Report No 631/1/01


Various studies have suggested that in order to ensure access to adequate sanitation facilities for all in the country within the constraints of the country's financial resources, it will be necessary to use a mix of levels of service1, an option which (ignoring costs of pollution2) is significantly cheaper than high levels of service throughout: At the Water and Sanitation 2000 workshop in 1991 a scenario was proposed in which some 50% of sanitation systems in the urban areas of the country by the year 2000 would be ventilated improved pit (VIP) latrines (Jackson, 1991). Subsequently, the Municipal Infrastructure Investment Framework (MIIF) study (Ministry in the Office of the President and the Department of National Housing, 1995) proposed a programme of infrastructure provision that would eliminate much (but not all) of the backlog within 5 to 7 years and would match service levels with predicted household income levels in 10 years (i.e. by the year 2005). This programme would result in a 55:25:20 distribution nationally between full, intermediate and basic levels of service.

Both studies therefore have envisaged a significant amount of on-site sanitation in use in the urban areas of South Africa for the foreseeable future. However, a concern that is often raised in relation to the use of on-site sanitation is the potential pollution of water resources that is associated with these systems. This concern about environmental impact of on-site sanitation systems appears to be serious enough to persuade some decision-makers in the urban areas of the country to opt for the provision of full water-borne sanitation where, but for this concern about environmental impact, on-site sanitation might have been used, thereby foregoing the significant potential cost saving in the construction, operation and maintenance of the service.

There is therefore a need to translate the environmental impact of sanitation systems (and on-site sanitation in particular) into financial terms so as to enable a comparison of these systems to be made, which includes not only the cost of the construction, operation and maintenance of the systems, but also the cost of their respective environmental impacts.

This study:

1 A basic level of service for sanitation would comprise on-site sanitation (e.g. a VIP latrine), while an intermediate level of service would comprise simple water-borne sanitation. Simple water-borne sanitation may include on-site systems such as the LOFLOS (low flush on-site sanitation system, also referred to by some as an aquaprivy). A full level of service would comprise full water-borne sanitation. A basic level of service is sometimes referred to as a low level of service, while a full level of service is referred to as a high level of service. Lower levels of sanitation service therefore tend to be on-site services, whereas higher levels of service tend to be off-site services.
2 The term 'pollution' or 'pollutant' is used where the concentrations exceed acceptable levels. Otherwise the term 'contamination' or 'contaminant' is used.

Conclusions from this study are as follows:

Re-statement of conclusions from previous studies

The following conclusions from previous studies require re-stating:

  1. All sanitation systems contaminate the environment to some extent, although characteristics of the contamination may differ between different levels of service (LOS).
  2. More specifically:

  3. The subsoil conditions of on-site sanitation systems need to be permeable enough, and the soakaways need to be big enough to ensure that effluent does not surface, but remains in the sub-surface.
  4. If it does not remain in the subsurface, but surfaces onto the ground surface, it firstly poses a direct health risk from microbiological contaminants for the users of the sanitation system (and their immediate community), and secondly is also susceptible to being washed off the surface by rainfall directly into surface watercourses.
  5. If it does remain in the subsurface, then there is minimal health risk from microbiological contaminants to users (and their immediate community). The microbiological contaminants are generally filtered out in the subsurface within a short distance (of the order of metres). Phosphorus is generally adsorbed in the subsurface and travels very little distance at all. Nitrogen in the form of nitrate is removed to varying degrees in the subsurface, depending on the conditions there. The remaining nitrate acts like a tracer, and remains in the subsurface. While very little nitrogen may be removed once it has been transported into a zone of the subsurface that is poor in organic material, the transport of contaminants is very slow.
  6. While phosphorus tends to be removed efficiently in the subsurface for on-site sanitation systems, for water-borne systems phosphorus (even if a substantial proportion is removed in the treatment process) is discharged directly to the surface watercourse. Discharge into the surface watercourse is virtually immediate, and can be a significant source of contamination.


  1. Cost of environmental impact needs to be added to the cost of provision of sanitation infrastructure (i.e. construction, operation and maintenance) in order for a fair comparison to be made of different levels of service.
  2. The method is a useful planning tool, but needs further refinement in input data. It is very important to remember what one needs data for. Data for decision-making makes demands which can be unexpectedly at variance with scientific endeavour. A key characteristic of the method proposed in this study is the objective of getting into the right 'ball-park' with the costs. If they are borderline, then they require further investigation.
  3. There is a strong case for environmental planning to be done at provincial level -or even more broadly -although priorities and trade-offs may need to be made at the local level.


  1. In Gauteng, 60% of the population live in the Vaal Barrage catchment area; 25% in the Hartbeespoort Dam catchment area, 12% in the combined catchments of Rietvlei, Bon Accord, Roodeplaat and Boskop Dams. (Based on figures for 1990; the proportions are unlikely to change much in the period up to 2010). 85% of the Gauteng population therefore falls within the Vaal Barrage and Hartbeespoort Dam catchments. In terms of cost of impact, the major impact of sanitation is felt in the above catchments in the same order.
  2. Almost as critical as the fact that most of the Gauteng population falls within two impoundment catchments is the fact that most of the population is concentrated in two sub-catchments: Crocodile/Jukskei/Hennops Rivers (A2H012 weir) and Klip River.
  3. In terms of LOS, the provincial boundaries are critical. Low LOS are on the fringes or just beyond the boundaries of Gauteng. This highlights the difference between Gauteng and Region H. Virtually all sanitation in Gauteng is water-borne.
  4. There is a temptation to suggest that because virtually all sanitation in Gauteng is water- borne, one should simply stick to water-borne sanitation throughout and not bother with a small percentage of on-site sanitation. The response to that is that it is really an anomaly of administrative boundaries. Across the boundary of the province, the situation is very different. The principles that are applied in Gauteng need to be consistent with other areas, particularly those just across the boundary. If that is not done, then one may find policies in one area undermining policies in neighbouring areas (which are under different jurisdiction; as has been the case in parts of Kwazulu-Natal where the administration has been so fragmented in the past ). The need for some kind of cross- boundary consistency has been recognised by the demarcation board in the setting up of cross-boundary (i.e. provincial boundary) local authorities.
  5. Because on-site sanitation currently makes up such a small percentage of the sanitation in the impoundment catchment, it is not really feasible (at the level of resolution of the data) to identify the contribution of on-site sanitation to the pollutant load with the desirable level of certainty. One could provide upper and lower bounds (from the diffuse source loads, together with theoretical analysis).

Water quality

  1. Taking account of the range of permeabilities found in Ivory Park and Orange Farm, as well as the fact that the rate of movement in the unsaturated zone of the subsurface is significantly lower than that for the saturated zone, the rate of movement of contaminants in the subsurface in Gauteng can be taken to be of the order of I to 10m/a. The rate of movement of contaminants by wash-off from the ground surface, transport within the surface watercourse to an impoundment in Gauteng can be of the order of 50km in 2-3 days, or a month at the most. There is difference of about 5 orders of magnitude between the two rates. Subsurface movement of contaminants, with the geological conditions encountered in Gauteng (taking Ivory Park and Orange Farm as typical), is therefore unlikely to impact on water resources in any significant way within a 10 year frame. Surface wash-off and transport of contaminants, on the other hand, is likely to impact well within a 1 year time frame.
  2. There was considerable variability of the contaminant load data - both from the wastewater treatment works (WWTW) as well as from diffuse sources. There was also a poor relationship between discharge from sewage treatment works and discharge into the lake -which was surprising. In particular, 'spikes' in contaminant loading from the WWTW in the Hartbeespoort Dam catchment could not be identified at the entrance to the lake (at weir A2H012), even with lag effects being taken into account. Also of note is the fact that the concentration of PO4-P entering Hartbeespoort Dam at weir A2H012 was virtually the same as the concentration of effluent leaving the Northern works (WWTW) some 30km away.
  3. The effect of the wetlands on the Klip River was not investigated in depth in this study. The effect of wetlands on the nutrient loading on the Vaal Barrage - certainly compared with Hartbeespoort Dam - may be very significant.
  4. The existing (REM) models for both nutrient budget and nutrient-algae poorly described lake response in Hartbeespoort Dam over the past 10 years.
  5. Accounting only for phosphorus, a (modified).nutrient-algae model adequately (for the purposes of this study) described the lake response. This implies that even if the lake is nitrogen-limited at certain select times, the effect of phosphorus is overriding.
  6. By comparison with water-borne sanitation discharges -even from well-functioning WWTW meeting the special standard of 1mg/l PO4-P -pollution from on-site sanitation is negligible. The 'wild card' is grey water; although the effect is not completely random in that if the contaminants remain in the subsurface, it isn't a problem. It needs some serious attention. A controlled experiment may be the best approach to further investigation. Pillay by her assumptions suggested that it was negligible. Ashton and Grobler in their Botshabelo study identified it as a critical question, and presented a range of scenarios.
  7. Nitrate contamination of groundwater will occur. In Gauteng, contamination of groundwater has already occurred (e.g. in Soshanguve). Groundwater is certainly a strategic resource. Dolomitic areas need special consideration. However, fractured rock aquifers are small.
  8. It has been assumed in this study that one is only concerned with human wastes i.e. that one is able to address the problem of inorganic salts, refractory organics, heavy metals etc by other means - and at source.
  9. Because WWTW effluent standards are concentration-related (e.g. 1mg/l PO4-P), one needs to keep an eye on growth of household water consumption (and hence sewage flow) for the full water-borne (WB) LOS. The reason is that if the flow volume doubles (for the same concentration of contaminants), then the mass load doubles (while still meeting the effluent standard). That can have a serious effect on the receiving impoundments. Mass load may well be a more appropriate measure for monitoring contaminant levels than concentration.
  10. In terms of environmental impact, there is little difference between basic (e.g. the VIP) and intermediate on-site sanitation systems (e.g. the LOFLOS).


  1. For all water treatment - both surface water and groundwater - the cost is a step-wise function as one moves to new processes in the treatment train, with deteriorating raw water quality.
  2. Costs of surface water treatment: For the smaller sized -i.e. 50Ml/d -plant capacity (2000 costs) the cost sequence for different process combinations is estimated as follows:
  3. For a large capacity works (i.e. 200Ml/d) water treatment costs are about 20-25% lower than the above figures.

  4. Maximum additional cost of surface water treatment to deal with poor quality raw water roughly doubles the costs of conventional treatment. For a relatively small works (50Ml/d) the magnitude of the increase (from 30c/kl to 65c/kl) amounts to about 35c/kl. For a larger works (200Ml/d), the proportional increase would remain about the same, but the magnitude of the increase would be somewhat less - about 30c/kl.
  5. Costs of groundwater treatment: Treatment ~f groundwater resources is considerably more expensive than the treatment of surface water resources -with groundwater treatment ranging between R 1.60/kl and R3.15/kl depending on plant capacity and process.
  6. Because groundwater is generally not treated before use (in certain cases, it may be disinfected), the additional cost of treatment due to poor raw water quality for groundwater is suggested to be the full cost of treatment. This may not be entirely reasonable, but is suggested as a very worst case scenario for the purposes of this study.
  7. Assuming this to be the case, the additional cost of groundwater treatment is somewhere between 4 and 9 times the additional cost of surface water treatment (30-35c/kl).
  8. Costs of provision (i.e. construction, operation and maintenance) of the different levels of service of water supply and sanitation in Gauteng (2000 costs) are:
  9. Assuming a maximum yield (assumed to be natural MAR) for the Gauteng portion of the catchment of the Vaal Barrage downstream of Vaal Dam - essentially consisting of the catchments of the Suikerbosrant River (C21 ) and the Rietspruit/Klip River (C22) - of 275Mm3/a at an additional unit cost of treatment of 30c/m3, the total additional cost of treatment will amount to R82.5, say R83million/a. For a population of 4.5 to 5million people in the catchment (in 2000), this translates to only R 18, say R20/cap.a. At the lower concentrations of contaminants, it will increase the use of PAC (say 7c/kl), which amounts to less than R4/cap.a.
  10. A key requirement is that the (clean) water imported from Lesotho Highlands should not be mixed with (contaminated) water from the Vaal Barrage (requiring more sophisticated treatment processes).
  11. Similar calculations to those for surface water can be made for the use of groundwater resources. Assuming a maximum yield (assumed to be the Groundwater portion of the Groundwater Harvest Potential) for the Gauteng portion of the catchment of the Vaal Barrage downstream of Vaal Dam - essentially consisting of the catchments of the Suikerbosrant River (C21 ) and the Rietspruit/Klip River (C22) - of 125Mm3/a at an additional unit cost of treatment of R 1.90/m3, the total additional cost of treatment will amount to R237.5, say R238million/a. For a population of 4.5 to 5million people in the catchment (in 2000), this translates to only R53, say R50/cap.a.
  12. For the Gauteng portion of the catchment of the Vaal Barrage downstream of Vaal Dam, the sustainable yield of groundwater is about half that of the surface water. The additional cost of treatment of groundwater due to deteriorated raw water quality is about five times the equivalent additional cost of treatment for surface water. Translated to a cost per person per year, the additional cost of treatment of groundwater is therefore about 2.5 times the equivalent cost for surface water treatment. However, with current utilisation of groundwater of this catchment only about 6% (7.5Mm3/a) of the maximum yield, this cost is still far from being realised.
  13. By comparison with the costs of a higher level of service of water supply and sanitation, the maximum additional costs of treatment are small. For surface water, the additional cost of treatment (R20/cap.a) is only about 15% of the difference in cost between a basic and a full level of service (essential use) (R130/cap.a). If water usage for the full level of service increases towards convenience use, then the relative cost of treatment will drop even more. For groundwater, the additional cost of treatment (R50/cap.a) is about 40% of the level of service cost difference.
  14. In summary, even conservative (i.e. high) estimates of additional treatment costs (either surface water or groundwater), fully (i.e. very conservatively) assigned to pollution from sanitation systems, are still well less than half the cost difference between basic and full (essential use) levels of service of water supply and sanitation (based on the particular catchments used in the analysis).

Application of model to Hartbeespoort Dam

  1. From 1990 sanitation figures, there are not that many people with inadequate sanitation in Gauteng. Most are in the 'fringe' areas. These will start to be included in the cross- boundary municipalities; but if one is looking at Gauteng only, there isn't a massive problem. 90% have full water-borne sanitation.
  2. Based on estimated population figures and effluent flows from WWTW, water usage for sanitation in the Hartbeespoort Dam catchment appears to be considerably higher than (about double) the figures originally estimated for that level of service. The variation in flow is large enough to warrant more than one level of service for water-borne sanitation (e.g. low level use and high level use); although it is likely that there will be less variation in per capita contaminant loads than in per capita flows. It is unclear whether these changes in water usage occur evenly, in which case more than two (say three i.e. low, medium and high) levels of service for water-borne sanitation may be necessary, or whether there is a step-wise change, in which case the two levels of service may suffice.
  3. A not unreasonable flow/TP relationship, which aggregated both point and non-point sources, could be identified at weir A2H012. Such a flow/load relationship could not be established for Northern works alone; nor could it be established for weir A2HO12 minus Northern works. This is anomalous. A possible explanation is that there is in-stream sedimentation of P, and transport of this P into Hartbeespoort Dam is more dependent on general stream flow than on discharge from the WWTW. This is also supported by the fact that spikes of PO4-P discharged from WWTW are not evident in the flows entering Hartbeespoort Dam; neither is any clear lag evident between discharge from WWTW and entry into Hartbeespoort Dam (in contrast to what Pillay found for Inanda Dam). To draw any further conclusions would require more detailed analysis.
  4. The response of the lake to contaminant loads is not static. In particular, it appears that either an algal species shift or a change in the response of the algae to nutrient load can be triggered by events such as floods or droughts. These changes in lake response overshadow any changes in nutrient loading. The increased incidence of algal blooms in Hartbeespoort Dam since February 1996 has not been as a result of increased contaminant loading, but rather as a result of changed lake response, which appears to have been triggered by the high flows and contaminant loads of February 1996. Once the shift had occurred, the lake did not revert to its earlier response characteristics.
  5. In terms of allocation of the cost of pollution, approximately half of the increased cost of surface water treatment as a result of deteriorated water quality could be attributed to sanitation systems, and most of this to full water-borne sanitation.

General comments + conclusions.

  1. While in themselves, most of these findings are not entirely new (virtually all of them are based on existing data), it is the implication or significance of a finding in one area ( e.g. planning) for another area (e.g. water quality) that is particularly noteworthy.
  2. Water-borne sanitation (WB) discharges directly to the surface watercourses. It is currently the major contributor to pollution (primarily phosphorus) from sanitation systems in Gauteng.
  3. Even effluents meeting the effluent quality standards have a major impact on water bodies; and, added to that is the fact that a number of the sewage treatment works do not meet the standards at all times.
  4. The situation in Gauteng with respect to sanitation provision and consequent pollution over the next 10 years appears to be slow to change: It appears that it is water-borne sanitation that is -and will continue -to have the major effect on lakes in Gauteng (with a bit of a 'wild card' being diffuse load washed off the surface). Although one can get significant changes in demographics, settlement patterns and LOS at a local (i.e. municipal) level in a relatively short space of time (say, of the order of two or three years), it takes a fairly extended period of time (say, of the order of a decade or two) to change the overall patterns of a large area such as Gauteng.
  5. Unless the 'polluter pays' principle is established, there is little incentive to use a cheaper system than full water-borne systems.
  6. With return flows from WWTW making up such a large proportion of the flow into impoundments such as Hartbeespoort Dam, it is becoming difficult to separate out issues of quality from issues of quantity. More specifically, some service providers may prefer to use (and treat) return flows of poor quality rather than import expensive but excellent quality water through inter-basin transfer schemes.
  7. Environmental impact is one of several factors to be considered in the choice of level of service of sanitation. It is important not to confuse these different factors. Environmental impact should not be given as the reason for not using a particular system, when in fact the reason is motivated by other considerations, such as promotion of equity among users.

In the light of the above conclusions, the following recommendations are made:

  1. That the method of pollution costing proposed in this study be adopted as an input to deciding whether or not to use on-site sanitation;
  2. That policy regarding sanitation use be set at provincial level - or higher i.e. at national level - and that DWAF (as custodian of the country's water resources) issue permits for the use of on-site sanitation;
  3. That a workshop be held to publicise the results of this work, and to identify priority areas for implementation and further development of the principles proposed in this study.

Critical issues that require further investigation include the following:

  1. The environmental impact of grey water discharged to the ground surface;
  2. The mechanisms surrounding changes in the response of algae to nutrient loads -and possible interventions to control this;
  3. The stages at which new water treatment processes need to be introduced to deal with deteriorating raw water quality;
  4. Quantitative estimates of the costs of loss of recreation and property value as a result of deteriorated impoundment water quality;
  5. Clearer identification of the characteristics of natural resources (including both quantity and quality) and the 'ownership' of these.