In 2012, New York, USA enacted the Sewage Pollution Right to Know (SPRtK) Act, which requires public notification of untreated and partially treated sewage discharges. With the passing of this law, New York joined 12 other states that have similar laws but none as comprehensive as New York's. As part of the SPRtK Act requirements, aggregated sewage discharge reports (SDRs) are made available on the web. For this study, we made use of one year's worth of SDRs to identify spatial and temporal patterns in sewage discharge incidents. The SDRs were strongly associated with the type of municipality, density and age of the treatment plant. New York has some of the oldest infrastructure in the USA, and this law enables the state environmental agency to document instances of failure and take corrective action. Proper implementation of the law would place information in the hands of the people and protect public health.

Introduction

Nearly three-quarters of the households in the USA are connected to centralized sewers. The remaining rely on decentralized forms of treatment such as septic tanks (Vedachalam et al., 2015). Primarily, there are two types of sewer systems: combined sewer systems (CSSs) and separate sanitary and stormwater systems (SSSs). CSSs are sewer systems that are a set of underground pipes that carry residential and industrial sanitary sewage at all times and additionally, may carry stormwater during rain events. During heavy precipitation, the combination of sanitary sewage and stormwater can overwhelm the capacity of municipal treatment plants leading to overflows. Such locations are designated combined sewer overflows (CSOs). CSSs are typically found in the older cities of Northeastern and Midwestern USA. As of 2004, there were approximately 828 CSSs in the USA with 9300 CSO discharge points that release more than 3 million m3 of untreated sewage to surface water bodies annually (United States Environmental Protection Agency (USEPA), 2004).

Compounding the problem of sewage management is the aging of water infrastructure in the USA. Complicated urban networks with old pipes magnify the scale of action. This situation is not unique to the USA, as other countries like Australia (Hardisty et al., 2013), Canada (Mirza & Haider, 2003), and Germany (Hummel & Lux, 2007) face some of the same issues as well. Specific to the USA, decentralized governance structure and the scale of funding needed to reshape the water infrastructure landscape has prevented strong federal action on this issue, while allowing for flexibility in how individual states respond (Rahm et al., 2013).

Despite the limitations, sustained action over the past two decades by the US Environmental Protection Agency, state governments and local environmental groups has resulted in several cities reducing their incidences of CSOs and a move toward separating sanitary sewage and stormwater. The impetus for this separation came from the USEPA's CSO Control Policy issued in 1994 that mandated communities to sharply reduce or eliminate CSOs to meet the goals set forth in the Clean Water Act (USEPA, 1994). Under this policy, cities with CSOs were required to establish short-term and long-term control plans (LTCPs) to manage the CSO discharges (Mee, 1997).

As a result of this policy and ensuing consent decree agreed upon between the USEPA, Department of Justice and the involved communities, cities with CSOs were left with three choices. Tibbetts (2005) lists two of the choices, but over time, communities have increasingly exercised a third option. They are: (1) increase capacity of wastewater treatment plants to accommodate the stormwater during heavy precipitation without the threat of overflow or discharge, (2) separate the stormwater and wastewater by constructing new stormwater drains that do not interact with the wastewater lines at any point, or (3) control the entry of stormwater into the combined systems in a decentralized manner through green infrastructure techniques such as rain gardens, infiltration ponds, porous pavements, etc.

SSSs are the other type of sewer systems found in cities that were settled much later. Right from the time of design, these systems separate stormwater from sanitary sewage. As a result, there are no overflows during wet weather events. It is no surprise that the separation between stormwater and sanitary sewage is one of the proposed actions for cities dealing with CSOs. However, these sanitary systems may have overflows due to pipe blockages, infiltration of groundwater, going beyond the designed capacity, etc. Such overflows are termed sanitary sewer overflows (SSOs). Figure 1 provides a visual representation of the differences between CSSs and SSSs during dry and wet weather.
Fig. 1.

(a) Combined sewer systems and (b) separate sanitary and stormwater systems, during dry and wet weather. Note: POTW refers to publicly owned treatment works, also known as wastewater treatment plants. Source:USEPA (2004).

Fig. 1.

(a) Combined sewer systems and (b) separate sanitary and stormwater systems, during dry and wet weather. Note: POTW refers to publicly owned treatment works, also known as wastewater treatment plants. Source:USEPA (2004).

Although the intense scrutiny of CSO discharges has forced federal action, SSOs are largely monitored by state environmental agencies and local municipalities. The USEPA estimates that 23,000–75,000 SSOs occur every year nationally (USEPA, 2015). Age of the pipe, capacity utilization, weather and the type of wastewater are some of the factors that influence whether and how often a system may experience SSO discharges. Individual states have taken various actions toward monitoring and enforcement to bring down the incidences of SSOs. In 2012, New York enacted the Sewage Pollution Right to Know (SPRtK) Act. The law requires that discharges of untreated and partially treated sewage discharges be reported by publicly owned treatment works (POTWs) and publicly owned sewer systems (POSSs) within two hours of discovery to the state environmental regulatory agency, New York State Department of Environmental Conservation (hereafter, NYSDEC) and local health departments, and within four hours of discovery to the public and adjoining municipalities (NYSDEC, 2015). In passing this law, New York joined a dozen other states in the nation that have similar public notification laws, including the neighboring state of Connecticut which approved a similar law earlier that year. The law became effective on May 1, 2013.

New York's law

The importance of clean drinking water cannot be overstated (Cutler & Miller, 2005). In addition, clean recreational water is an integral part of New York's economy and way of life, with thousands of large and small lakes, rivers and estuaries enveloping the state's landscape. Even small amounts of sewage discharge into drinking or recreational waters can release significant amounts of disease-causing pathogens, which can result in short-term and chronic illnesses. There was a sustained effort from environmental groups, particularly Riverkeeper – an organization focused on clean water in the Hudson River and the New York City region, to convince the state legislature to pass a sewage notification law. In June 2012, the Assembly and the Senate formally approved the SPRtK Act through resolutions A.10585A and S.6268D, respectively. Governor Andrew Cuomo signed the bill on August 9, 2012. The law became effective on May 1, 2013.

The law requires the notification of the discharge of untreated and partially treated sewage within two hours to the DEC and the local health department. This notification is required to include the following information: volume of discharge and extent of its treatment, expected duration of discharge, reason for the discharge and steps taken to contain it. The law also requires that the notification be sent to the public and chief elected officials of municipalities where the incident occurred and others that may be affected by the discharge. Further, the law requires the POTWs and POSSs to submit a more detailed report within five days (5-day report). As part of the law, NYSDEC is required to maintain and display all SDRs on a public website (NYSDEC, 2014a). A key aspect of this law is that it only pertains to SSOs, and excludes reporting of CSOs. In New York, there are 76 CSS facilities and nearly 1,000 CSO locations (NYSDEC, 2015). The NYSDEC maintains an easily accessible ‘Google’ map and a geographic information system layer that provides information about the location of these CSOs, general facility information, and receiving waterbodies affected by CSOs on its website (New York State, 2011). According to the NYSDEC, monitoring of CSOs is included within the LTCPs designed by individual municipalities, hence the law is intended to report only SSOs (Joe D. Mura, personal communication, November 5, 2014).

Public notification of sewage discharge

Just a few months before the New York law was approved by the legislature, the Connecticut General Assembly approved Public Act No. 12–11, ‘An Act Considering the Public's Right to Know of a Sewage Spill,’ which also required public notification in case of sewage discharges (State of Connecticut, 2012). The Connecticut law was more expansive than the one in New York as it covered CSO discharges, along with SSOs.

By requiring public notification of sewage discharge, New York and Connecticut joined 12 other states that already have laws on this issue. American Rivers (2007) published a comprehensive comparison of the public notification laws in 11 of these states, including a table that listed the various features of the states' laws. We include Michigan (which passed a law in 1994, but was missing from the list compiled by American Rivers), Connecticut and New York in that table to present a complete comparison of the state laws that exist today (Table 1). American River (2007) graded the notification system based on its extent and scope. Public notification programs in some states such as Alabama and South Carolina do not require notifications to the public, while some programs apply for only certain types of spill or in certain parts of the states (e.g., North Carolina). Experiences from the notification laws in other states reveal that making the law expansive is not sufficient. Despite several provisions, laws such as the ones in Alabama and Georgia score very low on the degree of implementation. Based on the metrics included in the table, the New York law is the most comprehensive sewage notification law currently existing in any US state. The authors make no attempt to rate the implementation of the laws in Michigan, Connecticut and New York, since no objective rubric was used to rate implementation by American Rivers in other states. In February 2013, a bill requiring public notification of SSO discharges was introduced in the US Senate by Sen. Frank Lautenburg (New Jersey) and Sen. Sheldon Whitehouse (Rhode Island) (Library of Congress, 2013). The Bill, S. 396 – Sewage Overflow Community Right-to-Know Act, was modeled after the New York law. The Senate Bill was not taken up for a floor discussion or voted on, and expired at the end of the Congressional term in January 2015.

Table 1.

Sewage notification laws across the USA.

ALIAGAKYMDMINCORSCTNVAWACTNY
Agency to be notified 
 State Environmental Agency (within 24 hours) − − 
 Health Department − − − − − 
 Public − − − − − 
Notification methods 
 Media − − − − − 
 Overflow signs − − − − − − − − − − 
 Annual report from each plant − − − − − − − − − − − 
 Direct notification (phone/internet) − − − − − − − − − − − − − 
 Degree of implementation: Lowest (1) – Highest (5)    
ALIAGAKYMDMINCORSCTNVAWACTNY
Agency to be notified 
 State Environmental Agency (within 24 hours) − − 
 Health Department − − − − − 
 Public − − − − − 
Notification methods 
 Media − − − − − 
 Overflow signs − − − − − − − − − − 
 Annual report from each plant − − − − − − − − − − − 
 Direct notification (phone/internet) − − − − − − − − − − − − − 
 Degree of implementation: Lowest (1) – Highest (5)    

Primary source:American Rivers (2007).

Notes: The abbreviations are: AL: Alabama, IA: Iowa, GA: Georgia, KY: Kentucky, MD: Maryland, MI: Michigan, NC: North Carolina, OR: Oregon, SC: South Carolina, TN: Tennessee, VA: Virginia, WA: Washington, CT: Connecticut, NY: New York. The text of the Michigan law was obtained from Michigan Legislature (1994). CT and NY are the most recent laws, approved in 2012.

In this article, we use one year's worth of SDRs posted by NYSDEC to identify patterns in sewage discharge incidents, including spatial and temporal characteristics. Further, we aggregate the reports by municipality and regress the per capita number of SDRs on structural and socio-economic characteristics of those municipalities to identify factors that explain heterogeneity in SDR incidences.

Methods

The SDRs were obtained from the NYSDEC website (see http://www.dec.ny.gov/chemical/90321.html), which hosts a summary of all discharge incidents in a sortable spreadsheet updated regularly. The time period of our analysis was one year: May 1, 2013 to April 30, 2014. Each report included information on 16 parameters relating to the time, location and nature of the incident. See Table S1 (available with the online version of this paper) for a summary of the information included in each SDR.

A total of 1,594 SDRs were identified within our study period. Duplicates and unusable entries were removed from the dataset, yielding 1,541 usable entries. The entries removed included identical duplicates (26 entries), duplicates with minor differences in time, spelling, and capitalization (15), near duplicates where one of the entries had missing information (9), and those entries where no discernable information on date of the incident was available (3). We had intended to obtain the 5-day reports for a subset of sewage discharge incidents to compare the details of the event with those provided in the 2-hour report. However, we were unable to obtain a reasonable sample to compare the two types of reports.

For the second part of the study, we aggregated the SDRs at the municipal level. To do so, the reports were assigned to a particular municipality in NY based on ‘town/city’ and ‘closest landmark’ features in the raw data. New York has three types of municipalities – city (C), town (T), and village (V). We analyzed the aggregated SDRs to identify factors that explain the heterogeneity in incidence of sewage discharges. The raw SDR counts for each municipality are impacted by a few outliers. A Box-Cox transformation test was unable to recommend a suitable transformation1. Hence, we decided to normalize the SDR counts by population to obtain normalized SDR. The different SDRs belonging to the same municipality were aggregated using the merge function in R through the use of data frames. An ordinary least squares (OLS) regression model was designed to test the explanatory power of independent variables in describing normalized SDR. The independent variables included demographic, physical and socio-economic characteristics of municipalities, along with measures of POTW performance, which were obtained from a variety of sources (Table 2). Definitions and summary of the variables are presented in Tables 3 and 4, respectively.

Table 2.

Data sources.

DatabaseData typeSource
SDRs SDR incidences NYSDEC (2014a
American Fact Finder Physical and socio-economic characteristics of NY municipalities US Census Bureau (2014)  
Fiscal Stress Monitoring System Fiscal score (0–100) New York State Office of State Comptroller (OSC) (2014)  
POTW Master database POTW characteristics NYSDEC (2014b)  
DatabaseData typeSource
SDRs SDR incidences NYSDEC (2014a
American Fact Finder Physical and socio-economic characteristics of NY municipalities US Census Bureau (2014)  
Fiscal Stress Monitoring System Fiscal score (0–100) New York State Office of State Comptroller (OSC) (2014)  
POTW Master database POTW characteristics NYSDEC (2014b)  
Table 3.

Variable definitions.

VariableDescription
NORMSDR Number of SDRs per 100,000 residents 
POP Population of the municipality from the American Community Survey (ACS) 2013 estimates. 
POPGROWTH Change in population (in percent) between 2000 and 2013 
VILLAGE Dummy for village 
TOWN Dummy for town 
CITY Dummy for city 
DENSITY Population per square mile of land area 
WATERAREA Area of municipality designated as open water (sq. mi) 
MHI Estimated annual median household income 
FISCALSCORE A measure of fiscal stress obtained by aggregating several fiscal indicators. Key: 0.00–0.45: No designation, 0.45–0.59: Susceptible to Fiscal Stress, 0.55–0.649: Moderate Fiscal Stress, and 0.65–1.00: Significant Fiscal Stress. 
POTWYEAR Year POTW built for municipality; if multiple plants service the same municipality then median is taken. 
FLOWUTIL Flow utilization of the POTW obtained by dividing actual flow by the design flow capacity of all POTWs serving the municipality 
VIOL Number of violations per million gallons per day of flow 
VariableDescription
NORMSDR Number of SDRs per 100,000 residents 
POP Population of the municipality from the American Community Survey (ACS) 2013 estimates. 
POPGROWTH Change in population (in percent) between 2000 and 2013 
VILLAGE Dummy for village 
TOWN Dummy for town 
CITY Dummy for city 
DENSITY Population per square mile of land area 
WATERAREA Area of municipality designated as open water (sq. mi) 
MHI Estimated annual median household income 
FISCALSCORE A measure of fiscal stress obtained by aggregating several fiscal indicators. Key: 0.00–0.45: No designation, 0.45–0.59: Susceptible to Fiscal Stress, 0.55–0.649: Moderate Fiscal Stress, and 0.65–1.00: Significant Fiscal Stress. 
POTWYEAR Year POTW built for municipality; if multiple plants service the same municipality then median is taken. 
FLOWUTIL Flow utilization of the POTW obtained by dividing actual flow by the design flow capacity of all POTWs serving the municipality 
VIOL Number of violations per million gallons per day of flow 
Table 4.

Variable summary.

VariableObservationsMinMaxMedianMeanStd. Dev.
NORMSDR 180 0.20 698.21 29.78 74.69 115.18 
POP 180 286 8,405,837 9,205.5 74,065.67 628,775.40 
POPGROWTH 180 −17.11 62.41 1.24 1.92 9.05 
VILLAGE 180 0.43 0.50 
TOWN 180 0.37 0.48 
CITY 180 0.21 0.41 
DENSITY 180 34.8 27,012.4 1,630.35 2,400.47 3,095.78 
WATERAREA 180 272.1 0.16 5.40 27.12 
MHI 180 23,336 175,733 52,423 59,013.69 27,306.51 
FISCALSCORE 163 0.708 0.14 0.17 0.16 
POTWYEAR 154 1903 1999 1968 1964.37 17.59 
FLOWUTIL 156 0.17 1.28 0.67 0.67 0.20 
VIOL 156 1,666.67 1.21 27.90 142.78 
VariableObservationsMinMaxMedianMeanStd. Dev.
NORMSDR 180 0.20 698.21 29.78 74.69 115.18 
POP 180 286 8,405,837 9,205.5 74,065.67 628,775.40 
POPGROWTH 180 −17.11 62.41 1.24 1.92 9.05 
VILLAGE 180 0.43 0.50 
TOWN 180 0.37 0.48 
CITY 180 0.21 0.41 
DENSITY 180 34.8 27,012.4 1,630.35 2,400.47 3,095.78 
WATERAREA 180 272.1 0.16 5.40 27.12 
MHI 180 23,336 175,733 52,423 59,013.69 27,306.51 
FISCALSCORE 163 0.708 0.14 0.17 0.16 
POTWYEAR 154 1903 1999 1968 1964.37 17.59 
FLOWUTIL 156 0.17 1.28 0.67 0.67 0.20 
VIOL 156 1,666.67 1.21 27.90 142.78 

Statistical models have been used in the literature to assess various aspects of sewage pollution. Most literature has focused on CSO pollution. Lee & Bang (2000) used a regression model to quantify the relationship between pollutant load (biochemical oxygen demand, suspended solids, etc.) and runoff. Wirahadikusumah et al. (2001) used Markov-chains-based models in conjunction with nonlinear optimization to model the deterioration of combined sewers. More recently, a binary logistic regression model was used to assess sewer deterioration using data from Austria (Fuchs-Hanusch et al., 2015). We are not aware of any prior work using SSO data from a large geographic region to assess explanatory factors. Despite the limitation of using data from one US state, the authors believe this is a novel work that adds to our understanding of water infrastructure management.

Results and discussion

For administrative purposes, NYSDEC divides the state into nine regions, with each region comprising of several counties (see http://www.dec.ny.gov/about/50230.html for a visual representation). Despite having just 7 percent of the state's population, Region 9, which includes counties in the western part of the state, was home to more than two-thirds of the SDRs (Figure 2). Of those reports, 87 percent come from a single county, Erie. Erie is home to Buffalo, an old industrial city that has an aging infrastructure. However, Buffalo was not the leading contributor to the Region 9 SDRs. The bulk of the reports came from the Town of Cheektowaga (population 87,665). Normalization by population yielded similar results. Responding to a news reporter, the Town Engineer from Cheektowaga diagnosed the problem as,
Fig. 2.

Frequency of reports by NYSDEC region.

Fig. 2.

Frequency of reports by NYSDEC region.

‘Cheektowaga's sanitary sewer system consists of aging clay pipes with joints every few feet’. Rainwater and melted snow seep into the pipes and when they fill, they overflow into local creeks. ‘It's wet-weather impact,’ [William] Pugh said. ‘Let's face it; these pipelines are 60-some-years old’ (Orr, 2014).

Figure 3(a) shows the incidence of SDRs across New York. A small number of municipalities in the western region record a large percentage of the overall SDRs. Normalizing the SDR incidence by population mitigates the outlier effect caused by a few municipalities like Cheektowaga, but the western region still dominates the rest of the state (Figure 3(b)).
Fig. 3.

(a) Incidence of sewage discharges. (b) Incidence of sewage discharges normalized by population (per 100,000 residents).

Fig. 3.

(a) Incidence of sewage discharges. (b) Incidence of sewage discharges normalized by population (per 100,000 residents).

Next, the SDRs were aggregated on a monthly basis (Figure 4). The months of June and October recorded above-average incidences. When precipitation data for the corresponding months are superimposed on the same plot, we observe a pattern2. There is high correlation between sewage discharge incidences and precipitation (R2 = 0.80, p < 0.001), and the incidences of sewage discharges increase during wetter months. During the months of April and October, Erie County experienced above-average precipitation (Cornell University, 2016). Except for April 2013, monthly precipitation in New York during the study period was within 75–125 percent of the historical average.
Fig. 4.

Frequency of reports by month.

Fig. 4.

Frequency of reports by month.

This hypothesis is confirmed when evaluating the primary reason for the discharge incident. Weather conditions far outweigh other factors that cause the discharge incident (Figure 5). A closer look at the SDR log reveals wet weather or heavy rain as the main descriptors of the ‘weather conditions.’ Other factors included insufficient system capacity and blockage due to intrusion by tree roots in cracks and fissures in the pipes.
Fig. 5.

Stated reason for the discharge incidents.

Fig. 5.

Stated reason for the discharge incidents.

Next, we aggregated the SDRs at the municipal level. There were 180 municipalities that recorded one or more incidents of sewage discharge. They included 77 villages, 66 towns and 37 cities. New York has 551 villages, 932 towns and 62 cities3. This suggests that only about 10 percent of the municipalities appear in our dataset. Cities are over-represented in our data, while towns are under-represented. A few municipalities, notably Cheektowaga (330), Hamburg (159), and West Seneca (124) impact the overall SDR count. Therefore, SDR counts were normalized by population to run the regression model.

The OLS regression estimates for SDRs are presented in Table 5. The columns present several variations of the model: basic (1), basic + fiscal score (2), basic + POTW performance measures (3), and full (4) model. We first discuss the results of the basic model (1). Population is positive and significant, indicating that despite normalizing SDRs by population, larger municipalities record higher rates of SDR than their smaller peers. Population growth rate is not significant, which suggests that sewage discharges are not related to recent population changes (in- or out-migration), but may be a result of longstanding issues related to infrastructure maintenance. As compared to villages, towns and cities are likely to record lower normalized SDRs, significant at the 1 percent level. Further, the physical attributes of the municipality play an important role. Low density municipalities record higher incidence of SDRs. Somewhat surprisingly, having a smaller water area in the municipality is also associated with high rates of SDR. Additionally, high SDRs are weakly associated with low median household income (MHI).

Table 5.

Regression estimates of normalized SDRs.

Variable(1)(2)(3)(4)
POP 2.28 × 10−5** 5.88 × 10−5 2.55 × 10−5*** 3.52 × 10−5 
(9.59 × 10−6(6.24 × 10−5(9.16 × 10−6(6.79 × 10−5
POPGROWTH −0.64 −0.33 −0.28 5.16 × 10−3 
(1.01) (1.07) (0.97) (1.03) 
VILLAGE – – – – 
TOWN −71.21*** −74.48*** −92.66*** −99.30*** 
(25.41) (28.19) (27.12) (31.19) 
CITY −81.24*** −73.45*** −79.20*** −72.10*** 
(17.49) (16.67) (18.50) (17.23) 
DENSITY −7.94 × 10−3** −8.14 × 10−3−9.65 × 10−3*** −1.05 × 10−2*** 
(3.71 × 10−3(4.19 × 10−3 (3.42 × 10−3(3.91 × 10−3
WATERAREA −0.20** −0.24** −0.13 −0.14 
(0.09) (0.11) (0.09) (0.13) 
MHI −6.00 × 10−4−6.12 × 10−4 −4.58 × 10−4 −4.40 × 10−4 
(3.26 × 10−4(3.64 × 10−4(3.45 × 10−4(3.93 × 10−4
FISCALSCORE  −49.10  −26.52 
 (44.43)  (45.28) 
POTWYEAR   0.72* 0.74* 
  (0.42) (0.44) 
FLOWUTIL   26.38 55.11 
  (59.75) (0.02) 
VIOL   −0.02 −0.01 
  (0.03) (0.02) 
CONSTANT 172.59*** 180.40*** −1,249.37 −1,306.97 
(32.29) (39.16) (805.89) (840.39) 
N 180 163 154 137 
F-stat (prob) 13.64 (0.000) 8.45 (0.000) 10.20 (0.000) 5.08 (0.000) 
R2 0.16 0.16 0.19 0.19 
Variable(1)(2)(3)(4)
POP 2.28 × 10−5** 5.88 × 10−5 2.55 × 10−5*** 3.52 × 10−5 
(9.59 × 10−6(6.24 × 10−5(9.16 × 10−6(6.79 × 10−5
POPGROWTH −0.64 −0.33 −0.28 5.16 × 10−3 
(1.01) (1.07) (0.97) (1.03) 
VILLAGE – – – – 
TOWN −71.21*** −74.48*** −92.66*** −99.30*** 
(25.41) (28.19) (27.12) (31.19) 
CITY −81.24*** −73.45*** −79.20*** −72.10*** 
(17.49) (16.67) (18.50) (17.23) 
DENSITY −7.94 × 10−3** −8.14 × 10−3−9.65 × 10−3*** −1.05 × 10−2*** 
(3.71 × 10−3(4.19 × 10−3 (3.42 × 10−3(3.91 × 10−3
WATERAREA −0.20** −0.24** −0.13 −0.14 
(0.09) (0.11) (0.09) (0.13) 
MHI −6.00 × 10−4−6.12 × 10−4 −4.58 × 10−4 −4.40 × 10−4 
(3.26 × 10−4(3.64 × 10−4(3.45 × 10−4(3.93 × 10−4
FISCALSCORE  −49.10  −26.52 
 (44.43)  (45.28) 
POTWYEAR   0.72* 0.74* 
  (0.42) (0.44) 
FLOWUTIL   26.38 55.11 
  (59.75) (0.02) 
VIOL   −0.02 −0.01 
  (0.03) (0.02) 
CONSTANT 172.59*** 180.40*** −1,249.37 −1,306.97 
(32.29) (39.16) (805.89) (840.39) 
N 180 163 154 137 
F-stat (prob) 13.64 (0.000) 8.45 (0.000) 10.20 (0.000) 5.08 (0.000) 
R2 0.16 0.16 0.19 0.19 

Note: The dependent variable is normalized number of SDRs. Parameter estimates are coefficients (robust standard errors in the parenthesis). *p < 0.1, **p < 0.05, ***p < 0.01. The number of models refer to (1) basic, (2) basic + fiscal score, (3) basic + POTW performance measures, and (4) full model.

In model (2), we add an additional control for fiscal stress faced by the municipality. The addition of fiscal score only marginally changes the estimates. Type of municipality and water area are significant at the same levels, while density is significant only at the 10 percent level. The effect of population is no longer seen in the model. This is likely due to the fact that some of the largest cities in our dataset (New York, Yonkers, Binghamton, etc.) were not evaluated on this metric, and as a result, do not have a fiscal score and thus drop out of the analysis. More importantly, fiscal score by itself is not significant in predicting the incidence of SDRs.

In model (3), we add POTW performance measures to the basic model (1). The POTW indicators are somewhat dated and missing several observations. The addition of these controls does not change the estimates obtained in the basic model (1) much. Population, type of government and density are significant. Water area and MHI are not significant. Year of POTW construction is significant, although in an unexpected way. Newer POTWs are weakly associated with higher SDRs. Flow utilization and violations rates do not impact SDRs.

Model (4) is our full model and includes the basic model (1), fiscal score and POTW performance measures. The estimates are similar to those seen in models (2) and (3). Municipal type, and density are significant at 1 percent level, and POTW year is significant at the 10 percent level. Population growth, water area, MHI, flow utilization, violation rate and fiscal score are not significant in estimating SDR rates.

Except for municipality size which is affected by the inclusion of fiscal score, nearly all the variables are consistent across the four models. Although not a formal condition, villages and cities in New York are more thickly populated than towns. Due to their dense urban core, they are also more likely to offer centralized water and sewer services. This very likely explains why villages record more SDRs than towns, many of whom rely on decentralized methods such as individual septic tanks for wastewater management. We have no explanation for why cities, which often have denser utility networks than villages, record lower SDRs. Having controlled for the type of municipality, we find that low density municipalities report higher SDRs. Such a result may seem counter-intuitive in light of our earlier observation that villages record higher SDRs than towns. A simple explanation may be at hand. Sparsely populated municipalities have long stretches of pipes that provide ‘last mile’ coverage to the residents. These pipes are also likely to be placed alongside or under roads, railway lines and bridges, making them highly vulnerable to failures. Indeed, a similar result was observed in the case of drinking water systems in the Mohawk-Hudson watershed of New York where low-density water districts were more likely to issue a boil water advisory during an extreme weather event (Vedachalam et al., 2014).

We have no formal hypothesis on the lack of significance exhibited by POTW characteristics such as flow utilization and violation rate. It is possible that there is little correlation between the physical attributes of a treatment plant and those of the sewer lines. Perhaps even more surprising is the weak, but positive significance exhibited by POTW age. One would expect older treatment plans to fail more than newer ones, however we observe the opposite here. There are several possible explanations. We have data only on the age of the plant, and not of the pipes. As suggested earlier, those two need not align closely. Second, the treatment plants could have undergone major upgrades since their construction, for which we have no corresponding data. As a result, older plants may perform better than some of their newer counterparts. An additional possible explanation could come from the size of the plant. In New York, plants built after 1980 have been typically smaller in size (Rahm et al., 2012). Given this association between age and size, it is possible that significance of the variable POTWYEAR is, in effect, suggesting that smaller plants record higher incidences of SDRs. Although we have data on size by treatment plant, demarcating them by municipality is a complex task since there are several instances where one municipality is served by multiple plants and where one plant services multiple jurisdictions.

This analysis, though unique in its use of a new dataset, is far from perfect. It uses data from one state during a single year. As it was the first year of the law's implementation, data recorded was not uniform, which resulted in some observations getting discarded. Aggregating data over a few years will result in more municipalities entering the model, making it more robust and representative. Although a few outliers affected the raw count of SDRs, the log transformation was a suitable alternative employed to preserve the data integrity. More complete data on independent variables such as age of the POTW, age and status of the sewage pipes, and municipality budget spent on infrastructure management would have improved the model. Although data on financing requests made by POTWs (or their respective municipalities) under the Clean Water State Revolving Fund are available (NYSEFC, 2014), processing of such data is complicated by the lack of one-to-one mapping between POTWs and municipal boundaries.

Despite the limitations, this model is an attempt to expand our understanding of sewage overflows in a US state that has some of the oldest infrastructure. The SPRtK Act implemented in New York is one of the most comprehensive of such laws in the USA, and even served as a blueprint for a national legislation that was introduced in the US Congress, but never voted on. The success of this law's implementation could make it a model not only nationally, but for other countries that may experience a surge in sewage overflows due to increasing climate volatility, excessive sewage volume or simultaneous pipe failures. Such a law would fit within the frameworks of several countries that have ‘freedom of information’ or ‘right to know’ acts, keeping local agencies and the public informed of contamination or hazards in the local water or recreational source.

Conclusion

The SPRtK Act allowed New York to join a dozen other states in the USA that have laws requiring public notification of sewage discharges. In the absence of an overarching federal law governing sewage discharges, such laws provide protection to the public from unintended contact with contaminated water and guarantee public health. The provisions contained within such laws, as well as their implementation, vary widely across various states. As per the guidelines designed by American Rivers (2007), New York's law is the most comprehensive sewage discharge notification law in the USA. Although the law does not cover CSO discharges, it is a model legislation, and one that bears striking resemblance to a bill introduced in the US Senate in 2013.

The New York law also requires that the 2-hour reports be publicly available, which became the basis for this study. We used SDRs over a 1-year period to assess spatial and temporal patterns. Nearly 90 percent of the reports came from a single county, the bulk of which were from a single town. Aging infrastructure was cited by municipal officials as the primary reason for the large number of reports from a single municipality. SDRs were strongly correlated with precipitation, with a large number of reports recorded in the wetter months of June and October. Inability to obtain the 5-day reports prevented us from comparing the information contained in the 2-hour reports to that provided in the 5-day reports.

Aggregating SDRs by municipality revealed additional insights. The type of municipality and density are significant variables that explain the variation in SDRs. The age of POTW is a weakly significant factor. Area of water and MHI are significant in some of the models, but are not consistent throughout. Rate of population growth, fiscal stress, and POTW characteristics such as flow utilization and violations do not impact SDRs. These results are based on SDRs recorded in one year. In subsequent years, more data will be recorded in the online database, ensuring a robust analysis. NYSDEC has already implemented the NY-alert system that sends out a warning message through email and text to subscribers when a sewage discharge occurs in the subscriber's area. With the online reporting and the alert system, the law is being implemented as originally intended. It remains to be seen if regular reporting and public awareness can lead to lower incidence of SSOs statewide.

Acknowledgements

This work was supported in part by the New York State Environmental Protection Fund via the Hudson River Estuary Program of the New York State Department of Environmental Conservation. Emily Vail and Koon Tang assisted in obtaining the 5-day reports.

1

Test for Box-Cox transformation checks for suitability of transforming the left hand side (dependent variable) to inverse, logarithmic or linear (no transformation) form.

2

Since Erie County accounted for over 75 percent of the SDRs, precipitation data for the county (Buffalo Niagara International Airport) was used to determine correlation. Precipitation data was obtained via the Northeast Regional Climate Center, NRCC (2014). See: http://climodtest.nrcc.cornell.edu/.

3

Not all municipalities have their own wastewater treatment plant. In order of likelihood, cities, villages and towns are likely to have their own treatment plant, or connect to a nearby one.

References

References
American Rivers
(
2007
).
What's in your water? The state of public notification in 11 U.S. states
.
Cornell University
(
2016
).
Monthly Maps: Percent of normal precipitation. Northeast Regional Climate Center. http://www.nrcc.cornell.edu/regional/monthly/monthly.html
(Accessed February 5 2016)
.
Fuchs-Hanusch
D.
Günther
M.
Möderl
M.
Muschalla
D.
(
2015
).
Cause and effect oriented sewer degradation evaluation to support scheduled inspection planning
.
Water Science and Technology
72
(
7
),
1176
1183
.
Hardisty
P. E.
Sivapalan
M.
Humphries
R.
(
2013
).
Determining a sustainable and economically optimal wastewater treatment and discharge strategy
.
Journal of Environmental Management
114
,
285
292
.
Hummel
D.
Lux
A.
(
2007
).
Population decline and infrastructure: the case of the German water supply system
.
Vienna Yearbook of Population Research
5
,
167
191
.
Lee
J. H.
Bang
K. W.
(
2000
).
Characterization of urban stormwater runoff
.
Water Research
34
(
6
),
1773
1780
.
Library of Congress
(
2013
).
S.396 - Sewage Overflow Community Right-to-Know Act. 113th Congress (2013–2014). Available at: https://www.congress.gov/bill/113th-congress/senate-bill/396.
Mee
S.
(
1997
).
Negotiated rulemaking and combined sewer overflows (CSOs): consensus saves ossification?
Boston College Environmental Affairs Law Review
25
,
213
245
.
Michigan Legislature
(
1994
).
Natural Resources and Environmental Protection Act (Excerpt): Act 451 of 1994. Section 324.3112a. Available at: https://www.legislature.mi.gov/%28S%28j02yhmaywpt4xsvbllghgqvd%29%29/mileg.aspx?page=getObject&objectName=mcl-324-3112a.
Mirza
M. S.
Haider
M.
(
2003
).
The State of Infrastructure in Canada: Implication for Infrastructure Planning and Policy. Technical report dated March 27. http://www.regionomics.com/infra/Draft-July03.pdf
.
New York State
(
2011
).
GIS dataset: State Pollution Discharge Elimination System. Available at: http://gis.ny.gov/gisdata/inventories/details.cfm?DSID=1010.
NYSDEC
(
2014a
).
Sewage Discharge Reports
.
NYSDEC
,
Albany, NY
.
NYSDEC
(
2014b
).
Descriptive Data of Municipal Wastewater Treatment Plants
.
NYSDEC
,
Albany, NY
.
NYSDEC
(
2015
).
Sewage Pollution Right to Know. Available at: http://www.dec.ny.gov/chemical/90315.html
(
Accessed May 28 2016)
.
NYSEFC
(
2014
).
2014 Final CWSRF Intended Use Plan. Available at: http://www.efc.ny.gov/Default.aspx?tabid=112.
Orr
S.
(
2014
).
Sewage Release Reports Improve, More Work Needed
.
Democrat & Chronicle
,
Rochester, NY
.
OSC
(
2014
).
List: Local Governments – 2-13 Data
.
New York State Office of the State Comptroller
,
Albany, NY
. Available at:
Rahm
B. G.
Vedachalam
S.
Shen
J. X.
McConnell
R. D.
Riha
S. J.
(
2012
).
A new assessment of wastewater infrastructure in the Hudson and Mohawk Basins
. In
Universities Council on Water Resources (UCOWR) Annual Conference
,
Santa Fe, NM
.
Rahm
B. G.
Vedachalam
S.
Shen
J.
Woodbury
P. B.
Riha
S. J.
(
2013
).
A watershed-scale goals approach to assessing and funding wastewater infrastructure
.
Journal of Environmental Management
129
,
124
133
.
State of Connecticut
(
2012
).
An Act Considering the Public's Right to Know of a Sewage Spill. Substitute Senate Bill No. 88, Public Act No. 12-11. Available at: http://www.cga.ct.gov/2012/ACT/PA/2012PA-00011-R00SB-00088-PA.htm.
Tibbetts
J.
(
2005
).
Combined sewer systems: down, dirty, and out of date
.
Environmental Health Perspectives
113
(
7
),
A465
A467
.
US Census Bureau
(
2014
).
Quickfacts
.
Washington, DC.
Available at:
USEPA
(
1994
).
Combined Sewer Overflow (CSO) Control Policy
.
United States Environmental Protection Agency
,
Washington, DC
. Available at:
USEPA
(
2004
).
Report to Congress: Impacts and Control of CSOs and SSOs. EPA 833-R-04-001. August 26, 2004. http://cfpub.epa.gov/npdes/cso/cpolicyreport2004.cfm
.
USEPA
(
2015
).
Sanitary Sewer Overflows and Peak Flows
.
United States Environmental Protection Agency
,
Washington, DC
.
Available at: http://water.epa.gov/polwaste/npdes/sso/ (Accessed May 28 2016)
.
Vedachalam
S.
Vanka
V. S.
Riha
S. J.
(
2015
).
Reevaluating onsite wastewater systems: expert recommendations and municipal decision-making
.
Water Policy
17
(
6
),
1062
1078
.
Wirahadikusumah
R.
Abraham
D.
Iseley
T.
(
2001
).
Challenging issues in modeling deterioration of combined sewers
.
Journal of Infrastructure Systems
7
(
2
),
77
84
.

Supplementary data