Phosphorus inactivation of incoming storm water to reduce algal blooms and improve water clarity in an urban lake

ABSTRACT

Wagner KJ. 2017. Phosphorus inactivation of incoming storm water to reduce algal blooms and improve water clarity in an urban lake. Lake Reserv Manage. 33:187–197.

A phosphorus (P) inactivation system was installed at Morses Pond in Wellesley, Massachusetts, in 2008, to treat storm inflows from 2 tributaries draining an urban watershed during late spring and early summer to minimize P loading and algal blooms. Modification and simplification of that system over 9 years has resulted in an efficient system for reducing P concentrations in the lake, decreasing algal blooms and increasing water clarity. Secchi transparency in the last 3 summers has been the highest on record, although precipitation has been below average in those years. On the basis of dollars per kilogram of P removed, total system cost has been well below that expected of watershed best management practices and near the low end of the cost range for P inactivation systems installed elsewhere.

KEYWORDS:

• Algae
• P inactivation
• storm water
• water clarity

Management of lakes in urbanized areas presents challenges. The developed nature of the watershed predisposes associated waterbodies to high pollutant loading. Control over land use and pollutant trapping opportunities are usually limited, yet the value of urban lakes is often high because they provide wildlife habitat, recreational opportunities, and even supply drinking water in areas with dense human populations and potentially large user groups. Frequent management activities characterized as maintenance techniques may be necessary to fulfill valued functions on a continuing basis.

Morses Pond is a shallow lake (mean depth = 2.5 m, maximum = 6 m) that covers 42.3 ha mostly in the Town of Wellesley with a small portion in the Town of Natick, Massachusetts (ENSR 2005), an urbanized area just west of Boston. Morses Pond is an important indirect source of public drinking water for the town through adjacent wells and is a major recreational resource, providing swimming, non-motorized boating, fishing, and shoreline hiking.

Morses Pond is fed by a 2100 ha watershed of mostly residential and commercial land (Fig. 1). Water enters the lake primarily through tributaries (Fig. 2), including Bogle Brook, Jennings Brook, and Boulder Brook, which respectively contribute 65%, 19%, and 9.5% of the water entering Morses Pond (ENSR 2002). These tributaries contribute large loads of contaminants during storm events to a 6 ha northern area of Morses Pond, separated from the rest of the lake by islands and acting as an initial stilling basin. Between 9.8 and 16.6 million cubic meters of water enters the lake each year, resulting in an average detention time of 25 to 40 days. Flushing is more rapid in late winter and spring, however, and water can remain in the lake for most of the summer.

Figure 1. Morses Pond location and watershed in eastern Massachusetts.

Loading of P is roughly proportional to the water load; the 3 tributaries account for about 95% of the P load, with at least 75% of that load entering during wet weather (ENSR 2005). Direct drainage from Wellesley and Natick to other parts of Morses Pond contributes to a much lesser extent, and atmospheric and groundwater loading is small. Internal loading to the lake is also minor, and historic whole-lake aluminum treatments provided only brief improvement (ENSR 2005). Morses Pond discharges into Paintshop Pond, which leads to Lake Waban and ultimately to the Charles River.

Because of the importance of Morses Pond as a multiple use resource, the Town of Wellesley has actively worked toward the management, improvement, and protection of the lake. Past in-lake management efforts have included the use of algaecides (copper sulfate), P inactivation (using aluminum sulfate), weed harvesting, and dredging. Monitoring has been performed almost every year since 1981, providing a useful database for management decisions. A plan for storm water management was prepared under the National Pollutant Discharge Elimination System Phase II regulations, promulgated under the Federal Clean Water Act, and related town bylaws were passed to control contaminant loading to water resources. Additionally, a comprehensive plan was prepared in 2005 (ENSR 2005) for the management of Morses Pond, incorporating input from diverse interest groups and establishing a management approach that included a P inactivation system.

Years of monitoring show that P concentrations <0.02 mg/L in the main body of the lake rarely support algae blooms and maintain water clarity suitable for contact recreation (ENSR 2005). The risk of cyanobacteria blooms and clarity <3 m increases substantially at higher P concentrations. It was hypothesized that if P levels could be lowered to near 0.01 mg/L going into summer, along with normally reduced summer inflow, the P concentration would remain <0.02 mg/L through the primary contact recreation period. Before P inactivation, values <0.02 mg/L have been observed only in dry springs when runoff inputs are reduced. Watershed management and/or treatment of incoming runoff is needed to achieve values <0.02 mg/L during wetter periods.

Educational and regulatory efforts have been directed at the entire Morses Pond watershed (ENSR 2005), and a reduction in P content in lawn fertilizers is expected to yield lower P concentrations in runoff (Lehman et al. 2013), but it was evident during management plan formulation that any short-term gain would require in-lake actions. Control of algae has historically been accomplished by copper treatments or rare whole-lake aluminum treatments, but the comprehensive plan included development of a P inactivation system to treat incoming storm water from the northern tributaries. This system is the subject of this analysis.

Material and methods

Phosphorus inactivation system

A P inactivation system was established in spring 2008 and has operated in late spring and early summer through 2016. The initial system included a trailer with a mounted a generator, a diesel powered compressor, and a pair of diaphragm pumps in a protective housing. Pumps were connected to a pair of chemical storage tanks off the trailer and a manifold with attached hoses running into the lake. The pumps could be run from the generator or by direct electrical feed from an adjacent town pump station. Valves on the manifold allowed direction of chemicals and air to any of 4 sets of feed lines into the lake. The chemical pump station was initially portable, stationed for the treatment period at the Wellesley Dale Street Pump Station, but in 2015 it was made a permanent station without the trailer. Four sets of lines initially ran from the pump station into the north basin (Fig. 3), each set consisting of an air feed line and 2 chemical feed lines.

Two lines with single diffusers and sets of chemical ports near the end of each line ran within the north basin to the mouths of Boulder Brook and Bogle Brook. Jennings Brook joins Boulder Brook just upstream of the inlet, so all 3 major streams were treated near the point of entry to the lake by the 2 inlet lines (#3 and #4). The other 2 lines, each with 4 diffusers and corresponding chemical ports, were spaced within the north basin itself to allow treatment of water in that basin. This setup allowed treatment if operation was not possible from the start of a storm, or if additional treatment in the basin became necessary. In 2013, dredging was completed in a portion of the north basin of Morses Pond to improve detention of incoming storm water, and use of the chemical feed lines in that basin was discontinued to allow settling in the newly deepened area. Mixing in the inlets was deemed sufficient to also eliminate the air feed, simplifying the system.

The P inactivation chemicals used for the treatment from 2008 through 2013 were aluminum sulfate (alum) and sodium aluminate (aluminate). Both are flocculating agents responsible for the inactivation of P, with alum creating acidic conditions and aluminate shifting the pH to a more basic level; these agents were added at a roughly 2:1 ratio (alum to aluminate, by volume) to balance the pH of treatments. One pump handled the alum at rates up to 320 L/h and the other moved aluminate at rates up to 200 L/h. Pumps were turned on manually, and manifold valves were adjusted to balance flows among feed lines in a rather inexact process.

Starting in 2014, polyaluminum chloride replaced the alum and aluminate to reduce corrosion of system parts and improve operator safety, and pumps were dedicated to a single tank and a single inlet treatment area, further simplifying the system. Because flows in Bogle Brook are larger than those in Boulder Brook, the larger pump (320 L/h) and the larger tank (7620 L) were dedicated to treatment of Bogle Brook and the smaller pump (200 L/h) and smaller tank (3810 L) were assigned to Boulder Brook. In 2016 the pumps were changed to peristaltic pumps, each with a capacity of 255 L/h, and the system was automated, turning on in response to rainfall above an adjustable threshold and controllable from a cell phone using Loggerlink from Campbell Scientific through a routine programmed by Blu-Dot Inc.

Aluminum chemicals were added to the north basin and/or inlets in mid-May through at least late June in response to storms in an attempt to achieve a target total phosphorus (TP) level in the south basin of <0.02 mg/L and preferably close to 0.01 mg/L at the start of summer. Higher flushing earlier in spring limits effectiveness of the treatment system, but treatment during what amounts to 1–2 flushing cycles at the end of spring was thought to be adequate to achieve desired conditions. Chemical supply was replenished by delivery as needed, and treatment ended as soon after the first weekend of July as the chemical supply was exhausted.

Monitoring program

Water quality has been assessed at the 2 main inlets to the north basin, at 2 points in the north basin (NB-1 and NB-2), at 2 more points in the transition zone (T-1 and T-2) just south of the islands that separate the north basin from the main body (south basin) of Morses Pond, at the surface and bottom of the deepest point in the lake (MP-1), and at 2 locations in the swimming area (B-1 and B-2) at the south end of the lake. Monitoring frequency varied, but samples were collected before treatment started in spring, in late June, and usually 2 more times over the summer. Additional samples were collected from the 2 main inlets during storms and from other stations as warranted by ambient conditions. The recreation department contracts for separate water quality characterization off the beach at the deepest point in the lake at least twice (Jun and Aug), with Secchi transparency assessed weekly from mid-June through mid-August of most years. Algae and zooplankton samples were collected with water quality samples at MP-1.

Figure 2. Phosphorus inactivation system layout and monitoring stations at Morses Pond, MA.

Samples were delivered the same day to a state certified lab where P was assessed by method SM 4500-P E (Rice et al. 2012). Additional water quality assessments, such as forms of nitrogen, were often made by standard methods from samples delivered to the laboratory. Secchi transparency was assessed with a viewing tube from a boat at MP-1. Algae and zooplankton were assessed microscopically according to standard methods (Rice et al. 2012). Field water quality including temperature, oxygen, conductivity, turbidity, and pH, was assessed with a calibrated Hydrolab DS5 multi-probe sonde in support of overall lake management but were not central to this assessment.

Results

System operation

Treatment was conducted in response to storm events in late spring and early summer, with treatment extending into August during a few years when the chemical tanks were filled in late June but conditions thereafter were relatively dry. Initial set up and testing occurred in 2008, with fewer treatment days and less aluminum applied than other years. Excluding 2008, the amount of aluminum applied ranged from 1555 kg in 2016 to 3055 kg in 2012 (May–Aug) with treatment occurring from 9 days in 2015 to 20 days in 2013 (Table 1). Although flows are not monitored daily at the 2 main inlets, the estimated concentration of aluminum at each inlet was between 1 and 3 mg/L when treatment was in progress.

Table 1. Operational data for the P inactivation system and precipitation record for Morses Pond, MA, 2008–2016.

Year	Applied Alum (L)	Applied Aluminate (L)	Aluminum Mass Applied (kg)	# of Treatment Days	May–June Precipitation (cm)	May–August Precipitation (cm)	Notes
2008	7620	3810	1018	5	15.7	42.3	Testing and adjustment phase, most treatment in July
2009	22868	11049	2998	16	15.0	40.8	Some elevated storm flow untreated
2010	15621	7925	2104	13	15.4	36.8	Additional chemical applied after early July
2011	19050	9430	2531	14	20.3	45.1	Some equipment failures. Additional chemical applied in August in response to bloom
2012	22860	11430	3055	19	17.4	36.5	Equipment problems hampered dosing during treatment
2013	23070	10611	2944	20	34.8	48.5	Very wet June (26.7 cm), unable to treat all storm flows; continued treatment through July
	Polyaluminum chloride (L)
2014	22803		1605	12	13.9	30.0	No treatment after first week of July, first year using polyaluminum chloride
2015	30099		2119	14	15.8	26.7	Leftover chemical used in summer, but little treatment after first week of July
2016	22098		1555	13	11.8	18.4	Only a little over half of the chemical was used by early July, remainder by 15 August

From 2008 through 2015, someone had to physically turn on the pumps and adjust valves to get chemicals flowing to the target areas. From 2008 through 2012, target areas could include 2 inlet injection points and 2 lines of in-lake injection, each with 4 actual discharge points. The process was labor intensive but considered necessary to gain experience with system performance and respond to problems of pumping and chemical distribution. Yet with storms occurring at any time of day and lasting for varying duration, operation was not always as timely as desired. Additionally, delivery of chemicals requires scheduling in advance, and larger or prolonged storms could exhaust the supply before replacement was possible. Further, wear on pumps and related fittings, especially associated with highly corrosive sodium aluminate, necessitated repairs and caused suboptimal system performance after the first few years until a change in chemicals was made.

The operational period to date can be divided into 3 distinct periods of 3 years each. Intermediate amounts of rain fell during May and June 2008–2010 compared to the other 2 periods (Table 1), and equipment generally functioned well in this early phase. The highest May–June precipitation occurred during 2011–2013, which experienced more equipment issues than the other 2 periods. The lowest May–June precipitation was observed during 2014–2016, and the change to polyaluminum chloride and simplification of system operation led to improved system performance.

The latest changes to the system seem to complete the development process at this site. The original polyethylene tanks remain and have performed well, but virtually all other equipment has been replaced. The trailer was replaced by a metal storage box with hinged panels on one side for complete access (Fig. 3), which sits in a fenced enclosure that also surrounds the 2 chemical storage tanks. This cubic (1.8 m) box houses the pumps, valves, and instrumentation that controls the system and is directly wired for power.

Figure 3. Current housing for pumps, valves, and instrumentation for the P inactivation system at Morses Pond, MA.

Polyaluminum chloride has been available for many years, but improved formulations have made this the form of choice for this type of aluminum application. It has a lower viscosity than alum or aluminate, facilitating pumping over longer distances and eliminating problems with differential movement for alum and aluminate that unbalanced the intended discharge ratio for Morses Pond. Low corrosivity of polyaluminum chloride lessens equipment maintenance issues. In laboratory jar tests using Bogle Brook stormwater, polyaluminum chloride removed 76% to 88% of TP at concentrations of 1 to 3 mg/L. Removal increased at larger concentrations, but greater treatment was not considered necessary to achieve program goals and would have required larger tanks and pumps to dispense enough polyaluminum chloride.

With 2 years of operational success and stability achieved in 2014 and 2015, the system was finally automated for 2016. A rain gauge was added on top of the adjacent pump station and wired to a control system in the metal storage box housing the pumps. At an easily adjustable rainfall threshold (usually set at 0.25 to 0.64 cm), the pumps are turned on and aluminum solution is fed into the lines leading to each inlet for an adjustable time (set at 4 h). If it is still raining at the end of the pumping period and the threshold is exceeded, another 4 h pumping period is triggered. A smartphone application can track progress and override the system to start or end pumping.

The maximum pumping rate is set for the larger Bogle Brook inlet and half that for the smaller Boulder Brook inlet, resulting in the larger and smaller tanks being emptied at the same rate. Sensors in the tanks provide remaining chemical level, allowing tracking and advance delivery arrangements. Those sensors will also shut off the pumps if the tank levels decline to just a few centimeters remaining. The only equipment issue in 2016 has been failure of the rain gauge (unbalancing of the tipping mechanism), and monitoring of nearby weather stations allowed remote override and system operation as needed.

Water quality and algae

Water quality in Morses Pond varies over time and space and has largely been a function of weather pattern until the treatment system was installed. Conditions are still strongly influenced by precipitation, but the treatment system limits P inputs and complements the detention and wetland treatment functions offered by the north basin of the lake (Fig 2), which develops a dense plant community in the summer except in the recently dredged area (0.8 of 6 ha). TP in Bogle and Boulder brooks, with Bogle also including the input from Jennings Brook as overflow from Jennings Pond, averaged about the same (0.128 and 0.129 mg/L) with respective medians of 0.051 and 0.065 mg/L (Table 2). The range was wide for each, however, and storm flows yielded higher TP values, indicating that most loading occurs during wet weather.

Table 2. Statistical characterization of TP in Bogle Brook and Boulder Brook inflows from 2008 to 2016.

	Total phosphorus concentration (mg/L)
Inlet	Mean	Median	Max	Min	n
Bogle Brook	0.128	0.051	1.029	0.013	40
Boulder Brook	0.129	0.065	1.385	0.027	38

Ratios of total nitrogen to total phosphorus (TN:TP) for Bogle Brook averaged 17.5:1 with a median of 14.8:1, whereas the TN:TP ratio for Boulder Brook averaged 14.9:1 with a median of 9.5:1. This finding suggests that ratios will be <10:1 and favor cyanobacteria that can fix dissolved gaseous nitrogen a substantial amount of the time, and removal of P will not only help limit productivity but could raise TN:TP ratios in favor of other algae. Data were inadequate in the north basin to determine if treatment has had this result, and TN:TP ratios in the main body of the lake (at MP-1) were usually well above 20:1. The ratio at MP-1 did increase after treatment commenced, however, with a pre-treatment average of 39:1 versus a post-treatment average of 48:1 and a pre-treatment median of 23:1 versus a post-treatment average of 34:1.

The spatial and temporal pattern of TP within Morses Pond over the 9 years of treatment effort (Table 3) shows^{Figure 4} considerable variability, even before treatment each spring. In general, TP concentrations declined from the north basin to the swimming area, which is near the outlet, with the transition zone usually but not always intermediate. The values in early summer after treatment were routinely lower than before treatment starts, and late summer values after a period without treatment tended to rise from early summer values. Beyond those generalizations, the data are perhaps best considered in the three 3-year periods described earlier.

Table 3. Phosphorus concentration for pre-treatment, early summer, and late summer periods in 3 zones of Morses Pond, MA, during each year from 2008 to 2016, with observations on algae and plant abundance.

Year	Location	Pre-application TP (mg/L)	Early summer TP (mg/L)	Late summer TP (mg/L)	Algae observations
2008	North basin	0.028	0.018	0.013	Mats observed, some cloudiness
	Transition zone	0.031	0.022	0.014	Some cloudiness, brownish color
	Swimming area	0.021	0.012	0.012	No blooms reported, first year without copper treatment in some time
2009	North basin	0.035	0.040	0.063	Cloudy, some green algae mats
	Transition zone	0.035	0.039	0.045	Cloudy
	Swimming area	0.015	0.010	0.027	Generally clear, no blooms reported
2010	North basin	0.026	0.046	0.053	Cloudy, green algae mats evident
	Transition zone	0.028	0.021	0.032	Brownish color, minimally cloudy
	Swimming area	0.019	0.015	0.043	Generally clear, no blooms until late August (Dolichospermum)
2011	North basin	0.053	0.033	0.130	Cloudy, green algae mats evident
	Transition zone	0.048	0.029	0.095	Slightly brownish
	Swimming area	0.030	0.029	0.060	August bloom (Dolichospermum), dissipated after just a few days without treatment
2012	North basin	0.032	0.024	0.048	Very dense plant growth, associated green algae mats
	Transition zone	0.028	0.037	0.028	Brownish most of summer
	Swimming area	0.020	0.027	0.024	Had bloom in mid-July (Dolichospermum), treated with copper
2013	North basin	0.036	0.047	0.030	Water brownish, little visible algae; first year with newly dredged area within north basin
	Transition zone	No Data	0.078	0.032	Generally elevated turbidity, but much of it is not living algae
	Swimming area	0.024	0.033	0.028	Treatment lowered TP but not to target level; June flushing minimized algae biomass
2014	North basin	0.030	0.022	0.020	Dense plant growths and green algae mats outside dredged area, water fairly clear
	Transition zone	0.021	0.020	0.018	Dense plant growths, but water fairly clear
	Swimming area	0.012	0.013	0.017	Water clear; Secchi to bottom in swimming area, no blooms reported
2015	North basin	0.012	0.017	0.023	Dense plant growths and green algae mats outside dredged area, water fairly clear
	Transition zone	0.008	0.015	0.014	Dense plant growths, but water fairly clear
	Swimming area	0.005	0.005	0.014	Water clear; Secchi to bottom in swimming area, no blooms reported
2016	North basin	0.012	0.009	0.005	Very dense plant growths after mild winter, but water still clear
	Transition zone	0.019	0.016	0.005	Dense plant growths but water clear
	Swimming area	0.014	0.005	0.005	Water clear; Secchi to bottom in swimming area, no blooms reported

During 2008–2010, TP ranged from 0.026 to 0.035 mg/L in the north basin before treatment started and from 0.015 to 0.021 mg/L in the swimming area. In early summer, after the bulk of treatment was complete, north basin TP values were lower in only 1 of 3 years, but the values at the swimming area ranged from 0.010 to 0.015 mg/L, within the target range. TP values remained acceptable in all monitoring locations in 2008 but increased markedly with summer storms in 2009 and 2010. Flushing seemed adequate to prevent bloom formation in 2009 despite the higher TP, but the cyanobacterium Dolichospermum (formerly Anabaena) bloomed in late August of 2010 during a hot spell after a substantial untreated storm.

During 2011–2013, TP ranged from 0.032 to 0.053 mg/L in the north basin before treatment started and from 0.020 to 0.030 mg/L in the swimming area. This set of years provided more precipitation in May–June (Table 1) than any of the 3 previous years, and 2 years (2011 and 2013) provided the largest May–August precipitation recorded in this project period. By early summer the north basin TP level had been reduced somewhat in 2 of 3 years, but the swimming area TP remained unacceptably high at 0.027 to 0.033 mg/L. Dolichospermum blooms were observed in 2011 and 2012 but not in 2013, likely a consequence of rapid flushing during one of the wettest summers on record for the area. Treatment continued longer into the summer in each of these years, with more aluminum applied than any other year except 2009, but was unable to keep up with inputs, and late summer TP values were elevated at all monitoring sites.

During 2014–2016, spring–summer conditions were drier than in any of the previous years (Table 1). TP values before treatment were acceptable in the swimming area in all 3 years and only slightly elevated elsewhere in 2014 (Table 3). Treatment improved or maintained low TP levels into early summer, and desirably low levels were also found in late summer and required less aluminum than most other years to maintain desirable conditions. The combination of treatment and subaverage inflows minimized P loading to the main body of the lake, and no algae blooms were reported.

Direct algae analysis has been sporadic, but samples from before treatment system installation and since treatment was instituted (Fig. 4)^{Figure 5, Figure 6} indicate treatment effectiveness. Cyanobacteria are not the only problem algae in Morses Pond, and the probable problem threshold of 3 mg/L of algae biomass was often exceeded before treatment. Since treatment commenced, the probable problem threshold has not been exceeded in collected samples, and since the change to polyaluminum chloride and system overhaul during the 2014–2016 period, algae biomass has reached the potential problem threshold of 1 mg/L only once, and then barely and with no cyanobacteria present.

Figure 4. Algae biomass in Morses Pond before and after implementation of P inactivation. Each column represents the average summer biomass for the corresponding year, subdivided by major algal group. Algae data are not available for all years. Horizontal threshold lines represent potential and probable recreational impairment levels based on beach operations over many years.

Water clarity

Although the focus of the treatment program was on reducing algae biomass, the most visible result is increased water clarity, a result with the greatest perceived importance to most lake users. Water clarity has tracked treatment success and generally been inversely related to TP concentration (Figs. 5 and 6), with clarity from the 2008–2010 period representing an improvement over most pretreatment years, clarity in the 2011–2013 period not distinctly different from pre-treatment years, but clarity in the 2014–2016 period representing the highest observed in Morses Pond. Note that copper was used before P inactivation to combat blooms but only once since P inactivation commenced, so pre-treatment clarity would have been lower than depicted in Figures 5 and 6, and the correlation between TP and Secchi transparency will be compromised, although that relationship remains strong. Early summer and average summer Secchi transparency values were tightly linked, with an R² of 0.84 in a linear regression, but not following a 1:1 slope. Average summer Secchi values were lower than early summer values, so targeting a high early summer value was necessary to maintain desired conditions throughout summer.

Figure 5. Average TP concentration and Secchi disk transparency over time in Morses Pond, MA. Values represent the summer average for the corresponding year. Pre-treatment and 3 treatment periods are identified based on equipment and chemical changes.

Figure 6. Relationship between average summer TP concentration and Secchi disk transparency before and after P inactivation in Morses Pond, MA. Four periods are represented by different symbols, including pre-treatment (2007 and earlier), early operation (2008–2010), a period of equipment issues (2011–2013), and the period of treatment system improvements (2014–2016).

Costs

The original system cost $133,000 to build and install in 2008, but some features were later deemed unnecessary, and an appropriate system could be constructed for about $100,000 today, even considering inflation. Capital improvements over the period of operation have cost about $60,000, including new pumps and automation of the system. The annual cost for aluminum solutions has ranged from $7200 (2015) to $22,700 (2009) with an average of close to $15,000 for 9 years. Maintenance costs are harder to quantify because the town has supplied skilled labor without external cost. About $10,000 per year is suggested based on the difference between a budgeted account value ($25,000) and the cost of chemicals, which are also charged to that account.

We assume that 75% of the TP entering the lake from the 2 main inlets is inactivated during May and June based on jar tests and the approximate portion of the inflow treated. This represents at least 15% of the total inflow from those inlets, based on the average May–June precipitation divided by annual total, which is projected to contribute 1274 to 2158 kg/yr of P (mean inflow concentration times range of inflow), suggesting that 143 to 243 kg of P are removed annually by treatment. Because the system treats mainly runoff with concentrations at least twice the overall average, however, an estimated 300 to 500 kg of P are removed each treatment season. Spreading the total capital cost to date over what is now a projected 20-year lifespan and adding an annual operational budget of $25,000, the annual cost for the system is about $34,650, and the cost per kilogram of P removed is between $69 and $116.

Discussion

Maintenance treatment with aluminum compounds is not a new approach to lake management; Harper et al. (1999) described a decade of experience 17 years ago, and a commercial P inactivation system that can be customized for lake use has been available for at least 5 years. Details of optimal treatment are yet to be worked out, and the use of aluminum to manage water quality in urban lakes is not a mature science. Each lake may present unique challenges, but decades of use in water and wastewater treatment (Randtke and Horsley 2012) provide guidance for lake applications. Key aspects of successful application include having a valid target condition for management, quantifying complementary P control functions in the watershed and lake, and understanding the necessary level and timing of treatment.

In Morses Pond, most water and P enters at the north end of the lake during storms and passes through a shallow basin with dense aquatic plants before entering the larger south basin, the focus of recreation, habitat management, and water supply interaction. Under dry conditions, detention and related P removal in the north basin with dilution in the larger south basin are adequate to minimize algae blooms in the southern lake area. With greater precipitation and runoff, however, natural processes are overwhelmed and treatment becomes necessary to limit P availability in the lake. The P inactivation system was designed to inactivate a sufficient amount of P to achieve a southern basin P concentration of <0.02 mg/L by early summer, with available data indicating that acceptable conditions would be maintained during most of the summer with limited or no further treatment.

The P inactivation system at Morses Pond has morphed over time as part of an adaptive management program, with monitoring and operational experience providing feedback on success and adjustments made to improve system effectiveness and efficiency. The operational history of the system can be conveniently divided into 3 periods based on system features and May–June precipitation.

During the first 3 years (2008–2010), the system worked well using alum and aluminate with considerable manual adjustment during application under presumably fairly typical precipitation levels. Algae blooms were prevented and water clarity was higher than before treatment was initiated.

During the second 3-year period (2011–2013), more precipitation fell in late spring and early summer, and equipment problems impaired treatment effectiveness. Algae blooms occurred in 2 of 3 years, and worse conditions may have been prevented at least partially by high flushing rates. Treatment may have helped, but water clarity was not higher than in the years before treatment began.

In the third 3-year period (2014–2016), changes in system design and lower than average precipitation allowed superior performance. Algae were lower and water clarity was higher than at any previous time monitored. Although completely separating the influence of treatment and weather with existing data is impossible, dry springs and summers occurred before treatment, resulting in algae blooms and low Secchi transparency. The high Secchi values during 2014–2016 are in the highest quartile for expected clarity in southern New England (Stephens et al. 2015), whereas the pre-treatment values are below the mean and median for this region. The potential for the system to successfully address P loading seems substantial, especially with recent modifications.

The system needs to operate more often with higher precipitation, creating the need for more chemicals and increasing wear on pumps and hoses, but the difference between the lowest and highest chemical use years is only one delivery (36% more), and most mechanical issues were a function of gradual deterioration of pump parts and pipe fittings from corrosive chemical contact over the first 3 years of the program. The system must operate at peak efficiency to address P loading during wet periods, and the recent adjustments were made for that reason. Continued operation is expected to build on the existing record.

Other factors complicate this analysis, but P inactivation is perceived as the main reason for improved conditions in Morses Pond. Dredging improved detention in the northern area and aided P control during 2014–2016 but was started in 2012 and completed in early 2013, with no clear benefit documented in those years. The lower spring precipitation in 2014–2016 reduced inputs, but inlet concentrations were still elevated. Although the system was successful in 2014–2016, the degree to which the current P inactivation system can control conditions in Morses Pond will remain uncertain until at least the next wet spring.

The cost of system development and operation equates to a unit cost of $69 to $116 for removal of each kg of P, well below costs listed for many competing watershed management techniques (USEPA 2015), including detention systems ($371 to $658), infiltration facilities ($7121 to $7443), or street sweeping ($3080 to $4840). P removal at Morses Pond is in the lower portion of the known range for storm water treatment systems applying aluminum ($74 to $253; ERD 2015), and the cost is believed to be offset by savings in drinking water treatment and increased revenue from recreational facility use. Even if the assumptions used to estimate the quantity of P inactivated are inaccurate and the actual cost per kilogram inactivated is as much as 100% higher, the cost of the Morses Pond system is within the range of other aluminum-based P inactivation systems and much lower than alternatives. This value makes P inactivation on a selective maintenance basis an attractive option in this urban watershed with limited jurisdiction over sources.

Acknowledgments

Field aid from Maxine Verteramo, Kathleen Burke, and Toni Stewart is gratefully acknowledged. The ESS Group is acknowledged for generating and supplying some of the data for this analysis. The comments of reviewers improved this manuscript and their efforts are appreciated.

Funding

The author acknowledges the support of the Town of Wellesley for the development of the comprehensive management plan and implementation of its elements for the last decade. The commitment of multiple departments as well as financial support has been essential to the success of this program.