National Academies Press: OpenBook

Watershed Management for Potable Water Supply: Assessing the New York City Strategy (2000)

Chapter: 6 Tools for Monitoring and Evaluation

« Previous: 5 Sources of Pollution in the New York City Watershed
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

6
Tools for Monitoring and Evaluation

The previous chapters have laid the groundwork for analyzing several aspects of New York City's watershed management strategy as embodied in the Memorandum of Agreement (MOA). In conducting its analysis, the committee identified components of New York City's strategy that correspond to the necessary components of a source water protection program described in Chapter 4. In some cases, this was relatively straightforward; in others, corresponding activities were more difficult to identify. The following six chapters consider how effectively New York City is carrying out source water protection and other related activities. Because of the study's broad statement of task, it was necessary to group particular programs and issues for analysis and discussion. Whenever possible, programs were grouped to correspond with specific necessary elements of a source water protection program.

The first step in any watershed management program is to set goals and objectives, both numeric and narrative. The stated goals of the New York City MOA are many. First, as with all other source water protection programs, one goal of the MOA is to comply with local, state, and federal statutes that protect drinking water quality. Thus, the City has developed an extensive water quality monitoring program, active disease surveillance, a Total Maximum Daily Load (TMDL) program, and a variety of programs for controlling and treating pollution. These programs, when carried out successfully, contribute toward the delivery of clean drinking water and the maintenance of healthy water supply reservoirs. However, the City's overall goals clearly go beyond those mandated by environmental laws. Supporting economic development within the watershed region is desired, as evidenced by the Watershed Agricultural Program, Watershed Forestry Program, and Watershed Protection and Partnership Programs.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Finally, a primary motivation for New York City to draft new watershed rules and regulations was to avoid filtration, which is considered to be an expensive water treatment option for this system. Although cost estimates of filtration differ greatly, all exceed the estimated costs of fully implementing the MOA.

This chapter considers two fundamental activities that inform decision-makers about the quality and safety of water from the Catskill/Delaware system. The first activity is water quality monitoring of all sections of the water supply system, including groundwater, streams, reservoirs, and the delivery system. The physical, chemical, and biological parameters being measured, their importance in assessing the condition of the water supply, and their role in water quality modeling are discussed. The Geographic Information System (GIS) developed for watershed inventory and other purposes is critically evaluated. Second, the safety of the drinking water system is considered by reviewing the role of active disease surveillance in New York City and by conducting a microbial risk assessment on the source water. Current activities that are evaluated include the detection of waterborne disease outbreaks and epidemiological studies for determining the proportion of illness attributable to drinking water.

WATER QUALITY MONITORING PROGRAM

The quality of the drinking water in New York City depends highly on the water quality of the 19 reservoirs of the Catskill/Delaware and Croton watersheds. Reservoir water quality is directly coupled to, and dependent upon, the loadings of pollutants from the individual drainage basins and from atmospheric deposition. Each of the drainage basins of the individual reservoirs combines a unique set of physical, chemical, and biological characteristics. These characteristics—including elevation, geomorphology, rock and soil composition and distribution, soil chemistry, rates of runoff and water residence times, types and extent of plant cover, and human modifications by land use, development, and waste releases—can vary markedly from basin to basin.

Because the volume of drinking water required by New York City is so large and entirely dependent upon the aggregate sources of these reservoirs, users rapidly realize changes in reservoir water quality. Therefore, any response by water managers and treatment operators requires rapid acquisition of information concerning changes to the reservoir ecosystems and their drainage basins. The regulatory steps that are taken to control various water quality parameters may be different. In addition, management steps may differ with varying seasonal, meteorological, and human influences. Thus, monitoring frequency must be flexible and respond to the rates of change that are observed for individual water quality parameters.

Four types of monitoring are discussed in Chapter 4: compliance monitoring, operational monitoring, performance monitoring, and monitoring to support modeling efforts. Efforts of the New York City Department of Environmental

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Protection (NYC DEP) in most of these areas are analyzed below, and recommendations for improvement, when necessary, are given. A full description of all monitoring efforts conducted by NYC DEP is available in the Water Quality Surveillance Monitoring report (NYC DEP, 1997a).

Compliance Monitoring

In order to comply with the Safe Drinking Water Act (SDWA) and the filtration avoidance determination, a variety of biological, chemical, and physical parameters are measured in the New York City source water reservoirs and the water distribution system (Table 6-1). In almost all cases, the New York City water supply has met the requirements for the physical and chemical parameters up to the present time. Compliance monitoring of fecal and total coliform measurements revealed increases in these parameters during the winters of 1991–1992 and 1992–1993 in the Kensico Reservoir. However, violations of the SDWA were avoided by bypassing the Kensico Reservoir (see Chapter 5 for details).

Operational Monitoring

Operational monitoring refers broadly to activities that are necessary for both short-term and long-term operation of the water supply system. This category, which encompasses much of NYC DEP's efforts, includes both routine activities and special projects (1) to follow changes in water quality and (2) to

TABLE 6-1 Frequency with which Parameters are Measured during SDWA Compliance Monitoring

Parameter

Catskill System

Delaware System

CAT(LEFF)a

CAT(EV)b

DEL18a

DEL19b

Turbidity

Continuous

Continuous

Continuous

Continuous

pH

Daily

Continuous

Daily

Continuous

Free Chlorine Residual

Not determined

Continuous

Not determined

Continuous

Total Coliform

Daily

Daily

Daily

Daily

Fecal Coliform

Daily

Daily

Daily

Daily

Temperature

Daily

Continuous

Daily

Continuous

a CAT(LEFF) and DEL18 are the effluent locations for the Catskill and Delaware systems, respectively, at Kensico.

b CAT(EV) and DEL19 are within the Catskill and Delaware aqueducts, respectively, just downstream of Kensico.

Note: SDWA compliance monitoring also includes some pesticides within the distribution system, which are not listed in this table.

Source: NYC DEP (1997a).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

identify sources of pollution that affect reservoir water quality. Monitoring of physical and chemical parameters is discussed first (organized by waterbody type) followed by a review of the microbial monitoring efforts that occur throughout the watersheds.

Regional Meteorology

Meteorological data, including air temperature, relative humidity, rainfall, snow depth, solar irradiance, photosynthetically active radiation (selected sites), and wind speed and direction, are measured at 22 stations throughout the Catskill/Delaware and Croton watersheds at one-minute intervals. This new (1998) network of meteorological stations has been established based on a reasonable set of criteria, including precipitation patterns, elevational gradients, and modeling requirements. Each station contains instrumentation for a large and complete set of meteorological parameters, and the one-minute interval frequency is satisfactory.

One could champion for greater meteorological data collection in a region such as the Catskill/Delaware watershed, where topography is quite heterogeneous. However, the new network is greatly improved over the previous system, which measured precipitation only, and it is adequate for most of the eutrophication and public health questions of concern.

Groundwater and Shallow Subsurface Monitoring

Regular groundwater monitoring has only occurred in the Kensico watershed, as this area has a high potential for contamination related to urbanization (NYC DEP, 1997b). Eighteen monitoring locations exist, consisting of 13 wells, some of which have multiple depths (ranging from 3.5 to 120 ft). These locations, which were selected in relation to geology, land and chemical use, and proximity to sewer lines, are reasonable for the Kensico watershed. All 13 wells were monitored for turbidity, pH, alkalinity, conductance, total and fecal coliforms, and nutrients on a monthly basis between April 1995 and April 1997, and static water levels were measured weekly. Analysis of monitoring data collected during that time period led to biannual sampling of all wells, which has continued to the present time. This frequency of sampling is sufficient for the deep subsurface but not for the shallow subsurface, which is influenced by seasonal variations. Other groundwater monitoring activities associated with special projects are discussed in a later section on performance monitoring.

Stream Monitoring

Monitoring of stream inputs is critical for determining reservoir water quality and managing reservoir operations. One of the first steps in understanding pollutant dynamics within short-detention-time reservoirs is to sample their tributaries

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

frequently and to construct input–output budgets of key physical and chemical parameters. There are two essential monitoring parameters for streams: (1) volume of influent water to compute loadings and for use in nutrient loading and productivity models and (2) concentrations of critical chemical parameters to evaluate the loadings of nutrients, potential toxic heavy metals, and organic compounds.

The locations of the 145 sampling sites throughout the Catskill/Delaware and Croton watersheds include primary river inflows, sites above and below selected wastewater treatment plants (WWTPs) and towns, and outflows from subcatchment basins. Grab samples are generally collected at biweekly intervals and are analyzed for temperature, pH, alkalinity, conductance, dissolved oxygen, major cations (Ca2+, Mg2+, Na+, K+), turbidity, color, suspended solids (SS), nutrients, total organic carbon (TOC), silica, chloride (Cl), trace elements, and total and fecal coliform, among others. Forty-eight (48) U.S. Geological Survey gauging stations measure water level (stage) continuously. The computerized data acquisition system being developed at present appears to be adequate.

Because samples are collected at fixed time intervals rather than on the basis of discharge, it is certain that major fluctuations in the loading of chemical and biological parameters are not captured. This issue, which pertains to stream sampling, precipitation measurements, and sampling of shallow subsurface flows, is discussed below with regard to certain parameters, and it is generally addressed later under a separate section titled Flow Proportional Monitoring.

Physical Parameters and Cations. Automation of stream monitoring can ease the transition from fixed frequency sampling to event-based sampling. Within the New York City watersheds, the monitoring of some stream parameters, most notably temperature, conductance, pH, and dissolved oxygen, could easily be automated. This is possible even at remote sites via telemetry of data to data acquisition centers. Although the initial expense to install automation would be high, costs could subsequently be reduced by decreasing the required personnel and by eliminating other analyses. For example, rather than being measured directly, the concentrations of Ca2+, Mg2+, Na+, and K+ could be obtained from strong correlations with conductance within 5 percent to 10 percent accuracy, which would be quite adequate for the purpose of determining hardness and reactivity (Otsuki and Wetzel, 1974). Alkalinity could also be continuously estimated with reasonable accuracy from the parameters measured automatically.

Turbidity, Color, Suspended Solids, Nutrients, Total Organic Carbon, and Silica. These parameters are currently measured at stream sites on a fixed biweekly or longer interval. The usefulness and validity of these data, particularly for use in nutrient loading models, are unclear. A fixed biweekly or longer sampling interval is marginally satisfactory for the monitoring of large reservoirs with moderately long (> six months) residence times. However, these sampling

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

intervals are not satisfactory for surface water influents and nonpoint shallow subsurface inflows. Fixed interval sampling will miss significant loading events and will substantially underestimate true loadings. It would be better to thoroughly sample stream inflows at fewer stations with automated discharge-mediated samplers, particularly close to the reservoir inlet mouth, than to sample many stations at infrequent, fixed intervals. In addition to the parameters currently measured, dissolved organic carbon (DOC) should also be measured in streams on a flow proportional basis, as suggested by others (ILSI, 1998).

Chloride, Trace Elements, and Toxic Compounds. Chloride and trace element analyses are adequate at present frequencies. Toxic compounds are not measured on a regular basis at stream sites. In 1997, concentrations of several pesticides and polychlorinated biphenyls (PCBs) were monitored at two stream sites in the Kensico watershed; compounds were detected at very low levels (NYC DEP, 1998a). Monthly sampling of streams for pesticides is a planned future activity of NYC DEP that is strongly supported by the committee as a way to determine presence/absence, establish ambient concentrations, and better pinpoint sources. Sampling should be timed to correspond with the application of pesticides and expected pesticide transport by stormwater from rainfall and snowmelt. This activity is needed most in the Cannonsville and Pepacton watersheds because of the density of pesticide application and in the Kensico and West Branch watersheds because of their proximity to the distribution system.

Macroinvertebrates. Monitoring of stream macroinvertebrates is done annually in August and September at 14 regular stream sites throughout the entire watershed region and at additional sites that vary in location. This occasional macroinvertebrate sampling is being used as a biotic index of relative stream ''health" based on indicator species. Based on monitoring results, the stream sites are classified as nonimpaired, slightly impaired, or severely impaired. Between 1994 and 1997, 29 sites were nonimpaired, 19 sites were slightly impaired for at least one year, and no sites were severely impaired (NYC DEP, 1999a). Because the present frequency of sampling is too low and the quantitative measures are too marginal to overcome high natural variance, the usefulness of these data in relation to the information accrued and the effort expended is questionable.

Aqueduct Monitoring

The aqueducts (or tunnels) are sampled at ten locations west of the Hudson River and at 11 locations east of the Hudson River, generally where water enters and exits the tunnels. The sampling interval is daily or weekly, depending on the proximity of the tunnel to Kensico Reservoir. Twenty-seven (27) different parameters are measured within the aqueducts for both operational and compliance monitoring purposes and to support the Process Control-Remote Monitoring pro-

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

gram. Those measured on a daily basis include odor, color, turbidity, temperature, specific conductivity, pH, free chlorine residual, fluoride, and total and fecal coliforms. In addition to the 27 parameters, some pesticides are being monitored on an annual basis at six aqueduct (key point) locations.

As part of the Process Control-Remote Monitoring Program, sampling of turbidity, pH, conductivity, temperature, free chlorine, and fluoride is automated at 13 aqueduct locations east of the Hudson River. NYC DEP plans to extend automated sampling to aqueduct key points west of the Hudson River when resources are available. The automated sampling is conducted at a significantly higher frequency than grab samples, and the results are continuously downloaded to chart recorders that display the data. The frequency and analytical techniques used for within-aqueduct analyses of water quality are adequate.

Reservoir Monitoring

Reservoirs receive collective loadings both from the atmosphere and from their drainage basins. The effects of these pollutant loadings on reservoir water quality are relatively low because loading volumes are significantly less than total reservoir volume. In addition, the residence time of reservoirs is increased compared to streams. Therefore, assuming that withdrawal volumes are relatively small in proportion to total reservoir volume, the frequencies of monitoring for most water quality parameters can be reduced from that of stream monitoring. This reduction, however, should not exceed the generation times of controlling processes, including biological processes.

The water supply reservoirs are sampled monthly from late March to early December, with biweekly sampling occurring at some reservoirs. Temperature, pH, dissolved oxygen, and specific conductivity are measured in situ with automated samplers at 1-m depth intervals. Other measured parameters can be found in NYC DEP (1997a). Present sampling includes a depth-integrated water sample (e.g., 1–3 m) from the euphotic or light-penetrating zone, the depth of which may or may not represent an integrated sample of the epilimnion1 during periods when the reservoirs are thermally stratified. Metalimnetic and hypolimnetic samples are taken in the deeper portions of the reservoirs. For the general purposes of assessing water quality in these moderately impacted reservoirs, the spatial sampling sites for monitoring are generally adequate.

1  

 Reservoirs with moderate to long retention times stratify thermally, with less dense, warner water overlying cooler, more dense water in summer. The water of the upper stratum, the epilimnion, is uniformly warm, circulating, and fairly turbulent. The epilimnion essentially floats upon a cold and relatively undisturbed bottom stratum, the hypolimnion. The intervening stratum between the epilimnion and the hypolimnion is the metalimnion, characterized by a steep thermal gradient from warm to cool water (decreasing at least 1°C per meter of increasing depth).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Growing Season Considerations. The in-reservoir "growing season" (March to December) is assumed to be the period during which the reservoir is most susceptible to water quality changes. Thus, there is little reservoir sampling during the winter, except for sampling of the aggregated outflows within aqueducts. This strategy may be acceptable at the present with low mesotrophic to moderate eutrophic conditions in the reservoirs. However, if the productivity increases further, major chemical and biological changes will occur in winter periods that must be monitored. For example, at this latitude 25 percent of the annual primary productivity can easily occur under ice cover of reservoirs.

Color, Secchi Depth, and Euphotic Depth. The euphotic depth refers to the depth of water that receives sufficient solar radiation to support net photosynthetic growth of phytoplankton. This depth is usually limited to the penetration depth of 0.1 percent to 1 percent of light reaching surface waters. Secchi depth is an estimate of water transparency and is equal to the approximate penetration depth of 15 percent of surface light. These parameters, along with color, give an indication of algal levels and other color-producing compounds. The color and clearness of the New York City reservoirs is often governed by dissolved organic matter (DOM). In the case of the Catskill/Delaware reservoirs, the DOM responsible for color and clearness originates largely from terrestrial and wetland sources of higher plant decomposition (allochthonous sources) rather than from inreservoir sources (autochthonous sources). This is because only under hypereutrophic conditions are light penetration and euphotic depth appreciably influenced by the densities of algae and cyanobacteria, and such conditions are not found in most of the Catskill/Delaware reservoirs.

Much of the loading of allochthonous DOM is directly correlated with precipitation events within the drainage basin. Once in the reservoir, this recalcitrant pool of DOM is biologically degraded at a relatively slow rate. Therefore, biweekly sampling of euphotic depth, Secchi depth, and color should be adequate.

Dissolved Organic Matter. Because DOM can serve as a precursor of trihalomethanes (THMs) and other chlorination byproducts, NYC DEP has spent considerable energy investigating the sources of DOM in the water supply reservoirs. These efforts have consisted of continuous reservoir monitoring as well as special activities to measure a variety of parameters, including DOC and THM formation potential (NYC DEP, 1997c; Stepczuk et al., 1998a–c). Although preliminary evidence suggests that autochthonous sources may predominate during certain times of the year (NYC DEP, 1997c), the slower degradability of allochthonous sources means they will be dominant for significant periods, especially during the winter (December through March). This potential switch in the dominance of different sources of DOM implies that monitoring must take place year round rather than just during the growing season.

DOM is almost always quantitatively expressed as DOC. In order to deter-

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

mine the pool of DOC available for reaction with chlorine, quantitative assays should be conducted using modern DOC analyses (heat oxidation followed by IR or coulometry to measure the CO2). Measurements of DOC made by the Carlo Erba (EA1108) instrument for CHN analyses (as used by NYC DEP in Stepczuk et al., 1998a) are not satisfactory because the instrument was not designed for such analyses. DOC measurements should occur in depth profiles at regular intervals with a frequency of at least every 2 weeks in the reservoirs. For making such measurements in streams and shallow subsurface sources, the frequencies for such analyses should be based on discharge, not time. It should be kept in mind when conducting such analyses that DOC is, in general, a poor surrogate measure of THM precursor concentration.

To complement DOC measurements, selective ultraviolet absorption spectrophotometry has been employed as a general index of DOM concentration in natural waters (American Public Health Association, 1998; Thurman, 1985; Wetzel and Otsuki, 1974). Ultraviolet absorption (UV254) permits a relative measure of the stable recalcitrant dissolved organic components derived from allochthonous sources. However, it is not particularly useful for detecting variations in autochthonous sources because DOM produced by phytoplankton is more labile by one or two orders of magnitude than allochthonous DOM. UV254 measures tend to exhibit little variation over time because the decomposition rates of allochthonous DOM of about one percent per day approximately balance allochthonous influxes. This trend has been borne out by NYC DEP studies; UV254 data collected in Kensico Reservoir effluents on an irregular basis were found to average around 0.036 OD with little variation (S. Freud, NYC DEP, personal communication, 1998). No correlations were found between UV254 and TOC or trihalomethane formation potential (THMFP) values.

Total DOC and spectral data yield no information about the sources of DOM. Only a detailed structural analysis of the compounds found in water samples can reveal the relative levels of autochthonous and allochthonous DOM. Allochthonous DOM is in large part composed of yellow organic humic acids of plant origin and consists of a mixture of fulvic acids, aromatic polyhydroxy carboxylic acids, and phenolic residues and polymers of these acids. Such compounds do not originate from phytoplankton algae and cyanobacteria.

The more complex organic compounds have been variously categorized on the basis of their structure and their solubility characteristics in acids and bases. Quantitative differentiation of the relative amounts of autochthonous and allochthonous DOM can be estimated by detailed organic chemical analyses using solid-state 13C-nuclear magnetic resonance and gas chromatography–mass spectrometry analyses (e.g., Wetzel et al., 1995; McKnight et al., 1997). Less demanding analyses can be used to estimate the likely proportions of aliphatic, aromatic, and "excess" carbon in a complex mixture of DOM from different sources by evaluating its elemental composition and carboxy1 content (Purdue, 1984; Wilson, et al., 1987). At best, these methods yield approximations of ratios

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

between autochthonous and allochthonous sources and should be coupled to more accurate estimates of external DOM loading to reservoirs. If the NYC DEP wants to determine the relative sources of DOM in the water supply reservoirs, these methods must be employed.

In addition to determining sources of DOM, NYC DEP may want to consider regularly measuring disinfection byproduct (DBP) formation potential in the reservoirs, particularly Kensico Reservoir. DBP yields (mg DBP/mg DOC) vary from source to source. Determining the seasonal DBP yields from the outflows of each reservoir would provide information that could help focus control efforts on the most important pollution sources.

Temperature, pH, Specific Conductance, and Dissolved Oxygen. In medium to large natural lakes in the temperate zone with depths greater than 10–15 m, temperature, pH, specific conductance, and dissolved oxygen can be evaluated adequately at biweekly intervals. However, caution should be used when applying these monitoring frequencies to reservoirs because they are subject to more rapid changes than natural lakes. Such rapid changes are most common when the collective reservoir volume is moderate or small in relation to inflows and outflows. In most of the New York City reservoirs, changes in inflow and outflow volume can be large in relation to total reservoir volume. Hence, two-week sampling intervals for these four parameters may not be adequate. In particular, sampling might be increased during the week of autumnal turnover,2 which is often quite predictable and is a time of major chemical redistribution.

Total Suspended Solids, Volatile Suspended Solids, and Turbidity. Total suspended solids, volatile suspended solids, and turbidity are moderately useful metrics to approximate the loading of organic and inorganic particles. Total suspended solids indicate the presence of inorganic and organic particulate matter, while volatile suspended solids reflect organic particulate matter. Both provide information about particle composition that cannot be derived from turbidity measurements.

In most natural lakes, levels of inorganic particulate matter are low in comparison to organic particulate matter. In reservoirs (including the New York reservoirs), however, inorganic particulate matter such as clays is sometimes found in high concentrations, reversing the ratio toward a dominance of inorganic matter. Whether this is true depends upon geomorphology, precipitation events, and other factors. Because levels of total suspended solids, volatile suspended solids, and turbidity are heavily dependent on precipitation events, the biweekly sampling of these parameters that is currently taking place is likely to be of

2  

 Autumnal turnover refers to loss of thermal stratification and complete water circulation.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

marginal operational value. However, the sampling will indicate which of the reservoirs are routinely problematic. The more rigorous monitoring done at the aqueducts is more responsive to the aggregate loadings from all reservoir sources and therefore should be continued.

Odor. It is assumed that odor measurements are precautionary and that odor is generally not a problem within the reservoirs because of low to moderate production of algae and cyanobacteria. Thus, biweekly sampling is adequate at this time.

Nutrients. Chemical analyses, particularly of nutrients, should be performed at frequencies commensurate with changes induced by loadings, biotic utilization and recycling, and losses. In the case of phosphorus, more than 90 percent to 95 percent is found within living and dead particulate organic matter. Soluble reactive phosphorus (SRP) cycles very rapidly (minutes to a few hours), as does much of the soluble organic phosphorus. Hence, the present biweekly measures of total phosphorus and monthly measures of SRP and total dissolved phosphorus (mostly organic) are likely to be adequate for the predictive modeling purposes.

Total nitrogen (inorganic and organic, particulate and dissolved) tells one relatively little in any functional sense because of the complexities of the oxidative–reductive interactions among the different chemical compounds. It is assumed (and it is likely correct, although algal bioassays have not been conducted) that nitrogen is not limiting phytoplanktonic productivity in these reservoirs except when phosphorus loading is excessive and the availability of phosphorus exceeds demand. Therefore, determinations of combined nitrogen (nitrate/nitrite and ammonium ion concentrations) are useful in relation to vertical intensities of bacterial metabolism, rates of hypolimnetic oxygen reduction, anoxia and related problems (such as sulfide production and iron reduction), and potential nitrogen limitation under eutrophic conditions caused by excessive phosphorus loading. Biweekly sampling frequency is generally adequate to follow seasonal changes in stratified reservoirs.

Chemicals. Chloride is a highly conservative ion of relatively minor limnological interest in freshwater inland lakes. Concentrations nearly always exceed biological requirements, they change little either spatially or temporally, and they are not a problem. Depending on the composition of road salt used in the watershed region, however, the reservoirs may have a chloride gradient. In order to assess this, the present sampling schedules are adequate. Bromide is currently not measured in the reservoirs. However, should New York City decide to install ozonation, bromide should be measured on a regular basis in the Kensico Reservoir using the IC method with a detection limit of 10 µg/L or less.

Sulfate ions can become reduced and depleted in the hypolimnia of productive stratified reservoirs, with the production of hydrogen sulfide and related

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

problems. Monthly sampling is likely to be adequate in the least productive reservoirs but should be increased to biweekly in the more productive reservoirs.

Silica (SiO2) concentration is an important parameter because of the dependence of desirable diatom and certain crysophyte algae on this substance. With increasing phosphorus loading and eutrophication, silica concentrations can be reduced to concentrations that are limiting to competitive growth of diatoms (<0.5 mg/L). Reduction of silica concentrations in the euphotic zone during productive thermally stratified periods to limiting levels is an important predictive parameter that should be monitored. Monthly sampling is adequate if silica concentrations exceed 5 mg/L, but sampling should be increased to biweekly when concentrations are less than 5 mg/L.

The present frequency of analyzing for major cations and trace metals appears to be adequate. As mentioned above for stream sampling, major cation concentrations can be estimated quite accurately by correlations with specific conductance to help reduce the frequency of direct analyses.

There is no regular monitoring of toxic chemicals in the water supply reservoirs except for pesticide monitoring that occurs annually at six reservoir locations immediately adjacent to aqueducts. Because of the expense associated with monitoring toxic chemicals in the reservoirs on a regular basis, it has been suggested that as a substitute measure, New York City sample fish tissues for bioaccumulating compounds (ILSI, 1998). The committee concurs that this type of activity should be undertaken, as long as the exposure of fish can be well characterized. Large native fish are highly preferable to stocked species. It should be noted that this type of assay is only effective for bioaccumulating organic compounds and mercury. It is not likely to detect pesticides (most of those used in the watershed do not bioaccumulate) and metals other than mercury. For this reason, the committee encourages the planned monthly sampling of pesticides in the reservoirs, particularly at Kensico Reservoir (NYC DEP, 1997a).

Chlorophyll. Chlorophyll concentrations can function effectively as a general indicator of phytoplanktonic biomass, and they can also be used as parameters for predictive models. The average generation times for algae in nature are 1–3 days, depending on the environmental conditions. Hence, it is reasonable to anticipate significant alterations of phytoplankton growth and productivity within a five-to seven-day period under good growing conditions. More frequent sampling (e.g., seven-day intervals at a minimum) would be an improvement, but for general modeling and evaluation purposes, the biweekly sampling is adequate. Monthly sampling is too long an interval and is not recommended for any reasonable evaluation, even in "nonkey" reservoirs.

Algae. The sampling of algae, cyanobacteria, and zooplankton at the generic levels is sufficient for a general overview of microbial developments, and biweekly sampling is adequate for the evaluation processes under way at this time.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Sedimentation. Since 1993, sediment samples have been collected in the Ashokan and Schoharie reservoirs in order to calculate a sediment mass balance in these reservoirs and to determine particle settling rates. Sediment traps, which are outfitted with between one to three sampling depths, are set early in the year and are collected on a weekly or biweekly basis. Samples are analyzed for total and volatile suspended solids. A sediment trap that can measure resuspension is currently being designed for the Ashokan Reservoir and should be operational soon.

The sediment trap data, as presently collected, will not permit an estimation of a sediment mass balance in the reservoirs. Rather, a mass balance calculation requires sediment loading data from the influents and effluents to the reservoirs collected during storm events. Sedimentation data can be of some use in estimating filling rates, but such accretion data can be obtained more effectively by paleolimnological methods. The current sedimentation data permit only minimal interpretations of sediment accretion over the long term. For these reasons, the present use of sediment traps in the monitoring program has no useful application.

Zebra Mussels. Zebra mussel monitoring occurs at 60 sites in the reservoirs by way of settling plates, water quality sampling, and shoreline and structure inspection. Although they have yet to be detected, should zebra mussels appear, they could cause major water quality problems. New York City's primary goal with its monitoring program is detection of zebra mussels, and its efforts are adequate for this purpose.

Distribution System Monitoring

New York City has two types of sampling sites in its distribution system: (1) compliance sites, which are located on distribution mains equal to or smaller than 20 inches with connections directly serving the public, and (2) surveillance sites from reservoirs, shafts, pumping stations, trunk mains, and wells within the distribution system. Table 6-2 presents the distribution of sites in the boroughs of New York City.

Each month, NYC DEP collects on average 946 compliance samples and 473 surveillance samples, resulting in the sampling of individual sites every 9–14

TABLE 6-2 Sampling Sites in New York City Distribution System

Type

Bronx

Brooklyn

Manhattan

Queens

Staten Island

Total

Compliance

46

70

56

85

29

286

Surveillance

31

27

32

113

19

222

Total

77

97

88

198

48

508

 

Source: NYC DEP (1998b).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

days under normal conditions. Because most chemical parameters would not be expected to change in concentration very rapidly in a system as large as New York City's, sampling every 9–14 days is adequate. In addition to those compounds measured for SDWA compliance purposes (Table 6-1), the distribution system is also sampled for a variety of other important parameters, including E. coli, orthophosphate, total organic carbon, total organic halides, haloacetic acids, additional pesticides, and PCBs, many of which are not yet regulated.

It should be noted that because of their location, the surveillance sites may not be representative of the water actually consumed (unlike the compliance sites). Furthermore, all sampling devices are subject to inspection, predisinfection flushing, and disinfection prior to sample collection—treatment that consumers do not generally provide to their tap water (NYC DEP, 1998b). Such treatment is performed to remove stagnant water from the sampling device and eliminate the possibility of external contamination.

Wastewater Treatment Plant Monitoring

It is apparent from available data that a major source of pollutant loading to the water supply reservoirs emanates from wastewater treatment plant (WWTP) effluents (NYC DEP, 1998c, 1999a). According to 1997 and 1998 end-of-year compliance reports for the individual plants, WWTP effluent in the Catskill/Delaware watershed often contains high (noncomplying) levels of phosphorus, coliform bacteria, and other pollutants. Phosphorus is the primary concern at plants utilizing secondary sewage treatment because these processes, when functioning properly, almost always reduce pathogenic microorganisms and particulate and dissolved organic matter to acceptable levels. However, they are not particularly effective in the removal of nutrients.

At the present time, monitoring of WWTP effluent occurs weekly at all six City-owned plants in accordance with State Pollutant Discharge Elimination System (SPDES) permits and twice monthly via grab samples at all 35 non-City-owned plants. Flow rates are generally measured by the plant operator. Measured parameters include the common physical, chemical, and biological parameters, but do not include metals and other toxic compounds (NYC DEP, 1997a). It has been suggested that plant effluents be monitored for toxic organic compounds and metals on an annual basis (ILSI, 1998). In the committee's opinion, such an activity is expensive and of limited usefulness, given that toxic contaminant input is infrequent and usually occurs in slugs (which can be identified if they cause plant upsets). Such compounds often end up in plant residuals, further limiting the information to be gained by effluent monitoring.

As with many of the monitoring activities previously discussed, monitoring of WWTP effluents should be based on volume discharges rather than on a fixed frequency. Although there is some disagreement about the response of WWTP effluent volumes to storm events, the committee is not convinced that precipita-

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

tion events are unrelated to larger WWTP pollutant loadings because the necessary data to resolve this issue are not available. Depending on the extent of stormwater infiltration into nearby sewer lines, WWTPs could play a significant role in initiating, stimulating, or supporting algal blooms or other polluting events in nearby reservoirs. Only through event-based sampling can the monitoring program accurately assess this possibility.

Microbial Monitoring

The microbial monitoring programs being conducted by NYC DEP (beyond compliance monitoring) are extensive, although they are focused on a few parameters (NYC DEP, 1998b). Total and fecal coliform bacteria, enteric viruses, Giardia, Vibrio, and Cryptosporidium samples are regularly collected at over 50 locations in the watershed, including tributaries, WWTP effluents, reservoirs, aqueducts, and the compliance points for raw and chlorinated water (which are also used for coliform sample collection). The sampling frequency varies from daily to monthly, and attempts are made to monitor at least one storm event per month. Storm event sampling has been automated at some sites located near the discharge of relatively urban subbasins. In addition to routine pathogen monitoring noted above, New York City has also occasionally measured other pathogens in water samples.

Methods. The procedures being used for sample collection and analysis of virus, bacteria, and protozoan samples are in general accordance with methods employed elsewhere in the United States. However, there are some important limitations to New York City's current methodology, especially for the protozoa. The City's standard method to generate (oo)cyst concentration data, which has not changed significantly since sampling began in 1992, is a slightly modified version of the ASTM P229 method and is similar to the ICR method (NYC DEP, 1997d). The City routinely samples approximately 300 gallons of source water to measure (oo)cysts. Whether the entire pellet generated from this sample is examined depends on water quality; however, enough of the pellet is always examined to result in a detection limit of at least 0.7 or less (oo)cysts/100L (Stern, 1998). Overall recovery efficiency is measured quarterly and varies between 30 percent and 70 percent.

Data from the pathogen studies suggest that improvements to the methods are needed. First, the very high number of nondetects makes it difficult to determine characteristic (oo)cyst concentrations emanating from each catchment in the Catskill/Delaware watershed. Ideally, at least 12 monthly nonzero or 24 twice-monthly nonzero measurements are required in order to establish characteristic concentrations for each source. The volumes required to do this may be large. Second, a determination of recovery efficiency should be a routine part of the weekly pathogen sampling in the watershed. Although the 30 percent to 70

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

percent recovery efficiencies measured by New York City are possible given the purity of the source water, other studies suggest that recovery efficiencies are substantially lower (perhaps as low as 5 percent) (Clancy et al., 1994; J. Ongerth, University of South Wales, personal communication, 1999; Shepherd and Wyn-Jones, 1995), and this should be investigated further by NYC DEP. Third, New York City compares counts of ''confirmed" and "presumed" (oo)cysts, based on the microscopic observation of internal cell structures. This method has not been proven to be a true measure of pathogen viability. Finally, in the documentation examined, there is little evidence of determining comparability between laboratories based on the use of split samples (NYC DEP, 1997d; Stern, 1998).

Fortunately, there is ongoing work in methods development for pathogen detection (Clancy et al., 1997; Falk et al., 1998), and New York City should take a higher profile in testing and using these approaches in order to address the deficiencies described above. Some of the new techniques, such as cell culture (Slifko et al., 1997), offer more promise in terms of increased sensitivity, specificity, and assessment of viability. For example, most current viability work on oocysts has used infectivity as an endpoint, and there is a growing sense that tissue culture methods may represent a more realistic approximation of viability. The use of these alternative methods in parallel with existing methods should allow New York City to test more samples with better precision (i.e., reduced variability among split samples) and accuracy (i.e., improved recovery and specificity of viable human infectious organisms). The City's involvement in the Environmental Protection Agency's (EPA) Supplemental Information Collection Rule Survey and evaluation study of the newly proposed Method 1622 for Cryptosporidium should be of considerable benefit in improving recovery efficiencies and increasing the number of positive (nonzero) samples (EPA, 1999).

Source of Pathogens. As discussed in greater detail in Chapter 12, the primary goals of the NYC DEP pathogen monitoring program should be to (1) determine sources of pathogens in the Catskill/Delaware watershed and develop quantitative source terms and compare the relative contributions of different sources or catchments, (2) measure the effect of best management practices on pathogen loadings, and (3) direct resources to the most polluting sources. In support of the first two goals, over 5,000 pathogen samples have been collected to date from watershed sites that are categorized as urban, agricultural, or undisturbed and from the effluent of WWTPs. Viruses have been detected in only five percent of all samples, while Giardia and Cryptosporidium are more common. Giardia is detected most frequently in the effluent of WWTPs (21 percent of all samples), followed by agricultural, urban, and undisturbed areas. Cryptosporidium is found most frequently in urban samples (18 percent), followed by agricultural areas, WWTP effluent, and undisturbed areas. Concentrations of these parameters vary widely, but average about 20 Giardia cysts/100 L and 1 Cryptosporidium oocyst/100 L. Although the data are limited in terms of

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

their usefulness when presented as percent positive, the (oo)cyst concentration data will eventually be needed for evaluation purposes and should continue to be collected.

The pathogen studies have been complemented by attempts to pinpoint actual animal or human sources of pathogens. NYC DEP and Cornell University have investigated both farm animals and wildlife (more than 6,000 specimens) for infection with Giardia and Cryptosporidium. To date, the prevalence of Cryptosporidium in farm animals and wildlife varies between 0.66 percent and 2 percent (NYS WRI, 1997). Research results also indicate that previously infected animals are more likely to become reinfected than previously uninfected animals (NYS WRI, 1997). There are currently no wildlife population data available to assess the relative contributions of different animal source terms to the overall pool of pathogenic protozoa in the Catskill/Delaware watershed. A study to determine the population of mammals with high protozoan infection rates has recently been initiated (NYC DEP, 1999a).

Although progress has been made, current efforts only minimally address the goals described above. In addition to identifying sources, the development of quantitative source terms (e.g., (oo)cysts per animal per time, or (oo)cysts per area per time) is greatly needed, including a determination of the variability of these source terms. Initially such terms could be developed for whole catchments by sampling reservoir effluents, with the eventual goal of developing such terms for urban, agricultural, and wildlife sources. It might be useful for NYC DEP to review and expand upon other attempts to "mass balance" a watershed for protozoa and to quantitatively compare the results in the Catskill/Delaware watershed with other investigations (Hansen and Ongerth, 1991; Ong et al., 1996; Ongerth, 1989).

A first approach to this problem for the Catskill/Delaware watershed has just been published by Walker and Stedinger (1999). The authors utilized local data on WWTP effluent volumes, farming and dairy practices, and dairy populations along with literature values of oocyst levels in sewage effluents and manures and decay rates of organisms under various land management practices. These data were used in watershed runoff models and reservoir water quality models to determine the impact of different oocyst sources on reservoir water quality in the New York City watershed. Although there are a number of data gaps and limitations in their analysis (which could alter the details of their conclusions), the study indicated that the contribution of oocysts in sewage far exceeded the contribution of oocysts in runoff emanating from dairy farms. As additional data are collected, it would be useful for NYC DEP to update and revise this approach and also to use this framework as a method for locating key sampling locations where model validation could occur.

Other Microbial Parameters. It is somewhat surprising that in an ongoing activity of the scope being carried out by New York City, only relatively few

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

microbial parameters are under investigation. There are a number of other pathogens and indicator organisms that should also be considered, as has been suggested previously (ILSI, 1998). E. coli, a more fecal-specific indicator than fecal coliform, should be regularly measured at all locations where fecal coliform is sampled. (E. coli is regularly measured in distribution system samples that are positive for total coliform.) Use of Clostridium perfringens as a water quality indicator, initially developed by Cabelli (1977), may also have merit. The relative insensitivity (Payment and Franco, 1993) of clostridia spores to die-off may make it a most advantageous indicator to track human and animal waste movement in the watershed. It has been used in this regard to track septic discharges in the marine environment (Paul et al., 1995).

The committee anticipates that with the development of techniques founded on modern molecular biology, there will be a number of different and complementary approaches to address source attribution. New York City should keep abreast of these technologies and become early adopters of such methods. In particular, work of Sobsey and colleagues indicates that genotyping F+ coliphage may provide a useful means of distinguishing between animal and human sources of contamination with specific typing indicators (Hsu et al., 1995). NYC DEP has been measuring for fecal streptococci in the Kensico Reservoir to help determine the relative contributions of human and avian sources of fecal bacteria to this important reservoir. It is not clear from available reports (NYC DEP, 1999b) whether the ratio of fecal coliform to fecal streptococci is being calculated to make determinations about bacterial sources. If so, this practice should be discontinued because this ratio is useless (APHA, 1998, pp. 9-74 to 9-75).

Other organisms that should also be considered for microbial monitoring include cyanobacteria capable of producing toxins. In mesotrophic and particularly eutrophic reservoirs, cyanobacterial blooms are common and during these blooms, population growth of other phytoplankton can be strongly suppressed. For example, the hydroxamate sidechrome compounds released by cyanobacteria to scavenge iron have growth-suppressing effects on algae. Certain cyanobacteria, particularly certain strains of Microcystis aeruginosa, produce microcystins and nodularins, a series of some 40 very toxic cyclic heptapeptide compounds. Effects of these compounds in humans are numerous and include gastroenteritis, liver damage, nervous system damage, pneumonia, sore throat, earache, and contact irritation of skin and eyes (Falconer et al., 1983; Carmichael, 1986; Codd et al., 1989; Turner et al., 1990; Harada et al., 1996; Ueno et al., 1996). Cyanobacterial toxins have been implicated in waterborne disease outbreaks in Pennsylvania (Carmichael et al., 1985) and Australia (E1 Saadi et al., 1995; Hayman, 1992; Soong et al., 1992). Although the potential chronic health effects of long-term exposure to cyanobacterial toxins in drinking water are unknown, the World Health Organization has recently proposed a provisional guidance value of 1 µg/L for the microcystin-LR microtoxin for the protection of drinking water (World Health Organization, 1998).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Flow Proportional Monitoring

This chapter has repeatedly stated that fixed-frequency sampling for both chemical and microbial parameters will not capture representative values for pollutant loading from streams, diffuse subsurface sources, and WWTPs. Support for this statement is abundant among limnological and environmental surveying studies. Box 6-1 discusses the inadequacies of fixed-frequency sampling specifically for pathogenic microorganisms. Numerous examples exist for other types of pollutants, particularly those in the particulate phase. Particulate-phase pollutants tend to increase in concentration during storm events because of the increased soil and sediment erosion associated with high flows.

Data collected from the Cannonsville watershed strongly support the superiority of event-based sampling over fixed-frequency sampling. Intensive sampling of the West Branch of the Delaware River during both base flow and storm events demonstrated that stream flow during storm events contains significantly higher concentrations of several pollutants than base flow, including all measured forms of phosphorus, total suspended solids, and total ammonia nitrogen (Longabucco and Rafferty, 1998). In addition to increased pollutant concentrations, storm events were also found to deliver the bulk (from 55 percent to 95 percent) of the annual load of almost all measured parameters. The study demonstrated that sampling based on a fixed frequency would considerably underestimate pollutant loading to the Cannonsville Reservoir via the West Branch of the Delaware River.

The extent of the discrepancy between fixed-frequency and event-based sampling regarding pollutant loadings depends on the particular pollutant, the length of the storm event, and the interval of fixed-frequency sampling. In general, errors in fixed-frequency sampling seem to increase nearly exponentially as the time interval between sampling events increases. The case study in Box 6-2 demonstrates the effect of increasing the frequency of streamflow sampling in three subwatersheds in Connecticut.

State-of-the-art monitoring that includes event-based sampling can substantially improve the effectiveness of watershed management. Fortunately, event-based sampling can be partially automated and coupled with nearly continuous automated monitoring of certain parameters (dissolved oxygen, pH, temperature, conductance, turbidity), although the initial expenses are high. Data from some automated monitoring can be downloaded by radio transmission to central processing stations. However, in other cases, collection and analysis of samples from proportional collectors requires personnel maintenance based on precipitation and snowmelt periods, which are unpredictable. A commensurate database management system is needed to make full and timely use of new, more detailed information about water flow and quality. Otherwise the volume of data generated by automated equipment can quickly overwhelm watershed managers and simply be archived instead of being actively used to support field operations. In

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

BOX 6-1
Storm Event Sampling of Pathogenic Protozoans

At the Metropolitan Water District of Southern California, use of innovative grab and "first flush" samplers was found to yield a substantially greater frequency of positive hits than the current Information Collection Rule (ICR) methodology (see Table 6-3). Although some of this may be due to the poor recovery associated with the classical ICR methods (Clancy et al., 1994), a substantial cause of this difference may be due to the "spiky" nature of oocyst (and cyst) prevalence in surface waters, particularly during storm events.

This finding is further supported by work of two other research groups. LeChevallier et al. (1997) conducted an intensive sampling campaign on a watershed and reported relationships between flow rates, turbidity

TABLE 6-3 Results of Different Types of Sampling Methods on Isolation Frequency for Protozoans in Water Samples

Protozoan

Sampling Method

Total

Filter (ICR)a N=87

First flushb N=20

Grabc N=21

% Positive

Ranged

% Positive

Ranged

% Positive

Ranged

Crypto

10

3–415

35

46–41,666

19

3.4–647

16

Giardia

29

2–119

60

25–16,666

19

42–2,428

32

Crypto or Giardia

28

 

45

 

19

 

38

Crypto and Giardia

6

 

25

 

10

 

9

a Sampled using current EPA methods (large-volume samples) during normal meteorological and flow conditions.

b Small-volume samples were collected during the periods of rapid initial flow increase following storm events.

c Small-volume samples were collected during normal meteorological and flow conditions.

d Range measured in 100 (oo)cysts /100L.

Source: Stewart et al. (1997). Reprinted from Proceedings of the 1997 AWWA Water Quality Technology Conference, by permission. Copyright © 1998, American Water Works Association.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

spikes, and peak levels of oocysts. As shown in Figure 6-1, peak oocyst levels may be found in proximity to peak flow and turbidity levels, in at least certain watersheds. Findings of protozoan peaks during wet weather periods are in accord with well-established findings with respect to indicator organism occurrences in runoff flows (Geldreich, 1978).

The "flashiness" of oocyst levels indicated by these two investigations shows that for the purposes of estimating the total loading of oocysts (number/day) to a waterbody, or estimating the total flux past a location, it is necessary to obtain samples during storm events (event-based sampling). For example, had samples not been collected during days 107–109 in Figure 6-1 (the storm event period), an underestimate of the maximum range of oocyst levels (and thus the flux) would have been made.

FIGURE 6-1 Variation in turbidity, river flow, and Cryptosporidium levels during a spring sampling campaign. Boxes highlight days on which samples were tested for Cryptosporidium. Source: LeChevallier et al. (1997). Reprinted from Proceedings of the 1997 AWWA Water Quality Technology Conference, by permission. Copyright © 1998, American Water Works Association.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

BOX 6-2
Integration of Fixed-Interval and Event-Based Water Quality Sampling: An Example from Current Watershed Research

This case study provides an example of how rapid advances in field, laboratory, computing, and communications equipment during the last decade have transformed the idea of event-based watershed monitoring from a logistically and financially impossible dream to an operationally feasible, albeit challenging, goal. This example shows part of the water quality sampling system used for the New Haven Watershed Project, which has a hydrogeologic setting and climate and watershed characteristics roughly comparable to the watersheds of the New York City water supply system.

Table 6-4 illustrates the effect of sampling frequency on the information content of water quality measurements. Stream sampling of three subwatersheds (rural, intermediate, and urban) was accomplished using an automated probe connected to a data logger. Four parameters—discharge, turbidity, dissolved oxygen, and nitrate-nitrogen—were compiled at 20 minutes, 24 hours, 7 days, and 14 days. The original 20-minute time interval was resampled to simulate the daily, weekly, and biweekly data collection.

As expected, the mean, standard deviation, maximum, and minimum reflect differences in the three classes of watershed and sampling frequency. Most notable are the maximum values for discharge and turbidity for all three sites. At the 20-minute time interval, discharge from the urban watershed exhibits the flashy response commonly associated with impervious cover. However, the observed response is an order of magnitude lower when data are collected at 7- and 14-day intervals. Similar trends are exhibited by turbidity data for all three subwatersheds.

Also note the difference between decreases in mean and maximum values as the sampling interval becomes larger and sample size is reduced (N = 1,800, 25, 4, and 2 for 20-minute, 24-hour, 7-day, and 14-day sampling intervals, respectively). If, by chance, the rising limb of a major storm event was sampled during routine biweekly data collection, it is likely to be eliminated as an outlier or influential case during subsequent statistical analyses. Turbidity is directly correlated with suspended particulate matter including pathogens and nutrients adsorbed to sediment particles. Hence, the ability to detect and respond to water quality problems may be seriously limited by the present monitoring system.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

TABLE 6-4 A Comparison of Stream Flow and Selected Water Quality Data (June 4–29, 1998) at Four Sampling Frequencies for Three Subwatersheds of the Quinnipiac and West Rivers near New Haven, Connecticut

Sampling Interval

Stats

Discharge (mm/day)

Turbidity (NTU)

Rural

Inter.

Urban

Rural

Inter.

Urban

20 mins.

Mean

1.6

1.8

4.2

2.4

8.2

7.2

 

SD

0.6

0.8

2.3

4.4

9.5

3.8

 

Max

5.2

6.1

38.1

80.9

93.6

99.3

 

Min

0.8

0.9

1.6

0.6

0.6

3.8

24 hours

Mean

1.6

1.9

4.4

1.9

7.7

7.0

 

SD

0.6

0.9

2.6

1.2

7.8

1.6

 

Max

3.0

4.1

14.3

6.0

37.9

11.1

 

Min

0.9

1.1

2.2

0.8

1.0

5.0

7 days

Mean

1.4

1.5

4.3

1.8

4.6

6.8

 

SD

0.4

0.2

1.7

0.7

3.4

1.6

 

Max

1.7

1.7

6.8

2.6

8.0

8.2

 

Min

1.0

1.4

3.0

1.0

1.2

5.1

14 days

Mean

1.68

1.6

5.4

1.4

4.6

6.9

 

SD

0.02

0.1

1.9

0.6

4.8

1.8

 

Max

1.69

1.7

6.8

1.8

8.0

8.2

 

Min

1.67

1.5

4.1

1.0

1.2

5.7

Notes: General land cover statistics: RURAL subwatershed (#1050, 5.00 km2) is 91 percent forest, 4 percent developed, 3 percent fields, and 2 percent water and wetlands; INTERMEDIATE subwatershed (#2020, 5.21 km2) is 70 percent forest, 20 percent developed, 7 percent fields, and 3 percent water and wetlands; URBAN subwatershed (#1005, 3.92 km2 is 1 percent forest, 85 percent developed, 6 percent fields, and 8 percent water and wetlands. The 20-minute interval data were generated with automated probes. The 24-hour, 7-day, and 14-day data were resampled from the 20-minute interval set.

addition, a backup methodology is needed because failure of automated systems can cause gaps in data needed for predictive modeling. For these reasons, flow proportional monitoring can require a substantial investment of personnel and monetary resources.

Samples collected by flow proportional monitoring must be properly preserved and retrieved in accordance with the vulnerability of the assays being

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

D.O. (mg/L)

NO3–N (mg/L)

Rural

Inter.

Urban

Rural

Inter.

Urban

7.3

9.0

5.7

0.19

1.15

1.24

1.3

1.4

1.8

0.05

0.29

0.23

11.3

11.4

11.0

0.27

1.53

1.70

5.1

6.4

2.9

0.11

0.47

0.74

7.2

9.1

5.3

0.11

1.16

1.25

1.3

1.4

1.6

0.05

0.32

0.24

8.9

10.9

8.7

0.27

1.53

1.70

5.2

6.6

3.2

0.11

0.47

0.74

7.3

9.2

5.6

0.19

1.28

1.27

1.4

1.7

2.2

0.06

0.17

0.19

8.4

10.9

8.8

0.27

1.51

1.42

5.4

7.0

3.5

0.12

1.13

1.00

7.6

9.8

4.4

0.14

1.15

1.13

1.1

1.5

1.2

0.03

0.02

0.18

8.4

10.9

5.3

0.16

1.16

1.25

6.9

8.8

3.5

0.12

1.13

1.00

performed. For example, samples collected for phosphorus analysis should be stored on ice and processed for analyses within 8–12 hours of collection. The use of flow proportional sampling for microorganisms (indicators such as coliforms and pathogens such as Cryptosporidium) is technically feasible with the use of refrigerated field samplers. However, there are a number of logistical and economic issues that may limit the applicability of this approach. For example, the

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

storage of samples for more than 24 hours is not generally acceptable, and so field samplers would need to be serviced on a daily basis. Sources of electric power (for refrigeration) for the sampler would be required. There are approaches that have been used for "first-flush" collection devices that may be attractive for use at a limited number of sites (Stewart et al., 1997).

It is suggested that fixed-frequency sampling be continued for a reasonable length of time (minimally one year) after the proportional sampling system is operational. By this means, interpretation of old fixed-frequency sampling data can be analyzed for evaluations of comparative reliability. In some cases, where the relation between discharge and the water quality constituent is well established, it may be reasonable to measure some parameters at a longer time interval. Once a sufficient concurrent record of fixed-frequency and event-based monitoring exists, a wide range of advanced statistical techniques can be used for retrospective modeling and analyses of historical data (Clarke, 1994; Haan, 1977; Hosking and Wallis, 1997).

For some time, NYC DEP has recognized the need to include event-based sampling in their water quality monitoring program (NYC DEP, 1997a), and its importance has also been repeatedly stressed by other sources (ILSI, 1998; Longabucco and Rafferty, 1998). Event-based sampling is on the rise in the streams of the New York City watersheds. Between April 1995 and April 1996, storm event sampling was conducted in a tributary of the Cannonsville Reservoir to determine nutrient loading for calibration and validation of water quality models. Samples from 18 storm events were collected and analyzed. In 1996, turbidity and suspended solids were measured during four storm events in the Ashokan basin, including the Esopus Creek, which is known to contribute to turbidity in the Catskill System. This increased to sampling eight events at 25 sites in 1998. To support further development of reservoir models, storm event sampling for nutrients is being conducted in all of the major tributaries west of the Hudson River. Finally, streams in the Kensico Reservoir basin are being sampled intensively for coliforms and turbidity, as mentioned below under the special section on the Kensico Reservoir. It is hoped that the efforts described above will lead to the rapid establishment of automated (when possible) event-based sampling, especially for the Kensico watershed and for those reservoir basins with impaired water quality (such as Cannonsville).

Performance Monitoring

In many water supply systems, compliance and operational monitoring make up the vast majority of all monitoring activities. However, in large, complex systems in which watershed land use varies, additional monitoring is needed to determine the effectiveness of management practices and policies for reducing nonpoint source pollution. For the purpose of this report, this additional monitoring is roughly classified as performance monitoring.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Performance monitoring is usually conducted on a limited basis to determine the effectiveness of a practice or a technology. It often consists of multiple data sets that are analyzed by conducting a mass balance around the particular practice or area of interest. Performance monitoring can occur over a wide range of scales, from a single farm best management practice (BMP) to an entire reservoir or watershed. Because of natural variability and the time lag between BMP implementation and system response, substantial time may pass before performance monitoring can detect changes in water quality around large (watershed-scale) areas. Performance monitoring results accrue more quickly as the monitored area diminishes in size. The types of performance monitoring that are most important for the New York City watersheds are discussed below, with indications as to whether this monitoring is taking place. Suggestions for improved performance monitoring are found in this chapter and in the discussion of individual programs in Chapters 810.

Shallow Subsurface Monitoring

Perhaps the most important type of performance monitoring for a source water protection program is monitoring of shallow subsurface flow. This monitoring should occur in multiple locations throughout the watershed, but it is particularly important in the riparian zones surrounding the reservoirs and major tributaries. These areas are critical for sequestering nonpoint source pollution from stormwater runoff and are often the sole barrier of protection to nearby waterbodies. The goal of this performance monitoring is to determine the retention efficiencies of the shallow subsurface for a variety of pollutants. Such analyses should be performed under a spectrum of differences in soil types, vegetation cover, precipitation, elevational gradients, and topography. In addition to riparian zones, performance monitoring of the shallow subsurface should be conducted around on-site sewage treatment and disposal systems (OSTDS) and their drainfields, above and below certain agricultural BMPs, and above and below urban stormwater BMPs.

Although some performance monitoring of the shallow subsurface has occurred in the New York City watersheds, it has been relatively infrequent. Given data on groundwater flow rates in the Kensico watershed, there is a general assumption that subsurface flows and their attendant pollutant loadings to the reservoirs are minor in comparison to surface (river) runoff. Such assumptions are likely invalid given quantitative studies in regions of similar geology and topography [e.g., Mirror Lake ecosystem of Hubbard Brook, NH, studies (Likens, 1985; Likens and Bormann, 1995)]. These studies, which are based on real data, not on models, have shown that composite pollutant loadings cannot be based on river discharge volumes alone.

To our knowledge, ongoing performance monitoring of the shallow subsurface is associated with five main activities in the New York City watersheds.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Eight of the groundwater monitoring wells in the Kensico watershed are being used to determine whether exfiltration from sewer lines is affecting water quality in the shallow subsurface (NYC DEP,1999a). Monitoring of total and fecal coliforms from these wells has not detected significant increases in these parameters over other groundwater wells in the watersheds, suggesting that exfiltration of sewage is not important. The Septic Siting Study, which is designed to determine the fate and transport of pathogens introduced into septic systems, is measuring pollutant concentrations in septic system drainfields and control plots at six locations (two in each watershed). A more detailed description of this MOA-mandated study is found in Chapter 11. Researchers at the U.S. Geological Survey are determining how pollutant loadings in shallow subsurface flow respond to a complete removal of forest vegetation in a small tributary of the Neversink. Cornell University research in support of the Watershed Agricultural Program has measured phosphorus concentrations in shallow subsurface affected by manure application. And finally, performance monitoring of wetlands east of the Hudson River (in the West Branch and Boyd's Corner watersheds) is planned for the near future (NYC DEP, 1999a). Phosphorus, suspended solids, dissolved organic carbon, and other parameters will be measured in the inflow and outflow of 9–15 reference wetlands and in the shallow subsurface. These activities should continue and be supplemented by more wide-scale performance monitoring of riparian buffer zones, septic system drainfields (particularly in the Kensico watershed), and the shallow subsurface above and below agricultural and urban stormwater BMPs.

Other Performance Monitoring

The shallow subsurface is not the only important target of performance monitoring. Any individual BMP is a logical candidate for performance monitoring, particularly BMPs that have not been used extensively or for which there are no published pollutant removal rates. The 44 urban stormwater BMPs that are being installed in the Kensico watershed (described in Box 6-3) should undergo such performance monitoring subsequent to their installation. Performance monitoring can also be conducted on a larger-scale, in order to assess the ''performance" of a region or subwatershed. This larger-scale strategy is being used by NYC DEP in monitoring agricultural areas. Two locations, the Robertson farm and Shaw Road (a control site), are undergoing stream monitoring to compute the difference in pollutant loading emanating from these sites. Similarly, a comparison of conditions at six pristine or disturbed wetland sites located partially within the Croton watershed is being conducted to determine the relationship between macroinvertebrate communities and wetland characteristics (NYC DEP, 1999a).

Performance monitoring does not necessarily require the acquisition of new data. In some cases, data collected during compliance and operational monitoring can be used to determine the effectiveness of a pollution reduction scheme.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

BOX 6-3
The Kensico Reservoir and Watershed

About 90 percent of the total New York City water supply is derived from the City's six West-of-Hudson reservoirs in the Catskill Mountains and the Upper Delaware River Basin (the Catskill/Delaware system). Water from those reservoirs is delivered via the Catskill and Delaware aqueducts, which cross beneath the Hudson, to a common destination at Kensico Reservoir, 15 miles north of New York City. There, water from both aqueducts mixes and, after 15–25 days of storage time, it is chlorinated and delivered to the City and other user communities. The state of water quality in Kensico Reservoir is thus critical: if it becomes degraded, it would contaminate the high-quality water derived from the other Catskill/Delaware reservoirs. Although both aqueducts are designed to permit bypassing of Kensico, bypassing is undesirable because the settling time, mixing, and flow control that Kensico provides would be lost.

Kensico Reservoir, with a capacity of 30 billion gallons, was placed in service in 1915 with the damming of the Bronx River. Although most water in Kensico Reservoir is delivered by the two trans-Hudson aqueducts, it also receives runoff from its immediate drainage area of 13.1 square miles. Largely forested, this area contains about 1,500 dwelling units, a number of office parks and other commercial facilities, as well as transportation facilities. The latter include Westchester County Airport, Interstate 684, and state highway 22, which borders the reservoir and crosses it at one point. Stormwater from each of these facilities and structures drains into Kensico.

Recent violations in stormwater runoff quality from the airport, the proposed widening of Routes 22 and 120, and new office construction have made the Kensico watershed the subject of much discussion (Marx and Goldstein, 1999). These projects pose serious threats to water quality in Kensico Reservoir if stormwater management is not undertaken to counteract associated increases in pollutant loading. In particular, wetlands bordering the reservoir will likely be removed to accommodate roadwidening, thereby reducing the ability of the surrounding buffer land to diminish pollutant loading from stormwater. If such projects go forth, stormwater management must (1) control all new pollutant loadings, (2) compensate for losses in nonpoint source pollutant control via wetland functioning, and (3) mitigate current problems with runoff from the Westchester County Airport.

To confront these challenges, NYC DEP has developed the Kensico Stormwater Management Plan, which currently includes 44 stormwater facilities in 16 subbasins to reduce loadings of turbidity and fecal coliform bacteria (and associated pollutants) to the reservoir. The stormwater

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

BMPs will be constructed first in the vicinity of Malcolm Brook and Young Brook, which are major tributaries to the reservoir near the Catskill aqueduct effluent chamber. All facilities should be operational by the year 2000. Performance monitoring of these BMPs and the shallow subsurface around the airport and the proposed road widenings should be one of NYC DEP's highest priorities.

In addition to stormwater runoff control, several other preventive measures are being taken to reduce the impact of watershed activities on Kensico reservoir water quality. An 850-ft turbidity curtain was installed in 1995 at the mouths of Malcolm and Young brooks to prevent direct transit of sediment to the Catskill aqueduct effluent chamber. So far it has been effective in preventing water at that effluent chamber from exceeding 5 NTU. As mandated by the MOA, maintenance dredging is occurring in the Kensico Reservoir to remove accumulated sediments near both aqueduct effluent chambers. Approximately 980 cubic yards of sediment will eventually be removed, dewatered, and shipped to a facility in Ohio.

Although there are no wastewater treatment plants in the Kensico watershed, the area is crossed by various sewer lines leading to treatment plants elsewhere. NYC DEP is conducting video inspection of sewers to find and repair leaks. As described elsewhere in this report, groundwater monitoring has not detected any exfiltration of sewage from these pipes to the surrounding subsurface. There are approximately 580 septic systems in the watershed, of which 210 lie within 300 ft of the reservoir or 100 ft from tributary streams. NYC DEP, as mandated by the MOA, is responsible for detecting failing septic systems within the watershed and in assisting those using septic systems to connect to sewers where feasible.

For example, operational monitoring data could be used to determine the reduction in pollutant concentration between the influent and effluent of an entire reservoir. Although there appears to be a large database taking shape that would permit this type of analysis, the only known attempts have been conducted for the West-of-Hudson reservoirs using percent detection data instead of concentration and flow data (Bagley et al., 1998; Stern, 1996). If NYC DEP wants to conduct this type of large-scale mass balance with accuracy, data need to be collected over long time periods and during storm events to reduce both variability and the risk of systematic sampling bias.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Because of the serious consequence of hazardous waste spills within the watershed, NYC DEP has developed a Kensico Spill Response Program. Four incidents in 1998 prompted action under the program; none were found to negatively affect water quality (NYC DEP, 1999a). Part of the success of this program is dependent on alert residents who notified NYC DEP of potential problem activities. Citizen involvement in monitoring and preventing pollution in the watershed is the backbone of the Kensico Environmental Enhancement Program (KEEP), a joint effort of NYC DEP and community groups. In 1998, KEEP held workshops on environmental protection, it patrolled the reservoir and its perimeter, and it distributed educational information to local residents suggesting how they can help protect water quality in the Kensico Reservoir.

These activities and others (such as the gull mitigation efforts discussed in Chapter 5) contribute to high water quality in the Kensico Reservoir and subsequently in the finished drinking water. This is demonstrated by the fact that the Kensico Reservoir has not been bypassed since 1994 (Delaware aqueduct) and 1995 (Catskill aqueduct). Because West Branch Reservoir is similarly positioned within the Catskill/Delaware system, these special types of activities should be considered for possible implementation in the West Branch watershed as well. Under normal conditions, water from the Delaware aqueduct (55 percent of the total City supply) passes through the West Branch Reservoir before reaching the Kensico Reservoir. As discussed in Chapter 2, population (and, by association, development) is increasing more rapidly in the West Branch watershed than in the Kensico watershed, highlighting the importance of extra protection efforts for West Branch.

Monitoring in Support of Modeling Efforts

Water quality parameters vary significantly among reservoirs and their tributaries. Spatial variations arise from differences in the drainage basin characteristics of physiography, land use practices, and meteorology. In addition, water quality within individual reservoirs is uncertain because of variable basin morphometry, natural and managed hydrology, climatic conditions, and fluctuations in pollutant loadings. As noted earlier, temporal variations in water quality typically occur over small time scales. One of the purposes of water quality

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

modeling is to follow and predict these patterns of spatial and temporal variability in water quality parameters. If successfully designed, calibrated, and verified, water quality models can lead to reduced monitoring requirements over the long term.

New York City has developed and is using a large number of terrestrial runoff and water quality models for both the Catskill/Delaware and Croton reservoirs. Table 6-5 lists some of the models being used by New York City and the data requirements for each. The table includes models of reservoir water hydrodynamics and quality, terrestrial models of surface water runoff and nonpoint source pollution, and precipitation and atmospheric deposition models. A long-term goal of these NYC DEP modeling efforts is to link models representing different processes into one integrated model of watershed functioning. It should be noted that pathogen models and agricultural models (as suggested earlier in this chapter) are not yet being developed to interface with the current runoff and water quality models. Development of pathogen fate and transport models will be critical to moving beyond the simplified presence/absence evaluation that characterizes the NYC DEP pathogen studies.

This report does not review and critique each water quality model being used by New York City. However, when a model is being used to comply with a regulatory program, the quality of the data used as input to the model must be taken into consideration. Thus, the report addresses the quality of the data required for calibration and verification of the Generalized Watershed Loading Function (GWLF), the Vollenweider model, and the Reckhow model, all of which are used for the Total Maximum Daily Load (TMDL) Program. Because a significant modeling effort is necessary to calculate TMDLs, the data acquired to support these models must be of sufficient quantity and high quality. In addition, the report reviews the hydrothermal and nutrient/phytoplankton models being considered for possible use during the Phase III TMDL process. These discussions are found in Chapter 8.

Monitoring of the Kensico Watershed

As the collection point for Catskill/Delaware system water prior to chlorination and distribution, the Kensico Reservoir is critical in controlling the quality of the New York City drinking water supply (Figure 6-2). The residence time afforded by travel through the Kensico Reservoir allows particulate phase pollutants to settle, and it supports greater degradation or inactivation of microbial pathogens. Because polluted runoff and other contamination from the Kensico watershed (and to a lesser extent the West Branch watershed) has the potential to adversely affect the quality of water emanating from the six West-of-Hudson reservoirs, a significant fraction of all monitoring efforts and pollution prevention activities are conducted in Kensico. Box 6-3 describes this unique reservoir and watershed in more detail, discussing the special activities being undertaken to

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

TABLE 6-5 Selected Water Quality Models Used in New York City

Model Name

Model Type

Data Requirements

Reckhow Model

Surface Runoff Water Quality

• Stream and WWTP flows

• Average pollutant concentrations in stream flows (for validation), WWTP effluents

• Land cover/land use

• Number of septic systems

• Export coefficients for various nonpoint sources

• Soil retention coefficient

Generalized Watershed Loading Function (GWLF)

Surface and Subsurface Runoff Water Quality

• Precipitation

• Air temperature

• Stream and WWTP flows

• Average pollutant concentrations in stream flows, WWTP effluents

• Septic system pollutant loads

• Land cover/land use

• Mean elevation and slope

• Soil chemistry data

Vollenweider Model

Reservoir Water Quality

• Pollutant concentration in reservoir

• Retention rates for pollutants in reservoir

• Reservoir residence time

• Mean reservoir depth

Hydrothermal Model

Reservoir Water Quality

• Meteorological dataa

• Geometric data (reservoir topography)

• Stream flows, all outflows, and water surface elevation

• Reservoir temperature profiles with depth, stream temperatures, aqueduct temperature

• Hydrodynamic and kinetic coefficients

• Structural details

Eutrophication Model (also known as the nutrient/phytoplankton model)

Reservoir Water Quality

• Meteorological dataa

• Stream flows, all outflows, and water surface elevation

• Reservoir temperature profiles with depth, stream temperatures, aqueduct temperature

• Light extinction

• Average pollutant concentrations in the epilimnion and the hypolimnion, in stream flows, and in outflows

a Meteorological data include air temperature, dew point temperature, wind velocity, and cloud cover (or solar radiation).

Sources: Doerr et al. (1998), NYC DEP (1998d), and Owens (1998).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

FIGURE 6-2 Kensico Reservoir and watershed. Courtesy of the NYC DEP.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

identify, quantify, evaluate, and mitigate sources of chemical and biological pollution within the Kensico watershed that affect its water quality.

Monitoring in the Kensico basin is intensive and frequent and encompasses some 35 physical, chemical, and biological water quality parameters. A significant quantity of all compliance and operational monitoring for the Catskill/Delaware system occurs in the Kensico watershed because of its proximity to the distribution system. Performance monitoring also takes place at a high frequency in the Kensico watershed in comparison to the other New York City watersheds. For example, Kensico is the only watershed in which groundwater monitoring is conducted on a regular basis. Table 6-6 lists all of the special monitoring activities occurring in the Kensico watershed beyond the compliance monitoring required by the Surface Water Treatment Rule (SWTR) and the standard operational monitoring that takes place in all Catskill/Delaware reservoir basins.

Given its influence on the quality of drinking water in New York City, Kensico Reservoir and its watershed should continue to be the focus of intensive monitoring by NYC DEP. As discussed in Box 6-3, the most serious local water quality threat to the reservoir appears to be pollutant-laden stormwater from roads, bridges, and the Westchester County Airport. Monitoring the performance of urban stormwater BMPs in the vicinity of these pollution sources should be one of NYC DEP's highest watershed management priorities.

Conclusions and Recommendations for the Monitoring Program

The complexity of the multiple interacting reservoir ecosystems of the New York City water supply imposes major monitoring demands to allow for effective management and responses to problems. In general, NYC DEP has been performing these formidable tasks excellently. Several recommendations for improvement are made below.

  1. Monitoring should be conducted on the basis of discharge-mediated volume rather than on fixed-frequency intervals for stream, shallow subsurface groundwater, WWTP effluent, and precipitation analyses. Eventbased or flow proportional sampling is needed to capture rapid variation in water flow and quality. Some parameters of sampling can be automated, but care is needed to avoid problems associated with storage of samples collected by flow proportional sampling.

  2. Shallow subsurface and groundwater parameters should be monitored regularly throughout the reservoir watersheds. This is particularly important given the prominent role of agriculture in some areas, the high density of OSTDS, and the potential for leaking sewer lines through the watersheds. Biannual sampling is sufficient for deep groundwater but not for shallow subsurface monitoring. Such routine groundwater sampling should be integrated with

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

TABLE 6-6 Special Monitoring Efforts in the Kensico Watershed Beyond Regular Compliance and Operational Monitoringa

Type of Monitoring

Parameters Measured

Frequency of Sampling

Reservoir

• Sediment cores and sediment samples: metals, pesticides, polyaromatic hydrocarbons, radioisotope dating

• One-time events in several locations

 

• Fecal streptococci

• NA

Stream

• Flow

• Event-based

 

• Fecal coliforms

 

 

• Protozoan pathogens

 

 

• Turbidity

 

 

• Suspended solids

 

 

• Total phosphorus

 

 

• Sediment and river water sampled for pesticides and hydrophobic organic compounds

• June 1997, June 1998

Groundwater/ Shallow Subsurface

• Total and fecal coliforms

• Biannual

 

• Physical parameters

 

 

• Nutrients

 

 

• Synthetic organic compounds

• One time event with followup

Wetlands in the Malcolm Brook Subwatershed

• Dye test study to determine travel time through wetlands

• Biannual

Bird Populations and Bird Fecal Matter

• Fecal coliforms

• Weekly

 

• Fecal streptococci

• NA

Forest Regeneration Study

• Seedling survival

• Biannual between 1995 and 1997

Notes: Entries indicate both aquatic and terrestrial monitoring efforts beyond those made for the other basins in the Catskill/Delaware system.

NA = not available

a Compliance monitoring is indicated in Table 6-1 and operational monitoring is discussed previously in the chapter.

Source: NYC DEP (1997a).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

direct experiments (performance monitoring) on the efficacy of OSTDS and riparian buffer zone development and management.

  1. With the exception of dissolved organic carbon, the analytical methods employed for physical and chemical parameter monitoring in both reservoirs and streams are generally adequate. Some additions to the evaluations of dissolved organic matter are suggested. The quality assurance/quality control methods are generally adequate for chemical and biological limnological parameters. They should be rigorously maintained, since they are critical for long-term analyses and continuity. If methods are changed or improved, both the old and new techniques should be performed simultaneously on the same samples for a period (e.g., two years) in order to evaluate the comparability of old and new data sets.

  2. Although methods currently used for pathogen monitoring are acceptable, New York City should take a more active role in the development and use of new and improved methods for pathogen detection. The goal of these efforts should be to achieve significantly higher numbers of nonzero measurements for (oo)cysts, to assess overall recovery efficiency on a weekly basis, to determine viability, and to build into the routine sampling plan split samples for these developing protocols. Methods for pathogen monitoring will change in the future, and the use of split samples will enable future analysts and managers in accurately interpreting historical data.

  3. The Pathogen Studies should have as a goal the estimation of source terms [e.g., (oocysts/acre)/day or (oocysts/calf)/day] for various catchments, animals, agricultural and urban activities, and farm waste management. The Pathogen Studies as currently conducted (and expressed as percent detection from different land uses) do not obviously lead to a greater understanding of pathogen dynamics at the watershed level. In this regard, NYC DEP could benefit by assimilating work done on pathogen inactivation under variable environmental conditions and by making use of approaches to watershed modeling developed for nonpathogens (e.g., nutrients).

  4. Additional microbial parameters should be considered for inclusion in routine water quality monitoring programs, in particular E. coli, coliphage, Clostridium perfringens, and cyanobacteria that produce toxins. Coliphage may be particularly useful for source typing in reservoirs and streams and for assessing potential groundwater contamination by human fecal pollution emanating from leaking sewerage. Clostridium perfringens can be used to assess the potential viability of protozoa in surface waters distant from the point of contamination. Fecal streptococci or enterococci may be useful as supplements

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

to coliforms as indicators of pollution. However, the outmoded and ambiguous ratio of fecal coliform to fecal streptococci should not be used.

  1. If NYC DEP wants to determine the effectiveness of managed riparian buffer zones, septic system drainfields, and agricultural and urban stormwater BMPs, performance monitoring using paired measurements is needed. The Septic Siting Study is an example of this type of monitoring currently under way in the watershed. Performance monitoring of septic systems and new urban stormwater BMPs is strongly recommended in the Kensico watershed. As suggested in Chapter 9, performance monitoring of agricultural BMPs for pathogen and nutrient removal would greatly enhance the Watershed Agricultural Program.

  2. Monitoring and other special pollution prevention activities in the Kensico Reservoir and watershed should continue at their present high intensity given the importance of this basin in controlling Catskill/Delaware drinking water quality. The goal of these activities should be to demonstrate no net increase in pollutant loading to the reservoir in the face of future development, road expansion, and airport use. Similar measures should be considered for the West Branch Reservoir and watershed, which is experiencing population growth at a rate greater than that in the Kensico watershed.

GEOGRAPHIC INFORMATION SYSTEMS

A Geographic Information System (GIS) is a relational database that supports mapping and spatially distributed modeling. Such systems can store data on a wide variety of watershed characteristics, including topography, waterbodies, roads, vegetation and soil type, land use, and such discrete points as buildings, wells, and septic systems. GIS also allows new data sets (e.g., flow path) to be derived from available data sets (e.g., slope), which is a valuable function for mapping of watershed areas and for a variety of NYC DEP watershed programs.

Over the past several years, NYC DEP has developed a GIS program at its Valhalla, NY, office with network links to field offices. Primary data layers have been obtained from the U.S. Geological Survey National Mapping Center or have been developed under contract. NYC DEP has the required hardware, software, and, most importantly, the staff expertise to make full use of state-of-the-art GIS capabilities to support the complex management actions included in the MOA. In addition, water quality monitoring data can be analyzed using the GIS to determine if management strategies are meeting their objectives.

Notwithstanding its considerable accomplishments to date, there are opportunities within NYC DEP to improve the application of GIS to watershed monitoring and management. As currently applied, land cover (e.g., forest, grassland, and wetland) and land use (e.g., residential, pasture, or row-crops) data, compiled

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

from Landsat imagery, are combined in one layer. This limits the ability of analysts and managers to differentiate between areas with similar appearances (spectral reflectance in a satellite image) but markedly dissimilar uses and pollution potential (e.g., alpine meadow, hayfield, pasture, or lawn). Additional field and integrated image analyses are needed to refine the characterization of human use and potential impacts. Such analyses could employ more Global Positioning System (GPS) data as well as conventional aerial photographs, color infrared aerial photographs, digital orthophotography, and multiseason/multispectral satellite imagery. Land cover/land use data are currently used to support several watershed management programs, including the Land Acquisition Program and the TMDL Program.

GIS is a particularly cost-effective means of evaluating wastewater treatment options. Soil, terrain, and flow path data can be used to assess the suitability of areas for residential development with OSTDS. Similarly, the location and spatial pattern of existing or potential residential development, again combined with other watershed features, can be used to assess the feasibility and costs of centralized wastewater treatment, package plants, or advanced OSTDS. A cooperative study by NYC DEP and the Catskill Watershed Corporation (CWC), fully supported by the GIS group, would greatly advance this key task.

As part of watershed planning and management efforts, citizens need to be more engaged. Because the debate about how to manage land is more substantive and productive when all stakeholders have the same information, the GIS database should be accessible to the public. A web site with ''read-only" layers, a database directory, and basic documentation would enable interested residents to study watershed characteristics and formulate ideas and alternatives.

Conclusions and Recommendations for the GIS

  1. In light of its central importance to watershed modeling and management, land cover and land use data should be separated, refined, and regularly updated. As currently operated, NYC DEP's GIS does not differentiate between these two types of data sets.

  2. Management of domestic wastewater should be fully supported by spatially distributed modeling and GIS. A cooperative effort involving NYC DEP, CWC, and state and county health departments would advance this goal.

  3. The GIS database should be made available, perhaps via a web site, to interested citizens, communities, nonprofit organizations (e.g., CWC, Catskill Center for Conservation and Development, and many others), and scientists (e.g., the Catskill Institute for the Environment). GIS maps, which can display varying degrees of complexity, can illustrate how individual actions might affect downstream water quality.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

DISEASE SURVEILLANCE AND PUBLIC HEALTH PROTECTION

Water quality monitoring is a powerful and direct way of revealing the presence of contaminants in drinking water. However, it is not a particularly good predictor of the likelihood of human illness as a result of drinking water consumption because (1) pollutants in drinking water often exist at extremely low concentrations that are below detection limits, (2) it is difficult to know the quality and quantity of water actually ingested, (3) individual responses to contaminated water vary considerably, and (4) it is hard to distinguish the harmful effects of contaminated water from those of other vehicles such as food. For these reasons, water quality monitoring programs should be complemented by active disease surveillance and monitoring for disease outbreaks to determine whether water consumption presents a disease risk for the public.

The New York City Department of Health (NYC DOH) and NYC DEP maintain a Waterborne Disease Risk Assessment Program that encompasses both active disease surveillance and outbreak detection. Acute diseases of microbial origin are the focus of this program, rather than chronic conditions that have been linked to long-term exposure to chemical contaminants (such as trihalomethanes). The objectives of New York City's program are to (1) determine rates of giardiasis and cryptosporidiosis, along with demographic and risk factor information on case patients, (2) track diarrheal illness to assure rapid detection of outbreaks, and (3) determine the contribution (if any) of tap water consumption to gastrointestinal disease. In addition to active disease surveillance and outbreak detection, the program includes special studies, information sharing, and public education.

The Waterborne Disease Risk Assessment Program is meant to provide evidence of the lack of waterborne disease outbreaks in New York City, as required under the SWTR. The program also compares reported cases of giardiasis and cryptosporidiosis or surrogate measures of diarrheal disease incidence in the City's population with measures of water quality. However, although temporal trends in diarrheal disease may coincide with trends in water quality parameters, determining whether an observed pattern of disease is associated with exposures to drinking water contaminants requires a specifically designed epidemiological study.

Active Disease Surveillance

NYC DOH staff regularly contact 84 clinical laboratories to collect case reports on stool specimens positive for Giardia and Cryptosporidium. For each case, an interview is conducted and/or medical charts are reviewed for information on demographic features and potential risk factors. Active surveillance for giardiasis in New York City began in July 1993, and case interviews were discontinued in August 1995. Prior to 1993, there was a passive surveillance program for giardiasis, and responsibility for reporting recognized cases rested with the

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

health care provider or laboratory. Active disease surveillance resulted in a marked increase in the reported number of giardiasis cases—from less than 300 cases per year during 1982–1988 to 2,456 cases in 1994 (the first year of active surveillance). Active surveillance of cryptosporidiosis in New York City began in November 1994, and case interviews and chart reviews have been conducted since January 1995. Active surveillance for cyclosporiasis began in June 1996 after a series of foodborne outbreaks were associated with this agent.

Active disease surveillance provides information on endemic rates of infection and demographic patterns in infection rates (such as residence, age, gender, and ethnicity). For cryptosporidiosis cases, additional information is collected on the immune status of the case and on known or suspected risk factors for infection, such as animal contact, high-risk sexual behavior, contact with a child enrolled in daycare, or international travel. Cases are also queried about their source of drinking water (e.g., plain tap water, filtered tap water, boiled tap water, or bottled water).

Endemic Rates of Giardiasis and Cryptosporidiosis

Active surveillance data for both giardiasis and cryptosporidiosis indicate that case rates per 100,000 have been decreasing since these data were first collected in 1994–1995 (see Figure 6-3 for cryptosporidiosis rates). Giardiasis

FIGURE 6-3 Reported cases of cryptosporidiosis in New York City, 1994–1998. Source: NYC DEP (1999c).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

has declined from 33.5 reported cases per 100,000 in 1994 to 25.7 cases per 100,000 in 1998. A total of 1,881 cases of giardiasis were reported in 1998, with children less than 10 years of age experiencing the highest rates of infection. Geographically, the highest case rate was in Manhattan. These age and geographic patterns have been observed for several years (NYC DEP, 1999c). Interview data from giardiasis cases during July 1994 to May 1995 indicated that the most commonly reported exposure risk was travel outside the United States, with other prevalent risk factors being contact with a household member having diarrhea, recent immigration, and adult male homosexual activities (NYC DEP, 1997e). Because most of the cases were in children, it is surprising that contact with childcare centers was not reported as an exposure risk.

Between 1995 and 1998, cryptosporidiosis rates dropped from 6.5 reported cases per 100,000 to 2.8 cases per 100,000 (NYC DEP, 1999c). In 1995 and 1996, the majority of reported cryptosporidiosis cases were among persons with AIDS, and the overall decline in cryptosporidiosis between 1995 and 1998 is due to a decline in cases in this subpopulation. This trend coincides with the increased use nationwide of newer antiretroviral therapies (Miller, 1998). Reported cryptosporidiosis among immunocompetent persons in New York City has increased from 71 cases in 1995 to 116 cases in 1998; the reasons for this rise are unclear. In addition, case rates for immunocompetent persons peak each year in late summer, a seasonal pattern observed elsewhere (MacKenzie, 1998). From 1995 to 1998, the most commonly reported exposure for both immunocompromised and immunocompetent persons was contact with an animal. For cryptosporidiosis cases among immunocompetent persons, foreign travel and recreational water contact were the second most commonly reported exposures.

For both giardiasis and cryptosporidiosis cases, the vast majority of cases reported drinking unboiled tap water as their primary source of drinking water. However, the significance of this exposure cannot be determined because there is no information on exposure histories in a suitable control population, such as persons who consume only bottled water. In addition, it is difficult to compare the New York City case rates with rates of giardiasis and cryptosporidiosis in other parts of the United States because there are few communities that have similar active surveillance for these diseases. Preliminary laboratory-based active surveillance data from six states in 1998 showed that cryptosporidiosis rates ranged from 1.3 to 3.5 per 100,000 (V. Dietz, CDC, personal communication, 1998), suggesting that the endemic rates of cryptosporidiosis in New York City are similar to those observed elsewhere.3

3  

 It should be noted that these 1998 case rates differ from those found in Chapter 5, Table 5-2, which relied on 1997 data.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Limitations of Active Disease Surveillance

Although much more sensitive than passive surveillance, active surveillance suffers from significant underdiagnosis, making comparisons among different localities problematic. Underdiagnosis is caused by the combined effects of several factors, illustrated in Figure 6-4. The net result of these factors is an underestimation of the true disease rate by several orders of magnitude. The greatest loss of information is due to the fact that most cases (94 percent in the Milwaukee Cryptosporidium outbreak) do not seek medical attention, and most health care providers do not routinely collect and test stool specimens from patients with gastroenteritis. Because most cases of gastroenteritis are only given supportive care, knowledge of the etiology of the disease does not affect treatment and is therefore rarely sought. These limitations apply to active disease surveillance programs nationwide.

To illustrate these effects, Table 6-7 compares the expected number of endemic Cryptosporidium cases reported per month among immunocompetent persons in New York City to the expected number of reported Cryptosporidium cases in a situation where there is an outbreak affecting one percent of the population. Reporting-loss estimates used in this table are based on the 1993 Milwaukee cryptosporidiosis outbreak, during which only 1 in every 22,000 cases of cryptosporidiosis were reported to the Health Department prior to recognition that the outbreak was waterborne (Juranek, 1999). (The Foodnet surveillance system of the Centers for Disease Control and Prevention (CDC) has made similar observations of the percent loss in this series of steps.) The estimate of six reported cases of endemic cryptosporidiosis per month is similar to that observed by the NYC DOH active surveillance system. That is, for 37 months of surveillance data, the number of reported cryptosporidiosis cases per month ranged from 1 to 17 with an average of 6.4. Table 6-7 shows that an outbreak affecting one percent of the population would result in the recognition of an additional nine cryptosporidiosis cases. It is clear that a large portion of the population (perhaps as much as ten percent) would have to become infected during an outbreak in New York City before a significant rise in the number of reported cryptosporidiosis cases would be detected by active surveillance (MacKenzie, 1998).

In addition to underestimating cases, active disease surveillance suffers from problems of timing. In the case of cryptosporidiosis, there are approximately 7–14 days between time of exposure and occurrence of symptoms. Delays in seeking medical care, and the time needed for laboratory diagnosis and reporting, would cause any observed peak in cryptosporidiosis rates detected by active surveillance to occur at least two weeks after a problem in water quality (assuming water is the causative agent). For example, surveillance data collected from the Milwaukee cryptosporidiosis outbreak showed that the peak in laboratory-reported cases occurred 15 days after the peak in turbidity reported at the faulty

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

FIGURE 6-4 Sources of loss of information during active disease surveillance. Note: In the case of cryptosporidiosis, the health care provider must specifically request that a stool specimen be tested for Cryptosporidium because this is not included in routine parasitic examinations of stool specimens. *A survey of U.S. laboratories in 1994–1995 indicated that only five percent of stool specimens submitted for routine "Ova and Parasite" exam are tested for Cryptosporidium (Boyce et al., 1996). Recent data from a survey of clinical labs in New York State indicate that about 18 percent of laboratories now include Cryptosporidium testing as part of the routine "Ova and Parasite" exam (J. Miller, NYC DOH, personal communication, 1998). Adapted from Frost et al. (1996) and Juranek (1997).

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

TABLE 6-7 Comparison of Expected Number of Reported Endemic Cryptosporidiosis Cases vs. Expected Number of Reported Cryptosporidiosis Cases in a Simulated Outbreak

Assumptionsa

Endemic Situation

Simulated 1% Outbreak

Total population = 7,000,000

 

 

Baseline diarrhea rate in immunocompetent persons = 11% prevalence monthly

770,000 cases of diarrhea

 

1% of population infected with Cryptosporidium in simulated outbreak

 

70,000 cases of cryptosporidiosis

2% of cases tested for any enteric pathogen

15,400 diarrhea cases

1,400 cryptosporidiosis cases

25% of these tested for "O&P" (ova and parasites)

3,850 diarrhea cases

350 cryptosporidiosis cases

5% of these tested for Cryptosporidium

193 diarrhea cases

18 cryptosporidiosis cases

3% of gastroenteritis cases test positive for Cryptosporidium

6 reported cryptosporidiosis cases per month in immunocompetent persons

 

50% of outbreak-associated cryptosporidiosis cases test positive

 

9 recognized cryptosporidiosis cases

a Estimates by MacKenzie (1998).

water treatment plant (Proctor et al., 1998). This time lag reinforces the argument that this type of active surveillance system is limited in recognizing and managing an outbreak in real time and can only provide retrospective information.

Outbreak Detection

Aware of the fact that active, laboratory-based surveillance is neither sensitive enough nor rapid enough to detect waterborne disease outbreaks, New York City has developed three independent and complementary diarrheal disease-monitoring systems as part of an Outbreak Detection Program. These systems collect information during the earliest steps in Figure 6-4 and thus detect more

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

potential cases of gastrointestinal disease than does the active surveillance program. Because the three systems are independent, their results can be compared to determine if an observed trend in a single system is confirmed by the other systems or is a reporting artifact. Although the sensitivity of these systems is potentially far greater than that of active surveillance, in some cases (monitoring of medication sales and number of stool specimens examined), the data collected are surrogate measures for cases of gastroenteritis and do not have the specificity of laboratory-confirmed infections.

Monitoring of Antidiarrheal Medications

Several reports in the literature indicate that outbreaks of enteric disease have been accompanied by increased sales of antidiarrheal medications (Rodman et al., 1998; Sacks et al., 1986). For example, Angulo et al. (1997) found that sales increased by 600 percent during an outbreak of salmonellosis. Monitoring antidiarrheal medication sales has been embraced by New York City as a sensitive, real-time indication of increased gastrointestinal disease and as one approach for detecting disease outbreaks. One of its advantages is that it provides some information on the total population experiencing diarrheal symptoms (step 1 in Figure 6-4), most of whom will not seek medical attention (step 2 in Figure 6-4).

In May 1995, New York City started collecting information on weekly shipments of Imodium from a major distributor to 1,265 pharmacies including 564 (about one-third of all pharmacies) in New York City. In addition, the weekly sales of 22 antidiarrheal medicines at a chain of 38 pharmacies located in the City have been monitored since February 1996. Both these monitoring activities provide information on the annual patterns of antidiarrheal medication sales at the distribution and retail level.

According to information provided by NYC DOH, the weekly sales of antidiarrheal medication have been relatively stable since 1996 (Miller, 1998). Peaks have been noted and attributed to promotional sales, including a large sales peak in November 1997.

Monitoring of Clinical Laboratories

To complement monitoring of antidiarrheal medicine sales, since November 1995 NYC DOH has received daily information by fax from three clinical laboratories (including the largest laboratory in the metropolitan area) on the number of stool specimens submitted for bacterial and parasitic testing. One of the three labs notes whether the stool specimens were submitted specifically for Cryptosporidium parvum. The reported number of routine stool specimens examined is relatively constant with modest seasonal variation, as shown in Figure 6-5. Because this outbreak detection system addresses steps 5 and 6 illustrated in Figure 6-4, it is less likely to detect increased disease rates com-

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

FIGURE 6-5 Number of stool specimens submitted for ova and parasite testing at one New York City laboratory. Courtesy of the NYC DEP.

pared to monitoring of antidiarrheal medication sales, which addresses steps 1 and 2 in Figure 6-5.

Nursing Home Monitoring

The third outbreak detection system is a nursing home surveillance system that involves 11 nursing homes in all five boroughs. Like monitoring of antidiarrheal medicine sales, this system provides information on step 1 in Figure 6-4. A pilot program, begun in March 1997, has continued to expand to include approximately 1,800 nursing home residents. Each nursing home provides daily information by fax on the number of new gastrointestinal illness cases among the residents. The population served by these homes includes elderly and AIDS patients who drink tap, filtered, or bottled water. Tap water drinkers from both the Croton and Catskill/Delaware systems are represented. Although surveillance data are limited because of the newness of the program, results so far indicate very low numbers of new gastrointestinal illness cases per day (NYC DEP, 1999c).

Validating the Outbreak Detection Programs

When a peak is observed in one of these three systems, the reporting source is contacted about the possibility of a reporting artifact. If a reporting artifact is

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

ruled out, then the results of the different outbreak detection systems are compared. If similar increases are observed, then NYC DOH contacts childcare centers, school nurses, and hospital emergency departments to determine if a corresponding rise in the number of clinical cases of gastroenteritis at these institutions has been noticed. Data from the outbreak detection program are also compared to water quality data to observe any simultaneous trends.

Two such follow-up investigations occurred in 1998. First, in March 1998, there was an increase in the sales of antidiarrheal medication and in the number of stool specimens submitted to clinical laboratories for examination. The timing of this rise coincided with the annual spring rotavirus season, although the etiology of this illness could not be confirmed because stool examination for rotavirus is rarely performed in clinical settings. During the same time period, three outbreaks of gastrointestinal illness were reported at New York City nursing homes, two of which were confirmed to be caused by rotavirus. Second, in October 1998, low levels of Cryptosporidium were detected in the Catskill effluent chamber of the Kensico Reservoir. However, the outbreak detection systems did not document any increased diarrheal disease in the population.

Because trends in all three outbreak detection datasets have been relatively stable, there have been few opportunities to determine their sensitivity to increases in gastrointestinal disease and compare the three systems to one another. More time must pass before the adequacy of these systems can be fully evaluated. In addition, because the actual sensitivity of these outbreak detection programs is currently unknown, it is premature to evaluate the adequacy of the response of the NYC DOH. Criteria for such evaluations should include (1) how peaks in these reporting systems are operationally defined (e.g., a certain percent rise in incidence over a specified time period); (2) how quickly, aggressively, and consistently peaks are investigated; and (3) under what circumstances these investigations include checking source water quality and informing the public.

Response to Consumer Complaints

As in many other water systems, NYC DEP and NYC DOH respond to consumer complaints about water quality by collecting and testing a water sample from the consumer's home. Although New York City performs a number of microbial and chemical tests, it does not analyze for the presence of cysts or oocysts, which would be infeasible given the needed volumes of tap water. If a consumer reports having diarrhea and thinks water may be the cause, arrangements can be made to obtain a stool specimen and test it at the NYC DOH for ova and parasites and bacterial enteric pathogens (A. Ashendorff, NYC DEP, personal communication, 1998).

There does not appear to be any systematic tracking of consumer complaints of water-related illness or water quality. This is valuable information that should

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

be recorded and reviewed, preferably using the GIS, to determine if any spatial or temporal patterns are evident. Consumer complaints in Milwaukee were one of the first indications that a serious problem with water quality and waterborne disease was occurring.

Boil-Water Advisories

A boil-water advisory is a public health measure that, if implemented promptly, can successfully reduce the risk for potentially serious diarrheal and other waterborne diseases among persons whose water supply has been contaminated by microbial pathogens (CDC, 1995). In New York City, it is not clear what concentration of contaminants triggers a boil-water advisory or other public health response. According to NYC DEP, should any elevated level of Cryptosporidium be detected in the water supply, NYC DEP, working with the City and State Departments of Health and EPA, will evaluate factors included in the 1995 CDC guidance to determine what action, if any, to take in response (Ashendorff, 1999). Developing a plan of action for issuing boil-water advisories and responding to changes in pathogen concentrations in source waters are concerns for water supplies across the country. New York City should take the lead in creating such plans by working with state and federal partners to formalize a decision-making process.

Information Sharing and Public Education

A fundamental element of any public health surveillance system is to share its findings with those who provide the data. Active disease surveillance data are regularly compiled into quarterly and annual reports. In addition to these reports, presentations on the surveillance system findings are made to health care professionals in the New York City area. NYC DOH staff also participate in national and international conferences that focus on drinking water and waterborne disease issues.

Two public outreach efforts by NYC DOH deserve mention. Special reports of the City Health Information publication that focus on Cryptosporidium have been mailed to health care professionals to draw attention to the need to request specific laboratory testing for Cryptosporidium when this infection is suspected. In addition, announcements have been faxed to area hospitals, health care providers of HIV/AIDS patients, and organizations serving persons with HIV/AIDS informing them of the findings of low levels of Cryptosporidium in the source water and the significance of these findings. The CDC has encouraged local public health authorities and water utility officials to develop coalitions with health care providers and advocacy groups for immunosuppressed persons in order to communicate important public health information and decide what

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

specific actions should be taken when oocysts are detected in municipal water (CDC, 1995).

Epidemiologic Studies

Active disease surveillance measures rates of specific infections, both endemic and epidemic. Outbreak detection systems monitor numbers of cases of gastrointestinal illness. In both instances, it can be difficult, if not impossible, to demonstrate that drinking water is the vehicle of disease transmission. In order to link contaminated source water with disease, epidemiologic studies must be conducted. As described below, New York City has supported such studies twice during the 1990s. Both of these studies utilized cases identified from the active disease surveillance program.

Giardiasis Case-Control Study

In June and July 1995, NYC DOH compared 120 patients suffering from giardiasis with a control group of 120 persons (NYC DEP, 1997e). The controls and patients were matched by age, gender, neighborhood of residence, and primary language. Interviews with all participants included questions about highrisk activities, food intake, water intake, and immune status. The study could detect no difference in the proportion of tap water drinkers among giardiasis cases versus controls. Thus, drinking water was not implicated as a source of Giardia.

NYC DOH study results contrast with other Giardia case-control studies. First, the NYC DOH study did not detect any association between disease and commonly recognized risk factors for giardiasis such as travel outside the United States and swimming in freshwater. Contact with childcare centers, another common risk factor, is not mentioned. Furthermore, the study reported that giardiasis cases were less likely than controls to have household members suffering from diarrhea, a surprising finding. These results directly contrast with those of other researchers. A case-control study of residents in New Hampshire that involved 273 cases and 375 controls found that giardiasis was significantly associated with a recent history of drinking untreated surface water (e.g., during camping or hiking), a history of swimming in a lake, pond, or any natural body of freshwater, and contact with a person thought to have giardiasis (Dennis et al., 1993). An earlier study using 171 giardiasis cases and 684 controls found that having a family member in day care, having a family member with diagnosed giardiasis, travel outside the United States, and camping were significant risk factors (Chute et al., 1987). It is possible that the smaller size of the New York City study did not permit a thorough examination of all risk factors for giardiasis.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Cryptosporidiosis Cross-Sectional Study

A cross-sectional study was conducted to examine the prevalence of cryptosporidiosis among HIV-infected persons in New York City and identify risk factors for cryptosporidiosis in this population. Between October 1995 and July 1997, 405 HIV-positive patients at the New York Hospital–Cornell Medical Center were interviewed, 379 blood samples (sera) were collected, and 331 stool specimens were collected. Only 1.2 percent of the stools tested positive for Cryptosporidium, while 28 percent of the blood samples were seropositive for a Cryptosporidium antibody (NYC DEP, 1998e).

Study subjects were queried about a variety of potential risk factors. Interviews revealed that the drinking water habits of the HIV patients were variable: 45 percent did not drink tap water at home but few completely avoided tap water, and 21 percent had recently changed their water consumption habits (Soave et al., 1998). Because of the low percentage of stools that tested positive for Cryptosporidium, the study was not able to examine the association between drinking water habits and enteric infection. In addition, although a sizable percentage of the study participants had antibodies against Cryptosporidium, seropositivity was not correlated with any of the potential risk factors for infection, including drinking water habits. Like the giardiasis case-control study, this study was limited by the small size of the study population.

Additional Epidemiologic Studies

Stool Survey for Cryptosporidium. Since September 1995, the NYC DOH has tested all stool specimens collected by Child Health Clinics (which serve approximately 80,000 children) for Cryptosporidium. The goal of this study is to collect information on the prevalence of Cryptosporidium in children. The number of samples tested per year has ranged from 3,400 to 5,400, and the detection rate has ranged from zero to 0.09 percent. Although it is reassuring to see such a low prevalence, there are some important unanswered questions. The study population should be defined in terms of age distribution, geographic distribution, and current gastrointestinal symptoms. Whether the specimens are fresh or preserved will also affect study results. Finally, these data should be compared to rates of Cryptosporidium detection in stool specimens submitted for ova and parasite exams as part of active surveillance. If stool surveys for Cryptosporidium are continued, further information about the source and quality of the specimens and how these results compare to other measures of Cryptosporidium prevalence in New York City and elsewhere, is needed to interpret the significance of the results from this activity.

Serologic Studies. Additional epidemiologic studies would be useful to help determine whether there is a relationship between consumption of New

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

York City tap water and endemic enteric illness or infection (as recommended by Craun et al., 1994). A cross-sectional or longitudinal serologic study could be conducted to examine seroprevalence or seroincidence of Cryptosporidium in New York City tap water drinkers compared either to residents who drink bottled water or to residents of other communities with filtered surface water or groundwater (Griffiths, 1999, Box 4-2). Although there is still some uncertainty about the best methods for determining seropositivity, recent advances have been made in this area, and the results of other Cryptosporidium serologic studies are encouraging (Frost et al., 1998a,b; Griffiths, 1999). Such studies could help explain the difference between the low rates of Cryptosporidium detected by the active surveillance program (0.2 per 100 HIV-infected persons in 1998) and the higher rate of seropositivity observed in the study of HIV-infected patients (28 percent).

Time Series Studies. Because of the wealth of water quality monitoring data, NYC DOH and NYC DEP should consider conducting time series studies that compare patterns of illness with various water quality parameters over time (as in Schwartz et al., 1997). This could be easily accomplished with the data that are already collected in the active disease surveillance, outbreak detection, and water quality surveillance monitoring programs.

Cohort Studies. Because it has large amounts of accumulated data and an active professional staff, NYC DOH is in an excellent position to conduct a randomized household intervention trial using some type of in-home water treatment device (as in Payment et al., 1991, 1997). This type of study compares the rate of enteric symptoms among households consuming tap water with those among households consuming tap water that receives additional home treatment (e.g., reverse-osmosis filtration or UV disinfection). Ideally, the study design should allow consideration of the separate contributions of source water quality and distribution system water quality to enteric illness or infection.

Although such studies are complex and expensive, they are the cleanest and most definitive way to examine the role of waterborne disease. Cohort studies compare two distinct populations, one using the water source to be evaluated and the other using some other very high-quality water. The longitudinal nature of such studies allows one to look at health effects over time and examine the temporal relationship between water exposure and disease. Because household intervention studies are conducted in single communities, community-to-community differences in other risk factors are not an issue. Finally, household-to-household differences in risk factors for gastrointestinal illness (such as children in childcare centers and travel) should be randomly distributed between the two study groups. Therefore, any difference in gastrointestinal illness rates can be attributed to differences in water quality.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Conclusions and Recommendations

A panel of experts reviewed New York City's Waterborne Disease Risk Assessment Program in April 1998 and individual panel members suggested several ways to enhance or supplement the current surveillance and outbreak detection programs and further examine the safety of New York City tap water. The following recommendations are targeted at the three primary goals of the program: (1) to determine rates of giardiasis and cryptosporidiosis, (2) to track diarrheal illness to detect outbreaks, and (3) to determine the contribution of tap water to gastrointestinal illness. The results are summarized briefly in Table 6-8 and are explained in greater detail below.

TABLE 6-8 Recommendations for Improving the New York City Waterborne Disease Risk Assessment Program

Program

General Limitations

Recommendations

Goal 1, 2, or 3?

Active Disease Surveillance

1. Low sensitivity because only a small percentage of stool specimens are tested.

2. Slow

All stool specimens should be tested. Peaks should continue to be aggressively analyzed.

1 and 2

Antidiarrheal Medicine Monitoring

Some peaks not associated with illness.

Continue aggressive and timely investigation of peaks.

2

Clinical Laboratories Monitoring

1. Some stools are from patients without infectious disease symptoms.

2. Addresses later steps in Figure 6-4.

Of marginal use for detecting outbreaks, but useful for comparison with other detection systems. Could be useful in determining the percentage of stools positive for Crypto.

2

Nursing Home Monitoring

May have a high background of gastrointestinal upsets because of multiple transmission routes in institutions.

Continue aggressive investigation of peaks and comparison of data from one home to data from other homes.

2

Epidemiological Studies

Previous studies limited in sensitivity and design.

Intervention (cohort) study

3

Response to Consumer Complaints

Sampling of water or individual likely to occur long after exposure event.

Do not sample water for protozoa, but do flag health complaints and set up GIS database for tracking complaints.

None

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
  1. NYC DOH and NYC DEP are making strong efforts to conduct active surveillance for giardiasis and cryptosporidiosis and to monitor levels of gastrointestinal illness in New York City. Such efforts are expected given the size of the service population and the fact that this is an unfiltered surface water supply. Since 1994, there has been substantial progress in the development of several complementary outbreak detection systems.

  2. Health care providers and laboratories should make Cryptosporidium testing part of all routine stool examinations. Active disease surveillance is too insensitive and slow to detect waterborne disease outbreaks on a real-time basis. Data from other communities suggest that only 1 in 22,000 cases are diagnosed and reported via disease surveillance. Sensitivity can best be improved by promoting enhanced collection and testing of stool specimens for Cryptosporidium. The median cost of such tests is $52 for those New York City laboratories that do not routinely test for Cryptosporidium (D. Warne, NYC DEP, personal communication, 1999). The 140 percent increase in testing stools for Cryptosporidium, observed between 1995 and 1997, is encouraging.

  3. NYC DOH should determine the lowest incidence of disease that can be detected by the current outbreak detection program and increase the sensitivity by studying specific populations. It is not clear what size of outbreak can be detected using the combined system of monitoring sentinel nursing home populations, sales of antidiarrheal medications, and submission of stool specimens. To increase the sensitivity and population base covered by these systems, the following additional approaches could be used: (1) monitoring school absenteeism related to gastroenteritis in sentinel schools (see Rodman et al., 1998), (2) monitoring Health Maintenance Organization nurse hotline calls to track increased incidence of gastroenteritis, (3) monitoring hospital emergency room visits for gastroenteritis, and (4) monitoring gastrointestinal symptoms in a network of sentinel families.

  4. NYC DOH and NYC DEP should develop a plan of action to define spikes in surveillance and outbreak detection data and trigger investigation of these peaks after eliminating the possibility of reporting error. There appears to be no systematic approach for defining peaks in active surveillance or outbreak detection data. Investigation of peaks should continue to include comparisons with water quality data. Substantial resources are currently being directed toward collecting data on water quality and health measures (both disease surveillance and outbreak detection). Linking and critically reviewing these data sets may lead to recognition of relationships between water quality and health. Analysis of the Milwaukee cryptosporidiosis outbreak indicated a strong temporal relationship between trends in finished water turbidity and trends in emergency room visits for gastrointestinal illness, customer complaints, laboratory diag-

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

noses of Cryptosporidium infections, prevalence of diarrhea in nursing homes, and absentee rate in schools (Proctor et al., 1998).

  1. NYC DOH and NYC DEP should develop a plan of action for tracking and investigating consumer complaints about water quality and water-related illness that goes beyond collecting and testing a household water sample for routine parameters. Such a system, which would optimally be linked to the GIS, may lead to earlier recognition of geographic and temporal problems with water quality both at the source and in the distribution system.

  2. To determine the role of tap water as a vehicle of infection, NYC DOH should conduct additional epidemiological studies. Two relatively small epidemiological studies of waterborne disease have been conducted in New York City, neither of which implicated drinking water as a risk factor for disease. It is questionable whether the Giardia case-control study had sufficient power to detect an association between exposure and disease for any risk factor of giardiasis. In the cryptosporidiosis cross-sectional study, only four laboratory-confirmed cases of Cryptosporidium infection were detected, making it impossible to evaluate the relationship between drinking water and risk of cryptosporidiosis. Several possible study designs, including serologic studies and intervention studies, are described in this chapter.

Surveillance systems alone, no matter how sensitive, cannot provide evidence of the risk of endemic waterborne disease. Well-designed and well-conducted epidemiologic studies can examine disease patterns revealed by the surveillance system (such as the rise of cryptosporidiosis among immuno-competent individuals) and provide valuable documentation of the safety of the water supply.

MICROBIAL RISK ASSESSMENT

Epidemiological methods provide one tool for estimating potential disease impacts from pathogens in water. A complementary tool is the use of quantitative microbial risk assessment. Although risk assessment is not currently being practiced by New York City on a regular basis, there are sufficient data being collected to use this technique to estimate potential disease impacts. The objective of this section is to outline the process and to perform a quantitative risk assessment focusing on Cryptosporidium. Cryptosporidium is selected as the target organism of interest because it is presently the most resistant pathogen to disinfection, with minimal inactivation by free chlorine alone. In contrast, other pathogens (Giardia and enteric bacteria and viruses, for example) would be expected to be substantially reduced by current treatment.

This risk assessment follows the general methods developed by the National

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Research Council (NRC, 1983) and provided in Appendix C. The following inputs are needed to conduct this assessment:

  • Water ingestion per day (V)

  • Oocyst concentration at point of ingestion (C)

  • Dose-response relationship for Cryptosporidium f(V•C)

In accordance with prior risk assessments, each day of exposure (consumption of water) is considered to result in a statistically independent risk of infection (Regli et al., 1991; Haas et al., 1993).

In performing this risk assessment, it is assumed that the oocyst levels in the Kensico Reservoir raw water (at the CATLEFF and DEL 18 sampling locations) reflect concentrations generally experienced at the point of consumption. This assumption is reasonable because free chlorine has been generally regarded as being of minimal efficacy in inactivating oocysts (Korich et al., 1990; Ransome et al., 1993; Finch et al., 1998). It is possible that at the periphery of the distribution network in New York City, sufficient time of contact with chlorine may occur to provide some inactivation of oocysts.

Determination of Input Variables

Water Ingestion

Tap water ingestion was modeled using the lognormal distribution for total tap water consumption developed by Roseberry and Burmaster (1992). In this study, the natural logarithm of total daily tap water consumption was found to be normally distributed with a mean of 7.492 (corresponding to an arithmetic mean of 1.95 L/day) and a standard deviation of 0.407. The study was based on data collected from 5,605 persons over a three-day period from 1977 to 1978 as part of the Nationwide Food Consumption Survey. Water consumption rates for New York City have not been measured, and although it does not explicitly reflect the New York City population, the Roseberry and Burmaster analysis is widely used (see the Exposure Factors Sourcebook published by the American Industrial Health Council, Washington DC). It should be noted that the default water consumption generally used by EPA in microbial risk assessment is 2 L/day, which is slightly greater than the mean developed in Roseberry and Burmaster.

Oocyst Concentration

Initial examination of oocyst levels measured at the two source water points (DEL 18 and CATLEFF) indicates a number of interesting features (Figure 6-6). First, the levels of oocysts are quite variable, as is common for many microbial data sets. Second, the densities appear to be higher during the earlier portion of

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

FIGURE 6-6 Total oocyst concentrations in Kensico raw water samples.  Data courtesy of the NYC DEP.

the data record than in the more recent part of the data record. The reasons for this difference are not clear. Third, there is a substantial amount of data where no oocysts were detected. The mean detection limit for these nondetects was 0.721 oocysts/100 L.

The overall mean oocyst concentration (treating the ''nondetects" as zeros) was 0.26 and 0.31 oocysts/100 L for the CATLEFF and DEL 18 locations,

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

respectively. Of the 292 samples taken at each location, only 45 samples at CATLEFF and 48 samples at DEL 18, respectively, were above individual daily detection limits. Of these samples, only 18 and 21, respectively, were above 0.721 oocysts/100 L (the average detection level for the nondetects). This pattern is not unusual in protozoan monitoring data, and it presents a level of complexity in assessing the risk posed by exposure to these organisms.

Detection Limit Considerations. The significant number of samples with concentrations close to or below the average detection limit must be taken into account when estimating mean oocyst densities and distribution. There are several methods that may be used when dealing with below-detection-limit data (Haas and Scheff, 1990). Two basic approaches are employed here:

  1. Observations that are below-detection-limit are treated as if they had values equal to the detection limit, half the detection limit, or zero. The arithmetic mean of the revised data is then computed by simple averaging. These alternatives are called "fill in" alternatives.

  2. The method of maximum likelihood is used. In this approach, the data are presumed to come from a particular distribution (e.g., lognormal), and standard methods for analyzing data with a single censoring point are used. A likelihood function is formulated with a contribution equal to the probability density function for all quantified values, and equal to the cumulative distribution function (up to the detection limit) for all below-detection-limit values. The values of the distribution parameters that maximize the resulting likelihood are accepted as the best estimators. If the lognormal distribution is used, and the parameters mln(x) and sln(x) represent the mean and standard deviation of the log-transformed densities, then the arithmetic mean (mx) determined by the property of the lognormal distribution is:

To develop the distribution for oocyst concentrations at the point of ingestion, all data present in Figure 6-6 for CATLEFF and DEL 18 were examined. Using maximum likelihood, and treating all observations less than or equal to 0.721 oocysts/100 L as being "censored" (for all censored observations, 0.721 oocysts/100 L was regarded as being the detection limit), the parameters of lognormal distributions were determined.

The maximum likelihood method is preferred over the fill-in method because

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

TABLE 6-9 Mean and Standard Deviation of Best-Fitting Normal Distribution for Natural Logarithm of Oocyst Levels (#/100 L) in Kensico Samples (1992 to July 1998)

Location

Oocyst Levels

Mean of Natural Logarithm

Standard Deviation of Natural Logarithm

CATLEFF

–2.752

1.828

DEL 18

–3.210

2.177

it is less subject to bias, especially when a large number of censored observations exist. Numerical studies have shown that this method, and closely related probability regression methods, are reasonably robust to deviations from an assumption of exact log-normality (Helsel and Cohn, 1988; Haas and Scheff, 1990).

Lognormal Distributions of Oocyst Data. Table 6-9 gives the parameters of the best-fitting lognormal distributions to the entire data record at each station. There is some underprediction at the extreme tails of the distribution; however, in general the fit is adequate. Investigation of alternative distributions (gamma, Weibull, and inverse Gaussian) did not yield fits superior to the lognormal distribution. The goodness of fit to the lognormal was acceptable as judged by a chi-squared test. Figure 6-7 provides a plot of the exceedance probabilities for the fitted and observed distributions.

Although the time series plot (Figure 6-6) suggests a potential correlation between the oocyst levels at CATLEFF and DEL 18, the rank correlation coefficient (for observations in excess of the detection limit only) is –0.18. This is lower in absolute value than a correlation that would have substantial influence on the results of a Monte Carlo risk assessment (Smith et al., 1992; Bukowski et al., 1995), and so it has been ignored in this computation.4

Comparing Fill-in Methods to the Maximum Likelihood Method. Figures 6-8 and 6-9 summarize the annual arithmetic averages at the CATLEFF and DEL 18 stations, respectively. The first column shows the mean oocyst concentration for the entire 1992–1998 data set, which was computed using the "fill-in" methods. Each subsequent column shows the mean oocyst concentration for an individual year of data. (For 1992 and 1998, these averages are for portions of the year.) The four bars correspond to the method of maximum likelihood and the three different fill-in methods. The "imputed arithmetic mean" (i.e., mx) is computed from the maximum likelihood estimates of mln(x) and sln(x); in more

4  

 The effect of including this slight negative correlation would have been to slightly reduce the estimated risk of infection.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

FIGURE 6-7 Complementary cumulative distribution for fitted (lines) and observed (individual points) Kensico oocyst densities.

FIGURE 6-8 Summary of mean oocyst levels (#/100 L) at CATLEFF estimated by different methods. *The high proportion of nondetects in these years makes it impossible to use the maximum likelihood method.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

FIGURE 6-9 Summary of mean oocyst levels (#/100 L) at DEL 18 estimated by different methods. *The high proportion of nondetects in these years makes it impossible to use the maximum likelihood method.

recent years, it was not possible to estimate the maximum likelihood mean densities at both locations and all times, since too few (less than two) observations above a detection limit were available.

The bias related to the "fill-in" methods using the detection limit and half the detection limit is quite evident in more recent years, where the oocyst levels were generally below detection. Both these fill-in methods may overestimate total oocyst concentration in the source water. Regardless of the methods used, in 1992 and 1993 average oocyst levels were higher than they were in more recent years.

Assessing Relative Contributions from CATLEFF and DEL 18. The two oocyst concentrations (CATLEFF and DEL 18) represent contributions dominated by Catskill and Delaware flows at Kensico. In combining these contributions to assess exposure, they were flow weighted by the average daily flows to the Hillview reservoir—521 and 800 mgd, for CATLEFF and DEL 18, respectively (NYC DEP, 1997a). In other words, the combined concentration entering the New York City distribution system can be estimated as the sum of 0.394

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

[which is equal to 521/(521+800)] times the CATLEFF concentration and 0.606 times the DEL 18 concentration.

Dose-Response Relationship

The dose-response relationship for infection of human volunteers with C. parvum oocysts has been found to be exponential, as given in Equation 6-2. V is the volume of water consumed, C is the oocyst concentration, p is the risk of infection from a single exposure, and k is the best-fit dose response parameter (or median infectious dose—238 oocysts) (Haas et al., 1996).

The confidence distribution to the natural logarithm of k, as determined by likelihood theory (Morgan, 1992), was found to be closely approximated by a normal distribution with a mean of 5.48 and a standard deviation of 0.32.

This dose-response relationship, and the variability of k, is only for the particular strain used in the underlying human volunteer studies. There is known to be variability in the infectivity of differing strains of Cryptosporidium parvum (Chappell, 1999) of perhaps an order of magnitude. However quantitative dose-response relationships for these other strains are not yet available. Hence strain-to-strain variability cannot be accounted for in a quantitative sense at this time.

Risk Assessment Calculation

Given a single value of water consumption (V), oocyst concentration (C), and the dose-response parameter (k), the risk of infection to an individual can be computed by application of Equation 6-2. To consider the distribution of risk that incorporates uncertainty and variability in each of the input parameters, this computation needs to be performed a large number of times. For each repetition, a new set of random samples (for water consumption, oocyst concentration at each location, and the dose-response parameter) must be obtained. Individual calculations using these sets of random samples are combined to reveal an estimated distribution of risk. This technique is termed the application of Monte Carlo methods to risk assessment, and it has received wide use (Burmaster and Anderson, 1994; Vose, 1996).

Two types of results are presented below. First, the daily risk estimate is calculated for each individual year, given a single water dose, dose-response parameter, and average oocyst concentration. Four oocyst concentrations are used, representing the different methods for considering data points below the detection limit. The purpose of this exercise is to observe trends in the risk

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

estimate over time. The second set of results shows the range of estimated risk, taking into account uncertainty in all of the input parameters. This range is generated using the combined data from 1992 to 1998.

Results: Point Estimates

Point estimates for the daily risk infection from Cryptosporidium are presented in Table 6-10. The volume of water consumed was 1.95 L/day and k was 238. Calculations were done using both the total (1992–1998) data set and for each year individually, and they do not take into account variability in any of the input parameters. Table 6-10 presents results of four different methods used to determine the average oocyst concentration: the maximum likelihood method and the three "fill in" methods. Depending on which of the methods was used, the daily risk ranges from zero to 12.2 × 10-5. The table demonstrates that the risk estimate has dropped between 1993 and 1996.

When fewer than two quantified (above-detection-limit) observations existed, the maximum likelihood method could not be used (as in the cases of data for 1995 and later). The bias inherent in the fill-in methods is particularly evident for 1995 and subsequent data. That is, risk values derived from using the full detection limit and half detection limit methods are likely to be overestimates of the true risk, while those derived from using a zero detection limit method are likely to be underestimates.

TABLE 6-10 Computed Point Estimates for the Daily Risk of Infection from Cryptosporidium (all numbers × 10-5)

Year

Maximum Likelihood Method

Fill-in Methods

Detection Limit

Half Detection Limit

Zero Detection Limit

all years

3.2

7.1

4.7

2.3

1992

10.7

7.8

6.5

5.3

1993

10.8

12.2

10.9

9.7

1994

3.1

6.9

4.6

2.2

1995

*

5.7

3.0

0.2

1996

*

5.7

2.9

0.09

1997

*

5.6

2.9

0.1

1998 (Jan–June)

*

5.6

2.9

0

* Could not be estimated because fewer than two quantified observations are available.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Results: Monte Carlo Simulation

Although useful, point estimates of risk do not reveal the degree of uncertainty in the risk estimate. Monte Carlo simulations are currently the most rigorous way to take uncertainty into account during risk assessment, assuming high quality data are available. Summary statistics on 10,000 iterations of the Monte Carlo model are shown in Table 6-11. For this computation, the entire oocyst concentration database (1992–1998) was used.

The main result of the analysis is that the mean individual daily risk is estimated as 3.4 × 10-5. For an exposed population of 7.5 million, this would translate into an estimated 255 infections per day. The range of the 95 percent confidence limits would translate into an estimated range of infections per day from 2.6–1,643. It should be noted that the results of the Monte Carlo analysis bracket the range of point estimates observed by considering each year's data set separately, whether maximum likelihood or "fill-in" methods are used.

As part of this computation, a sensitivity analysis was conducted. The rank correlation of the individual daily risk with the various input parameters was computed (Table 6-12). The densities of pathogens in the two effluent flows

TABLE 6-11 Summary of 10,000 Monte Carlo Trials on Kensico Risk Assessment: Daily Risk of Cryptosporidium Infection (× 10-5)

Statistic

Individual Daily Risk

Daily # of Infectionsa

Mean

3.4

255

Median

0.7

53

Standard Deviation

19.8

1,485

Lower 95% confidence limit

0.034

2.6

Upper 95% confidence limit

21.9

1,643

a Based on an exposed population of 7.5 million persons.

TABLE 6-12 Rank Correlation of Input Parameters with Daily Risk of Infectiona

Input Parameter

Rank Correlation with Daily Risk

DEL 18 oocyst density

0.61

CATLEFF oocyst density

0.56

Water consumption

0.24

Dose-response "k" value

–0.20

a Rank correlation is the correlation between two sets of data when the individual observations in each set are replaced by the rank in that set.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

from Kensico have the greatest correlation with the estimated daily risk. The other inputs (water consumption, dose-response parameter) contribute only a minor amount to the uncertainty and variability of the estimated risk. This suggests that to reduce the degree of uncertainty and variability in the estimated risk, attention should be paid primarily to obtaining better (more precise) estimates of the effluent oocyst concentrations.

Caveats

Dose-Response Relationship

The above risk assessment has a number of caveats that should be taken into account in making a decision based on these results. First, the dose-response relationship was obtained from a study on healthy volunteers who were believed to have no prior history of cryptosporidiosis (Dupont et al., 1995). Prior exposure to Cryptosporidium may result in reduced susceptibility (Okhuysen et al., 1998). On the other hand, the elderly, children, and persons with lowered immunity (e.g., those on antirejection drugs after organ transplantation, recipients of cancer chemotherapy, and persons with HIV infection) are in general more susceptible to infectious diseases such as cryptosporidiosis (Gerba et al., 1996). The same populations may also suffer more severe symptoms than the general population, as was demonstrated during the cryptosporidiosis outbreaks in Milwaukee (Hoxie et al., 1997) and Las Vegas (Goldstein et al., 1996).

Secondary Infection

The above risk analysis does not incorporate consideration of community-level impacts, such as the formation of secondary cases. Such cases refer to individuals not directly infected by water exposure but who are exposed to other infected individuals. A consideration of secondary infections can raise the risk estimate. For example, in the 1993 Milwaukee outbreak, 4.2 percent of households with one or more ill persons also contained one or more secondary cases (MacKenzie et al., 1995). In a foodborne (apple cider) outbreak of cryptosporidiosis, for each primary case, there were one-third as many secondary cases (Millard et al., 1994).

Oocyst Viability and Recovery

The above analysis used total oocysts found in the Kensico Reservoir raw water sampling locations to assess exposure. Not all of the total oocysts represent viable, human, infectious forms. Although several methods are available, there is no rapid, inexpensive, and reliable method for determining oocyst viability on a routine basis. One frequently used method is to calculate viability based on the

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

ratio of "confirmed" to "presumed" oocysts, terms that are applied to oocysts during microscopic observation. However, this viability analysis is not scientifically defensible because it has never been conclusively established that oocyst collection and processing steps do not themselves result in loss of microscopic features required for confirmation. Simple shaking of oocysts with sand has been found to cause loss of internal structure of oocysts (Parker and Smith, 1993).

The efficiency with which the sampling methodology can recover oocysts can be quite low (Clancy et al., 1994). Oocyst recovery in New York City currently ranges from 30 percent to 70 percent (Stern, 1998). In prior risk assessments for protozoa, it has been assumed that the two errors (errors related to viability and to recovery) roughly cancel (Regli et al., 1991; Rose et al., 1991).

Endpoints

The focus of this risk assessment is on infections. It should be recognized that the end-point of clinically confirmed human illness may be substantially less than this. Based on human feeding studies, only about half of all infections progress to frank illness (Dupont et al., 1995). Even if illness occurs, in normal healthy individuals symptoms may be mild and not cause medical attention to be sought. If the endpoint of illness is used, the risk assessment would predict a lower number of symptomatic cases.

The endpoint for this risk assessment—infection—is different than the endpoint measured by active disease surveillance—frank illness that is diagnosed and reported. This is one of the reasons that the risk estimate predicts a higher rate of infection than is observed in the active disease surveillance program. Other factors, such as the limited sensitivity of active disease surveillance and the contribution of other vectors such as food, are also responsible for discrepancies between the risk assessment and measured surveillance rates.

Time Period

The exposure estimate in the risk analysis used the entire data record at the two locations. If the lower oocyst levels, which have been seen in more recent years (Figure 6-6), are assumed to be more typical of future oocyst levels, then a lower exposure and consequently a lower risk would be estimated.

Strains of Cryptosporidium

It is noted that the present dose-response relationship derives from a single set of studies on a single oocyst strain (the "Iowa" or "Harley Moon" strain). Information currently being analyzed suggests that other strains of oocysts may have higher and lower infectivities and different dose-response curves than the

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

''Iowa" strain (Chappell, 1999). This represents an additional source of potential variability in the risk assessment that should be amenable to quantitation.

Other Sources of Water Supply

The above analyses assumed that the sole oocyst loading reaching consumers was from the Catskill/Delaware supply. Although this represents the dominant source of drinking water for New York City, some residents are primarily served by water from the Croton system. To the degree that the (current) quality of raw water from the Croton system with respect to oocysts is different than the Catskill/Delaware, the numbers in Table 6-11 may under-or overestimate the total risk to some consumers of New York City water.

Impact of Treatment and Watershed Management

If the watershed management programs described elsewhere in this report are successfully implemented, the level of oocysts in the Catskill/Delaware supply may decrease. Although it is not yet possible to quantitatively forecast the magnitude of this decrease, a reduction in risk to consumers is expected. A primary motivation for conducting microbial risk assessment in water supply systems that pursue watershed management should be to determine the contribution of watershed management to overall risk reduction.

Quantifying the impacts of other treatment processes on the risk estimate is a more straightforward task. Ozonation and particle removal will decrease the oocyst levels in drinking water, and it is possible to determine the magnitude of this reduction based on pilot-scale testing. Given standard treatment efficiencies, a properly functioning water filtration plant is expected to achieve at least a 2-log removal of Cryptosporidium oocysts (Nieminski and Ongerth, 1995).

For both treatment processes and watershed management activities, a 1-log reduction in oocyst concentration translates directly into a 1-log reduction in the risk estimate because the dose-response relationship is linear at low dose. In other words, any process that reduces oocyst levels in the Kensico Reservoir by a factor of ten will reduce the risk estimate to 0.34 × 10-5 per person per day, or 25.5 infections per day, in a population of 7.5 million persons.

Conclusions and Recommendations on Risk Assessment

1. Based on the committee's risk assessment using data from 1992 to 1998, the current daily risk of Cryptosporidium infection in New York City is 3.4 × 10-5, with a 95 percent confidence interval ranging from 3.4 × 10-7 to 21.9 × 10-5. EPA has stated that less than one microbially-caused illness per year

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

per 10,000 people is a reasonable policy (EPA, 1989).5 This risk level, which corresponds to 10-4 per year or 2.7 × 10-7 per day, is smaller than the lower 95 percent confidence interval of the estimated daily risk for New York City based on the Catskill and Delaware supplies (Table 6-11). It is also below the point estimates for risk during individual years. Hence, based on the assumptions used, the calculated risk of cryptosporidiosis would appear to be in excess of the frequently propounded acceptable risk level.

The calculated risk estimate must be considered in conjunction with the caveats listed above. The risk estimate does not take into account measurements of oocyst viability and recovery, secondary infection, multiple strains of Cryptosporidium, or multiple dose-response relationships. The endpoint of the risk assessment was assumed to be infection, several years of data were used, and only the Catskill/Delaware system was considered. Finally, the impacts of watershed management on the risk estimate are not quantified.

2. It is recommended that a Cryptosporidium risk assessment be performed on a periodic basis for New York City. The goal of these efforts should be to help determine the contribution of watershed management (vs. other treatment options and management strategies) to overall risk reduction. Data that are sufficient for these purposes are currently collected as part of the NYC DEP Pathogen Studies. As new methods for oocyst recovery, detection, speciation (bird vs. human vs. animal), and viability become available, the risk assessment methods used in this report should be improved upon. Depending on the frequency of monitoring, risk assessment can be calculated for varying time periods to assess potential high-risk exposure times, such as during certain seasons and during storm events.

Prior to commencing this regular effort, a decision must be made as to what level of risk is deemed to be acceptable to the regulatory agencies, the City, and the affected parties. This level should be arrived at after full and open discussion with the various stakeholders. Should an annual risk level of greater than 10–4 be regarded as acceptable by NYC DEP or other relevant risk managers, then the risk estimates computed in this report can be compared to such alternate yardsticks.

3. An ongoing program of risk assessment should be used as a complement to active disease surveillance. Risk assessment allows one to ascertain the level of infection implied by a very low level of exposure that would go undetected by active surveillance, thus acting as a complementary source of information

5  

It should be noted that although the 10-4 risk level was developed for giardiasis, it is the only EPA-endorsed value available with which to compare current risks of cryptosporidiosis. As suggested in the second recommendation, New York City should determine an acceptable risk level before undertaking regular risk assessments.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

about public health. In combination, risk assessment and active disease surveillance data could be used to estimate the proportion of gastrointestinal disease cases attributable to drinking water (as in Perz et al., 1998). These estimates could then be validated by comparison with epidemiological data from a cohort study. In general, periodic risk estimates should be examined for concordance with prior computed risks and observed illness rates when formulating subsequent water treatment and watershed management decisions.

REFERENCES

American Public Health Association (APHA). 1998. Standard Methods for the Examination of Water and Wastewater. 20th Edition. Clesceri, L. S., A. E. Greenberg, and A. D. Eaton (eds.). Washington, DC: American Public Health Association.

Angulo, F. J., S. Tippen, D. J. Sharp, B. J. Payne, C. Collier, J. E. Hill, T. J. Barrett, R. M. Clark, E. E. Geldreich, H. D. Donnell, and D. L. Swerdlow. 1997. A community waterborne outbreak of salmonellosis and the effectiveness of a boil water order. Am. J. Pub. Health 87(4):580–584.

Ashendorff, A. 1999. NYC DEP. E-mail memorandum to the National Research Council dated January 6, 1999.


Bagley, S. T., M. T. Auer, D. A. Stern, and M. J. Babiera. 1998. Sources and fate of Giardia cysts and Cryptosporidium oocysts in surface water. Journal of Lake and Reservoir Management 14(2-3): 379–392.

Boyce, T. G., A. G. Pemberton, and D. G. Adiss. 1996. Cryptosporidium testing practices among clinical laboratories in the US. Ped. Infect. Dis. J. 15:87–88.

Bukowski, J., L. Korn, and D. Wartenberg. 1995. Correlated inputs in quantitative risk assessment: The effects of distributional shape. Risk Analysis 15(2):215–219.

Burmaster, D. E., and P. D. Anderson. 1994. Principles of good practice for the use of Monte Carlo techniques in human health and ecological risk assessment. Risk Analysis 14(4):477–481.


Cabelli, V. J. 1977. Clostridium perfringens as a Water Quality Indicator. Bacterial Indicators/Health Hazards Associated with Water. A. Hoadley and B. Dutka. Philadelphia, PA: ASTM.

Carmichael, W. W., C. L. A. Jones, N. A. Mahmood, and W. C. Theiss. 1985. Algal toxins and water-based diseases. Crit. Rev. Environ. Control 15:275–313.

Carmichael, W. W. 1986. Algal toxins. Adv. Bot. Res. 12: 47–101.

Centers for Disease Control and Prevention (CDC). 1995. Assessing the public health threat associated with waterborne cryptosporidiosis: report of a workshop. MMWR 1995; 44(RR-6): 1–19.

Chappell, C. 1999. Presentation at Heath Effects Stakeholder Meeting for the Stage 2 DBPR and LT2ESWTR. USEPA. February 1999, Washington, DC.

Chute, C. G., R. P. Smith and J. A. Baron. 1987. Risk factors for endemic giardiasis. Am. J. Public Health 77(5):585–587.

Clancy, J. L., W. Gollnitz, and Z. Tabib. 1994. Commercial labs: How accurate are they? Journal of the American Water Works Association 86(5):89–97.

Clancy, J. L., C. R. Fricker, C. R. Fricker, and W. Telliard. 1997. New US EPA Standard Method for Cryptosporidium Analysis in Water. AWWA Water Quality Technology Conference. Denver, CO: American Water Works Association.

Clarke, R. T. 1994. Statistical Modeling in Hydrology. New York, NY: John Wiley & Sons.

Codd, G. A., S. G. Bell and W. P. Brooks. 1989. Cyanobacterial toxins in water. Water Science and Technology 21:1–13.

Craun, G., G. Birkhead, S. Erlandsen, F. Frost, W. Jakubowski, D. Juranek, R. Soave, C. Sterling, and B. Ungar. 1994. Report of New York City's Advisory Panel on Waterborne Disease Assessment. NYC DEP.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Davis, L. J., H. L. Roberts, D. D. Juranek, S. R. Framm and R. Soave. 1998. A survey of risk factors for cryptosporidiosis in New York City: drinking water and other exposures. Epidemiol. Infect. 121:357–367.

Dennis, D. T., R. P. Smith, J. J. Welch, C. G. Chute, B. Anderson, J. L. Herndon and C. F. vonReyn. 1993. Endemic giardiasis in New Hampshire: a case-control study of environmental risks. J. Infect. Dis. 167(6):1391–1395.

Doerr, S. M., E. M Owens, R. K. Gelda, M. T. Aurer, and S. W. Effler. 1998. Development and testing of a nutrient-phytoplankton model for Cannonsville Reservoir. Journal of Lake and Reservoir Management 14(2-3):301–321.

Dupont, H. L., C. C. Chappell, C. R. Sterling, P. C. Okhuysen, J. B. Rose, and W. Jakubowski. 1995. Infectivity of Cryptosporidium parvum in healthy volunteers. New England Journal of Medicine 332(13):855.


El Saadi, O., A. J. Esterman, S. Cameron, and D. M. Roder. 1995. Murray River water, raised cyanobacterial cell counts and gastrointestinal and dermatological symptoms. Med. J. Aust. 162:122–125.

Environmental Protection Agency (EPA). 1989. Federal Register 54:27486 (June 29, 1989)

EPA. 1999. Method 1622: Cryptosporidium in Water by Filtration/IMS/FA. EPA-821-R-99-001. Washington, DC: EPA Office of Water.


Falconer, I. R., A. M. Beresford, and M. T. Runnegar. 1983. Evidence of liver damage by toxin from a bloom of the blue-green alga Microcystis aeruginosa. Med. J. Aust. 1(11):511-514.

Falk, C. C., P. Karanis, D. Schoenen, and H. M. Seitz. 1998. Bench Scale Experiments for the Evaluation of a Membrane Filtration Method for the Recovery Efficiency of Giardia and Cryptosporidium from Water. Water Research 32(3):565-568.

Finch, G. R., L. L. Gyurek, L. R. K. Liyanage, and M. Belosevic. 1998. Effects of Various Disinfection Methods on the Inactivation of Cryptosporidium. AWWA Research Foundation and American Water Works Association. Denver, C.O.

Frost, F. J., G. F. Craun, and R. L. Calderon. 1996. Waterborne disease surveillance. Journal of the American Waterworks Association 88(9):66–75.

Frost, F. J., A. A. de la Cruz, D. M. Moss, M. Curry, and R. L. Calderon. 1998a. Comparisons of ELISA and Western blot assays for detection of Cryptosporidium antibody. Epidemiology and Infection 121(1):205-211.

Frost, F. J., R. L. Calderon, T. B. Muller, M. Curry, J. S. Rodman, D. M. Moss, and A. A. de la Cruz. 1998b. A two-year follow-up survey of antibody to Cryptosporidium in Jackson County, Oregon following an outbreak of waterborne disease. Epidemiology and Infection 121(1):213–217.

Frost, F. J., G. F. Craun, and R. L. Calderon. 1996. Waterborne disease surveillance. Journal of the American Waterworks Association 88(9):66–75.


Geldreich, E. E. 1978. Bacterial populations and indicator concepts in feces, sewage, stormwater and solid wastes. In Berg, G. (ed.) Indicators of Viruses in Water and Food. Ann Arbor, MI: Ann Arbor Science.

Gerba, C. P., J. B. Rose, and C. N. Haas. 1996. Sensitive populations: Who is at the greatest risk? International Journal of Food Microbiology 30(1-2):113–123.

Goldstein, S. T., D. D. Juranek, O. Ravenholt, A. W. Hightower, D. G. Martin, J. L. Mesnik, S. D. Griffiths, A. J. Bryant, R. R. Reich, and B. L. Herwaldt. 1996. Cryptosporidiosis: An outbreak associated with drinking water despite state of the art water treatment. Annals of Internal Medicine 124(5):459–468.

Griffiths, J. K. 1999. Serological testing for Cryptosporidiosis. Presentation at U.S. EPA Office of Ground Water and Drinking Water Stage 2 Microbial/Disinfection Byproducts Health Effects Workshop. February 10-12. Washington, DC.


Haan, C. T. 1977. Statistical Methods in Hydrology. Ames. IA: Iowa State University Press.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Haas, C. N., and P. A. Scheff. 1990. Estimation of averages in truncated samples. Environmental Science and Technology 24:912–919.

Haas, C. N., J. B. Rose, C. Gerba, and S. Regli. 1993. Risk assessment of virus in drinking water. Risk Analysis 13(5):545–552.

Haas, C. N., C. Crockett, J. B. Rose, C. Herba, and A. Fazil. 1996. Infectivity of Cryptosporidium parvum oocysts. Journal of the American Water Works Association 88(9):131–136.

Hansen, J. S., and J. E. Ongerth. 1991. Effects of time and watershed characteristics on the concentration of Cryptosporidium oocysts in river water. Applied and Environmental Microbiology 57(10):2790–2795.

Harada, K., M. Oshikata, H. Uchida, M. Suzuki, F. Kondo, K. Sato, Y. Ueno, S.Z. Yu, G. Chen and G.C. Chen. 1996. Detection and identification of microcystins in the drinking water of Haimen City, China. Nat. Toxins 4(6):277–283.

Hayman, J. 1992. Beyond the Barcoo—probably human tropical cyanobacterial poisoning in outback Australia. Med. J. Aust. 157:794–796.

Helsel, D. R., and T. A. Cohn. 1988. Estimation of descriptive statistics for multiply censored water quality data. Water Resources Research 24(12): 1997–2004.

Hosking, J. R. M., and J. R. Wallis. 1997. Regional Frequency Analysis: An Approach Based on L-Moments. London: Cambridge University Press.

Hoxie, N. J., J. P. Davis, J. M. Vergeront, R. D. Nashold, and K. A. Blair. 1997. Cryptosporidiosis-associated mortality following a massive waterborne outbreak in Milwaukee, Wisconsin. American Journal of Public Health 87(12):2032–2035.

Hsu, F. C., Y. S. Shieh, J. van Duin, M. J. Beekwilder, and M. D. Sobsey. 1995. Genotyping male-specific RNA coliphages by hybridization with oligonucleotide probes. Applied and Environmental Microbiology 61(11):3960–3966.

International Life Sciences Institute (ILSI). 1998. Comprehensive Watershed Monitoring: A Framework for the New York City Reservoirs. Washington, DC: ILSI Risk Science Institute.


Juranek, D. 1997. CDC. E-mail from Dennis Juranek to the National Research Council Committee.

Juranek, D. 1999. CDC. E-mail from Dennis Juranek to the National Research Council Committee, May 1999.


Korich, D. G., J. R. Mead, M. S. Madore, N. A. Sinclair, and C. R. Sterling. 1990. Effects of ozone, chlorine dioxide, chlorine, and monochloramine on Cryptosporidium parvum oocyst viability. Applied and Environmental Microbiology 56(5):1423–1428.


LeChevallier, M. W., W. Norton, M. Abbaszadegan, T. Atherholt, and J. Rosen. 1997. Variations in Giardia and Cryptosporidium in Source Water: Statistical Approaches to Analyzing ICR Data. In AWWA Water Quality Technology Conference. Denver, CO: American Water Works Association.

Likens, G. E., (ed). 1985. An Ecosystem Approach to Aquatic Ecology: Mirror Lake and Its Environment. New York, NY: Springer-Verlag.

Likens, G. E., and F. H. Bormann. 1995. Biogeochemistry of a Forested Ecosystem. 2nd Edition. New York, NY: Springer-Verlag.

Longabucco, P., and M. Rafferty. 1998. Analysis of material loading to Cannonsville Reservoir: Advantages of event-based sampling. Journal of Lake and Reservoir Management 14(2-3):197-212.


MacKenzie, W. R., W. L. Schell, B. A. Blair, D. G. Addiss, D. E. Peterson, N. J. Hoxie, J. J. Kazmierczak, and J. P. Davis. 1995. Massive outbreak of waterborne Cryptosporidium infection in Milwaukee, Wisconsin: Recurrence of illness and risk of secondary transmission. Clinical Infectious Diseases 21:57–62.

MacKenzie, W. R. 1998. Presentation given at the Risk Assessment Workshop for the NRC Committee to Review the New York City Watershed Management Strategy. Atlanta, GA. April 14–15, 1998.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Marx, R., and E. A. Goldstein. 1999. Under Attack: New York's Kensico and West Branch Reservoirs Confront Intensified Development. New York, NY: Natural Resources Defense Council.

McKnight, D. M., R. Harnish, R. L. Wershaw, J. S. Baron, and S. Schiff. 1997. Chemical characteristics of particulate, colloidal, and dissolved organic material in Loch Vale Watershed, Rocky Mountain National Park. Biogeochemistry 36:99–124.

Millard, P., K. Gensheimer, D. G. Addiss, D. M. Sosin, G. A. Beckett, A. Houck-Jankoski, and A. Hudson. 1994. An outbreak of cryptosporidiosis from fresh-pressed apple cider. Journal of the American Medical Association 272(20):1592–1596.

Miller, J. 1998. Presentation given at the Risk Assessment Workshop for the NRC Committee to Review the New York City Watershed Management Strategy. Atlanta, GA. April 14–15, 1998.

Morgan, B. J. T. 1992. Analysis of Quantal Response Data. London: Chapman and Hall.

National Research Council (NRC). 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, DC: National Academy Press.

New York City Department of Environmental Protection (NYC DEP). 1997a. Water Quality Surveillance Monitoring. Valhalla, NY: NYC DEP.

NYC DEP. 1997b. Results of Groundwater Monitoring in the Kensico Reservoir Watershed. July 1997. Valhalla, NY: NYC DEP.

NYC DEP. 1997c. The Relationship Between Phosphorus Loading, THM Precursors, and the Current 20 µg/L TP Guidance Value. December 1997. Valhalla, NY: NYC DEP.

NYC DEP. 1997d. Pathogen Laboratory: Standard Operating Procedure. January 1997. Valhalla, NY: NYC DEP.

NYC DEP. 1997e. Waterborne Disease Risk Assessment Program 1996 Annual Report. Corona, NY: NYC DEP.

NYC DEP. 1998a. Kensico Watershed Study Annual Research Report: April 1997–March 1998. Valhalla, NY: NYC DEP.

NYC DEP. 1998b. Distribution Sampling Plan. March 26, 1997. Valhalla, NY: NYC DEP.

NYC DEP. 1998c. DEP Pathogen Studies of Giardia spp., Cryptosporidium spp., and Enteric Viruses. January 31, 1998. Valhalla, NY: NYC DEP.

NYC DEP. 1998d. Calibrate and verify GWLF models for remaining Catskill/Delaware Reservoirs.

NYC DEP. 1998e. Waterborne Disease Risk Assessment Program 1997 Annual Report. Corona, NY: NYC DEP.

NYC DEP. 1999a. Filtration Avoidance Report for the Period January 1 through December 31, 1998. Corona, NY: NYC DEP.

NYC DEP. 1999b. Kensico Watershed Study Semi-annual Progress Report. Valhalla, NY: NYC DEP.

NYC DEP. 1999c. Waterborne Disease Risk Assessment Program 1998 Annual Report. Corona, NY: NYC DEP.

New York State Water Resources Institute (NYS WRI). 1997. Science for Whole Farm Planning: Cornell University Phase II Twelfth Quarter and Completion Report.

Nieminski, E. C., and J. E. Ongerth. 1995. Removing Giardia and Cryptosporidium by conventional treatment and direct filtration. J. Amer. Waterworks Assoc. 9:96–106.


Okhuysen, P. C., C. L. Chappell, C. R. Sterling, W. Jakubowski, and H. L. DuPont. 1998. Susceptibility and serologic response of health adults to reinfection with Cryptosporidium parvum. Infection and Immunity 66(2):441–443.

Ong, C., W. Moorehead, A. Ross, and J. Isaac-Renton. 1996. Studies of Giardia spp. and Cryptosporidium spp. in two adjacent watersheds. Applied and Environmental Microbiology 62(8):2798–2805.

Ongerth, J. E. 1989. Giardia cyst concentrations in river water. Journal of the American Water Works Association 81(9): 81–86.

Otsuki, A., and R. G. Wetzel. 1974. Calcium and total alkalinity budgets and calcium carbonate precipitation of a small hard-water lake. Arch. Hydrobiol. 73:14–30.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Owens, E. M. 1998. Development and testing of one-dimensional hydrothermal models of Cannonsville Reservoir. Journal of Lake and Reservoir Management 14(2-3):172–185.

Parker, J. F. W., and H. V. Smith. 1993. Destruction of Oocysts of Cryptosporidium parvum by Sand and Chlorine. Water Research 27(4):729–731.

Paul, J. H., J. B. Rose, J. Brown, E. A. Shinn, S. Miller, and S. H. Farrah. 1995. Viral tracer studies indicate contamination of marine waters by sewage disposal practices in Key Largo, Florida. Applied and Environmental Microbiology 61:2230–2234.

Payment, P., and E. Franco. 1993. Clostridium perfringens and somatic coliphages as indicators of the efficiency of drinking water treatment for viruses and protozoan cysts. Applied and Environmental Microbiology 59(8):2418–2424.

Payment, P., L. Richardson, J. Siemiatycki, R. Dewar, M. Edwardes and E. Franco. 1991. A randomized trial to evaluate the risk of gastrointestinal disease due to consumption of drinking water meeting current microbiological standards. Am. J. Public Health 81(6):703–708.

Payment, P., J. Siemiatycki, L. Richardson, G. Renaud, E. Franco and M. Prevost. 1997. A prospective epidemiological study of gastrointestinal health effects due to the consumption of drinking water. Int. J. Env. Health Res. 7:5–31.

Perz, J. F., F. K. Ennever, and S. M. Le Blancq. 1998. Cryptosporidium in tap water: Comparison of predicted risks with observed levels of disease. American Journal of Epidemiology 147(3):289–301.

Proctor, M. E., K. A. Blair, and J. P. Davis. 1998. Surveillance data for waterborne illness detection: an assessment following a massive waterborne outbreak of Cryptosporidium infection. Epidemiol. Infect. 120:43–54.

Purdue, E. M. 1984. Analytical constraints on the structural features of humic substances. Geochim. Cosmochim. Acta 48:1435–1442.


Ransome, M. E., T. N. Whitmore, and E. G. Carrington. 1993. Effect of disinfectants on the viability of Cryptosporidium parvum oocysts. Water Supply 11:75–89.

Regli, S., J. B. Rose, C. N. Haas, and C. P. Gerba. 1991. Modeling the risk from Giardia and viruses in drinking water. Journal of the American Water Works Association 83(11):76–84.

Rodman, J. S., F. Frost, and W. Jakubowski. 1998. Using nurse hotline calls for disease surveillance. Emerging Infectious Diseases 4(2):329–32.

Rose, J. B., C. N. Haas, and S. Regli. 1991. Risk assessment and the control of waterborne giardiasis. American Journal of Public Health 81:709–713.

Roseberry, A. M., and D. E. Burmaster. 1992. Log-normal distributions for water intake by children and adults. Risk Analysis 12(1):99–104.


Sacks, J. J., S. Lieb, L. M. Baldy, S. Berta, C. M. Paton, M. C. White, W. J. Bigler, and J. J. Witte. 1986. Epidemic campylobacteriosis associated with a community water supply. Am. J. Pub. Health 76(4):424-8.

Schwartz, J., R. Levin and K. Hodge. 1997. Drinking water turbidity and pediatric hospital use for gastrointestinal illness in Philadelphia. Epidemiology 8(6):607–609.

Shepherd, K. M., and A. P. Wyn-Jones. 1995. Evaluation of different filtration techniques for the concentration of Cryptosporidium oocysts from water. Water Science and Technology 31(5-6):425–429.

Slifko, T. R., D. Friedman, J. B. Rose, and W. R. Jakubowski. 1997. An In vitro Method for Detecting Infectious Cryptosporidium Oocysts with Cell Culture. Applied and Environmental Microbiology 63(9):3669–3675.

Smith, A. E., P. B. Ryan, and J. S. Evans. 1992. The effect of neglecting correlations when propagating uncertainty and estimating the population distribution of risk. Risk Analysis 12(4):467–474.

Soave, R., L. J. Davis, and D. Juranek. 1998. Meeting of the Infectious Disease Society of America.

Soong, F. S., E. Maynard, K. Kirke and C. Luke. 1992. Illness associated with blue-green algae. Med. J. Aust. 156:67.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×

Stern, D. 1996. Initial Investigation of the Sources and Sinks of Cryptosporidium spp. and Giardia spp. within the watersheds of the New York City Water Supply System. In New York City Water Supply Studies. Proceedings of a Symposium on Watershed Restoration Management. American Water Resources Association. Herndon, VA.

Stern, D. 1998. NYC DEP. E-mail memorandum to the National Research Council dated November 16, 1998.

Stepczuk, C. L., A. B. Martin, P. Longabucco, J. A. Bloomfield, and S. W. Effler. 1998a. Allochthonous contributions of THM precursors in a eutrophic reservoir. Journal of Lake and Reservoir Management 14(2-3):344–355.

Stepczuk, C. L., A. B. Martin, S. W. Effler, J. A. Bloomfield, and M. T. Auer. 1998b. Spatial and temporal patterns of THM precursors in a eutrophic reservoir. Journal of Lake and Reservoir Management 14(2-3): 356–366.

Stepczuk, C. L., E. M. Owens, S. W. Effler, J. A. Bloomfield, and M. T. Auer. 1998c. A modeling analysis of THM precursors for a eutrophic reservoir. Journal of Lake and Reservoir Management 14(2-3):367–378.

Stewart, M. H., D. M. Ferguson, R. De Leon, and W. D. Taylor. 1997. Monitoring Program To Determine Pathogen Occurrence In Relationship To Storm Events And Watershed Conditions. In AWWA Water Quality Technology Conference. Denver, CO: American Water Works Association.

Thurman, E. M. 1985. Organic Geochemistry of Natural Waters. Martinus Nijhoff/Dr. W. Junk Publishers, Dordrecht.

Turner, P. C., A. J. Gammie, K. Hollinrake and G. A. Codd. 1990. Pneumonia associated with contact with cyanobacteria. Br. Med. J. 300:1440–1441.


Ueno, Y., S. Nagata, T. Tsutsumi, A. Hasegawa, M. F. Watanabe, H. D. Park, G. C. Chen, G. Chen, and S. Z. Yu. 1996. Detection of microcystins, a blue-green algal hepatotoxin, in drinking water sampled in Haimen and Fusui, endemic areas of primary liver cancer in China, by highly sensitive immunoassay. Carcinogenesis 17(6):1317–1321.


Vose, D. 1996. Quantitative Risk Analysis: A Guide to Monte Carlo Simulation Modeling. New York, N.Y.: John Wiley.


Walker, F. R., Jr., and J. R. Stedinger. 1999. Fate and Transport Model of Cryptosporidium. Journal of Environmental Engineering 125(4):325–333.

Warne, D. 1998. NYC DEP. Memorandum to the National Research Council dated November 1998.

Wetzel, R. G., and A. Otsuki. 1974. Allochthonous organic carbon of a marl lake. Arch. Hydrobiol. 73:31–56.

Wetzel, R. G., P. G. Hatcher, and T. S. Bianchi. 1995. Natural photolysis by ultraviolet irradiance of recalcitrant dissolved organic matter to simple substrates for rapid bacterial metabolism. Limnol. Oceanogr. 40:1369–1380.

Wilson, M. A., A. M. Vassallo, E. M. Purdue, and J. H. Reuter. 1987. Compositional and solid-state nuclear magnetic resonance study of humic and fulvic acid fractions of soil organic matter. Anal. Chem. 59:551–558.

World Health Organization. 1998. Guidelines for Drinking Water Quality. Second Edition. Addendum to Volume 2: Health Criteria and Other Supporting Information. Geneva, Switzerland: World Health Organization.

Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 206
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 207
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 208
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 209
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 210
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 211
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 212
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 213
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 214
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 215
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 216
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 217
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 218
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 219
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 220
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 221
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 222
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 223
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 224
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 225
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 226
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 227
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 228
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 229
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 230
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 231
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 232
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 233
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 234
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 235
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 236
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 237
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 238
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 239
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 240
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 241
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 242
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 243
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 244
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 245
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 246
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 247
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 248
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 249
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 250
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 251
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 252
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 253
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 254
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 255
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 256
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 257
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 258
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 259
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 260
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 261
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 262
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 263
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 264
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 265
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 266
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 267
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 268
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 269
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 270
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 271
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 272
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 273
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 274
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 275
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 276
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 277
Suggested Citation:"6 Tools for Monitoring and Evaluation." National Research Council. 2000. Watershed Management for Potable Water Supply: Assessing the New York City Strategy. Washington, DC: The National Academies Press. doi: 10.17226/9677.
×
Page 278
Next: 7 Land Acquisition and Land Use Planning »
Watershed Management for Potable Water Supply: Assessing the New York City Strategy Get This Book
×
Buy Paperback | $65.00 Buy Ebook | $54.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

In 1997, New York City adopted a mammoth watershed agreement to protect its drinking water and avoid filtration of its large upstate surface water supply. Shortly thereafter, the NRC began an analysis of the agreement's scientific validity.

The resulting book finds New York City's watershed agreement to be a good template for proactive watershed management that, if properly implemented, will maintain high water quality. However, it cautions that the agreement is not a guarantee of permanent filtration avoidance because of changing regulations, uncertainties regarding pollution sources, advances in treatment technologies, and natural variations in watershed conditions.

The book recommends that New York City place its highest priority on pathogenic microorganisms in the watershed and direct its resources toward improving methods for detecting pathogens, understanding pathogen transport and fate, and demonstrating that best management practices will remove pathogens. Other recommendations, which are broadly applicable to surface water supplies across the country, target buffer zones, stormwater management, water quality monitoring, and effluent trading.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!