The Water-Quality Monitoring Program for the Baltimore Reservoir System, 1981–2007—Description, Review and Evaluation, and Framework Integration for Enhanced Monitoring

Abstract

The City of Baltimore, Maryland, and parts of five surrounding counties obtain their water from Loch Raven and Liberty Reservoirs. A third reservoir, Prettyboy, is used to resupply Loch Raven Reservoir. Management of the watershed conditions for each reservoir is a shared responsibility by agreement among City, County, and State jurisdictions. The most recent (2005) Baltimore Reservoir Watershed Management Agreement (RWMA) called for continued and improved water-quality monitoring in the reservoirs and selected watershed tributaries. The U.S. Geological Survey (USGS) conducted a retrospective review of the effectiveness of monitoring data obtained and analyzed by the RWMA jurisdictions from 1981 through 2007 to help identify possible improvements in the monitoring program to address RWMA water-quality concerns.

Long-term water-quality concerns include eutrophication and sedimentation in the reservoirs, and elevated concentrations of (a) nutrients (nitrogen and phosphorus) being transported from the major tributaries to the reservoirs, (b) iron and manganese released from reservoir bed sediments during periods of deep-water anoxia, (c) mercury in higher trophic order game fish in the reservoirs, and (d) bacteria in selected reservoir watershed tributaries. Emerging concerns include elevated concentrations of sodium, chloride, and disinfection by-products (DBPs) in the drinking water from both supply reservoirs. Climate change and variability also could be emerging concerns, affecting seasonal patterns, annual trends, and drought occurrence, which historically have led to declines in reservoir water quality.

Monitoring data increasingly have been used to support the development of water-quality models. The most recent (2006) modeling helped establish an annual sediment Total Maximum Daily Load to Loch Raven Reservoir, and instantaneous and 30-day moving average water-quality endpoints for chlorophyll-a (chl-a) and dissolved oxygen (DO) in Loch Raven and Prettyboy Reservoirs. Modelers cited limitations in data, including too few years with sufficient stormflow data, and (or) a lack of (readily available) data, for selected tributary and reservoir hydrodynamic, water-quality, and biotic conditions. Reservoir monitoring also is too infrequent to adequately address the above water-quality endpoints.

Monitoring data also have been effectively used to generally describe trophic states, changes in trophic state or conditions related to trophic state, and in selected cases, trends in water-quality or biotic parameters that reflect RWMA water-quality concerns. Limitations occur in the collection, aggregation, analyses, and (or) archival of monitoring data in relation to most RWMA water-quality concerns.

Trophic, including eutrophic, conditions have been broadly described for each reservoir in terms of phytoplankton production, and variations in production related to typical seasonal patterns in the concentration of DO, and hypoxic to anoxic conditions, where the latter have led to elevated concentrations of iron and manganese in reservoir and supply waters. Trend analyses for the period 1981–2004 have shown apparent declines in production (algal counts and possibly chl-a). The low frequency of phytoplankton data collection (monthly or bimonthly, depending on the reservoir), however, limits the development of a model to quantitatively describe and relate temporal variations in phytoplankton production including seasonal succession to changes in trophic states or other reservoir water-quality or biotic conditions.

Extensive monitoring for nutrients, which, in excessive amounts, cause eutrophic conditions, has been conducted in the watershed tributaries and reservoirs. Data analyses (1980–90s) have (a) identified seasonal patterns in concentrations, (b) characterized loads from (non)point sources, and (c) shown that different seasonal patterns and trends in nutrient concentrations occur between watershed tributaries and downstream reservoirs. A lack of data for total nitrogen and (or) available phosphorus limits direct comparisons of temporal or spatial variations in nutrient availability (comparable forms or ratios) between watershed tributaries and reservoirs.

Eutrophic conditions in the shallow water layer (30 feet in depth or less) in each reservoir have been assessed with four Carlson Trophic State Indices (TSIs)—derived from concentrations of chl-a, total phosphorus (TP), or DO, and (Secchi disc) transparency data. The frequency of eutrophic conditions for the entire period from 1982–2000 differed within each reservoir, and among the reservoirs, depending on which TSI index was used. The use of each index to compare trophic conditions among the reservoirs, however, possibly is biased because of the manner by which TSI data were collected, aggregated, or analyzed. In addition, no analyses of these indices were encountered that assessed possible trends in the frequency of eutrophic or mesotrophic conditions during this period.

Analyses of suspended-sediment data (1982–mid-90s) indicate that tributary concentrations and loads varied markedly within a year, and from year to year, but were clearly highest in wet years. Most sediment is carried by storm- as opposed to dry-weather (low) flow. Sediment transport has reduced reservoir capacity by 3 to 11 percent, and remains the major source of the TP load to the reservoirs. The role of this sediment as a source of available phosphorus (unmeasured) for phytoplankton production, however, has not been adequately addressed.

Manganese and iron are frequently monitored in water-supply intake waters during reservoir stratification and initial turnover. Elevated concentrations of these metals often occur at the supply intakes following their release from reservoir bed sediments under anoxic conditions, which can result from the decomposition of algal bloom residues. Monitoring in the reservoirs is too infrequent (monthly to bimonthly) to provide sufficient advanced warning of their occurrence at the intakes.

Elevated concentrations of mercury in game fish in the reservoirs are considered the end result of atmospheric deposition and beyond the control of RWMA jurisdictions. The submergence of terrestrial plants established on reservoir bed sediments exposed during droughts could enhance methylmercury production and biological uptake during reservoir recovery. However, this cannot be determined by conventional synoptic monitoring for mercury in game fish.

Fecal coliform bacteria have occurred at elevated counts in selected reservoir watershed tributaries, but counts in supply-reservoir intake waters consistently have been below the State recreational water-contact standard. Depending on results from synoptic surveys conducted by RWMA jurisdictions in the watersheds, the State could require routine monitoring of bacteria in the tributaries.

Among emerging concerns, trihalomethanes (THMs) and haloacetic acids (HAAs) are DBPs created by chlorination that are present in the drinking-water distribution systems of both supply reservoirs. Analysis of DBP data (2003–08) by the USGS indicates that the total concentrations of THMs and HAAs could exceed Federal standards under a pending rule change on approximately 19 percent and 40 percent of the sampling dates, respectively, at one or more monitoring stations in each water-distribution system. THM concentrations in drinking water varied seasonally, whereas HAA concentrations did not. There was little correlation between total concentrations of THMs and HAAs at a given monitoring station, or between monitoring-station concentrations of either DBP and total organic carbon (TOC) in intake waters. Monitoring of TOC alone will not identify intake waters associated with high concentrations of DBPs after chlorination.

In 2003, sodium and chloride concentrations at supply intakes were three-to-four-times greater than in the 1970s. Concentrations generally peaked during the winter months. Watershed and reservoir monitoring do not include the collection of sodium data. Monitoring also is too infrequent to provide either advanced warning of elevated sodium and chloride concentrations at the supply-reservoir intakes, or timely information on reductions in their concentrations if management activities are implemented to reduce road-salt use—the suspected source of the recent increases.

Projected changes combined with the inherent variability in climate in the Mid-Atlantic region indicate more intense storms with heavy precipitation and more frequent drought conditions. These changes imply increases in storm-borne contaminants (nutrient, sediment, salt, and bacterial loads), which could adversely affect reservoir water quality, particularly during recovery from drought conditions. Monitoring of stormflow does not appear to be adequate to address climate change and variability.

The 2007 Baltimore Reservoir System monitoring program could be improved in three major areas: (a) the monitoring design framework, (b) the temporal and spatial resolution of water-quality assessments in the major tributaries and reservoirs, and (c) the management and archival of data. Improvements in the framework design could include adoption of a quantitative phytoplankton model, such as the Phytoplankton Ecology Group model. Such models describe intra-seasonal, seasonal, and annual variations in phytoplankton abundance and succession. The model data can be analyzed in relation to temporal variations in nutrients or TSIs. The characterization of these biotic and water-quality conditions could be evaluated in relation to temporal variations in climate by the collection of climatic and water-quality data that reflect the full range in tributary flows and reservoir hydrodynamics within a year and from year to year. The minimal monitoring data would include daily temperature (mean), daily precipitation (total and type), continuous or partial records of streamflows depending on the type of tributary monitoring station, and daily water levels, withdrawals, and releases from each reservoir. To aid in this evaluation, the monitoring framework could incorporate the routine use of statistical and modeling methods to help define, aggregate, analyze, and interpret data.

Improvements in spatial and temporal assessments of water-quality conditions could be realized with two major and selected minor modifications to historical monitoring. First, to quantify water-quality conditions for the full range in tributary flows in the reservoir watersheds, sampling could include 3 to 15 pre-defined high (or storm-) flows per year at each of seven stations—three historical stations in each of the two supply reservoirs and one new station on a tributary to Prettyboy Reservoir. Pre-defined base-flow conditions could be sampled at each station on a monthly fixed time interval. Second, two fixed-station continuous monitors could be established in each reservoir to provide daily 5-foot-depth-increment profiles for selected parameters—water temperature, DO, pH, specific conductance, chl-a, turbidity, and depth of measurement. Data from these monitors could be transmitted to water-treatment staff to provide advanced warning of potential problems with supply intake waters.

A comprehensive quality-assurance program and plan (QAPP) with clear lines of responsibility could help ensure collection of the correct type and quality of data. The QAPP would include the following: (a) clear and concise definitions of the data and data-quality requirements for each water-quality concern; (b) field and laboratory methods and analytical procedures to obtain and provide the required data; (c) procedures to archive, clearly remark, and qualify data, including quality-assurance and control data; (d) procedures to routinely evaluate collected data in relation to data requirements; and (e) procedures to modify and document changes in field and laboratory methods.

First posted September 8, 2011