biochemical). National Council of the Paper Industry for Air and Stream Improvement, Inc. Special Report No. 87-06. May 6, 1987.

dry season surveys monitored the mainstem and forks of the Skagit River, as well as the variety of loading sources to the river in the study area. Sources include municipal wastewater treatment plants (WWTPs), combined sewer overflows (CSOs), urban stormwater, drainage district pump stations, and tributary creeks.

Conditions during surveys varied widely. The Skagit River during the December 27-28 survey were near flood level, and was also very high during the February 21-22 survey. Some surveys occurred during rainfall or after recent rains, while others took place following short or long periods of dry weather. The effect of these conditions on the data will be analyzed in detail in the final report.

Attachment A gives a detailed summary of the data QA/QC analysis. Data collected in the Lower Skagit TMDL study are usable subject to certain qualifications:

• Qualifiers are provided in the data tables that place conditions on the laboratory data.
• Data variability must be taken into consideration when interpreting results and applying data to other analyses.
• Data from certain locations may have been taken under unusual conditions. Some of these conditions are described below or in Attachment A. The unique circumstances of sampling must be taken into consideration in evaluating the data.

Data from the Skagit River and its tributary creeks (Hansen, Nookachamps, and Carpenter/Fisher Creeks) and some storm drain and pump station sources (Northern State Hospital Drain, Tributary at Riverfront Park, Brickyard Creek, and Gages Slough) were taken directly from flowing surface waters and should be representative of their quality. One exception is that during the December 1994 survey, the Skagit River was flooding and forcing water upstream in Nookachamps Creek, so it is uncertain whether this sample represents the river, the creek, or some mixture of each.

Data from certain storm drain and pump station sources (South Sedro-Woolley Storm Drain, Frontage Road Pump Station [Kulshan Creek], Division Street and Park Street CSOs, Britt Slough Pump Station) were collected some of the time from water flowing directly into the Skagit River, and at other times from water that was backed up, or flowing into or held in a wet well. These conditions were mostly controlled by the height of the river. Other pump stations (Freeway Drive, Westside, Conway, and Rexville) were monitored solely from wet wells or drainage basins, and their representativeness as inputs to the river would depend on whether inflows to the pump stations were causing active pumping to the river. This report does not evaluate whether these sources could have been actively discharging at or near the time of sampling. Prior to use of these data in a detailed analysis, conditions at these locations must be evaluated to determine whether the observed quality is representative of an actual discharge.

Tables with the data collected during the Lower Skagit TMDL Study are provided in Attachment B. The tables are organized as follows:

Some significant features of the data worth noting are:

• No Skagit River DO measurement fell below 9 mg/L. This is not consistent with the results of the 1992 survey reported in Entranco (1993), where low DO was found in the Skagit River at several locations. (Data from that report was determined to be of poor quality.)
• Hourly DO data collected by the Hydrolab® Datasonde 3® (DS3) found a curious drop in DO of about 1 mg/L at the higher high tide during the dry season surveys. This pattern can be seen in Figure 2. This phenomenon is associated with stagnant conditions during a high neap tide, and is consistent with similar conditions found in the Snohomish River (Cusimano, 1995).
• During the summer surveys, the South Fork Skagit River showed slightly higher levels of ammonia nitrogen and bacteria as compared to the North Fork and upstream mainstem stations. The reason for this difference is unknown, but it may bear some relation to the dip in DO observed. The final report will include a more detailed analysis of these observations.
• During the September 1995 survey, Kulshan Creek (at the Frontage Road Pump Station) showed unusually high levels of ammonia, nitrate/nitrite, and total persulfate nitrogen; elevated FC bacteria; and low DO compared to other surface waters.
• FC bacteria levels in the CSOs were extremely high, exceeding 100,000 cfu/100mL during several surveys.

The data from the study were analyzed for compliance with the Water Quality Standards. Data were also compared to NPDES permit limitations for the WWTPs. (Comparison to permit limitations is done for informational purposes; values above the permit limitations may not necessarily be permit violations.) The Class AA Standards were applied to the Skagit River above Sedro-Woolley, and Class A standards were applied to all other surface waters. The results of that analysis are presented in the following tables in Attachment C:

The focus of the Lower Skagit TMDL is on compliance with the state Water Quality Standards for fecal coliform (FC) bacteria and dissolved oxygen (DO). The final report will provide an analysis of pollutant loading and compliance with the FC and DO standards under both current and future conditions. However, a simple review of the data reveals some significant patterns.

One goal of the September 1994 reconnaissance survey was to determine whether saline water from Skagit Bay reached the lower boundaries of the study area. Although the river flow is halted, and perhaps slight upstream flow occurs for a short time, no noticeable increase in salinity was found. Therefore, freshwater criteria were applied throughout the study area. Possible impacts of Skagit River bacteria loading on Skagit Bay marine waters will be evaluated in the final report.

To evaluate compliance with the Water Quality Standards and for comparison to permit limits for FC, a "daily" or "weekly" geometric mean was calculated from the data for each survey. A "monthly" geometric mean was calculated from the means of three consecutive wet season surveys (a four-week averaging period) and the two dry season surveys, for a total of seven averaging periods. The monthly mean was compared to either the Class A or AA geometric mean criteria or the monthly average permit limits. The maximum daily or weekly geometric mean from each averaging period was compared to the Class A or AA 10th percentile criteria or the weekly average permit limitation.

Table C.1 highlights in bold values that exceed the FC criteria or permit limits. The following results are of interest:

• The Skagit River fell below the FC criteria, except during the December 1994 flood conditions, when the Class AA criteria were exceeded at the upstream end of the study area.
• FC levels in the Burlington and Big Lake WWTPs fell below their permit limitations for bacteria at all times. FC levels at the Sedro-Woolley WWTP were above the monthly average permit limitations during two averaging periods - late February through late March 1995, and early March through early April 1995; and exceeded the weekly average permit limitations during the late March and early April surveys. (Note that samples at the Sedro-Woolley WWTP were taken from the manhole nearest the outfall, which is not the sampling location designated in the NPDES permit.) At the Mount Vernon WWTP, FC levels were above the monthly average permit limit during the late September through early October 1995 averaging period; and exceeded the weekly maximum permit limit during both the late September 1995 and early October 1995 surveys.
• Surface waters that exceeded the FC criteria included Hansen Creek, Tributary at Riverfront Park, South Sedro-Woolley Storm Drain, Brickyard Creek, Nookachamps Creek, Kulshan Creek (Frontage Road Pump Station), and the Rexville Pump Station.

To evaluate turbidity standards, values at mainstem and tributary survey stations were compared to "background" levels. For this analysis, background for the Skagit River stations was the upstream sampling site near Sedro-Woolley; and for tributary stations the sources were categorized and the lowest value taken as background for comparison with other stations in each category. The standards were evaluated for each survey.

Table C.2 displays a "NO" for values that exceed the turbidity criteria. The following results are of interest:

• The Skagit River did not exceed turbidity standards during the study. However, the station above Sedro-Woolley exceeded 200 NTU during the December 1995 survey, and exceeded 70 NTU during the late February survey, both high water conditions.
• Surface waters that exceeded turbidity standards more than once during the study included: Hansen, Nookachamps, and Brickyard Creeks; The Britt Slough, Conway, and Rexville Pump Stations; the Tributary at Riverfront Park; and the South Sedro-Woolley Storm Drain.

All surface waters were in compliance with the Water Quality Standards for DO, temperature, and pH, except Kulshan Creek where one measurement fell below the Class A DO criterion. BOD₅ levels fell below permit limits for all WWTPs.

Several conclusions and recommendations result from this preliminary examination of the Lower Skagit TMDL data:

• FC levels at the Sedro-Woolley and Mount Vernon WWTPs were above their permit limits. If these values differ from the permittee's effluent monitoring results, the reason for the inconsistency should be investigated.
• Problems with FC bacteria and turbidity are widespread, and appear to be largely associated with nonpoint sources. FC bacteria will be evaluated in more detail in the Lower Skagit TMDL final report.
• High turbidity and violations of Class AA FC bacteria standards were found in the Skagit River above Sedro-Woolley. Further investigation of water quality problems in the Upper Skagit River should be considered as part of the next Watershed Needs Assessment
• High bacteria levels in CSOs point to the need to continue the program to correct those discharges.
• Results point to the South Fork Skagit River as a critical location for DO in the river. The final report will investigate current and future DO conditions, and evaluate the loading capacity and possible allocation alternatives for ammonia nitrogen and BOD₅.

References

All field monitoring meters were calibrated according to manufacturer's specifications. Temperature thermistors in the meters are factory calibrated. All meters were calibrated for DO and pH at least weekly, and for specific conductance at least monthly.

All meters were checked against standards before, during, or after each survey. Meter temperatures were spot-checked with a Standard mercury thermometer, and no readings fell outside the target accuracy of 0.2 ^oC specified in the QAPP. When paired readings of DS3 and Surveyor 2 (S2) meters were evaluated, the root mean square standard deviation (RMS-SD) of the pairs was 0.16 ^oC.

A problem that was encountered while taking pH measurements was that readings from the Skagit River tended to drift upwards if the probe remained in the water. The probes in the S2 and Orion® meters were not designed for the low ionic strength of the Skagit River's water, and the migration of ions across the surface of the probes caused a gradual increasing error. Grab pH measurements in the Skagit River should probably be considered overestimates, especially for multiple sequential readings.

To measure the variability (bias and precision) of field data, sequential duplicate samples were taken of laboratory parameters, and field verification samples were taken for comparison with field measurements. To compare paired values, the standard deviation (SD) and coefficient of variation (CV, sometimes called the percent relative standard deviation, or %RSD) were calculated for each pair, and the root mean square calculated for groups of data. In general, the RMS-SD gives a measure of the magnitude of differences, while the RMS-CV gives a percent difference. In interpreting QA/QC results it is important to note that a pair of values that are low in magnitude with a low SD may have a high CV, while a pair of values high in magnitude with a high SD may have a low CV.

DO measurements using the Winkler method sampled as replicate pairs had an overall RMS-SD of 0.17 mg/L. By survey, only the October 1995 survey met the QAPP target accuracy for the Winkler method of 0.1 mg/L, and the poorest accuracy was 0.22 during the September 1995 survey. Although exceeding target accuracy at times, this data will be used with consideration given to the variability.

DO meter readings were paired with Winkler field verification measurements to evaluate the accuracy of the meters. For each DS3 meter during each survey, the average residual ranged from -1.18 to 0.30 mg/L and the RMS-SD ranged from 0.08 to 0.86 mg/L. For S2 and YSI® meter readings for each survey, the average residual ranged from -0.39 to 0.35 mg/L, and the RMS-SD ranged from 0.05 to 0.28 mg/L. Comparing paired S2 and DS3 readings, the average residual and RMS-SD were -0.27 and 0.42 mg/L, respectively. Using a t-test at the 0.05 confidence level, the S2 and DS3 DO readings were significantly different.

To improve the compatibility of DO data, all DO results from meters were adjusted by the average residual between the Winkler and meter results for each meter during each survey. The RMS-SD for adjusted DS3 data improved to 0.12 mg/L overall, and no worse than 0.22 mg/L for any meter. The RMS-SD for adjusted S2 and YSI® data improved to 0.13 mg/L overall, and no meter exceeded 0.18 mg/L. The RMS-SD between adjusted DS3 and S2 readings improved to 0.17 mg/L, and the t-test detected no significant difference between adjusted data at the 0.05 confidence level. Because of this adjustment, DO data from different meters are considered comparable for use in analysis, with consideration given to the estimated variability.

All laboratory analyses were performed within specified holding times. Data reported with qualifiers should be used with caution. All other data reported by Ecology's Manchester Laboratory may be used without qualification.

The data quality objectives for laboratory results were for no parameters to exceed 20% RMS-CV, except for results near the reporting limit and FC bacteria results which should not exceed 50% RMS-CV. All parameters met this objective, except for FC which slightly exceeded the objective with a RMS-CV of 51.5%.

Due to the depth of the manhole used for FC samples at the Sedro-Woolley WWTP, some FC samples were taken from a bucket grab, while other were taken from a bottle lowered in a holder. To check the comparability of these two sampling methods sequential duplicates were taken using both methods during the September 1995 survey. The means of duplicates for the two methods were identical.

For ultimate BOD analyses, all water blanks and sugar standards met QA/QC requirements. Water+nutrients blanks showed increased BOD over water blanks, which deviates from QA/QC requirements, but probably did not detract significantly from the accuracy of results. Some WWTP samples went anoxic during the analysis, but examination of the time series does not suggest that results were affected. Ultimate BOD results are deemed acceptable for use.