OR/17/056 Groundwater vulnerability and risk assessment

Lapworth, D J, Stuart, M E, Pedley, S, Nkhuwa, D C W, and Tijani, M N. 2017. A review of urban groundwater use and water quality challenges in Sub-Saharan Africa. British Geological Survey Internal Report, OR/17/056.

Urban recharge

To date very few studies have focussed on urban recharge processes in Africa. Urbanisation affects both the quantity and quality of underlying aquifer systems by (Morris et al., 2003):

• Radically changing patterns and rates of aquifer recharge
• Initiating new abstraction regimes
• Adversely affecting quality

Recharge patterns can be affected by modifications to the natural infiltration system, by changes in natural drainage (for example by less permeable surfaces), and by any water supply network. Water may also be imported from outside of the city (Morris et al., 2003) adding to the volume of water that will recharge local groundwater. The net effect can be a rise in the total volume of recharge. This is most pronounced where on-site sanitation or amenity watering is important, and in arid and semi-arid climates where natural recharge is enhanced. Higher groundwater levels can impede drainage and result in groundwater flooding.

Borst et al. (2013) calculated a water balance for Nablus, Palestine where wastewater is discharged untreated into neighbouring wadis. This indicated that recharge by wastewater was as much as 50% of that from precipitation and that nitrogen pollution was about 60% of the agricultural load.

There are methods available to help to identify water inputs to groundwater. Vázquez-Suñé et al. (2010) proposed a series of tracers to estimate the proportion of water of different origins in the Barcelona, Spain, city aquifers. A preliminary hydrological model is required to choose appropriate compounds. They used a combination of major ions, Cl, SO₄, residual alkalinity, pollution indicators, total N, B, F, Zn, and isotopes δ³⁴S, δ¹⁸O, δ²H and a method employing mixing ratios. Their analysis identified leakage from water supplies and the sewage network, rainfall, runoff and surface water infiltration as the main sources of recharge to the Barcelona City aquifers. Kumar et al. (2011) used tritium and stable isotopes to identify recharge zones and sources of water in Delhi, India. They found that groundwater was being recharged by surface water in the dry season and also by precipitation in the monsoon.

Aquifer vulnerability

In recognition of the importance of protecting groundwater resources from contamination, techniques have been developed for predicting which areas are more likely than others to become contaminated as a result of human activities at the land surface. Once identified, areas prone to contamination can be subjected to certain use restrictions or targeted for greater attention. The fact that some areas are more likely than others to become contaminated has led to the use of the terminology ‘groundwater vulnerability to pollution’. Some authors view it as an intrinsic characteristic of the subsurface matrix. Others have associated vulnerability with the properties of individual contaminants or contaminant groups, or specific set of activities at the land surface (Sililo et al., 2001).

The vulnerability of groundwater to pollution depends upon:

• The time of travel of infiltrating water
• The contaminant attenuation capacity of the soil and geological materials through which the water and contaminants travel

Some assessments take into account the soil zone. The soil is a biologically active zone and many pollutants can be attenuated. For pollutants which are surface spread or applied, such as pesticides or artificial and organic fertilisers, this can be significant. It may also be important for above-ground latrines and solid waste disposal.

Sililo et al. (2001) set out a methodology for rating properties of the soil in Southern Africa as part of their assessment of groundwater vulnerability. In conclusion to their study Sililo and co-workers note the importance of soil properties in determining groundwater vulnerability; for example, in Windhoek, Namibia, the thin soils developed on the underlying Kuiseb schist and amphibolites and fracturing of the bedrock make the aquifer particularly vulnerable. Similarly Conboy and Goss (2000) found that Zimbabwean soils and geological settings did not offer much protection possibly due to deeply weathered joints and fractures. Mapani and Schreiber (2008) assessed the vulnerability of groundwater to a number of urban pollutants including pesticides, oil spills, toxic chemical spills, solid waste dumping, septic tanks and fertiliser applications. Soil properties also have been recognised as important for groundwater vulnerability in the karst Kanye wellfield, SE Botswana (Alemaw et al., 2004).

The unsaturated zone represents the next line of defence against contamination reaching the groundwater and needs to be fully considered in evaluation of risks both in terms of thickness and travel time. Flow in the unsaturated zone is predominantly vertical but there can be spreading on less permeable layers. Natural flow rates in the unsaturated zone of almost all formations do not generally exceed 0.2 m/d in the short term, and less when averaged over longer periods. Water flow and pollutant penetration rates in fractured formations may be an order of magnitude higher (Lawrence et al., 2001). Rock type, degree of consolidation and fracturing are key factors, especially for pathogens. In the saturated zone, attenuation will continue but may be at a lower rate because water moves more rapidly. In this zone, dispersion and dilution play an important role in reducing contaminant concentrations.

Early experiments using lithium bromide tracers in weathered basement geology in Botswana to model the movement of contaminants from a pit latrine to a borehole showed rapid transport times and good recovery, implying fracture flow with little diffusion (Lewis et al., 1980). Taylor et al. (2009) carried out a study in the weathered basement in Zimbabwe using E.coli bacteriophage and forced gradient solute tracer experiments. The tracer was largely unrecovered, and detection at the pumping well show that groundwater flow velocities exceed that of inert solutes and are consistent with statistically extreme flow pathways.

Table 6.1 shows typical permeability values for aquifer materials. Sands and gravels which have large well-connected spaces between the grains make good aquifers whereas silts and clays which also have large porosity but little interconnection transmit water very poorly. Fractured rocks transmit water very readily. Table 6.2 shows typical ranges for transmissivity values for both the fractured basement and weathered regolith from SSA. These typically vary by at least three orders of magnitude spatially and vertically (Taylor and Barrett, 1999). The greater the subsurface travel time the greater the opportunity for contaminant attenuation. Aquifer vulnerability can be defined into four broad classes (Table 6.3). Extreme vulnerabilities are associated with highly fractured aquifers which offer little chance for contaminant attenuation. The likely vulnerabilities of a range of broad categories of aquifer types are shown in Table 6.4.

Many large cities abstract water from thick sedimentary sequences e.g. the Karoo aquifers in South Africa, Tanzania, Mozambique, Malawi and Zambia and the regional Iullemeden aquifer system in Mali, Niger and Nigeria. These are often layered with complex flow patterns. Deeper groundwater reserves may be several thousand years old and of naturally good quality; however they may be mineralised at depth. Mountain valley sediments can be thick and also variable in permeability and are similar to the above. Laterite soils are widespread across a large part of SSA. The base of laterite soils have low vertical but high lateral permeability which means that contaminants can be transported significant distances (Bonsor et al., 2013). The variety of aquifer types and properties means that source identification and tracking can be challenging in these environments.

Table 6.1    Permeability values for different rock types
(from Lawrence et al., 2001).

Lithology	Range of likely permeability (m/d)
Silt	0.01–0.1
Fine silty sand	0.1–10
Weathered basement (not fractured)	0.01–10
Medium sand	10–100
Gravel	100–1000
Fractured rocks	Variable, 10s or 100s possible

Table 6.2    Transmissivity values for fractured basement
and weathered regolith (from Taylor and Barrett, 1999).

T range (m2/d)	Location (reference)
Fractured crystalline rock
5–60	Botswana (Buckley and Zeil, 1984)
0.8–90	Zimbabwe (Houston and Lewis, 1988)
0.07–250	Uganda (Howard and Karundu, 1992; Taylor and Howard, 1998, 1999)
Weathered regolith
0.2–20	Malawi (Chilton and Foster, 1995)
0.04–170	Uganda 1996 (Taylor and Howard, 1996); (Taylor and Howard, 1999 )
1–60	Zimbabwe (Chilton and Foster, 1995)

Table 6.3    Aquifer vulnerability classes (from Lawrence et al., 2001).

Vulnerability class	Definition
Extreme	Vulnerable to most water pollutants with relatively rapid impact in many scenarios
High	Vulnerable to many pollutants except those highly adsorbed and/or readily transformed
Low	Only vulnerable to most persistent pollutants in the very long-term
Negligible	Confining beds present with no significant groundwater flow

Weathered basement aquifers occur over large areas of Africa. These have no primary porosity with water being present in both the weathered and fractured layers. The thickness of the weathered layer controls the time of travel and hence the vulnerability (Figure 6.1).

The karst dolomites which underlie Lusaka and cities in the Copperbelt, such as Kabwe, are extremely vulnerable, particularly with the shallow water levels typical of this area. There have been a number of approaches to assessing karst aquifers. Andreo et al. (2006) used the combination of an intrinsic vulnerability map and contaminant properties to prepare specific vulnerability maps for faecal coliforms and petroleum compounds (BTEX) and applied this to a karst case study area in Southern Spain.

Table 6.4    Pollution vulnerability of principal hydrogeological environments (from Lawrence et al., 2001).

Hydrogeological environment		Travel time to saturated zone	Attenuation potential	Pollution vulnerability
Thick sediments associated with rivers and coastal regions	Shallow layers Deep layers	Weeks–months Years–decades	Low–high High	High Low
Mountain valley sediments	Shallow layers Deep layers	Months–years Years–decades	Low–high Low–high	Low–high Low–high
Minor sediments associated with rivers	Shallow layers Deep layers	Days–weeks	Low–high	Extreme
Windblown deposits	Shallow layers Deep layers	Weeks–months Years–decades	Low–high High	High Low
Consolidated sedimentary aquifers	Sandstones Karstic limestones	Months–years Days–weeks	Low–high Low	Low–high Extreme
Weathered basement	Thick weathered layer (>20 m) Thin weathered layer (<20 m)	Weeks–months Days–weeks	High Low–high	Low High

File:OR17056fig6.1.jpg

Figure 6.1    Water movement to wells and boreholes in weathered basement aquifers: a) thin weathered layer with fast movement in fractures; b) thick weathered layer (from Lawrence et al., 2001).

Intrinsic vulnerability was derived by two similar methods:

• The PI method which uses the Protection factor related to all the layers between the ground surface and the soil and the degree by which this is bypassed by karst features, such as swallow holes, the Infiltration factor.
• The COP method which uses the Concentration of flow (the surface conditions that control water movement to zones of rapid infiltration), Overlying layers and Precipitation (both quantity and intensity).

These methods both classified large areas of the Sierra de Líbar as having elevated vulnerability, with the PI method giving greater areas as very high due to the epikarst development. These classifications were validated using tracer tests and hydrographs. Specific vulnerability was assessed by modification of the O factor to taken pollutant attenuation into account.

Aquifer vulnerability mapping was used as a planning tool and applied to the coastal sand aquifer at Calabar, Nigeria, where uncontrolled disposal of domestic, industrial and agricultural wastes have caused groundwater contamination (Edet, 2013). The DRASTIC method used seven parameters (depth to groundwater table, net recharge, aquifer media, soil media, topography, influence of vadose zone and hydraulic conductivity), to produce vulnerability maps. Documented nitrate concentration in hand-dug wells and boreholes are in agreement with vulnerability zones. A recent study undertaken by BGR within the Lusaka region used the DRASTIC method to map aquifer vulnerability. Tilahun and Merkel (2010) also used DRASTIC to assess and map groundwater vulnerabilities in areas of Dire Dawa, Ethiopia. The study successfully identified area of low, medium and high vulnerability.

Ibe et al. (2001) assessed the vulnerability of the Owerri aquifer in southeast Nigeria in order to develop a protection strategy. They compared the results of several models, including GOD, Siga, Legrand and DRASTIC. This confirmed the vulnerable nature of the sandy and gravelly sequences which underlie the city of Owerri and the area to the southwest.

GIS based vulnerability models: e.g. GOD, LeGrand and DRASTIC. The GOD model was developed by Foster (1987) and uses a combination of ratings from three factors; groundwater confinement, overlying strata and depth to groundwater. The LeGrand model (LeGrand, 1964) uses a similar rating method. These earlier models formed the basis of parameter weighting and rating methods of which the most commonly used model is DRASTIC. DRASTIC was developed by Aller et al. (1987) and uses seven parameters; depth to groundwater, net recharge, aquifer material, soil material, topography, retardation in vadose zone and hydraulic conductivity. A DRASTIC vulnerability index is obtained by computing linear combinations of the rating and weights for each factor.

Local pathways

As well as moving through the body of the aquifer contamination can occur via pathways resulting from the design and construction of the supply or its deterioration with time (Figure 6.2).

Localised contamination is a very common cause of the decline in quality of groundwater supplies, and is frequently illustrated by the rapid rise in the concentration of contaminants following rainfall events (Ishii and Sadowsky, 2008). It can occur either where contaminated water:

• Is in direct contact with the head-works of boreholes, wells and springs and where pathways exist that allow this to mix with the water supplies
• Has infiltrated into the sub-surface in the close vicinity of a borehole, well or spring moves along fast horizontal pathways to the supply (Lawrence et al., 2001).

Rapid pathways in high risk terrains

Figure 6.2 summarises the key pathways in high risk settings such as fractured basement terrains with lateritic soil, highlighted in orange, including surface and subsurface pathways for migration of pollutants from sources to receptors. Very rapid horizontal pathways exist in the shallow tropical soil zone (5), which may be laterally extensive, providing transmissivities in excess of 300 m²/day. Rapid vertical pathways also exist due to the presence of natural macro-pores e.g. from burrows and tree roots (6), which can reach significant depths in places.

File:OR17056fig6.2.jpg

Figure 6.2    Pathways for local contamination into groundwater supplies in high risk basement settings (after Lapworth et al., 2015).

Combined, these more rapid pathways make shallow wells and spring sources particularly vulnerable to contamination and are increased during high water table conditions or when soil infiltration capacity is exceeded. Horizontal saturated groundwater flow, both in the lower permeability horizon above the weathered basement (7) and in the weather basement and fractured basement (8) is a pathway which can affect deeper groundwater sources such as boreholes. These pathways are slower and longer and provide the greatest attenuation potential for hazards.

In areas with red tropical soils and laterites groundwater flow exhibits extremely high permeability characteristics, i.e. very rapid transient pathways may operate for short periods of time and show sudden changes in permeability. The combination of high rainfall and the prevalence of these types of tropical soils suggest that a significant part of Sierra Leone, and neighbouring regions, may be susceptible to these types of extreme hydraulic flow conditions. This, combined with the fact that diffuse open defecation is widespread, cast doubt on the simplistic use of single minimum separation distances from particular hazard sources, and requires further investigation.

Surface pathways include surface runoff (1) which can contaminate surface waters and poorly constructed wells, bypass pathways for contamination of well and spring collectors by ropes, buckets used to draw water (2). Shallow sub-surface pathways include vertical soil flow from surface (3) and subsurface sources (4) where there is hydraulic continuity, e.g. from a liquid discharge or from a buried source such as a pit latrine, cemetery or buried waste.

Localised contamination will result where:

• Potential contaminating activities are not excluded from the vicinity of the headworks;
• Sanitary protection measures employed in the headworks are insufficient;
• The design and construction of a groundwater supply is inadequate.

A summary of these factors is shown in Table 6.5:

Table 6.5    Pathways and indirect factors influencing contamination
of groundwater sources (from Lawrence et al., 2001).

Source type	Pathway factor	Contributing factors to contamination
Protected spring	Eroded backfill or loss of vegetation cover Faulty masonry	Lack of uphill diversion ditch Lack of fence Animal access close to the spring
Borehole	Gap between riser and apron Damaged apron	Lack of diversion ditch Lack of wastewater drain Animal access to borehole
Dug well	Lack of headwall Lack of cover Use of bucket and rope Gap between apron and well-lining Damaged apron	Lack of diversion ditch Lack of wastewater drain Animal access to dugwell Uncontrolled use

Dug wells are one of the lowest-cost forms of water supply. They are particularly vulnerable to contamination as it is difficult to ensure that the lining of the top layers is impermeable. The most common way of collecting water from open wells is the use of rope and bucket technology (Figure 6.6). Ropes and buckets are often left on the ground and are an important route for groundwater pollution by faecal coli forms (Cronin et al., 2006). When large diameter wells are abandoned, due to alternative sources or changes in groundwater tables, they are often then used to dispose of household waste material and become a subsequent source of contamination in the sub-surface.

File:OR17056fig6.3.jpg

Figure 6.3    Rope and bucket used to draw water.

Protection

Groundwater protection is complex in Africa, with many small sources being used, compared to the fewer larger sources characteristic of Europe (Robins et al., 2007). Nevertheless, some areas lend themselves to conventional zoning of land-use, such as the dolomites underlying Lusaka. In Lusaka many urban supply sources now have 1 km protection zones around them. A key recharge zone is the Lusaka Forest Reserve with land-use increasingly dominated by small-holdings with increased runoff and reduced recharge. Schemes such as ‘aquafarms’ are being proposed to protect and enhance recharge for municipal supplies in towns such as Kabwe, Central Province, Zambia.

Demlie et al. (2008) used both major ion chemistry and stable isotopes to assess groundwater occurrence and flow in a heavily urbanised area of Ethiopia, including Addis Ababa. They were able to identify three flow systems: shallow controlled by steep topography, intermediate with some recharge from the shallow system, and deep, getting meteoric recharge and interconnected with overlying water through faulting. Understanding these complex flow patterns is key to protecting groundwater.

In many countries well protection zones are defined using deterministic models based on Darcian flow assumptions. Taylor et al. (2004) reviewed evidence from field studies to determine the effectiveness of well protection zone methodologies. They concluded, citing evidence from natural and introduced microbiological tracers,that statistically extreme flow velocities must be considered, and that bulk macropore flow and presumed pathogen survival times (30–50 days) are not consistent with field observations. This is of particular relevance in fissured aquifers e.g. crystalline basement aquifers, as well as the karstic dolomite aquifers, such as those found in Zambia and at the base of lateritic soils which are widespread in SSA (Bonsor et al., 2013).

Frind et al. (2006) developed a methodology for deriving a protection zone for individual wells in a complex multi-layered aquifer. This assessed both the intrinsic vulnerability of the well and the impact of potential sources giving a quantitative risk assessment for water managers. Groundwater protection priorities for Dar-es-Salaam, Tanzania, were established using an empirical model (Mato, 2007). This used five factors to derive a protection score:

• Water quality — rating system developed using nitrate concentrations
• Aquifer yield — rated to provide a public supply (>40 m³/hour) as the top yield
• Vulnerability — developed using DRASTIC coupled to GIS,
• Use value of water — highest in areas where no piped supply and lowest in areas where groundwater is not used.
• Landuse — highest ease of implementing measures in urban/fringe areas. And lowest in unplanned and industrial areas.

The results indicated areas of the built-up city centre and other areas with a sewerage system had the highest priority for groundwater protection.

In Iganga and parts of Kampala, Uganda, analysis of data from protected springs, wells and boreholes in weathered basement rocks, showed that boreholes were less vulnerable to contamination than wells (Howard et al., 2002). The results indicate that a horizontal separation of 10 m between sanitation facilities and groundwater sources was adequate even in the rainy season. Nsubuga et al. (2004) looked at the effectiveness of protection of a number of springs used to provide water for settlements without a piped, treated supply in Kampala, Uganda. Both high and low income areas were assessed. The protection in place was shown to be ineffective with ammonium, nitrate, faecal coliforms and faecal streptococci detected at over the guideline levels for potable water. Concentrations were higher in the high-density settlements. Pit latrines and animal wastes were found within 5 m of the springs. Furthermore, these springs were vulnerable to a rapid deterioration in water quality immediately following rainfall (Ishii and Sadowsky, 2008), probably as a result of surface contamination accessing the water supply though the highly degraded spring protection (Figure 6.4).

File:OR17056fig6.4.jpg

Figure 6.4    Highly degraded spring protection in Kampala, Uganda.

Edet et al., (2012) collated results from systematic field sampling and analysis in Calabar, Nigeria, were integrated using descriptive statistics; correlation matrices; bivariant plots; geochemical modelling; and a mixing model to gain insight into the hydrogeochemical processes. The dominant processes controlling groundwater chemistry were found to be silicate weathering, cation exchange and human activity (waste disposal).

Kreamer and Usher (2010) set out the steps for improving groundwater protection in SSA by mirroring the strategies already adopted by more-industrialised nations, namely:

• Define acceptable risk for African populations and ecosystems.
• Establish numerical, heath-based guideline values (beyond World Health Organization guidelines).
• Initiate risk-based remediation predicated on improved site characterization.
• Create hydrogeological and water quality data storage systems that are accessible, electronic, and versatile.
• Formulate a common vision on monitored natural attenuation and technical impracticability.
• Encourage proactive management, leak detection systems, and early remedial action beyond ‘emergency’ response.
• Strengthen natural protected areas.
• Consider implementation of guidance and strategies from other countries.

Risk assessment

Risk assessment needs to demonstrate the presence of hazards:

• Source hazard
• Pathway hazard by which the risk is transmitted
• Receptor which is harmed

Trauth and Xanthopoulos (1997) described a process in which the impairment of groundwater quality due to non-point sources of pollution in urban areas could be assessed, using Karlsruhe as an example.

This included:

• Design of groundwater quality monitoring network
  

• Sampling point design
• Spatial distribution

• Sampled depth

• Determination and characterisation of the catchment areas
  

• Groundwater model to determine recharge area — groundwater recharge
• Anthropogenic pollutant emission risk-landuse

• Establish rules to link landuse and groundwater contamination on the basis of the observed data

They found a number of problems with this type of approach. In urban areas the different types of land use lie close together, and these change with time. The model did not sufficiently account for decreased recharge in paved areas and also did not deal very well with leaking sewers, and particularly the additional recharge from exfiltration.

Cissé Faye et al. (2004) assessed the vulnerability of the aquifer underlying the Thiaroye area of Dakar, Senegal, to pollution from urban development and other land uses and related these to aquifer characteristics using GIS. Water quality data were grouped according to landuse features. They found nitrate to be associated with urban areas where the water table was shallow and the aquifer was oxygenated. In contrast nitrate was very low in uninhabited areas, particularly reforested zones.

Love et al. (2004) used factor analysis to develop separate chemical signatures for uncontaminated water, agricultural activities, mining activities and potentially sewage inputs using two examples from Zimbabwe. Using the example of an iron ore mine the analysis found:

• Ca, Mg and HCO₃ representing the dolomitic water signature
• K and NH₄ representing fertiliser or livestock manure
• Na, Cl and SO₄ associated with the main mine dumps and workshops.

Whereas at a sewage disposal works serving Harare they found:

• Cr inversely associated with PO₄, Pb and Ni
• High NO₃, PO₄, minor Fe- sludge and effluent related
• Fe, Ni due to possible geological reason

Orebiyi et al. (2010) showed that high values of colour, turbidity, nitrate, iron, manganese lead, total suspended solids, phosphate, bacteria and total coliforms were related to pollution hazards in urban areas. In peri-urban areas the levels of these contaminant was lower. Robins (2010) propose an approach for the evaluation of vulnerability to pollution of African basement aquifers which uses a tick-box approach for field use for a list of parameters (Table 6.6). This assessment can be adapted to focus on the parameters important in the area of concern by weighting the score. It excludes parameters which are difficult to evaluate in the field, such as unsaturated zone properties.

There have been a number of approaches to risk assessing karst aquifers. Mimi and Assi (2009) described a two-stage assessment using GIS to make a risk assessment of an aquifer in Ramallah district of Palestine:

Hazard mapping:

• Definition and inventory of hazards — divided into infrastructural, industrial and agricultural activities
• Hazard data requirements — assessing harmfulness of each type of hazards — nature, location, characterisation, quantification
• Rating and weighting of hazards — allocation of value between 10 and 100

Risk mapping:

• Product of intrinsic vulnerability and hazard map.

Using this assessment, Mimi and Assi (2009) identified the main hazards as urban areas with leaking sewers, unsewered villages, wastewater discharge to surface water, solid waste disposal, petrol stations, quarrying and stone cutting and industries including pharmaceuticals, dairy, textiles, detergents and soft drinks. Overlaying this on vulnerability suggested that only 1% of the area was at high risk and 4% at moderate risk.

Ducci (1999) used a GIS-based risk mapping scheme for an area of southern Italy using a similar combination of a hazard and a vulnerability map and combined these with a third layer representing the socio-economic value of the resource.

Sanitary risk assessment

Sanitary risk inspections have been described by the WHO (WHO, 1997) as:

“A sanitary inspection is an on-site inspection and evaluation by qualified individuals of all conditions, devices, and practices in the water-supply system that pose an actual or potential danger to the health and well-being of the consumer. It is a fact-finding activity that should identify system deficiencies — not only sources of actual contamination but also inadequacies and lack of integrity in the system that could lead to contamination.”

Surveys are carried out using a series of simple questions, specific for the water source, designed to identify the common hazards and pathways to contamination that may be present at a small water supply, such as a hand-dug well, or protected spring. Normally, the survey is done at the same time as the water is sampled for analysis of faecal indicator bacteria and the combination of the two results provides a basis for the design of remedial actions to protect the source. However, the results of the surveys can be used on their own, and because they are simple to administer the WHO suggest that surveys should take precedence over analysis (WHO, 1997).

A sanitary risk assessment for open dug wells in the Maldives showed that only 6.4% of wells sampled met the WHO Guidelines for faecal coliforms (Barthiban et al., 2012). The most important factor was the separation between the well and any latrines. Due to the vulnerability of the hydrogeological setting, it was not possible to establish a safe distance. In contrast, Wright and co-workers could not find any correlation between thermotolerant coliforms and sanitary risk score from a survey of 263 wells surveyed over a six year period in Kisumu, Kenya (Wright et al., 2013); however, there was a significant correlation between the nitrate and chloride concentrations and the density of pit latrines in a 100 metre radius of the sample point.

In Bangladesh, tube wells in flood-prone areas were found to be vulnerable to contamination by faecal coliforms. This was poorly correlated with sanitary risk factors, but was better predicted by a history of inundation. In northern Mozambique, there was a high risk of faecal microbiological contamination related to local as opposed to aquifer pathways. The principal factors were stagnant water on and around the wellhead, loose base of the handpumps, cracks at the base of the handpumps, and buckets and rope that can become contaminated (Godfrey et al., 2005).

A more substantial approach for making a risk assessment of private water supplies has been introduced into the UK and is being implemented in England and Wales by local environmental health departments with the support of the Drinking Water Inspectorate (http://www.privatewatersupplies.gov.uk/private_water/21.html; accessed 01/08/2014). However, the principles are the same as the simpler sanitary survey and the results are used to inform the management of water supplies.

Table 6.6    Vulnerability scorecard for weathered
basement aquifers (after Robins, 2010).

Parameter	Indicator	Score
Topography	Flat	0
	With hollows	3
Depth to water	<5 m	3
	5–10 m	2
	>10 m	1
Clay zone thickness	Absent	3
	<2 m	2
	>2 m	0
Degree of weathering	Shallow <5 m	1
	Thick >5 m	3
Vegetation and land use	Sparse cover	2
	Farmland	0
	Livestock	1
Laterite present	Absent	0
	Patchy	1
	Continuous	2
Human dependence	High	2
	Other sources	2
	None	0
Polluting activities	Mining	3
	Fuel dumps	2
	Livestock	1
	Few	0
Total

Risk management

Tait et al. (2004) addressed the issue of borehole location in urban areas. The Borehole Optimisation System (BOS) was developed to predict groundwater quality at a potential new sites and to use this to determine the best sites for future development. This was a data-intensive model which integrated a Modflow catchment zone probability model with a GIS-based landuse model, and contaminant information to inform a probabilistic pollution risk model for a user defined borehole. This was applied to Nottingham as a case study. It made assumptions about steady-state abstraction and continuous pollution sources making it perhaps unsuitable for an evolving urban area.

Mapani (2005) assessed risks to groundwater quality and their mitigation for Windhoek, Namibia. Particular source hazards considered were cemeteries, sewage pipe leaks, chemical and petroleum spills, and amenity pesticides. As described in Aquifer vulnerability (Sililo et al., 2001) this aquifer is very vulnerable since the soil cover is thin and groundwater flow is through fractures. In order to manage the risk, Mapani (2005) recommends that:

• A minimum of 20 cm soil cover is maintained over surface expression of faults
• Recharge areas should be protected from further development
• Monitoring of petrol stations, landfills, cemeteries and the sewage system

Pegram et al. (1999) set out the key factors for characterising settlements which can be used to guide the identification of priority problems. These include:

• Water quality problem, priority water quality effects on health, ecology or water treatment, associated water quality problems, critical receiving water
• Settlement character — activities, infrastructure
• Institutional arrangements
• Socio-economic conditions

Using examples from urban centres in Nigeria, Ojo (1995) discuss the reduction of health risks from drinking water. They focus on the control and disposal of solid wastes and the need for public awareness of risks for groundwater and surface water protection.

Risks to health from use of small private supplies can also be managed by implementation of WHO water safety plans. Mahmud et al. (2007) assess their implementation in an area of Bangladesh. These comprised baseline assessments of water quality, sanitary condition and hygiene practices and implementation of actions, such as relocating pit latrines or a level of household-scale water treatment. In the UK, this approach is being used effectively to monitor and manage the quality of private water supplies.

Groundwater management

Groundwater has many advantages and is a significant water resource for SSA which can be developed for low capital expenditure compared to large surface water schemes, is generally good quality, has large storage and capacity, can buffer climate extremes and can be developed in close proximity to urban centres, again keeping infrastructure cost down. However, the resource is not without potential quantity and quality problems, and therefore must be managed effectively if it is to be used sustainably in the long-term. This includes assessing requirements for human use such as drinking water and irrigation as well as maintaining base flow for groundwater dependant ecosystems such as wetlands and rivers. There are different dimensions of sustainability including societal, economic and environmental considerations — sustainability is only achieved when there is a discourse between all three dimensions (Gauthier and Archibald, 2001).

Morris et al. (2003) distinguish between two contrasting philosophical standpoints for groundwater management, the ‘technical’ view proposes an incremental approach based on reinforcing existing institutions to address hydrogeological problems, and the ‘holistic’ approach in which coping strategies and technical measures are used together to resolve management issues. The holistic view has its roots in the idea that existing institutions and management systems are not adequate to resolve current and future groundwater problems and that difficult choices may need to be made regarding how the ultimately finite resource is used. The basis of effective groundwater management responses are outlined by Morris et al. (2003):

• Awareness of the status of groundwater (both quality and quantity)
• Understanding of the hydrogeology to be able to identify options to remedy the problem
• Water laws and rights in place that are accepted and clear
• Surveillance, to monitor adherence to regulation
• Awareness in governmental planning and society of the importance of groundwater

It is true to say that these are rarely all in place, and certainly water rights and laws can be difficult and unclear especially with regards to ownership, i.e. landowner vs common pool ownership Morris et al. (2003). The lack of basic data sets to underpin groundwater management options should not be underestimated in many parts of Africa.

Alkhaddhar and Hepworth (2001) set out a ten-point plan for managing groundwater in Dar-es-Salaam resulting from investigation of groundwater quality at a number of sites:

• Coordination of monitoring/sampling
• Hydrogeological investigation and modelling
• Establishment of protection zones
• Quantification and characterisation of current aquifer exploitation
• Promotion of alternative ways to ensure sustainable abstraction
• Promotion of groundwater protection
• Investigation and promotion
• Generation of funding to support activities
• Facilitation of effective communication
• Education and training

Guidance on the mitigation of groundwater pollution in developing countries is set out by Foster and Vairavamoorthy (2013) For sanitation they recommend a more integrated approach to urban water supply, mains sewerage provision, and urban land use to avoid persistent and costly problems, especially where local aquifers are providing the municipal water supply. Public administrations and water service providers can employ a number of simple measures to improve groundwater sustainability (e.g. Drangert and Cronin, 2004; Foster et al., 2010). These could include:

• Prioritising recently urbanised districts for sewer coverage to protect good quality groundwater and/or limiting the density of new urbanisation served by in-situ sanitation to contain groundwater nitrate contamination.
• Establishing groundwater source protection zones around all utility waterwells that are favourably located to take advantage of parkland or low density housing areas.
• Ensuring availability of ‘nitrate-dilution capacity’ by securing a stable source of high quality supply for blending.
• Involving residents in wastewater quality improvement by seeking cooperation on not discarding unwanted chemical products to toilets or sinks, and avoiding the use of particularly hazardous community chemicals.

Foster et al (2010) state that much more effort is needed to change attitudes towards wastewater reuse and associated energy and nutrient recovery, which can contribute positively to urban groundwater management. New technologies that promote wastewater as a resource need to be tailored to conditions in low-income countries, including low-cost membrane systems, hybrid natural and constructed wetlands and eco-sanitation, which separates urine from faeces and recovers both for reuse. This reduces the subsurface contaminant load. But large scale retro-installation in existing dwellings is not straightforward and it is not well suited for cultural groups who use water for anal cleansing.

Where there is significant industrial activity interspersed with public utility and private domestic wells, it is essential to carry out groundwater pollution surveys and risk assessments. Fuel storage facilities, chemical plants, paint factories, metallic and electronic industries, dry-cleaning establishments, leather tanneries, timber treatment, and waste tips can all discharge mobile, persistent, and toxic chemicals with potential to contaminate groundwater and thus need to be closely monitored. The intensity of subsurface contamination is not necessarily a function of the size of industrial activity. Often small, widely-distributed enterprises use considerable quantities of toxic chemicals and pose a major threat since they operate outside the formal registers and environmental controls.

Groundwater pollution surveys and risk assessments should be commissioned by the public health, environmental, or water resource agencies, in close liaison with water service utilities, using recommended protocols (Foster et al., 2002, reprinted 2007). A typical survey would involve the following steps:

• A systematic survey of existing and past industrial activity to assess the probability of different pollutant types contributing to subsurface contaminant load.
• A groundwater pollution hazard assessment considering the interaction between the subsurface contaminant load and local aquifer pollution vulnerability.
• A detailed groundwater sampling and analysis programme with the analytical parameters being guided by the above survey.

The results of such scientific survey and assessment work should guide policy by:

• Introducing pollution control measures including better constraints on handling and disposal of industrial effluents to reduce groundwater pollution risk.
• Increasing quality surveillance for selected utility wells and/or progressive investment to replace wells considered at greatest risk of serious pollution.
• Advising and warning private domestic well users of potential pollution risks, imposing use constraints, and in extreme cases forcing closure of wells.
• Designing a long-term focused groundwater monitoring programme to improve water quality, surveillance and security.