Water quality data for COD, BOD, N and P for the period 2001-2009 in the upper Citarum River at Wangisagara, Sapan, Cijeruk, Dayeukholot and Nanjung (Figure 1) was obtained through the West Java Regional Environmental Agency [39] [40] .

Three sources of pollutions were distinguished and assessed for 2010 and 2030, being (A) Domestic, (B) Industrial and (C) Agricultural (Table 1).

A. Domestic pollution:

Domestic pollution was determined in five steps.

1. Determination of specific per person pollution loads: Domestic specific water consumption rates followed the Indonesia guidelines [16] for 6 categories of urban area: 1) metropolitan (>1 million people), 2) large town (500,000 - 1 million people), 3) medium town (100,000 - 500,000 people), 4) small town (20,000 - 100,000 people), 5) village (3000 - 20,000 people) and 6) rural (<3000 people). An 80% return factor was used to estimate wastewater production from consumed water [41] . Metropolitan specific pollution loads were based on [19] .

2. Correction of pollution load with varying types of urban status: The relation between urban category and pollution loads was reflected using the study of [17] applying a greywater pollution load decrease between urban metropolitan and rural areas of 30% of COD and N and 50% of P, while for urban categories in between metropolitan and rural areas these were made relative to water consumption data (Table 1). Because of lack of detailed data, urban and rural black water pollution load rates were assumed the same.

a. Based on [17] - [19] ; b. Data obtained by authors from Food & Beverage (dairy, brewery) in Indonesia; c. Values depend on type of paper and pulping process and range from 1500 to over 20,000 mg/l COD [21] [48] . Applied values are based on experience of authors for Pulp and Paper South East Asia; d. Based on [47] ; e. Textile industry data were determined based on actual measurements of 21 textile industries in project area [37] and verified with [21] .

3. Correction of pollution reaching surface water bodies: Baseline pollution correction coefficients (included in Table 1) were based on [18] and were 100% (metropolitan and large towns), 92% (medium town), 83% (small town), 74% (village) and 65% (rural areas). Thus, only 83% of pollution generated in a small town is expected to reach the surface water. As specific information on these coefficients was lacking for the Upper Citarum Basin, two alternative scenarios were compared, being 1) where 100% of pollution entered the surface water, and 2) where half of the baseline value entered the surface water (i.e. 50% for metropolitan and large towns, 46% for medium town, 42% for small towns, 37% for village and 33% for rural areas).

4. Determination of pollution loads reaching the surface water for 2010 and 2030: Total specific pollutions loads per location reaching the surface water were calculated applying the specific pollution loads (combining step 2 and 3 above) on population developments obtained from the Java Spatial Model (JSM). JSM shows the population development for each urban category between 2010 and 2030 [16] .

5. Determination of the number of people with access to wastewater facilities in 2010: The pollution loads reaching the water bodies were corrected for interventions already in place. The 2010 wastewater access data were obtained from the statistical bureau of Indonesia (BPS) and were determined as 52%. 490,000 people were connected to the Bojong Soang WWTP (pond systems) in Bandung [42] .

B. Industrial pollution:

838 industries in the catchment area were categorized by location and type (Table 1) and water consumption (m³/d) in which data on ground and surface water consumption were obtained from the West Java provincial agency for Energy and Mineral Resources [43] and provincial agency for Water Resources Management [44] . Pollution loads were determined by effluent flow (using 80% return factor) and effluent concentrations (Table 1). Because reliable industrial pollution data is lacking [37] , an impact analysis was performed (Table 2). A distinction is made between 1) a best case scenario, 2) a baseline scenario and 3) a worst case scenario. These scenarios vary in terms of removal efficiency and percentage of industry having a WWTP (Table 2). COD removal efficiencies in the best case followed self-reported COD removal efficiencies by industries, whereas the worst case effluent COD values followed externally measured COD removal efficiencies [37] . N and P are not measured by industries and presented values were assumed, based on [21] . Reference [45] reports that 80% of the textile industries comply with the effluent standards, whereas the environmental office in nearby Cimahi mentions 3% [46] . Therefore, the baseline case assumes that 80% of the largest industries (consumption >2000 m³/d) treat their wastewater, while with decreasing water consumption this percentage decreases with a minimum of 25% (Table 2).

C. Agricultural pollution:

The 2010 and 2030 water demand for irrigation was based on [16] . Pollution discharged (Table 1) for rice and non-rice crops were based on [47] .

Domestic interventions:

Selection of type of domestic WWT facilities (Table 3) was based on the residential features following [19] . For off-site systems three scenarios were compared to identify the effect on the surface water quality and cost:

1. Simple Technology (ST): Anaerobic filter is applied for medium centralized systems and a conventional activated sludge (CAS) for centralized systems;

2. Advanced Technology (AT): Medium central and central systems apply a CAS with additional N, P removal;

a. Table 2 shows for example that in the best case scenario, 90% of the industries with a water consumption exceeding 2000 m³/d have a WWTP and removal efficiencies are 90% (COD), 95% (BOD), 90% (TN) and (50% (TP).

a. Current users in urban areas with a residential density between 25 - 100 pp/ha apply on-site systems, whereas all new development will be served by medium centralized system [50] ; b. Selection criteria are formulated based on the expected population status in 2020 (mid-term).

3. Resource Recovery technology (RR): Comprising Aerobic Granular Sludge (AGS) system [49] with sludge digestion, P-recovery as struvite and composting of produced sludge. The removal efficiencies of AT and RR are the same.

Associated investment and operational costs were based on [19] (see Table S1 of the OSI Section 1). The effects on discharged pollution loads reaching the surface water and associated investment costs of a 25%, 50% and 75% switch of households currently applying on-sites system to an off-site system were compared.

Industrial interventions:

Three industrial wastewater treatment types were formulated based on currently applied technologies [51] (see OSI, Section 3): 1) textile wastewater using reactive dyes (typically used for traditional batik), apply a CAS and activated carbon for color removal, 2) textile wastewater using non-reactive dyes apply CAS followed by Dissolved Air Flotation (DAF), and 3) other industries apply pre-treatment (DAF) and CAS. Future effluent values should meet at least current standards [46] defined as 80 mg/l COD, 20 mg/l BOD, 10 mg/l N and 10 mg/l P. Investment and operational costs were determined for different sizes of treatment capacities, based on available engineering cost standards (see OSI, Section 3).

Municipal Solid Waste (MSW) interventions:

Solid waste system selection interventions (Table 4) and their costs are based on [19] and distinguish home composting, landfilling and centralized and decentralized 3R application (see OSI Section 4).

A generic model package (RIBASIM) for simulating the behavior of river basins under various hydrological conditions was used to simulate the effect of different interventions on water quality development in the Upper Citarum River [38] [52] . Based on pollution loads produced in each defined catchment area and resulting water flows

concentrations are calculated. The RIBASIM model and defined catchment areas are further explained in OSI Section 5. The pollution loads entering the Upper Citarum River were varied, using 6 scenarios (Table 5).

The output of the 2010 RISBASIM average pollutant concentrations was calibrated based on the average measured concentration (step 1).

Five economic benefits of wastewater and solid waste management improvements were defined following [32] and [53] :

A. Health:

Averted costs of fecal-oral disease from improved on-site sanitation and wastewater management: An average disease reduction of 36% by on-site sanitation and an additional 20% by adding improved off-site facilities was applied [54] - [56] . The average annual health cost per 5 member family as a result of unimproved sanitation was US $316 [31] .

Associated averted health impacts (infectious diseases and skin complaints) of less exposure during flooding events: Reported health cases during a period of several flooding events (January-March 2009) were compared to the same period in a non-flood year (January to March 2010) and was scaled to reflect all the flooded communities in the Citarum River basin, resulting in an estimated 15,000 averted cases of diarrhea in an average year [53] . The economic value was estimated by multiplying the average number of additional cases per year by the unit cost of inpatient (hospitalized) and outpatient services, including productivity losses [31] .

B. Access time:

Value of time savings from reduced travel time and/or queuing for meeting sanitation needs: An average daily gain of 115 minutes per household with an annual value of US $95 per household is used [31] . Only the time of adults and school-aged children were included, valued at 30% and 15% of the hourly rate implied by the GDP per capita, respectively [57] . This figure was applied to the access gain afforded by on-site sanitation facilities of 45% of households for the period from 2010 until 2030.

C. Water:

Reduced drinking water treatment costs to households and industries: The total cost of water treatment (including both capital and operating costs) using surface water of a better quality source will decrease from 0.13 to 0.06 US $/m³ [16] . This saving was

a: Except for S6, where a MSW resource recovery based system is applied, all other cases apply a conventional MSW system (no resource recovery).

multiplied by the assessed annual production of water from surface water sources (207 million m³ for domestic and 70 million m³ for industrial consumers) in 2030.

Improved fish yields from farming in downstream lakes due to improved water quality: Data collected through interviews with the regional Fisheries Office showed a decrease in fish catch of 5,000 ton/year in recent years [58] . Fish kills in Saguling (Figure 1) related to discharge of untreated wastewater have been described by [12] and [23] . By 2030 the fish capture is estimated to increase by 8,000 metric tons per year [58] . The increase of improved water quality was assumed to account for one-third of this expected annual gain of farmed fish in the Citarum basin [53] . A market prices of fish of 1.5 US $/kg was used [58] .

D. Environment:

Reduced frequency of river and reservoir dredging due to improved sludge and waste management: An estimated 35 l/person/year of septic waste [9] and 11% and 17% of domestic urban and rural solid waste [59] accumulating to nearly 500 ktonne/year are currently discharged to the surface water and will be prevented from being disposed in the surface water in 2030 with the described interventions (Table 5). With a cost of dredging estimated at US $3.76 [16] per ton of sediment (assuming no degradation), the total annual cost averted was estimated.

Rise in land prices due to improved aesthetics of riverside and lakeside real estate: Currently the Citarum riverside area is not developed due to water pollution. However, the area is expected to become a place where riverside property could be developed for inhabitants, small businesses, and tourist facilities in a situation where water quality is improved. The current agricultural land price (10.7 US $/m²) in the vicinity of Bandung was used as a benchmark for current riverside land prices. The current market suggests that land prices can climb to 71.3 US $/m² in highly desirable locations [16] . In this study 50% of this increase is attributed to improved water quality. This value was multiplied by an estimated 50 ha of land that could be developed each year after the water quality improvements have occurred [53] .

Averted maintenance costs of hydro-electric facilities: Improved solid waste management would avert the current costs of US $0.1 million [16] to evacuate the solid and unmanaged sludge waste to avoid equipment damage in the hydroelectric facility [53] .

E. Recovery of resources:

In scenario 6, resource recovery was considered (see also Table S4 in the OSI, Section 4):

Ÿ Off-site wastewater systems: Production of energy (sludge digestion), struvite (from centrate) and compost (digested sludge composting).

Ÿ MSW: Energy and compost production from organic waste and recovery of plastics and paper.

Ÿ Industrial wastewater: industries with a water consumption exceeding 2000 m³/d reused 80% of the effluent, whereas for industries using 1000 - 2000 m³/d this was 50%.

To compare the production (recovery) of resources with the potential demand in the Upper Citarum River catchment area in 2030, the compost, struvite, plastic and paper demand in the whole of West Java obtained from [29] was corrected for people living in the Upper Citarum River basin area. The amount of recycled water from industries was compared to the total domestic and industrial water demand in 2030 in the catchment area [16] . Energy production from digestion is compared to the energy demand for domestic wastewater treatment in the area applying aerobic granular sludge technology [19] .

To relate benefits and costs to either wastewater or solid waste interventions, BCR’s were presented separately. To analyze the individual impact of domestic, industrial and resource recovery interventions the BCR of scenarios S3: (50% ST and S3: 50% AT), S4: (Industrial interventions only), S5: (50% ST; S5: 50% AT) and S6 (50% RR) were determined (see also Table 5). A sensitivity analysis was performed in which input values that have the highest anticipated impact were varied: (1) health and access time benefits reduced from 100% to 50%, (2) lifespan of all wastewater and solid waste facilities varied from 20 years to 15 and 40 years, and (3) resource selling price reduced to half baseline values [32] [53] . Health and access time benefits were all attributed to domestic intervention. Water quality and environmental benefits were attributed to the fraction of COD load discharged by domestic and industrial sources respectively.

Figure 3 shows the average 2000-2009 water quality from upstream to downstream locations. Maximum allowable concentrations are defined in class II standard [60] and are COD 25 mg/l, BOD 3 mg/l and P 0.2 mg/l. From Sapan on (Figure 1) all measured values exceed these standards. Concentrations in several Citarum branches passing high density urban areas, show COD values approaching 500 mg/l and pathogen levels as high as 10⁷ Units/100 ml [40] .

Current cumulative pollution loads in the Upper Citarum River basin of COD (585 tonne/d), BOD (264 tonne/d), TN (91 tonne/d) and TP (20 tonne/d) were determined as the baseline values (Table 6). The sensitivity analysis with variations in domestic pollution coefficient [18] and performance of industries shows considerable differences with the baseline scenario (Table 6) with COD loads varying between 325 and 688 tonne/d (see also OSI, Section 7).

The domestic pollution loads entering the Upper Citarum River depend on (1) the type of technology applied (simple versus advanced) and (2) the rate of current households applying on-site systems in urban areas that will switch to an off-site system (Figure 4). The use of advanced compared to simple technologies has a minor impact on COD removal in the range of 3% - 4%, but a major impact on N-removal in which a rate of

a. Total baseline values comprise domestic and industrial baseline loads + agricultural loads; b. Total minimum values add domestic low pollution correction coefficient and Industrial best case + agricultural loads; c. Total maximum values add domestic high pollution correction coefficient and Industrial worst case + agricultural loads.

25% households switching to off-site systems leads to a 29% difference and a rate of 75% households switching to off-site systems leads to a 37% difference (Figure 4).

When increasing the switch factor from 25% to 75%, the additional removed COD and N increased with 5% and 1% for simple technologies and 6% and 9% for advanced technologies. BOD removal follows the COD trend, whereas P removal follows the N trend. Thus, the application of advanced technologies or a higher rate of people switching from on-site system to off-site systems mainly affects the additional nutrient removal, while organic removal is less affected. The numeric values of this analysis and further elaboration on costs of interventions and their impact on water quality are described in the OSI, Section 7.

The industrial pollution load amounts to 28% of the total load (Table 6), but industrial interventions can result in a relatively large COD reduction (35%) compared to the combined domestic and industrial COD reduction (see also OSI, Section 7).

Figures 5(a)-(f) shows the effect of interventions on the year round average water quality at different locations. The location names are approximate locations, as RIBASIM calculates concentrations in defined segments of a river (see OSI, Figure S8). Without

additional interventions all concentrations will increase compared to the 2010 values (Figure 3) with values as high as 100 mg/l of COD (Figure 5(a)). The modeled pollutant concentrations in water entering Saguling reservoir (approximate location Nanjung) are 80 mg/l COD, and 7 mg/l TN and 1 mg/l TP. When applying S3 with 50% AT (Figure 5(b)) a considerable drop in all pollution concentrations is achieved, whereas the introduction of industrial interventions result in approximately 20% COD & BOD removal and about 4% N & P removal (Figure 5(c)). The combination of these interventions (S5: 50% AT; Figure 5(d)) results in concentrations of 30 mg/l for COD, 10 mg/l for BOD, 3.4 mg/l for TN and 0.7 mg/l for TP. The maximum removal scenario (Figure 5(f)) results in values approximating the class II standard (COD < 25, BOD < 3, P<0.2 mg/l). Comparing Figure 5(e) (ST) with Figure 5(f) (AT) shows limited impact on COD or BOD removal, while considerable extra N, P removal is shown when using advanced instead of simple domestic technologies.

To reach the desired water quality levels (class II) both industrial and domestic municipal interventions are needed. In addition, the applied off-site technologies should also include N and P removal, requiring more advanced and more costly technologies (see Figure 4) compared to the application of only anaerobic filters.

The maximum quantified economic benefits are US $430 million per year in which health benefits account for 39% (Figure 6). Health benefits largely result from reductions in fecal-oral diseases, since 1) the people without access to wastewater (on-site and Bojong Soang WWTP) facilities (48%) all have access by 2030 (55.2% of health benefits), and 2) people that have access to a well-managed off-site or fecal sludge management system increased from 7% to 73% (44.6% of health benefits). Associated averted health impact due to irregular flooding events is only US $0.3 million.

Convenience and time savings are among the top five reasons for having a latrine in the home area [31] . Based on [31] a mean annual gain of US $77 million was determined for an additional 45% of the population in 2030 having access to their own

latrine facilities. This estimate is conservative as 1) it excludes travel needs for urination purposes, and 2) time is valued conservatively at 30% of the GDP per capita at hourly values.

US $13.9 million of the total US $23 million reduction in water treatment cost will accrue to the public water utilities and their consumers, while industries are expected to benefit US $4.7 million annually. The value of farmed fish yields is expected to be US $4 million annually.

The combined environmental benefits (increased land value, reduced dredging, averted maintenance costs of hydro-electric facilities) amount to US $17 million, of which nearly 90% is attributed to increases in land value based on annual land sales. The benefits of reduced dredging (even assuming no decomposition or organic waste) have minor benefits.

Table 7 shows the estimated reuse benefits based on the per capita production features and resource values (Table S4 in OSI). 87% of the US $147 million yearly potential revenues are from municipal solid waste, 11% from domestic wastewater treatment and recovery and reuse of its resources and recycling, and 2% from industrial wastewater treatment and recycling. The potential demand for recoverable resources is higher than the potential supply through recovery (Table 7), ranging from a factor 13 for water to a factor 1.6 for plastic.

Following the anticipated benefits (Figure 6) and corresponding investment and operational costs (Table S6 in OSI, Section 8) the BCR was calculated (Figure 7). The BCR varied between the interventions. The highest BCR of 3.2 is achieved by implementing simple technologies (S3: 50% ST), in other words an economic return of US $3.2 is anticipated for each US $1 invested. Because of higher costs for AT compared to ST, the BCR is expected to be lower for the AT scenario (BCR in S3: 50% AT = 2.06). The lowest BCR (0.52) is found in scenario 4 (industrial interventions alone). A joint approach

tackling both domestic and industrial pollution results in a BCR ranging from 1.83 (S5: 50% AT) to 2.64 (S5: 50% ST). However, simple technologies were not found sufficient to improve the water quality to levels approaching class II, especially in terms of nutrient (N, P) removal (Figure 5).

The economic returns on combined wastewater and solid waste interventions are lower than the returns on wastewater interventions only (Figure 7). Economic costs related to absence of solid waste services are associated with unhygienic living conditions [31] , loss of tourism developments or value of land [61] . Loss of land value, however, contributes to only a fraction (4%) of total related economic impact (Figure 6) and on their own do not outweigh the estimated costs (see Table S6 in OSI) to establish the MSW management systems. The willingness of households to pay for solid waste collection and treatment services has been better established compared to wastewater services in Indonesia [62] . This may be attributed to direct visibility of improving solid waste management [63] . Consequently, there is a larger potential for recovering some of the costs through MSW tariffs paid by households compared with tariffs for wastewater services. Potential revenues from fees were excluded from the BCR analysis, but are relevant for development of a cost-effective wastewater and solid waste management system.

Additional benefits of resource recovery from MSW can be a driver for improving water quality. The BCR (including MSW) of scenario 5 (applying AT) is 1.19 and will increase to 1.65 by applying resource recovery (Table 8). The BCR of scenario 6 with MSW recovery is even higher than the BCR of Scenario 5 applying ST (1.49) showing that required additional costs to improve the water quality can be financed through the sale of resources recovered from solid waste. However, application of resource recovery from wastewater only results in a minor increase in BCR (from 1.83 to 1.85) compared to applying only advanced technology. Thus, from a financial perspective using existing market prices, the additional investments to recover resources from wastewater out

weigh the benefits by a small margin.

In case recovered resources are sold at only half the current market price (Table 8) the BCR of resource recovery (S6) is lower than for AT, but still higher than 1. The BCR may change depending on the lifespan of capital stock (Table 8). A lifespan of 40 years results in a BCR approaching 5 (S3: 50%_ST). A major part of the cost (Table S1 in OSI) is related to sewer system developments that have typically much longer potential lifespans (even up to 100 years) [64] and therefore it is likely the BCR will be higher than the baseline BCR of 3.2 for that same scenario (S3: 50%_ST).

Evaluating the economic performance of wastewater and solid waste interventions is a complex process, involving many variables and alternative combinations and coverage levels of interventions. Therefore a methodology was developed that combines several assessment methods and data sources in order to support decision making. The added value of the integrated approach allows for a nuanced view on interrelations compared to single cause-effect relations [65] . Thus the effects of different interventions on water quality, resource recovery potential, and related economic returns could be evaluated in parallel (Figures 5-7). This parallel evaluation provides significant benefits in a dynamic context [66] . It also addresses the need for a method that can quantitatively evaluate a set of sanitation alternatives to resolve trade-offs across sustainability dimensions (social, environmental, and economic) [67] .

The practical application of the integrated approach is first demonstrated in the analysis of contribution of pollution per sector (industry, domestic or agriculture) related to the pollution prevention costs. The large contribution of domestic pollution was unambiguous and confirmed in a sensitivity analysis (see also OSI, Section 7). Presented results are in line with findings of [13] who determined that households contributed 55%, industries 40% and agriculture 6% of BOD pollution entering the Saguling reservoir. Reference [11] demonstrated the importance of fertilizer use management to avoid future coastal eutrophication in Indonesian Rivers, which corresponds with the large nutrient load as a result of agricultural activities (25% for P) determined in the current study. Despite a relative low (28%) contribution of industrial COD pollution, 35% of COD can be reduced by industrial interventions, whereas the investment costs for industrial interventions are less than 10% of the domestic interventions (Table S6 in OSI). Further, the number of industries is only a fraction (~1%) of the number of households in the Citarum area and monitoring interventions would be much more practical than monitoring individual household connections. Thus, although COD pollution from industry is relatively small, it is more cost effective (>factor 5) than domestic, which may help a policy maker in prioritizing interventions.

Secondly, the integrated approach supports determination of cost-effective interventions. The added value of applying more advanced technologies or switching more people to a sewer system showed that required additional investments can be justified from the point of nutrient removal, but less so from COD removal (Figure 4). In addition, the use of software tools like RIBASIM to model and estimate the impact of discharged pollution loads on the anticipated water quality allows the policy maker to relate interventions and their cost to applicable water quality standards.

Thirdly, linking the resource recovery potential and its revenues to its potential demand may benefit formulation of policies or increase government involvement to foster financial sustainability of sanitation facilities [24] . The value of recoverable resources from solid waste has resulted in a very active, but informal waste recovery sector in Indonesia [68] [69] . In addition, the demonstrated potential recovery of resources exceeding the agricultural demand allows for selective marketing, focusing on safe reuse (e.g. on non-edible crops) [70] [71] . Electricity production from the joint wastewater and solid waste facilities is potentially higher than the demand for domestic wastewater and supports the potential for a joint development of wastewater and solid waste facilities [72] .

Fourthly, monetizing both direct use and indirect non-use values of sanitation implementation in relation to achievable surface water quality enables the formulation of a cost and environmental effective approach. The performed analysis demonstrated that the most cost effective scenario (S3: 50%_ST) with the highest BCR differs from the scenario reaching the required water quality (e.g. S5: 50%_AT). Therefore, a policy maker needs to prioritize between these two options. As a cost effective strategy, application of advanced technologies may be restricted to the most highly densely populated urban areas (where most pollution is produced). Alternatively, a phased approach in which first simple (low cost) technologies are implemented that are later replaced, converted or extended by systems that allow for nutrient removal [73] . Monetizing benefits may further help to raise funds from other sources or actors that benefit from improved water quality, such as residential project developers or tourism sites [32] .

The outcomes of the study were formulated in a planning document for the Indonesian government [53] and confirmed our hypothesis that quantification of tangible outputs using the presented approach can support policy-makers in the field.

The presented framework can be further extended given the following considerations:

Ÿ To assess the sustainability of interventions and ensure that pollution is being removed and not displaced, environmental emissions other than water pollution (COD, N, P), such as odor or greenhouse gasses may be included. The effect of greenhouse gasses emitted by low cost technologies (e.g. anaerobic filters or septic tanks) is excluded from the current evaluation.

Ÿ In the determination of the water quality, several assumptions were made that may affect obtained results and could be incorporated in a next phase (see also OSI Section 9). First, a connection between surface and ground water was assumed in which infiltrated septic tank effluent load directly influences the surface water quality. Second, RIBASIM model disregards biological conversion of pollutants in the surface water, whereas these are observed in the field [12] . Third, all interventions are assumed to be designed, constructed, operated and maintained correctly, which may be optimistic in view of current practice [37] [74] . Fourth, the effect of dumped solid waste on water quality is excluded. Finally, surface water pollution from animal manure was excluded.

Ÿ The low BCR of industrial interventions (0.52, Table 8) and the weak mandatory industrial regulation in Indonesia [75] may suggest limited possibilities to implement industrial pollution prevention. However, alternative means to spur Indonesian industries to comply with environmental standards such as public disclosure (the regular collection and dissemination of information about firms’ environmental performance) have been shown to be effective [76] .

Ÿ Aerobic technologies were used as industrial references, whereas the use of anaerobic technologies may result in lower investment and/or operational costs [77] [78] .

Ÿ Not all economic impacts were quantified in this study (see also OSI, Section 10), such as consumption of fish imbibing toxic wastes or otherwise infected [79] , reduced land subsidence and improved recreational values [30] [80] . In addition, long-term impacts on the river and population of industrially discharged toxins and heavy metals were excluded and would specifically increase the BCR of scenario S4 (industrial intervention). Although the current study focuses on a part of a river basin only and uses locally collected data, a further detailing on a city or community level may be required to prevent overgeneralization and misunderstanding of individuals’ preference tradeoffs [81] .

Ÿ Applying advanced technologies (AT) will improve water quality (Figure 5), but will not increase quantified health or welfare impact. At the same, anticipated long- term effects of reduced eutrophication and less impacted ecosystem functioning [11] were not quantified, whereas these would further increase the BCR.

Ÿ The BCR considers the overall societal perspective, whereas different costs and benefits are incurred and enjoyed by different stakeholders. Thus, the costs of domestic interventions are to a large extent paid for by the national and local governments (in Indonesia ~ 70%) and to lesser extent by individual households [50] , whereas industries typically pay the costs of the interventions themselves [37] . Benefits of improved water quality as a result of interventions can be either increased revenues (e.g. sale of recovered resources) or averted costs (e.g. lower water treatment costs) which benefit a single party, or are generalized to the population (e.g. averted health or time costs) which benefit society as a whole [30] . In the elaboration of a planning document, the incidence of costs and benefits should be further detailed. In addition, serious institutional challenges have been identified in providing access to sanitation in a development world’s context [82] . To implement the planned sanitation development, the responsible institutions need to be identified and budgets should be allocated. A methodology to do this, involving private households, local and national governments, was described by [50] .