Using local sources (roof runoff, stormwater, graywater, and onsite wastewater) to meet non-potable water demands can minimize potable water use in buildings and increase supply reliability. In 2017, an Independent Advisory Panel developed a risk-based framework to identify pathogen log reduction targets (LRTs) for onsite non-potable water systems (ONWSs). Subsequently, California's legislature mandated the development and adoption of regulations—including risk-based LRTs—for use in multifamily residential, commercial, and mixed-use buildings. A California Expert Panel was convened in 2021 to (1) update the LRT requirements using new, quantitative pathogen data and (2) propose treatment trains capable of meeting the updated LRTs. This paper presents the updated risk-based LRTs for multiple pathogens (viruses, protozoa, and bacteria) and an expanded set of end-uses including toilet flushing, clothes washing, irrigation, dust and fire suppression, car washing, and decorative fountains. The updated 95th percentile LRTs required for each source water, pathogen, and end-use were typically within 1-log10 of the 2017 LRTs regardless of the approach used to estimate pathogen concentrations. LRT requirements decreased with influent pathogen concentrations from wastewater to graywater to stormwater to roof runoff. Cost and footprint estimates provide details on the capital, operations and maintenance, and siting requirements for ONWS implementation.

  • Risk-based log reduction targets (LRTs) and model treatment trains were developed for onsite non-potable water systems (ONWSs) using new pathogen data.

  • Convergence of new LRTs with earlier values provides confidence for development of ONWS standards.

  • Similarity of pathogen distributions in the US, Europe, and Australia suggests LRTs could apply across a wide geographic region, though additional monitoring is recommended.

Graphical Abstract

Graphical Abstract
Graphical Abstract

Interest in onsite non-potable water systems (ONWSs) continues to grow globally as a way to reduce potable water usage and enhance water supply reliability. Communities from Europe to North America to Asia are increasingly collecting and treating decentralized waste streams, stormwater, roof runoff (i.e., rainwater), and other sources at building-scale for multiple end-uses including toilet flushing, irrigation, and cooling towers (Kehoe & Nokhoudian 2022). Numerous US states and regions are pursuing uniform regulations for the implementation of ONWS including Colorado, Hawaii, Minnesota, Washington, D.C., and the city and county of San Francisco. In California, the State Water Resources Control Board (State Water Board) Division of Drinking Water (DDW) is currently developing statewide ONWS regulations to comply with a legislative mandate (California State Water Resources Control Board 2022).

One key element of future regulations will be to specify risk-based pathogen reduction requirements necessary to meet established public health goals. The quantitative microbial risk assessment (QMRA) has been used to estimate adverse outcomes and public health risks (Haas et al. 1999; World Health Organization 2016) and aid in risk management, risk communication, and decision making (Beaudequin et al. 2015; Petterson & Ashbolt 2016). One risk management strategy is to specify treatment requirements to reduce pathogen concentrations in a source water down to acceptable levels based on a human health benchmark. A previous effort by Sharvelle et al. (2017) developed a framework for identifying the log reduction targets (LRTs) necessary to reduce the risk of infections associated with exposure to alternative ONWS source waters down to a threshold of 1 in 10,000 (10−4) infections/person/year. The LRTs were independently developed for three target pathogens including virus, protozoa, and bacteria. Given the dependence of the LRT requirements on the concentration of pathogens in the source waters, separate LRTs were established for onsite wastewater, graywater, stormwater, and roof runoff.

One key limitation highlighted by Sharvelle et al. (2017) was the relative dearth of data describing the distribution of pathogens in many ONWS source waters. The availability of data led to varying degrees of confidence in the resulting LRT requirements. Since 2017, however, several key studies have been conducted to better characterize pathogen concentrations in ONWS source waters. The use of the new pathogen data provides an opportunity to refine the LRT requirements for indoor uses and irrigation, and additional end-uses including fire suppression, car washing, and indoor decorative fountains.

This study presents the outcomes of an Expert Panel (Panel) that was convened to assist the State Water Board with ONWS regulatory development in California (Olivieri et al. 2021). The goal of the study was to utilize new, recently published pathogen datasets to refine the ONWS LRTs, compare the findings with previously developed LRTs, and recommend updated LRTs. The study also developed cost estimates for model treatment trains capable of complying with the new LRTs. The findings provide guidance for regulatory development in other jurisdictions by identifying recommended datasets and providing a consistent framework for establishing LRTs for ONWSs serving greater than 500 people.

Source waters

Four ONWS source waters were considered in the development of LRTs: (1) onsite wastewater, (2) graywater, (3) stormwater, and (4) roof runoff. Onsite wastewater is defined as wastewater containing inputs from decentralized blackwater (e.g., toilets and/or kitchen sources) and graywater sources (e.g., bathroom sinks, showers, bathtubs, clothes washers, and laundry sinks). Stormwater is defined as runoff from rainfall or snow that can be measured downstream in a pipe, culvert, or stream after a precipitation event and may be in an engineered feature, running over the ground surface, or percolating into the shallow subsurface and then resurfacing as seeps (National Research Council 2009). Stormwater is broadly used to include all urban runoff that is accessible to people and pets. Roof runoff is defined as the capture of rainfall from rooftops and its storage in barrels or cisterns (i.e., rainwater harvesting). For this study, runoff water captured from roof areas with public access was considered stormwater.

Source water pathogen concentrations

A set of screening criteria was used to evaluate approximately 40 papers and studies that measured pathogens in ONWS source waters (Olivieri et al. 2021). The criteria for ‘ideal’ pathogen datasets were based on Pecson et al. (2022) and Schoen et al. (2017):

  • Large sample size.

  • Monitoring of multiple locations over time.

  • Freshly collected samples (i.e., not stored prior to enumeration).

  • High method sensitivity.

  • Pathogens enumerated with methods that are compatible with dose–response functions.

  • Targets human-infectious strains or groups.

  • Report raw data including recovery and limit of detection.

The focus of the review was on papers published after the 2017 evaluation completed by Sharvelle et al. (2017). Additional screening criteria for the literature review included the following: (a) exclusive use of primary literature, (b) use of studies conducted in water matrices consistent with the four relevant ONWS source waters, (c) study locations preferentially in the US or Canada, and (d) study dates later than 2016 that were not included in the analysis by Sharvelle et al. (2017). The updated literature review identified both a recommended dataset for each source water/pathogen combination along with one or two alternate datasets. Summaries of the datasets are provided in Table 1 for onsite wastewater, graywater, stormwater, and roof runoff. Note that the datasets were described by distributions rather than point estimates to capture variability in the pathogen concentrations. Additional details on the modeling of the distributions are available in the Supplementary Material (see Modeling of ONWS Datasets section).

Table 1

Recommended (Rec.) and alternate (Alt.) datasets of pathogen concentrations for all source waters

Source water/pathogenDatasetData sourceDistributionaUnitsPercent detected
Municipal Wastewater (MW) and Onsite Wastewater (OW) 
Cryptosporidium Rec. (MW) Pecson et al. (2022)  Normal (1.7, 0.4) oocysts/L 98% (n=120) 
Alt. (OW) Kothari et al. (2020)  Normal (1.4, 0.6)b oocysts/L 12% (n=25) 
Giardia Rec. (MW) Pecson et al. (2022)  Normal (4.0, 0.4) cysts/L 100% (n=120) 
Alt. (OW) Kothari et al. (2020)  Normal (2.0, 0.9)b cysts/L 64% (n=25) 
Norovirus II Rec. (MW) Pecson et al. (2022)  Normal (4.0, 1.2) gc/L 72% (n=122) 
Alt. (OW) Kothari et al. (2020)  Normal (1.8, 2.3) gc/L 38% (n=13) 
Adenovirus Rec. (MW) Pecson et al. (2022)  Normal (2.8, 1.0) MPN/L 84% (n=122) 
Alt. (OW) Kothari et al. (2020)  Normal (4.9, 0.5) MPN/L 67% (n=15)c 
Enterovirus Rec. (MW) Pecson et al. (2022)  Normal (3.2, 1.0) MPN/L 95% (n=122) 
Alt. (OW) Kothari et al. (2020)  Normal (1.6, 0.3) MPN/L 14% (n=14) 
Graywater 
Cryptosporidium Rec. Pecson et al. (2022)  Normal (1.7, 0.4)d oocysts/L 98% (n=120) 
Alt. Jahne et al. (2017)  Modeled Dataset oocysts/L – 
Giardia Rec. Pecson et al. (2022)  Normal (4.0, 0.4)d cysts/L 100% (n=120) 
Alt. Jahne et al. (2017)  Modeled Dataset cysts/L – 
Norovirus II Rec. Pecson et al. (2022)  Normal (4.0, 1.2)d gc/L 72% (n=122) 
Alt. Jahne et al. (2020)  Normal (2.0, 0.1) gc/L 6% (n=50) 
Adenovirus Rec. Pecson et al. (2022)  Normal (2.8, 1.0)d MPN/L 84% (n=122) 
Alt. Jahne et al. (2020)  Normal (1.9, 0.4) gc/L 14% (n=50) 
Enterovirus Rec. Pecson et al. (2022)  Normal (3.2, 1.0)d MPN/L 95% (n=122) 
Stormwater 
Cryptosporidium Rec. Pecson et al. (2022)  Normal (1.7, 0.4)e oocysts/L 98% (n=120) 
Giardia Rec. Pecson et al. (2022)  Normal (4.0, 0.4)e cysts/L 100% (n=120) 
Norovirus II Rec. Pecson et al. (2022)  Normal (4.0, 1.2)e gc/L 72% (n=122) 
Adenovirus Rec. Pecson et al. (2022)  Normal (2.8, 1.0)e MPN/L 84% (n=122) 
Enterovirus Rec. Pecson et al. (2022)  Normal (3.2, 1.0)e MPN/L 95% (n=122) 
Roof runoff 
Giardia Option A Alja'fari et al. (2022)  Normal (−1.4, 0.3) cysts/L 5% (n=79) 
Option B Alja'fari et al. (2022)  Uniform (−0.7, 1.2) cysts/L 5% (n=79) 
Salmonella Option A Alja'fari et al. (2022)  Normal (−0.2, 0.6) cells/L 9% (n=79) 
Option B Alja'fari et al. (2022)  Uniform (0.9, 2.4) cells/L 9% (n=79) 
Campylobacter Option A Alja'fari et al. (2022)  Normal (−0.4, 0.2) cells/L 3% (n=79) 
Option B Alja'fari et al. (2022)  Point Estimate (0.6) cells/L 3% (n=79) 
Source water/pathogenDatasetData sourceDistributionaUnitsPercent detected
Municipal Wastewater (MW) and Onsite Wastewater (OW) 
Cryptosporidium Rec. (MW) Pecson et al. (2022)  Normal (1.7, 0.4) oocysts/L 98% (n=120) 
Alt. (OW) Kothari et al. (2020)  Normal (1.4, 0.6)b oocysts/L 12% (n=25) 
Giardia Rec. (MW) Pecson et al. (2022)  Normal (4.0, 0.4) cysts/L 100% (n=120) 
Alt. (OW) Kothari et al. (2020)  Normal (2.0, 0.9)b cysts/L 64% (n=25) 
Norovirus II Rec. (MW) Pecson et al. (2022)  Normal (4.0, 1.2) gc/L 72% (n=122) 
Alt. (OW) Kothari et al. (2020)  Normal (1.8, 2.3) gc/L 38% (n=13) 
Adenovirus Rec. (MW) Pecson et al. (2022)  Normal (2.8, 1.0) MPN/L 84% (n=122) 
Alt. (OW) Kothari et al. (2020)  Normal (4.9, 0.5) MPN/L 67% (n=15)c 
Enterovirus Rec. (MW) Pecson et al. (2022)  Normal (3.2, 1.0) MPN/L 95% (n=122) 
Alt. (OW) Kothari et al. (2020)  Normal (1.6, 0.3) MPN/L 14% (n=14) 
Graywater 
Cryptosporidium Rec. Pecson et al. (2022)  Normal (1.7, 0.4)d oocysts/L 98% (n=120) 
Alt. Jahne et al. (2017)  Modeled Dataset oocysts/L – 
Giardia Rec. Pecson et al. (2022)  Normal (4.0, 0.4)d cysts/L 100% (n=120) 
Alt. Jahne et al. (2017)  Modeled Dataset cysts/L – 
Norovirus II Rec. Pecson et al. (2022)  Normal (4.0, 1.2)d gc/L 72% (n=122) 
Alt. Jahne et al. (2020)  Normal (2.0, 0.1) gc/L 6% (n=50) 
Adenovirus Rec. Pecson et al. (2022)  Normal (2.8, 1.0)d MPN/L 84% (n=122) 
Alt. Jahne et al. (2020)  Normal (1.9, 0.4) gc/L 14% (n=50) 
Enterovirus Rec. Pecson et al. (2022)  Normal (3.2, 1.0)d MPN/L 95% (n=122) 
Stormwater 
Cryptosporidium Rec. Pecson et al. (2022)  Normal (1.7, 0.4)e oocysts/L 98% (n=120) 
Giardia Rec. Pecson et al. (2022)  Normal (4.0, 0.4)e cysts/L 100% (n=120) 
Norovirus II Rec. Pecson et al. (2022)  Normal (4.0, 1.2)e gc/L 72% (n=122) 
Adenovirus Rec. Pecson et al. (2022)  Normal (2.8, 1.0)e MPN/L 84% (n=122) 
Enterovirus Rec. Pecson et al. (2022)  Normal (3.2, 1.0)e MPN/L 95% (n=122) 
Roof runoff 
Giardia Option A Alja'fari et al. (2022)  Normal (−1.4, 0.3) cysts/L 5% (n=79) 
Option B Alja'fari et al. (2022)  Uniform (−0.7, 1.2) cysts/L 5% (n=79) 
Salmonella Option A Alja'fari et al. (2022)  Normal (−0.2, 0.6) cells/L 9% (n=79) 
Option B Alja'fari et al. (2022)  Uniform (0.9, 2.4) cells/L 9% (n=79) 
Campylobacter Option A Alja'fari et al. (2022)  Normal (−0.4, 0.2) cells/L 3% (n=79) 
Option B Alja'fari et al. (2022)  Point Estimate (0.6) cells/L 3% (n=79) 

aValues are log10 transformed. Normal distribution parameters listed as (mean, standard deviation). Uniform distribution parameters listed as (minimum, maximum).

b16 values did not include recovery data. For consistency, the average recovery of the dataset was applied.

cFive of 15 values were greater than the measured value. Greater than values were set to the reported value.

dA 2-log reduction in mean concentrations was applied to municipal wastewater data with the assumption that graywater contains 1% of the fecal load of municipal wastewater.

eMean values of distributions reduced by 1-log or 3-log to account for low or moderate dilution of municipal wastewater, respectively.

Municipal wastewater and onsite wastewater

The literature review identified only a small number of new datasets that characterized pathogen concentrations in onsite wastewater. To estimate concentrations, the Panel relied on untreated municipal wastewater (rather than onsite wastewater) based on the following factors: (1) the lack of reliable onsite wastewater data, (2) anticipated building-scale projects serving greater than 500–1,000 people, and (3) the robust available data on waterborne pathogens in municipal wastewater. The Panel identified Pecson et al. (2022) as a robust dataset including 120 culture- and molecular-based samples from five municipal wastewater facilities representing approximately 10 million people in California. The Panel also used a limited empirical dataset characterizing onsite wastewater to estimate LRTs (Kothari et al. 2020) and compared the results against estimated LRTs for untreated municipal wastewater.

Graywater

Insufficient data were available to make reliable estimates of graywater pathogen densities. Consequently, three approaches were assessed to identify potential treatment criteria: (1) estimating graywater pathogen concentrations based on a 10−2 dilution of untreated municipal wastewater assuming 100% occurrence; (2) simulated epidemiological dataset of Cryptosporidium and Giardia density and occurrence in the community based on Jahne et al. (2017), and (3) limited measured pathogen data (i.e., norovirus GII and adenovirus) provided in Jahne et al. (2020).

Stormwater

Given the limited pathogen information for stormwater, the Panel assumed that the range of stormwater pathogen concentrations could be bounded by a low dilution (10−1) and a high dilution (10−3) of untreated municipal wastewater. The dilution range is consistent with Schoen et al. (2017) who analyzed stormwater pathogen data collected by Bambic et al. (2011). The Panel relied on the same untreated municipal wastewater dataset to represent baseline pathogen concentrations, which were then diluted to estimate treatment requirements (Pecson et al. 2022). There were not sufficient empirical measurements of pathogens in stormwater to provide better estimates than the ‘sewage dilution’ approach.

Roof runoff

Limited pathogen information is available for roof runoff. The Panel relied on new data by Alja'fari et al. (2022) to represent bacteria (Campylobacter and Salmonella) and protozoa (Giardia). Due to the large number of non-detects (NDs) in the dataset, the data were evaluated as both log-normal (Option A) and uniform (Option B) distributions. Unlike the other source waters, the Panel did not identify a recommended and alternate dataset due to the limited availability of data.

Exposure

Exposures to ONWS waters were estimated based on anticipated end-uses and exposure events: (1) toilet flushing, (2) clothes washing, (3) accidental cross-connection, (4) unrestricted irrigation, (5) fire suppression, (6) car washing, and (7) indoor decorative fountains (Table 2). Additionally, three combinations of these end-uses were evaluated, all of which included toilet flushing, clothes washing, and accidental cross-connection as the base uses to represent combined indoor use. Indoor Use 1 included the three base exposures plus decorative fountains modeled as a uniform distribution. Indoor Use 2 included the base three exposures plus decorative fountains modeled as a log10-normal distribution. Indoor Use 3 included only the base three exposures.

Table 2

Summary of exposure assumptions including ingestion volume, frequency of exposure, and fraction of population exposed

End-useIngested volume (l/day)Use frequency (days/year)Fraction of population exposed
Toilet flushinga 3E-05 365 
Clothes washinga 1E-05 100 
Cross-connectiona 0.1 
Unrestricted irrigation and dust suppressiona 1E-03 50 
Fire suppressionb 2E-03 20 
Car washinga 1E-03 12 
Decorative fountainsc    
 Log10-normal N (−4.05, 3.81)d 50 
 Uniform U (6.00E-05, 3.79E-03)e 50 
Indoor use 1 Combined exposure due to toilet flushing, clothes washing, accidental cross-connection, and decorative fountains (modeled as a uniform distribution). 
Indoor use 2 Combined exposure due to toilet flushing, clothes washing, accidental cross-connection, and decorative fountains (modeled as a log10-normal distribution). 
Indoor use 3 Combined exposure due to toilet flushing, clothes washing, and accidental cross-connection. 
End-useIngested volume (l/day)Use frequency (days/year)Fraction of population exposed
Toilet flushinga 3E-05 365 
Clothes washinga 1E-05 100 
Cross-connectiona 0.1 
Unrestricted irrigation and dust suppressiona 1E-03 50 
Fire suppressionb 2E-03 20 
Car washinga 1E-03 12 
Decorative fountainsc    
 Log10-normal N (−4.05, 3.81)d 50 
 Uniform U (6.00E-05, 3.79E-03)e 50 
Indoor use 1 Combined exposure due to toilet flushing, clothes washing, accidental cross-connection, and decorative fountains (modeled as a uniform distribution). 
Indoor use 2 Combined exposure due to toilet flushing, clothes washing, accidental cross-connection, and decorative fountains (modeled as a log10-normal distribution). 
Indoor use 3 Combined exposure due to toilet flushing, clothes washing, and accidental cross-connection. 

aNRMMC 2006. For car washing, assumed similar exposure as garden irrigation.

dValues are log10 transformed. Distribution values listed as (mean, standard deviation).

eDistribution values listed as (min, max).

The exposure volumes due to toilet flushing, clothes washing, unrestricted irrigation and dust suppression, and accidental cross-connection were assumed to be the same as those used by Sharvelle et al. (2017). The volumes and frequency of use for these four exposures were adapted from the Australian guidelines for water recycling (NRMMC 2006) (Table 2). For all end-uses, it was assumed that 100% of the population was exposed to accidental ingestion with the exception of the cross-connection exposure which assumed 10% of the population would be exposed.

This study also evaluated three new end-uses including fire suppression, car washing, and indoor decorative fountains. For fire suppression, exposure volumes were based on a health risk assessment of occupational exposure for firefighters (Water Services Association of Australia 2004) and assumed recycled water would be used for fire-fighting 20 days/year. For car washing, exposures were estimated based on indirect ingestion that occurs during hand washing a car. Exposure volume was estimated to be similar to that of routine ingestion during garden irrigation reported in the Australian guidelines for water recycling (NRMMC 2006). One car washing event per month was assumed. For decorative fountains, exposure volumes were evaluated using both a uniform and a log10-normal distribution based on values from a study that measured ingestion from spray exposures (Sinclair et al. 2016). Exposures to decorative fountain spray was assumed to occur 50 days/year. For all new end-uses evaluated, it was assumed that 100% of the population was exposed to accidental ingestion.

Pathogen dose–response

This study utilized the same dose–response models as Sharvelle et al. (2017) for consistency. A summary of the models used is available in the Supplementary Material (see Pathogen Dose–Response Functions section and the summary of values used in Table ES-1).

QMRA model

LRTs were calculated to achieve a tolerable risk goal of 10−4 infections/person/year. This annual risk goal underlies the EPA's Surface Water Treatment Rule and is the explicit goal of California's potable reuse regulations for groundwater recharge and surface water augmentation (Hultquist 2016; DDW 2018). The QMRA model used to calculate the LRT that meets the annual risk goal uses the standard QMRA equations (Haas et al. 1999):
formula
where is the annual probability of infection; is the number of days of exposure; DR is the dose–response function for the reference pathogen; is the volume of water ingested on day n; is the pathogen concentration in the source water on day n; LRT is the log removal target.

A Monte Carlo analysis was used to capture inherent variability in pathogen concentrations and exposure volumes on a given day (Tables 1 and 2). Risk on a given day was calculated from the accumulated pathogen dose from all exposure types occurring on that day. A single LRT value was used for every day modeled in the year, and the resulting annual probability of infection was compared to the annual risk goal of 10−4 infections/person/year. LRT values were in 0.1 increments starting at 0 up to 15. The LRT value resulting in the smallest difference between the calculated annual probability of infection and the annual risk goal was stored as the LRT for a single year. This process was simulated 10,000 times to develop a distribution of LRTs for each water source, reference pathogen, and end-use. The 95th percentile value from the distributions was used to select a representative LRT in alignment with the approach used by Sharvelle et al. (2017). For the LRT recommendations, the 95th percentile value was rounded up to the nearest 0.5 log increment. This approach ensured that the recommended LRT met the annual risk goal of 10−4 infections/person/year and in most cases resulted in an annual risk well below 10−4.

To ensure that the current modeling approach produced results that are consistent with previous work, the method was used with the inputs and assumptions from Schoen et al. (2017) to ensure the same 95th percentile LRTs were calculated. For each water source, each pathogen, and each use type modeled, the current method was able to replicate the results from Schoen et al. (2017) within 0–0.4 LRTs (results not shown).

Capital and operation and maintenance estimates

A planning-level cost analysis was conducted to understand the impacts of the regulatory approach on ONWS implementation. Cost and layout estimates were developed for representative treatment trains that could be implemented to meet California's anticipated LRTs. For each source water, multiple system suppliers were consulted to provide a range of budgetary estimates for capital and operation and maintenance (O&M) costs along with site layout requirements. Two different system capacities were considered for each source water type to cover the typical range of building-scale systems used in California. The following system capacities were considered for estimating capital and operational costs: onsite wastewater (19,000 and 57,000 L/day), graywater (5,700 and 19,000 L/day), and stormwater and roof runoff (110 and 230 L/min).

Capital costs included the following:

  • All process equipment included in each treatment train, plus auxiliary equipment needed for the proper functioning of the unit processes (e.g., plumbing, electrical, and signal wiring within the treatment system).

  • Instrumentation and controls integration including for automatic shutdown and diversions.

  • Design fees, manufacturing, shipping, and installation support, where installation support includes system suppliers who oversee installation by a contractor. Installation is defined as all activities and components required to connect the ONWS treatment system to the building's structural, electrical, HVAC (heating, ventilation, and air conditioning), and plumbing systems such that it can be fully operational (producing water for in-building use).

Capital costs excluded the following:

  • Building site preparation including electrical service (estimates assume 120 and 480 V are available), premise plumbing, backflow prevention devices, HVAC (estimates assume standard building specifications), and structural components required to install and operate the ONWS system.

  • Pumping or piping outside of the boundaries of the ONWS system.

  • Distribution system plumbing.

  • Taxes.

  • Contractor fees for system installation, including mark-ups, overhead, and profit.

  • Project general conditions and contingencies.

O&M cost estimates included (1) system maintenance, replacement, and repair, (2) chemicals, (3) power, and (4) operations labor at one quarter of a full-time equivalent (FTE). The O&M estimates did not include the cost of compliance monitoring. Assumptions used for the O&M estimate are presented in Table 3.

Table 3

Assumed cost for O&M components

CategoryAssumed cost
Chemicals  
 Sodium hypochlorite $16/gal 
 Liquid ammonium sulfate $3/gal 
Power $0.17/kWh 
Labor $146,000/FTEa 
CategoryAssumed cost
Chemicals  
 Sodium hypochlorite $16/gal 
 Liquid ammonium sulfate $3/gal 
Power $0.17/kWh 
Labor $146,000/FTEa 

aFull time equivalent (FTE) is assumed to include base salary and benefits.

Site layout requirements will be affected by the space available within a building, e.g., a building's basement may have limited space with low ceilings that impact the size or type of treatment and storage options. To make layout considerations more uniform, it was assumed that the installation would take place at a greenfield site where the building does not constrain the installation of the ONWS project. Additional information on the cost and layout estimates is provided in Supplementary Material (see Cost and Layout Estimates section and Table ES-2).

Pathogen concentration data

The following sections present the Panel's rationale for selecting the preferred and alternate datasets for each source water type and highlight important departures from the approach used by Sharvelle et al. (2017).

Municipal wastewater and onsite wastewater

To estimate LRTs for onsite wastewater, the Panel used a recent dataset that characterized pathogen concentrations in raw municipal wastewater at five facilities in California (Pecson et al. 2022). This dataset was selected because it was one of the few that met the characteristics of an ‘ideal’ dataset. The key assumption underlying the use of this dataset is that concentrations in municipal wastewater have sufficient equivalence with onsite wastewater. Previously, it was hypothesized that onsite wastewaters may contain higher or more variable concentrations than municipal wastewater (O'Toole et al. 2014; Jahne et al. 2017). Jahne et al. (2020) quantified norovirus and adenovirus in an onsite wastewater, providing data (n=28) to compare against the results of the epidemiological model for onsite wastewater (Jahne et al. 2017). The onsite wastewater had a higher distribution of Norovirus GII (range: 5.2–7.9 log10 GC/L with 11/28 detects) than the Pecson et al. (2022) municipal dataset (range 3.8–7.9 log10 GC/L with 88/122 detects), but lower concentrations of adenovirus. Review of the Kothari et al. (2020) empirical building-scale data indicates that—with the exception of adenovirus—the concentrations of all pathogens of interest were lower than the municipal dataset. Based on the equivocal findings from the limited available data, the Panel concluded there remained insufficient data to confirm or refute the hypothesis that onsite wastewaters have higher or more variable concentrations. Overall, the Panel believed there was a high degree of certainty in the quality of the municipal dataset and a moderate degree of certainty in the equivalence between onsite wastewater and municipal wastewater (Table 4).

Table 4

Panel's assessment of the certainty of the assumptions needed to estimate onsite wastewater pathogen concentrations in the current approach and in Sharvelle et al. (2017) 

 
 

In contrast, the approach used by Jahne et al. (2017) to guide recommendations in Sharvelle et al. (2017) required five assumptions to estimate onsite wastewater pathogen concentrations. The lack of empirical data to confirm the modeling results generally caused the Panel to have lower certainty in the assumptions underlying this approach (Table 4). Based on its professional judgment, the Panel recommends the use of the municipal wastewater approach because it relies on (a) fewer assumptions and (b) greater certainty in the assumptions compared to the epidemiology-based modeled datasets.

The alternate dataset used empirical measurements of pathogen concentrations in an onsite system treating onsite wastewater (Kothari et al. 2020). While limited to 12–15 samples collected over an 8-month period, it is an empirical dataset from an onsite wastewater system in San Francisco (treating approximately 500–1,000 people) to compare against the municipal values.

Graywater

The recommended dataset for graywater concentrations was modeled as a modification of the municipal wastewater dataset of Pecson et al. (2022). Wastewater values were adjusted to account for lower fecal loads by assuming that graywater contained 1% of the pathogen concentrations of a municipal wastewater. This assumption was supported by Jahne et al. (2017) who found a 100:1 ratio when comparing fecal indicator bacteria concentrations in onsite wastewater and graywater sources. Similarly, Schoen et al. (2017) found that graywater LRTs were roughly two orders of magnitude lower than onsite wastewater. The second modification was to conservatively assume 100% pathogen occurrence in graywater, whereas the models and limited empirical data suggest less than 100% occurrence (Jahne et al. 2017; Jahne et al. 2020). The Panel felt a high degree of certainty in the quality of the municipal wastewater dataset, a moderate degree of certainty estimating the ratio of fecal loading between onsite wastewater and graywater, and lower certainty adapting municipal wastewater pathogen values for graywater based on those ratios (Table 5).

Table 5

Panel's assessment of the certainty of the assumptions needed to estimate onsite graywater pathogen concentrations in the updated approach and in Sharvelle et al. (2017) 

 
 

In addition, two alternate datasets were evaluated including the epidemiology-based approach to estimate Cryptosporidium and Giardia concentrations as described in Jahne et al. (2017). As with onsite wastewater modeling, the Panel wanted to assess the accuracy of the epidemiology-based estimates of graywater concentrations. A separate empirical dataset was used to estimate virus concentrations in graywater based on a molecular analysis of 50 graywater samples for adenovirus and norovirus GI and GII (Jahne et al. 2020). The study reported three detectable values for norovirus GII, zero for GI, and seven for adenovirus. While the data fell within the range of the modeling results, they are several orders of magnitude lower than the modeling results. Thus, these data provided preliminary evidence that the epidemiological model may overestimate the occurrence of both norovirus and adenovirus in graywater. This need for model tuning is normal and expected but highlights the uncertainty of the epidemiology-based model and the need for additional empirical verification.

The Panel's assessment of the certainty of the assumptions used in the two approaches is presented in Table 5. Because the municipal wastewater-based approach relies on fewer assumptions, the Panel recommends it over the epidemiology-based modeled datasets.

Stormwater

Stormwater pathogen concentrations were estimated as dilutions of municipal wastewaters with 10 and 0.1% dilutions used to encompass the range of potential impacts. These dilutions were based on Schoen et al. (2017) who evaluated measured concentrations of pathogens in stormwater (Bambic et al. 2011; McBride et al. 2013). The two dilutions bound the likely range of wastewater contamination in urban stormwater, which is highly variable, but poorly defined for pathogens (Chong et al. 2013; Nshimyimana et al. 2014). The 10% dilution estimate is supported by studies by Sercu et al. (2011) and McBride et al. (2013) who estimated wastewater concentrations in the range of 3–20%. Whereas Sharvelle et al. (2017) estimated wastewater pathogen concentrations using several national and international studies, the dataset from Pecson et al. (2022) is recommended for California regulatory development because it provides a more robust and tailored characterization of California wastewaters.

The Panel's assessment of the certainty of the assumptions used in the two approaches is presented in Table 6. Because the updated approach relies on a high-quality dataset better tailored to California wastewaters, the Panel recommends it over the approach presented by Sharvelle et al. (2017).

Table 6

Panel's assessment of the certainty of the assumptions needed to estimate stormwater pathogen concentrations in the updated approach and in Sharvelle et al. (2017) 

 
 

Roof runoff

The data presented in Alja'fari et al. (2022) are recommended as estimates of the pathogen concentrations in roof runoff because (a) the data include multiple, empirical measurements, (b) include four different pathogen gene targets, and (c) samples were taken at four sites across the US over multiple seasons. While limited, this dataset most closely met the requirements of the ‘ideal’ dataset of the published studies. In addition, detection frequencies and concentrations of pathogens were comparable to those measured in a similar study in Australia (Ahmed et al. 2010). As a result, the Panel judged this dataset to be of moderate certainty (Table 7). The Panel believed that this empirical approach had greater certainty in its assumptions compared to the previous model-based approach recommended by Sharvelle et al. (2017), which estimated bacterial pathogen concentrations based on the level of fecal indicator bacteria present within seagull feces.

Table 7

Panel's assessment of the certainty of the assumptions needed to estimate roof runoff concentrations in Sharvelle et al. (2017) and the current modified empirical approach

 
 

The Panel's assessment of the certainty of the assumptions used in the two approaches is presented in Table 7. Because the updated approach relies on an empirical dataset, the Panel recommends it over the approach presented by Sharvelle et al. (2017).

LRTs based on new pathogen data

The QMRA was used to establish the degree of pathogen log reductions needed to achieve the risk goal of 10−4 infections/person/year for healthy adults. The analysis provides a distribution of LRTs required to meet the annual risk goal for each water source type, each pathogen, and each exposure scenario resulting in ingestion. Because the LRTs needed to meet the risk goal are not point estimates but distributions of treatment performance, the 95th percentile value was used to select an LRT from the distributions (see Supplementary Material for Distribution of LRTs Required to Meet Risk Thresholds). Results based on both the recommended/alternate and optional (roof runoff) pathogen datasets are shown in Table 8 for both indoor use and irrigation/dust suppression. The resulting LRTs for the other end-uses are provided in the Supplementary Material (see LRT Results for ONWS Source Waters and Table ES-3–Table ES-8).

Table 8

95th Percentile LRTs for the use of multiple ONWS source waters based on both recommended and alternate pathogen datasets

 
 

Both the recommended and alternate datasets were used to test the sensitivity of the LRTs to assumptions about source water pathogen concentrations. Overall, there was alignment in the LRTs irrespective of the datasets used—in most cases, LRTs were within one log10 value of each other. Larger differences were observed between the graywater LRTs across the pathogen groups. This deviation was likely due to the differences in the approaches used to estimate the concentrations: a modified empirical approach based on municipal wastewater (recommended) and an epidemiological model-based approach (alternate). These two approaches did not lead to systematic differences, however, as neither approach had uniformly higher (or lower) LRTs.

Given the lack of roof runoff pathogen data, the Panel evenly weighed the results of both Option A and Option B. Generally, the highest LRTs required for indoor use ranged from 1.0 to 1.5 across the three pathogens evaluated with only one value out of 18 being greater than 1.5 (see Table ES-8 for the 18 indoor use scenarios evaluated). Similarly, unrestricted irrigation generally required 0.3–0.7 LRT across the pathogens. Despite the lack of data, the consistency of these LRTs for the two end-uses was noted. The Panel recommends that the more conservative LRT between Option A and B be used for roof runoff given the lack of data. The Panel believes that applying the criteria for Giardia alone will provide sufficient protection against pathogenic bacteria, particularly when using validated technologies that comply with approved protozoa crediting frameworks. Consequently, a separate bacterial LRT is not required at this time.

Beyond the results shown in Table 8, the analysis considered additional end-uses including fire suppression, car washing, and decorative fountains (see Supplementary Material Table ES-3–Table ES-8). Based on the analysis, fire suppression and car washing (when considered alone) would not require LRTs higher than those needed for unrestricted irrigation and indoor use. Consequently, ONWS water could also be used for fire suppression and car washing if they comply with the requirements for either unrestricted irrigation or indoor use. When considered alone, decorative fountains required LRTs higher than unrestricted irrigation, but lower than combined indoor use LRTs. The analysis also calculated LRTs for combined indoor use that included exposure due to decorative fountains as well as toilet flushing, clothes washing, and cross-connection. For the selected reference pathogens, the indoor use LRTs that included decorative fountains were the same as the indoor use LRTs that did not include decorative fountains except for the virus LRTs for onsite wastewater and stormwater that were 0.1-log10 higher and would have resulted in a 0.5-log10 increase in the LRT requirements when rounding up to the nearest 0.5-log10 value. For this reason, the Panel believes that the indoor use LRTs would be protective if decorative fountains were included as an indoor use. This is consistent with findings from Schoen et al. (2020).

Comparison of updated and previous LRTs

To compare the updated LRTs to those recommended by Sharvelle et al. (2017), a single LRT value was selected to represent the two groups—parasitic protozoa and enteric viruses. For the parasitic protozoa, the Panel selected Giardia as the benchmark pathogen because it consistently required LRTs that were equal to or greater than those needed for Cryptosporidium. Sharvelle et al. (2017) also selected the more conservative protozoan in developing LRTs; however, Cryptosporidium was the more conservative organism based on the datasets used in that study.

For the viruses, there was virtually no difference in the LRTs required for the two culturable viruses (enterovirus and adenovirus) across source waters and end-uses observed. Two dose–response functions were used for norovirus—in line with recommendations from Van Abel et al. (2017)—resulting in an approximate 3-log10 range of LRTs for each source water and end-use. Given the wide range of LRT results associated with norovirus LRT requirements, the results were included in the selection of virus LRTs but principally to help define the potential range of necessary reductions. To select a specific numerical target, the Panel used the LRT requirements for the two culturable viruses. The LRTs for the benchmark pathogens were rounded up to the nearest 0.5-log10 value to match the approach used in Sharvelle et al. (2017). A summary of the new and previous LRTs is presented in Table 9. Note that the differences are a direct result of the new source water pathogen concentrations, as the QA/QC confirmed that the method reproduced equivalent values when using the same datasets as Sharvelle et al. (2017). The LRT difference were mostly one log10 or less, though there were a small number that were either more than one log10 lower (e.g., protozoa requirements for wastewater) or more than one log10 higher (e.g., virus requirements for stormwater).

Table 9

Comparison of the LRTs using the updated pathogen datasets and those developed by Sharvelle et al. (2017) 

Source water and use typeUpdated enteric virusEnteric virus (2017)Updated parasitic protozoaParasitic protozoa (2017)Updated enteric bacteriaEnteric bacteria (2017)
Wastewater 
Indoor use 8.0 8.5 6.5 7.0 – – 
Irrigation/Dust 7.5 8.0 5.5 7.0 – – 
Graywater 
Indoor use 6.0 6.0 4.5 4.5 – – 
Irrigation/Dust 5.5 5.5 3.5 4.5 – – 
Stormwater (10% wastewater contribution) 
Indoor use 7.0 5.5 5.5 5.5 – – 
Irrigation/Dust 6.5 5.0 4.5 4.5 – – 
Roof runoff 
Indoor use – – 1.5 – – 3.5 
Irrigation/Dust – – 1.0 – – 3.5 
Source water and use typeUpdated enteric virusEnteric virus (2017)Updated parasitic protozoaParasitic protozoa (2017)Updated enteric bacteriaEnteric bacteria (2017)
Wastewater 
Indoor use 8.0 8.5 6.5 7.0 – – 
Irrigation/Dust 7.5 8.0 5.5 7.0 – – 
Graywater 
Indoor use 6.0 6.0 4.5 4.5 – – 
Irrigation/Dust 5.5 5.5 3.5 4.5 – – 
Stormwater (10% wastewater contribution) 
Indoor use 7.0 5.5 5.5 5.5 – – 
Irrigation/Dust 6.5 5.0 4.5 4.5 – – 
Roof runoff 
Indoor use – – 1.5 – – 3.5 
Irrigation/Dust – – 1.0 – – 3.5 

The wastewater LRTs had a high degree of similarity with the previous LRTs showing the consistency of the three approaches used for estimation: (1) the use of untreated municipal wastewater as a surrogate for onsite wastewater, (2) the use of empirical onsite wastewater data, and (3) the use of modeled onsite wastewater concentrations. The approaches generally resulted in a 0.5-log10 range of values, with the exception of the protozoa LRT for irrigation and dust suppression (1.5-log10). The Panel recommends the use of the municipal wastewater dataset for the reasons previously stated, but believe that the similarity between the municipal and onsite wastewater datasets supports the use of municipal wastewater data in ONWS settings treating 500 or more people.

Greater reproducibility of the LRTs was observed for graywater, which was evaluated with (1) an adapted, empirical wastewater dataset, (2) a modeled graywater dataset, and (3) an empirical graywater dataset. While the protozoa requirements for stormwater were equivalent between approaches, a greater discrepancy was observed for viruses. There was not a consistent pattern of one dataset requiring uniformly higher LRTs than the other. Because the framework for estimating LRTs was identical for both approaches, the use of the new wastewater pathogen dataset drove the higher viral LRTs in the updated approach (Pecson et al. 2022).

The roof runoff LRTs represent the largest change between the updated and previous approaches given that the previous bacterial requirement has been replaced with an updated protozoa requirement. This was the only source water for which the target pathogen group was modified, so a direct comparison of the change in requirements is not possible.

Model treatment trains

Model treatment trains capable of achieving the minimum LRTs using existing crediting frameworks and technologies with validated log reduction values (LRVs) are shown in Figure 1. The trains were developed considering not only pathogen control but other water quality objectives, such as the reduction of organics, suspended solids, color, odor, and microbial stability in the ONWS distribution system (e.g., via secondary disinfection). Both graywater and onsite wastewater have significant levels of biological oxygen demand (BOD), total suspended solids (TSS), and ammonia, and will require the addition of biological treatment and more robust filtration compared to roof runoff and stormwater.
Figure 1

Summary of model treatment trains capable of meeting LRTs for ONWS source waters.

Figure 1

Summary of model treatment trains capable of meeting LRTs for ONWS source waters.

Close modal

The Panel recommends the use of existing crediting frameworks to determine ONWS compliance with the LRT requirements (Pecson & Post 2020). Typical LRV credits for the unit processes in the model treatment trains are presented in Table 10.

Table 10

Typical LRV credits associated with unit processes in ONWS trains

Unit processEnteric virusParasitic protozoaNotes on LRV crediting
MBR 1.0 2.5 Based on framework developed in WRF 4997 for MBRs using the default Tier 1 crediting scheme (Salveson et al. 2021
MF/UF Based on US EPA Membrane Filtration Guidance Manual (EPA 2020
UV 1–6 1–6 Based on US EPA disinfection tables relating UV dose (in mJ/cm2) to log inactivation of virus and protozoa (EPA 2020
Free chlorine 1–6 0–3 Based on US EPA chlorine disinfection tables relating CT (in mg-min/L) to log inactivation of virus and Giardia [note no credits achievable for Cryptosporidium] (EPA 2003
Cartridge filter 1–2 Based on compliance with US EPA Long Term 2 Enhanced Surface Water Treatment Rule requirements for the control of Cryptosporidium by cartridge filters (EPA 2010
Unit processEnteric virusParasitic protozoaNotes on LRV crediting
MBR 1.0 2.5 Based on framework developed in WRF 4997 for MBRs using the default Tier 1 crediting scheme (Salveson et al. 2021
MF/UF Based on US EPA Membrane Filtration Guidance Manual (EPA 2020
UV 1–6 1–6 Based on US EPA disinfection tables relating UV dose (in mJ/cm2) to log inactivation of virus and protozoa (EPA 2020
Free chlorine 1–6 0–3 Based on US EPA chlorine disinfection tables relating CT (in mg-min/L) to log inactivation of virus and Giardia [note no credits achievable for Cryptosporidium] (EPA 2003
Cartridge filter 1–2 Based on compliance with US EPA Long Term 2 Enhanced Surface Water Treatment Rule requirements for the control of Cryptosporidium by cartridge filters (EPA 2010

Capital costs, O&M costs, and site layout estimates were developed for the model treatment trains and are presented in Table 11. Given the uncertainty inherent in cost estimates at the budgetary level, it is advisable to conservatively assume that the future cost and layout of the treatment train is closer to the higher range of the estimates. Using the upper bound costs for the onsite wastewater and graywater treatment trains results in an equivalent annual cost of $20,000 and $8700/ac-ft, respectively, assuming a 30-year lifespan and 5% interest. It is noted that these costs are an order of magnitude higher than the estimated cost to produce water for large-scale indirect potable reuse (Raucher & Tchobanoglous 2014); however, the costs are similar to the actual cost of urban water supply in some areas. It is likely that other design approaches with lower cost would be acceptable assuming they could meet the required LRTs with validated technology whose performance can be verified with continuous online monitoring.

Table 11

Cost and layout estimates for model ONWS treatment trains

Source water and ONWS capacityCapital cost estimate (USD)Annual O&M cost estimate (USD)Site layout estimate (m2)
Wastewater 
19,000 L/day $500,000–$650,000 $66,000–$70,000 60–93 
57,000 L/day $600,000–$950,000 $70,000–$85,000 70–140 
Graywater 
5,700 L/day $300,000–$400,000 $20,000–$50,000 9.3–46 
19,000 L/day $450,000–$500,000 $21,000–$66,000 15–70 
Stormwater (101 dilution) 
110 L/min $300,000–$350,000 $15,000–$50,000 19–46 
230 L/min $400,000–$450,000 $15,000–$50,000 28–46 
190,000-l storage tank $200,000–$350,000 N/A 50 
Roof Runoff 
110 L/min $200,000–$300,000 $15,000–$40,000 19–46 
230 L/min $250,000–$350,000 $15,000–$40,000 28–46 
190,000-l storage tank $200,000–$350,000 N/A N/A 
Source water and ONWS capacityCapital cost estimate (USD)Annual O&M cost estimate (USD)Site layout estimate (m2)
Wastewater 
19,000 L/day $500,000–$650,000 $66,000–$70,000 60–93 
57,000 L/day $600,000–$950,000 $70,000–$85,000 70–140 
Graywater 
5,700 L/day $300,000–$400,000 $20,000–$50,000 9.3–46 
19,000 L/day $450,000–$500,000 $21,000–$66,000 15–70 
Stormwater (101 dilution) 
110 L/min $300,000–$350,000 $15,000–$50,000 19–46 
230 L/min $400,000–$450,000 $15,000–$50,000 28–46 
190,000-l storage tank $200,000–$350,000 N/A 50 
Roof Runoff 
110 L/min $200,000–$300,000 $15,000–$40,000 19–46 
230 L/min $250,000–$350,000 $15,000–$40,000 28–46 
190,000-l storage tank $200,000–$350,000 N/A N/A 

Capital cost estimates rounded to the nearest $50,000 USD. ENRCCI=11,627 (January 2021).

It should be emphasized that there are also many economic, social, and environmental benefits that have driven ONWS implementation such as (1) helping communities defer capital costs for upgrading centralized infrastructure, (2) assisting utilities in balancing infrastructure investments rather than viewing onsite reuse systems as a competing priority, (3) fostering system resilience by improving the ability to respond to disruptions in water service delivery that may come from droughts or increased storm events, (4) generating environmental and community amenities when onsite systems integrate green roofs, wetlands, and rain gardens that can transform urban landscapes into more vibrant, natural public spaces, and (5) providing opportunities to engage the community in becoming an active partner in water management. Other benefits include advancing water use efficiency, diversifying and stretching water supplies, managing stormwater flows, inspiring innovation in technology, and creating opportunities for public-private partnerships that meet market demands (National Blue Ribbon Commission for ONWS 2018; Kehoe & Nokhoudian 2022).

One of the key limitations in developing risk-based LRTs is the relative lack of data characterizing pathogen concentrations in ONWS source waters. The goal of this effort was to re-evaluate and update the LRTs based on recent studies that provide significant new data and insight into pathogen distributions. A total of nine different pathogen scenarios were used to develop the LRTs including three protozoa scenarios (Giardia analyzed with a single dose–response function; Cryptosporidium analyzed with two), four virus scenarios (enterovirus and adenovirus evaluated with single dose–response functions; norovirus evaluated with two), and two bacterial pathogens (Campylobacter and Salmonella each evaluated with a single dose–response function). Furthermore, each source water was evaluated with at least two different frameworks (e.g., empirical vs. modeled data, untreated municipal vs. onsite wastewater) to assess the impact of source water data on LRT requirements.

The first key finding is that there is a high degree of alignment in the LRTs regardless of the different approaches used to estimate pathogen concentrations (Tables 8 and 9). While this fact alone does not prove the accuracy of the LRTs, the convergence of the values around a relatively narrow band provides greater confidence in the selection of LRTs and their lack of sensitivity to the approaches. Because regulatory development will require the selection of a single LRT, the Panel recommends the use of the updated LRTs for enteric virus and parasitic protozoa for the various source waters (Table 9).

Future efforts should continue to characterize pathogen concentrations in ONWS source waters. While a recent study has shown the relative insensitivity of population size on pathogen concentrations in municipal wastewater settings (Pecson et al. 2022), greater variability in pathogen distributions is expected for onsite wastewater and graywater in ONWS settings. Currently, there are only a small number of studies reporting empirical data on these source waters (Jahne et al. 2020; Kothari et al. 2020) making it difficult to estimate how the occurrence and distribution of pathogens may vary based on the size of the contributing population, building uses, occupancy patterns, etc. In the meantime, there was a strong coherence between the wastewater LRTs developed with empirically measured (a) municipal wastewater, (b) onsite wastewater (Kothari et al. 2020), and (c) the epidemiological model. As fewer datasets are available for ONWS graywater, a greater number of assumptions were required to estimate concentrations in graywater compared to wastewater. Additional monitoring campaigns should also be pursued for graywater using the criteria established for ‘ideal’ datasets (Schoen et al. 2017; Pecson et al. 2022).

Water Research Foundation project 5034 is providing new insight into the factors impacting pathogen concentrations in stormwater and should be leveraged in future efforts to confirm or update the LRTs (WRF 2022). In the meantime, the ‘wastewater dilution’ approach is the recommended framework for establishing stormwater LRTs. The source water whose pathogen concentrations are least well characterized is roof runoff. While the updated analysis considered both protozoa and bacterial pathogens, the final recommendation is to apply only the protozoa requirement, based on the premise that a validated treatment system credited with 1.0- to 1.5-log10 inactivation and/or removal of protozoa will provide similar (or higher) levels of control of bacterial pathogens.

One question that arises is whether the LRTs are applicable to California ONWS settings alone or can be more broadly applied to other geographic locations. LRTs for three of the four ONWS source waters (onsite wastewater, graywater, and stormwater) were developed using the municipal wastewater pathogen dataset from Pecson et al. (2022), which was collected from five utilities in California. A comparison of this California dataset with other high-quality wastewater datasets collected in the US (including Alabama, Arizona, California, Colorado, Florida, Michigan, North Carolina, Pennsylvania, Vermont, Wisconsin), Europe (Norway), and Australia showed a high degree of alignment between the various distributions (Pecson et al. 2021). The similarity of the California dataset with a wide range of other locations in the US, Europe, and Australia suggest that the LRT values may not be exclusive to California but are also applicable in other geographic locations. The fourth dataset characterizing roof runoff was also collected at four different locations across the US and showed consistency with recent findings from Australia, suggesting that it may also be appropriate for use in multiple geographic locations. Nevertheless, the authors recommend additional pathogen monitoring be conducted in different geographic regions to evaluate the similarities (and differences) in source water concentrations and identify the factors that may require site-specific modifications to the LRTs.

One practical limitation with specifying bacterial LRTs is the absence of bacterial crediting frameworks for most unit processes. Because the EPA's Surface Water Treatment Rules specified pathogen log reduction requirements for virus and protozoa, frameworks were established to credit unit processes for these two pathogen groups (EPA 1989). Bacterial removal requirements have typically relied on end-point monitoring (e.g., effluent total coliform monitoring), which confirms the low levels of bacteria in the treated effluent but does not quantify how much reduction took place through treatment. Nevertheless, many removal and disinfection processes capable of controlling Cryptosporidium and Giardia should provide equivalent or higher control against bacteria. One exception is that some filtration processes with effective pore sizes larger than 1 μm could allow passage of some bacteria while effectively removing Cryptosporidium and Giardia through size exclusion. Application of such filters as a singular unit treatment process for roof runoff may not ensure adequate removal of bacterial pathogens.

While pathogen control is the most critical public health goal, the ONWS treatment trains must achieve additional water quality and treatment goals that may necessitate the inclusion of additional unit processes. One key feature is the control of microbial regrowth in the distribution system, particularly organisms that can impact public health such as Legionella and other saprozoic species. While the focus of the LRTs is on the control of enteric pathogens entering in the ONWS source waters, treatment systems must also pursue strategies to limit the post-treatment regrowth of organisms (National Academy of Science 2020). The Panel recommends the use of a residual disinfectant concentration as the key indicator of distribution system water quality. A potential goal would be to maintain a chlorine (free chlorine or chloramine) residual of at least 0.2 mg/l at distal end-use locations to maintain Legionella pneumophila at concentrations below 10 colony forming units/mL (National Academy of Science 2020). This goal also supports additional removal of bacteria, providing particular advantages for control of bacterial pathogens in roof runoff.

  • Using recent updated pathogen data, LRTs were developed for four ONWS source waters producing water for indoor use (i.e., toilet flushing, clothes washing, and unintentional cross-connections), unrestricted irrigation, fire suppression, car washing, and decorative fountains.

  • LRT requirements were based on the evaluation of nine different reference pathogens.

  • Multiple frameworks for estimating pathogen densities were used to evaluate the sensitivity of the LRT requirements to this input, and compared to the previous LRT findings from Sharvelle et al. (2017).

  • The convergence of the LRTs around a relatively narrow band and the lack of sensitivity to the different approaches to estimate ONWS source water pathogen concentrations provide greater confidence in regulatory development of ONWS standards at the building-scale.

  • The similarity of the pathogen distributions across cities in the US, Europe, and Australia suggest that the LRTs could be applicable across a wide geographic region, though additional monitoring campaigns in ONWS source waters are recommended.

  • While ONWSs have potential advantages associated with achieving a high level of water use efficiency, the economic feasibility of these projects will depend, in part, on the cost savings from offset municipal water supply. An investigation of model treatment trains that can meet the required LRTs found that the total costs are sensitive to ONWS capacity and source water used.

The authors thank Sherly Rosilela from the Division of Drinking Water of the California State Water Resources Control Board, and the National Water Research Institute for administering the Panel, including Kevin Hardy, Suzanne Sharkey, and Mary Collins. The Panel thanks the following people for their comments and discussions during the development of the log reduction targets: Michael Jahne (US EPA), Charles Haas (Drexel University), Mark LeChevallier (Dr Water Consulting), and Mary Schoen (Soller Environmental). The Panel acknowledges the following people who helped develop cost estimates, layout estimates, and design criteria for the onsite non-potable water systems: Chris Allen (SUEZ Water Technologies), Ben Arnold (Aquacell Water Recycling), Daniel Bacani and Glenn Van Eekhout (Los Angeles Department of Public Health), Taylor Nokhoudian (San Francisco Public Utilities Commission), Amelia Luna (Sherwood Design Engineers), Kim Seay (Wahaso Water Harvesting Solutions), and Sarah Clark (Carollo Engineers).

This work was supported by the California State Water Resources Control Board contract number 20-011-400.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.

Ahmed
W.
,
Goonetilleke
A.
&
Gardner
T.
2010
Implications of faecal indicator bacteria for the microbiological assessment of roof-harvested rainwater quality in southeast Queensland, Australia
.
Canadian Journal of Microbiology
56
(
6
),
471
479
.
https://doi.org/10.1139/W10-037.
Alja'fari
J. H.
,
Sharvelle
S. E.
,
Brinkman
N. E.
,
Jahne
M.
,
Keely
S.
,
Wheaton
E.
,
Welty
J.
,
Sukop
M. C.
&
Meixner
T.
2022
Characterization of roof runoff microbial quality in four U.S. cities with varying climate and land use characteristics
.
Water Research
Accepted.
Bambic
D.
,
McBride
G.
,
Miller
W.
,
Stott
R.
&
Wuertz
S.
2011
Quantification of Pathogens and Sources of Microbial Indicators for QMRA in Recreational Waters
.
WERF
,
Alexandria, VA
.
Beaudequin
D.
,
Harden
F.
,
Roiko
A.
,
Stratton
H.
,
Lemckert
C.
&
Mengersen
K.
2015
Modelling microbial health risk of wastewater reuse: a systems perspective
.
Environment International
84
,
131
141
.
California State Water Resources Control Board
2022
SBDDW-22-001 Regulations for Onsite Treatment and Reuse of Nonpotable Water
. .
Chong
M. N.
,
Sidhu
J.
,
Aryal
R.
,
Tang
J.
,
Gernjak
W.
,
Escher
B.
&
Toze
S.
2013
Urban stormwater harvesting and reuse: a probe into the chemical, toxicology and microbiological contaminants in water quality
.
Environmental Monitoring and Assessment
185
(
8
),
6645
6652
.
DDW
2018
Regulations Related to Recycled Water. California Code of Regulations, Titles 22 and 17
.
EPA
1989
Surface Water Treatment Rule. 40 CFR 141.70-141.75
.
United States Environmental Protection Agency
,
Washington, DC
.
EPA
2003
LT1ESWTR Disinfection Profiling and Benchmarking Guidance Manual
.
United States Environmental Protection Agency
,
Washington, DC
.
EPA
2010
Long Term 2 Enhanced Surface Water Treatment Rule Toolbox Guidance Manual
.
United States Environmental Protection Agency
,
Washington, DC
.
EPA
2020
Innovative Approaches for Validation of Ultraviolet Disinfection Reactors for Drinking Water Systems
.
United States Environmental Protection Agency
,
Washington, DC
.
Haas
C. N.
,
Rose
J. B.
&
Gerba
C. P.
1999
Quantitative Microbial Risk Assessment
.
Wiley
,
New York
.
Hultquist
B.
2016
Basis for California's 12-10-10 log removal requirements paper presented at the 20th Annual WateReuse Research Conference, Denver, CO
.
Jahne
M. A.
,
Brinkman
N. E.
,
Keely
S. P.
,
Zimmerman
B. D.
,
Wheaton
E. A.
&
Garland
J. L.
2020
Droplet digital PCR quantification of norovirus and adenovirus in decentralized wastewater and graywater collections: implications for onsite reuse
.
Water Research
169
,
115213
.
Kehoe
P.
&
Nokhoudian
T.
2022
Onsite Water Recycling: an Innovative Approach to Solving an Old Problem
.
San Francisco, CA
.
Kothari
M.
,
Triolo
S.
,
Weeks
B.
&
Salveson
A.
2020
PureWaterSF: Building-Scale Potable Water Reuse Demonstration Project
.
Water Research Foundation
,
Alexandria, VA
.
National Academy of Science
2020
Management of Legionella in Water Systems
.
National Academies Press
,
Washington, DC
.
National Blue Ribbon Commission for ONWS
2018
Making the Utility Case for Onsite non-Potable Water Systems
.
National Research Council
2009
Urban Stormwater Management in the United States
.
National Research Council, D.o.E.a.L.S., Water Science and Technology Board
,
Washington, DC
.
NRMMC, EPHC & NHMRC
2006
AHMC, Australian Guidelines for Water REcycling: Managing Health and Environmental Risks (Phase 1)
.
Nshimyimana
J. P.
,
Ekklesia
E.
,
Shanahan
P.
,
Chua
L. H.
&
Thompson
J. R.
2014
Distribution and abundance of human-specific Bacteroides and relation to traditional indicators in an urban tropical catchment
.
J. Appl. Microbiol.
116
(
5
),
1369
1383
.
Olivieri
A.
,
Ashbolt
N.
,
Leverenz
H.
,
Pecson
B.
&
Sharvelle
S.
2021
On-Site Treatment and Reuse of Nonpotable Water – Technical Guidance
.
NWRI
,
Fountain Valley, CA
.
O'Toole
J.
,
Sinclair
M.
,
Barker
S. F.
&
Leder
K.
2014
Advice to risk assessors modeling viral health risk associated with household graywater
.
Risk Analysis
34
(
5
),
797
802
.
Pecson
B.
&
Post
B.
2020
Onsite Non-Potable Water System Guidance Manual (WRF 4909)
.
Water Research Foundation
,
Alexandria, VA
.
Pecson
B.
,
Darby
E.
,
Di Giovanni
G.
,
Leddy
M.
,
Nelson
K. L.
,
Rock
C.
,
Slifko
T.
,
Jakubowski
W.
&
Olivieri
A.
2021
DPR-2: Pathogen Monitoring in Untreated Wastewater
.
Water Research Foundation
,
Alexandria, VA
.
Pecson
B. M.
,
Darby
E.
,
Danielson
R.
,
Dearborn
Y.
,
Giovanni
G. D.
,
Jakubowski
W.
,
Leddy
M.
,
Lukasik
G.
,
Mull
B.
,
Nelson
K. L.
,
Olivieri
A.
,
Rock
C.
&
Slifko
T.
2022
Distributions of waterborne pathogens in raw wastewater based on a 14-month, multi-site monitoring campaign
.
Water Research
213
,
118170
.
Petterson
S. R.
&
Ashbolt
N. J.
2016
QMRA and water safety management: review of application in drinking water systems
.
Journal of Water and Health
14
(
4
),
571
589
.
Raucher
R. S.
&
Tchobanoglous
G.
2014
The Opportunities and Economics of Direct Potable Reuse
.
WaterReuse Research Foundation
,
Alexandria, VA
.
Salveson
A.
,
Trussell
S.
&
Linden
K.
2021
Membrane Bioreactor Validation Protocols for Water Reuse
.
Water Research Foundation
,
Alexandria, VA
.
Schoen
M. E.
,
Jahne
M. A.
&
Garland
J.
2020
Enteric pathogen treatment requirements for nonpotable water reuse despite limited exposure data
.
Environmental Science & Technology Letters
7
(
12
),
943
947
.
Sercu
B.
,
Van De Werfhorst
L. C.
,
Murray
J. L. S.
&
Holden
P. A.
2011
Sewage exfiltration As a source of storm drain contamination during dry weather in urban watersheds
.
Environmental Science & Technology
45
(
17
),
7151
7157
.
Sharvelle
S.
,
Ashbolt
N.
,
Clerico
E.
,
Hultquist
R.
,
Leverenz
H.
&
Olivieri
A.
2017
Risk-Based Framework for the Development of Public Health Guidance for Decentralized Non-Potable Water Systems: Final Report
.
Water Environment Federation
,
Alexandria, VA
.
Sinclair
M.
,
Roddick
F.
,
Nguyen
T.
,
O'Toole
J.
&
Leder
K.
2016
Measuring water ingestion from spray exposures
.
Water Research
99
,
1
6
.
Water Services Association of Australia
2004
Health Risk Assessment of Fire Fighting From Recycled Water Mains
.
Water Services Association of Australia
,
Melbourne
,
Australia
.
World Health Organization
2016
Quantitative Microbial Risk Assessment: Application for Water Safety Management
.
World Health Organization
,
Geneva, Switzerland
.
WRF
2022
Assessing the Microbial Risks and Impacts From Stormwater Capture and use to Establish Appropriate Best Management Practices
. .
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with no derivatives, provided the original work is properly cited (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Supplementary data