Modelling uncertainty under future climate change and socio-economic development is essential for adaptive planning and sustainable management of water resources. This is the first study in South Africa incorporating uncertainty within climate and development scenario modelling for understanding the implications on water availability through comparison of the resulting uncertainty. A Water Evaluation and Planning model application was developed for the Amatole system (South Africa), which consists of three catchments with inter-basin transfers. Outputs for three sets of scenarios are presented, namely development-only, climate-change-only and climate-and-development scenarios. Near future (2046–2065) development uncertainty was estimated from three scenarios (lower, intermediate and upper) and climate change uncertainty from nine downscaled global climate models under the A2 emissions scenario. Consideration of development increased the uncertainty associated with climate-change-only scenarios, particularly at low flows. Water deficits are projected in the future for the Amatole system as the present water infrastructure cannot meet water demands under the near future intermediate and upper development scenarios. The deficits are likely to be exacerbated by inclusion of environmental flows (not included in the model). The recommended strategy is that of adaptive management, in combination with continual monitoring of climate and development changes, for reducing future uncertainty.

INTRODUCTION

Reports by the Intergovernmental Panel on Climate Change (IPCC Fourth Assessment: Parry et al. 2007) have placed emphasis on freshwater resources, particularly their vulnerability. The development of management and adaptation measures is therefore critical and should recognise that water resources are fundamental to basic human needs and for facilitating present and future development projects. The climate in South Africa is naturally highly variable, and this, along with over-allocated water resources, results in a vulnerable water supply (Walmsley et al. 1999; DWA 2013). However, few projects based within southern Africa have attempted to quantify the additional effects of climate change to the range of issues currently causing the deterioration of South African water resources (such as development, mismanagement and pollution). This is despite suggestions that a combined evaluation of climate and non-climate risks is required when developing a decision-making system that incorporates climate risk and uncertainty (e.g., Willows & Connell 2003). The current study aims to quantify future changes from the current situation for a South African case study, and the resulting uncertainty in available water, from projected development and climate scenarios for the near future period.

Uncertainty in water resources

Uncertainty in our understanding of water resources exists in various forms including data (availability and accuracy), models (related to structure and parameter estimation), impacts of economic and social factors, and management approaches (Hughes et al. 2011). The need to acknowledge and to incorporate uncertainty in hydrological modelling, decision-making and policy has been emphasised in recent years (Pappenberger & Beven 2006; Beven & Alcock 2012). Cognisance and incorporation of this uncertainty in future predictions is necessary to facilitate adaptive planning and sustainability in the management of our resources.

The current study specifically addresses prediction uncertainty arising from various climate change models and future socio-economic development scenarios. Note that the basic uncertainties in the ability of the model to represent hydrological responses in the catchments have been ignored in favour of the much larger uncertainties associated with future climate and development scenarios (reviewed further under the Discussion section). The predictions of concern are for a South African case study for the near future period (2046–2065), specifically in terms of water availability and requirements. The current study has focussed on the near future period, with the expectation that adaptive planning over the next 30–50 years will build resilience and knowledge of the system, thereby allowing water services delivery institutions to plan for the far future period (2081–2100), for which the predictions are much more uncertain (e.g., Wolski et al. 2012). Following the presentation of the model, results for three sets of scenarios are presented (i.e., development-only, climate-change-only and climate-and-development scenarios). The current study discusses the implications for adaptive management in the context of the South African institutional framework, the planning instruments, the available information for water authorities and the recommended adaptive measures to deal with future uncertainty and decision-making risk.

Background to climate change and water resources in South Africa

Climate change is expected to affect water availability through its effects on the water cycle. In general, higher temperatures are expected to increase evaporation from the oceans and thus increase the global average rainfall (Jackson et al. 2001). However, regional patterns are expected to deviate from the macro global changes due to regional climatic differences, urban heat island effects on local temperatures resulting from land changes and land-use, and soil responses to higher temperatures at a regional level (IPCC Fourth Assessment: Christensen et al. 2007; IPCC Fifth Assessment: IPCC 2013).

While most global climate models (GCMs) project consistently increasing temperatures due to climate change, models are inconsistent in their predictions for rainfall changes. There have been some indications that rainfall patterns in South Africa are possibly starting to change, particularly in the western areas of the country (van Wageningen & du Plessis 2007), which is also supported by recent modelling studies using downscaled climate change data for southern Africa (Lumsden et al. 2009; Wolski et al. 2012). Hewitson & Crane (2006) used three empirically downscaled GCMs and mean atmospheric fields to generate future rainfall predictions over South Africa. They showed some convergence in precipitation predictions (although the magnitude of the prediction varied), with results indicating increased summer (December–February) rainfall in the central and eastern part of the country and decreased rainfall in the western section. Consequently, and of concern for many countries in southern Africa, the changes to precipitation and evaporation regimes due to climate change may affect water and food security.

Background to the study system

The Amatole system, consisting of the Buffalo, Nahoon and Kubusi rivers, is situated on the east coast of South Africa. Figure 1 shows the main catchments, location of the rivers, dams and major towns in the Amatole system. The system consists of seven reservoirs: four located on the Buffalo River (Maden, Rooikrans, Laing and Bridle Drift dams), two located on the Kubusi River (Gubu and Wriggleswade dams) and one located on the Nahoon River (Nahoon dam). A Department of Water Affairs (DWA) report concluded that the 2003 population in the Buffalo River catchment of 570,000 had less than 500 m3 per annum water available per individual (DWAF 2004a).
Figure 1

Map of the Amatole catchments in the Eastern Cape province (see location on the South African coast on the inset map).

Figure 1

Map of the Amatole catchments in the Eastern Cape province (see location on the South African coast on the inset map).

METHODS

The Water Evaluation and Planning (WEAP) model, developed by the Stockholm Environmental Institute, is a ‘water-accounting’ model and an Integrated Water Resource Management tool that can be used to investigate different scenarios of water use and resource development (Yates et al. 2005). WEAP simulates water supply and demand, river flow, storage, pollution generation, treatment and discharge, while considering competing users of water, including environmental flows. It can therefore facilitate trade-off analysis between users by assigning priority levels which define the order for supplying the demands of users, including the environment (Sieber & Purkey 2007). The WEAP model has been successfully applied in South Africa (Lévite et al. 2003; Arranz & McCartney 2007) and internationally (Assaf & Saadeh 2008; Bharati et al. 2009; Yates et al. 2009, 2013; Mehta et al. 2011) to model demand scenarios, environmental flows, hydropower production and inter-basin transfers.

Climate inputs and natural hydrology

Historical rainfall and evaporation/temperature data for input into the model were obtained for individual quaternaries (a fourth order hierarchical catchment definition that is the basic unit for water resources management) from the South African WR2005 database (Middleton & Bailey 2008). The WEAP model utilises a simplified model (Food and Agriculture Organization (FAO) rainfall–runoff model) to generate time series of monthly streamflows for the tributaries based on data on rainfall, catchment area, effective precipitation and monthly evapotranspiration. The natural hydrology simulations of the model were difficult to validate against observed flow data, as the majority of the gauges are located downstream of water use or below effluent return flow effects which have not been stationary over the period of available data.

Current water demands

The primary water demand users in the system are human settlements, industry, agricultural areas and alien vegetation. For simplification, the Amatole system was divided into three demand areas – upper, middle and lower Amatole. Current water requirements for the year 2005 (DWAF 2008: Table 4.13) were used for generating the current/reference scenario for comparison with future scenarios. The current water requirements were entered into the WEAP model as a stationary demand over the modelled years 1921–2005, to assess the effects of the natural variation in the hydrological regime on the water use and availability for the current level of water use. No attempts were made to represent historical variations in water use over the reference period of 1921–2005.

Water losses

Based on the DWAF (2008) estimates of losses in the water supply system (including leakage losses at the water treatment works and WWTW, reticulation losses, demand site consumption and conveyance losses), the total losses entered into the WEAP model were 44.6% in the upper Amatole, 40.9% in the middle Amatole and 44.0% in the lower Amatole. Water lost to invasive aliens was estimated from DWAF (2004b) to be approximately 3 × 106m3y−1 (equivalent to ∼5% of the current demand in the system).

Streamflow calibration

Daily streamflow data (from October 1979 to September 2005, depending on availability) for 11 gauging stations in the Amatole catchment (nine gauges on the Buffalo River and two on the Kubusi River (http://www.dwa.gov.za/Hydrology) were utilised for system calibration, and any gauges below dams were ignored in this process. WEAP results for the simulation of hydrology from rainfall and water use data were matched against recorded data by stream gauges to assess how accurately the model was simulating present day water quantity. Missing periods of data were patched using available data from nearby gauging stations using the methodology described in Hughes & Smakhtin (1996).

Future climate data

From a modelling point of view, in order to generate specific climate change projections at a regional scale, the global projections for southern Africa need to be downscaled for the region. Assessment of the impacts based on these regionalised projections is critical in the context of adaptation to climate change since adaptive planning requires information on expected changes, including an estimation of the uncertainty in these changes specific to the area of interest. The climate data consisted of statistically downscaled (methods discussed in Hewitson & Crane (1996, 2006)) daily rainfall plus maximum and minimum temperatures from nine GCMs (Table 1) for the SRES A2 emission scenario (IPCC 2000). The outputs of the regionally downscaled GCMs include baseline, near future and far future predictions. Since the baseline simulations for the nine GCMs are very different from each other and from the WR2005 rainfall data that have been compiled from observed rainfall station data (Middleton & Bailey 2008), the future projections require bias correction before being used within a hydrological model calibrated with historical climate data (Hughes et al. 2014). Nine downscaled and bias corrected climate projections were used in the WEAP model to incorporate uncertainty in climate projections for the near future (2046–2065). Note that the results for the near future scenarios described below have been compared with the current/reference scenario (with stationary demands for the year 2005) that was simulated by the model, and not the historical observed data which are non-stationary.

Table 1

GCMs used in the study

GCM abbreviation Source of GCM 
CCCMA Canadian Centre for Climate Modeling and Analysis 
CNRM France Centre National de Recherches Meteorologiques 
CSIRO Australian CSIRO Atmospheric Research 
GFDL USA NOAA Geophysical Fluid Dynamics Lab 
GISS USA Goddard Institute for Space Studies 
IPSL France Institut Pierre Simon Laplace 
MIUB Germany Meteorological Institute of the University of Bonn 
MPI Max-Planck Institute For Meteorology 
MRI Japan Meteorological Research Institute 
GCM abbreviation Source of GCM 
CCCMA Canadian Centre for Climate Modeling and Analysis 
CNRM France Centre National de Recherches Meteorologiques 
CSIRO Australian CSIRO Atmospheric Research 
GFDL USA NOAA Geophysical Fluid Dynamics Lab 
GISS USA Goddard Institute for Space Studies 
IPSL France Institut Pierre Simon Laplace 
MIUB Germany Meteorological Institute of the University of Bonn 
MPI Max-Planck Institute For Meteorology 
MRI Japan Meteorological Research Institute 

Future development data

The near future water requirements were estimated from the available data on expected increases in water demands for the years 2005–2030, which were then extrapolated to the middle of the near future period (i.e., water requirements for the year 2056; see Table 2 and Figure 2). The water requirements defined in DWAF (2008: Table 4.13), are presented as three scenarios: lower, intermediate and upper development, which represent the uncertainty in future development. These scenarios include changes from the current water requirements for the following users: population (increases for the upper Amatole, decreases for the middle Amatole and increases or decreases for the lower Amatole depending on the specific scenario), industry (increases in the lower Amatole) and agriculture (ranging from no agricultural demand to increases in the middle and lower Amatole areas). In total, under the intermediate development scenario, there is an expected increase of 33% in the water demand, and a two-fold increase in the current situation demand under the upper development scenario (Table 2). The population water requirements are the major contributor to the uncertainty in the future water requirements for the area (Figure 2). These water requirements were entered into the WEAP model as a stationary demand, similar to the current requirements, to simulate the system dynamics and assess the reliability of the water supply in the near future.
Table 2

Total water requirements (106m3y−1) for the three sectors (not including alien vegetation demands) under the current (2005) and the future development scenarios (lower, intermediate and upper) for the years 2046–2065

  Current Lower development Intermediate development Upper development 
Population 45.55 44.41 60.49 90.01 
Industry 14.32 16.24 20.57 25.55 
Irrigation 4.40 0.00 4.40 12.74 
Total 64.27 60.65 85.46 128.30 
% increase – −5.6% 33.0% 99.6% 
  Current Lower development Intermediate development Upper development 
Population 45.55 44.41 60.49 90.01 
Industry 14.32 16.24 20.57 25.55 
Irrigation 4.40 0.00 4.40 12.74 
Total 64.27 60.65 85.46 128.30 
% increase – −5.6% 33.0% 99.6% 
Figure 2

Projected uncertainty in the water requirements originating from the population, industry and irrigation sectors for the Amatole system in the near future (2046–2065). The values have been extrapolated up to the year 2065 from the values available for the years 2005 and 2030.

Figure 2

Projected uncertainty in the water requirements originating from the population, industry and irrigation sectors for the Amatole system in the near future (2046–2065). The values have been extrapolated up to the year 2065 from the values available for the years 2005 and 2030.

MODEL CALIBRATION RESULTS

The streamflow simulated by the WEAP model (e.g., Figures 3 and 4) generally matched the pattern of historical monthly streamflow variation in the system, despite the fact that performance and goodness of fit measures were generally very poor. The Nash–Sutcliffe efficiency (E) and efficiency with natural logarithmic values (ln E, which reduces the influence of high flows on efficiency calculation) were calculated for gauges in the upper reaches and tributaries of the Buffalo River that are relatively unimpacted by abstractions, to assess the errors in the model simulation of streamflow (Nash & Sutcliffe 1970; Krause et al. 2005). The E values ranged between −0.59 and 0.44 (n = 6), while the ln E varied between −0.68 and 0.40. The reasons for the difference between simulated and recorded flows could be uncertainties arising from various sources. These include uncertainty in the input climate data, model structure and calibrated parameters, uncertainty in the observed flow data and in the water user demands. An additional reason for the difference is the variation in the actual water user demands, which have changed over time, whereas stationary demands were entered into the WEAP model. Data for the recent past indicate that the historical demands for the Amatole system have increased by 65% from 1980 to 1996 (from 29.34 × 106 to 41.94 × 106m3y−1; BCM 2002: Table 12).
Figure 3

Simulated flow output of WEAP model relative to gauge data at R2H009 on the Ngqokweni River, a tributary in the upper sections of the Buffalo River, for the years 1980–2005: (a) monthly flows (Nash–Sutcliffe efficiency E 0.43; ln E of −0.45) and (b) flow duration curve.

Figure 3

Simulated flow output of WEAP model relative to gauge data at R2H009 on the Ngqokweni River, a tributary in the upper sections of the Buffalo River, for the years 1980–2005: (a) monthly flows (Nash–Sutcliffe efficiency E 0.43; ln E of −0.45) and (b) flow duration curve.

Figure 4

Simulated flow output of WEAP model relative to gauge data at R2H005 in the middle section of the Buffalo River for the years 1995–2005: (a) monthly flows (Nash–Sutcliffe efficiency E −0.45; ln E of 0.04) and (b) flow duration curves.

Figure 4

Simulated flow output of WEAP model relative to gauge data at R2H005 in the middle section of the Buffalo River for the years 1995–2005: (a) monthly flows (Nash–Sutcliffe efficiency E −0.45; ln E of 0.04) and (b) flow duration curves.

To compare simulated reservoir storage with the actual storage, water releases from the dams needed to be included in the actual storage values. Thus, the WEAP model simulations of reservoir storage + spills below the dam was compared with the actual data on storage + spills + releases that were obtained from the DWA (e.g., Figure 5 for Laing dam), and the patterns of change in simulated values were similar to the actual variation.
Figure 5

Monthly reservoir storage and spills below the dam as simulated by WEAP compared to actual values of storage + spills + releases for the Laing Dam on the middle Buffalo River (Nash–Sutcliffe efficiency E 0.19; ln E of 0.28).

Figure 5

Monthly reservoir storage and spills below the dam as simulated by WEAP compared to actual values of storage + spills + releases for the Laing Dam on the middle Buffalo River (Nash–Sutcliffe efficiency E 0.19; ln E of 0.28).

RESULTS FOR WATER USE AND AVAILABILITY

Current scenario water requirements and availability

The WEAP results indicate that the water user requirements can be met 100% of the time for all users under the current scenario, i.e., there is no deficit in meeting the demands at present, which matches the experience of the local water services provider (personal communication: Amatole Water Board, the local water services provider; DWA 2013).

Future development-only scenarios

The future development-only scenarios were run using the historical rainfall to assess the effects on water availability with no climate change. According to the model results, under the intermediate development scenario, deficits of 8–16% are expected for population and industrial water users. The median water deficits under the upper development scenario for population and industry users are higher at 27–44%, with larger water deficits in the lower Amatole section of the system compared to the upper and middle sections. There are no water deficits simulated under the lower development scenario, since the water supply requirements for the population, industry and irrigation sectors are projected to be lower (60.65 × 106m3y−1) than the current situation (64.27 × 106m3y−1; see Table 2).

Future climate-change-only scenarios

For the near future climate-change-only scenarios, the water demands were set at current levels and the monthly streamflows were simulated using the nine downscaled near future climate scenarios. An increase in streamflow relative to today was generally predicted, since the lower limit of the uncertainty band for flow (obtained from the minimum and maximum values for the nine climate change scenarios) overlapped with the current simulation. This is because an increase in rainfall is predicted for these catchments by most of the climate scenarios, which is not fully offset by increases in evapotranspiration losses. Thus, the water user requirements are expected to be met 100% of the time under all nine climate change scenarios. The seasonality of monthly average streamflows, however, is highly uncertain for the near future climate scenarios (examples in Figure 6). The simulations suggest a possible reduction in the low monthly flow volumes for the dry months of April–June. The median values for the near future uncertainty band in Figure 6 indicate similar seasonality to the present day simulation; however, the upper extreme of the uncertainty band shows a higher magnitude difference between the high flow months (October–November) and the low flow periods (May–June) compared to the current simulation.
Figure 6

Average simulated monthly flow at a reach (a) on the lower section of the Buffalo River and (b) at the Nahoon River estuary under current climate conditions (1921–2005) and under the near future climate scenarios (2046–2065) shown as a band of uncertainty (minimum and maximum values).

Figure 6

Average simulated monthly flow at a reach (a) on the lower section of the Buffalo River and (b) at the Nahoon River estuary under current climate conditions (1921–2005) and under the near future climate scenarios (2046–2065) shown as a band of uncertainty (minimum and maximum values).

Future climate-and-development change scenarios

To estimate the total uncertainty for the near future development scenarios under near future climate conditions, the results of 27 model runs (nine climate change scenarios, each in combination with three different socio-economic development scenarios) were assessed. The predicted streamflows showed greater uncertainty at low flows, particularly for the lower river reaches, in comparison to the results for climate change scenarios in isolation (e.g., Figure 7).
Figure 7

Flow duration curve at a reach on the lower section of Buffalo River under current climate conditions (1921–2005), under the near future climate-change-only scenarios and under the near future climate-and-development scenarios (minimum and maximum values of uncertainty are shown).

Figure 7

Flow duration curve at a reach on the lower section of Buffalo River under current climate conditions (1921–2005), under the near future climate-change-only scenarios and under the near future climate-and-development scenarios (minimum and maximum values of uncertainty are shown).

The storage volume of reservoirs was associated with larger uncertainty under the upper development scenario relative to the intermediate scenario under the nine near future climate change scenarios (cf. Figure 8(a) and 8(b)). The storage capacity under these scenarios overlaps with the current simulated storage due to an increase in the balance between rainfall and evaporative losses under the near future climate scenarios (resulting in increased water volume stored in the reservoirs) and increased water demand under the intermediate and upper development scenarios (leading to a reduction in stored amounts in the reservoirs).
Figure 8

Average simulated monthly reservoir storage for the Bridle Drift dam on the Buffalo River under the (a) intermediate and (b) upper development scenarios under nine near future climate conditions. The current simulated reservoir storage (for the years 1921–2005; black line) and the net capacity (dashed line) are shown for reference.

Figure 8

Average simulated monthly reservoir storage for the Bridle Drift dam on the Buffalo River under the (a) intermediate and (b) upper development scenarios under nine near future climate conditions. The current simulated reservoir storage (for the years 1921–2005; black line) and the net capacity (dashed line) are shown for reference.

The WEAP model predicted deficits in meeting the population and industrial demands, with median water deficits of 21.45 × 106m3y−1 and 87.94 × 106m3y−1 (equivalent to 25% and 69% of the total water requirements) for the intermediate and upper development scenarios, respectively (Table 3). These deficit amounts do not consider upgrades in water infrastructure or inter-basin water transfers that are already planned from the Wriggleswade dam on the Kubusi River to the Amatole system (see Figure 1). These infrastructure changes and the effect on the water deficits are discussed in the next section.

Table 3

Median water deficits (in 106m3y−1) for the Amatole system for the nine climate change scenarios for the years 2046–2065 using the water requirements for the intermediate and upper development scenarios without and with (in brackets) infrastructure upgrades and inter-basin water transfers

  Intermediate development Upper development 
Upper Amatole   
Population 1.57 (0.00) 6.94 (4.15) 
Industry 0.00 (0.00) 0.00 (0.00) 
Middle Amatole     
Population 0.86 (0.00) 4.01 (2.39) 
Industry 0.18 (0.00) 0.63 (0.37) 
Lower Amatole     
Population 13.60 (0.00) 57.95 (32.84) 
Industry 5.24 (0.00) 18.41 (10.44) 
  Intermediate development Upper development 
Upper Amatole   
Population 1.57 (0.00) 6.94 (4.15) 
Industry 0.00 (0.00) 0.00 (0.00) 
Middle Amatole     
Population 0.86 (0.00) 4.01 (2.39) 
Industry 0.18 (0.00) 0.63 (0.37) 
Lower Amatole     
Population 13.60 (0.00) 57.95 (32.84) 
Industry 5.24 (0.00) 18.41 (10.44) 

Future climate-and-development scenarios with upgrades to infrastructure and water transfers

The local water authorities in the Amatole system are planning infrastructural upgrades, which include increasing the production capacity of water treatment works, upgrading a waste water treatment works (WWTWs) and decommissioning of some of the smaller WWTWs plus water transfers of 18 × 106m3y−1 (equivalent to 30% of the present water demand) from the Wrigglewade dam on the Kubusi River to supplement the Bridle Drift and Laing dams on the Buffalo River (DWAF 2004b).

The WEAP model was rerun for the intermediate and the upper development scenarios under the nine near future climate change models with inclusion of future infrastructure upgrades. The results indicated that the infrastructure upgrades will provide sufficient water to meet the demands under the intermediate development scenario (which is 33% higher in water demand relative to the present demand). Although the water deficits will be greatly reduced for the upper development scenario, water deficits of 16–25% are still expected for the population and industry users. The median water deficit amount for the system under the upper development scenario with water transfers and upgrades is expected to be 50.18 × 106m3y−1 under the near future climate scenarios (see Table 3).

DISCUSSION

Climate change has been the focus of much debate and future planning (such as the IPCC Fourth Assessment: Parry et al. 2007; Mehta et al. 2011; IPCC 2012). The present study aimed to emphasise the need to consider both the climate and non-climate related changes, along with the associated uncertainty, to facilitate adaptive planning and sustainability of water resources. The analysis of Pappenberger & Beven (2006) suggesting seven reasons to use uncertainty analysis has been influential in stressing the need for estimating uncertainties as part of hydrological modelling studies.

The WEAP model with the single FAO rainfall/runoff method did not perform very well in terms of simulating individual months of flow volume. However, the seasonal distribution and flow duration curve characteristics were adequately reproduced by the model, and these are the most important issues that affect the yield of the system. In the present study, the authors found the WEAP model useful for comparing the water availability and use under scenarios of climate-change-only with scenarios that combine climate and development changes in the future. In the selected system, climate-and-development scenarios in combination resulted in greater uncertainty, particularly during low flow periods, relative to climate scenarios alone. The increased uncertainty in the low flows would not only impact on meeting the environmental flow requirements (discussed below), but also has implications for the frequency of water transfers needed and stricter management of reservoir storage for both water quantity and quality. Recent studies outside South Africa (e.g., Rochdane et al. 2012; Chavez-Jimenez et al. 2013; Paton et al. 2013) have analysed the effects of future climate and development on adaptive planning for meeting future water demands. The results of such studies are invaluable for local authorities to strategically and adaptively plan for the future.

Although environmental flows have not been implemented so far in the Amatole system, they are expected to be put in place in the future. Environmental Water Requirement (EWR) assessments have been conducted for various locations along the Buffalo, Kubusi and Nahoon rivers. The Amatole Reconciliation Strategy's review of the EWR shows that there are various factors preventing the EWR from being implemented, such as reduction in the system yield by up to 25%, or that high flow components of the EWR are not expected to be met due to discharge limitations of the dam outlets (DWAF 2008). Currently, finalisations of the river classifications (which define the balance between protection level for the resource and water for human use) and EWRs are still in progress, and their implementation would require the development of reservoir operating rules to facilitate the ecological flows. Thus, while acknowledging that the impacts of climate change are of concern for future water resources, longer periods of low flows in the streams, due to greater withdrawal of water to meet higher development demands, and worsened water quality are of equal, if not greater, concern in meeting the environmental flows and protecting the ecosystem services. This stresses the need for future planning incorporating both climate and development impacts together instead of in isolation.

The water authorities in the Amatole area have planned for increases in the available water supply in future through water transfers and increased capacity for treating water. However, the model suggests that despite these planned increases, there will still be a large deficit (of up to 50.18 × 106m3y−1) if the upper development scenario, which requires an increase of 100% in water supply over the current amounts, becomes reality. To meet this higher demand, measures such as more frequent inter-basin water transfers, water conservation practices and reduction in losses in the system would need to be targeted. For water services institutions in South Africa, the importance of water demand management (WDM) has been recognised as being critical for the future (DEA 2011, p. 90). For the Amatole system, the Buffalo City Municipality is expected to put water conservation and demand management measures in place that can save up to 6.2 × 106m3y−1 over the next 5 years, before any additional water is brought into the system (UWP 2012a, b). Through continual monitoring of the future development changes, in particular changes in the magnitude of population and industrial demands in the system, along with monitoring of climate trends (to reduce climate change uncertainty), adaptive planning and appropriate measures can be implemented to avoid water shortages in future.

Some important considerations when reviewing the results of this study are discussed below. The first, regarding not including the environmental flow requirements in the model, has been mentioned above and thus, it is noted that the results for future water deficits are conservative. The current study also estimated development demands for the near future by extrapolating projected demands for the year 2030; this assumption could be incorrect and demands might instead level off. Although the current water use by invasive aliens is not large relative to the current water requirements, this could either increase or decrease in the future depending on the success of the present management programme (Working for Water, WfW; https://www.environment.gov.za/projectsprogrammes/wfw) of the Department of Environmental Affairs. The uncertainty in all these variables emphasises the need for continual monitoring of the catchment (in terms of stream discharge, river and reservoir water quality, and reservoir storage) and of development changes, to reduce the uncertainty in the predicted scenarios for climate change and socio-economic development so that the Amatole system can be managed effectively, efficiently and sustainably. To move forward with the outputs of the present study, a decision-support system is being developed which will be based on thresholds of potential concern for water quantity and quality management for the Amatole system. This is based on a framework for incorporating uncertainty in modelling and assessment tools in South Africa that has been developed previously (Hughes et al. 2011).

Finally, the present study has ignored some sources of uncertainty, such as alternative emission scenarios, modelling uncertainty, and the use of different downscaling techniques (e.g., Paton et al. 2013) or social contexts, as they were not part of the aim of the present study. However, ongoing research at the Institute for Water Research is investigating ways of constraining uncertainty in hydrological modelling outputs using the Pitman model (Hughes 2013). The present study formed part of a larger project investigating uncertainties in hydrological model outputs, and a study under the project addressed the issues of model uncertainty using the Pitman hydrological model. The results suggested that the uncertainty in near future water resources availability arising from climate models (the same nine GCMs as the present study) is substantially higher than that from historical model simulations incorporating parameter uncertainty (Hughes et al. 2013, 2014).

DECISION-MAKING RISK AND UNCERTAINTIES

The results presented in this paper are based on the outputs of a South African Water Research Commission funded project conducted by the research team at the Institute for Water Research (Hughes et al. 2013; http://www.ru.ac.za/static/institutes/iwr/climate/). The project was overseen by a steering committee consisting of representatives of the Department of Water and Sanitation (previously the DWA), the local bulk water supplier involved in planning for future water supply with cognisance of the results of the present study, South African academics/researchers working in the field of climate change, and a consulting agency working with the municipality on water supply issues. The following is a summary of the discussions with the steering committee on the application of the results, and it is supplemented by a review of South African literature and documents which provide insight into the South African context for developing adaptation strategies to future changes.

It is widely accepted that water resources development and operational decisions need to be made in an environment of uncertainty associated with historical and future conditions. In the context of the present study, decision-making risk for future water availability can be defined as the possibility of failure to achieve objectives through some management plan due to uncertainty in the future conditions, such as failure in the expected yield of a water resources scheme, which is linked to the uncertainty in future predictions (Hughes et al. 2011). Note that uncertainty is not the same as a lack of confidence in a single prediction; it is a way of quantifying all likely (or probable) predictions so that critical knowledge gaps can be identified (and addressed in the future), and so that adaptive management approaches can be adopted (Tadross et al. 2011). Realistic ranges of uncertainty can be useful for identifying gaps in knowledge and assist with adaptive management of the water resources (Hughes et al. 2011). The explicit inclusion of uncertainty in decision-making is, however, a recent development (Lempert & Collins 2007; Beven & Alcock 2012). To manage water resources sustainably while considering the uncertainty surrounding climate change, the adaptive capacity of the system (i.e., ‘the ability or potential of a system to respond successfully to climate variability and change, and includes adjustments in both behaviour and in resources and technologies’; Adger et al. 2007, p. 727) is of major concern (Cutter et al. 2012). Beven (2011) proposed that in the face of uncertain change in the future, society needs to move ahead with developing an adaptation strategy based on a risk-averse or risk-accepting attitude, as doing nothing (which some societies choose) is ‘to be risk accepting (or even irresponsible)’. In South Africa, the precautionary principle has been included as part of the uncertainty in our knowledge about climate change and a ‘risk-averse and cautious approach which takes into account the limits of current knowledge about the consequences of decisions and actions’ is being promoted (RSA 2010, p. 6).

Developing responses (or adaptation strategies) to future climate and development changes involves three key issues: (1) the institutional framework in which responses can be developed; (2) the instruments for developing the responses; and (3) the information that is available to inform the responses. The first issue of the institutional framework includes the national strategies and legislative framework within which the water services delivery institutions operate in South Africa, as well as the local framework and the specific functions of the individual institutions. In some areas of South Africa, the functions of water services delivery are the sole responsibility of a water services authority, such as a municipality, while in other areas (such as the Amatole system), the functions and responsibilities are shared between the water services authority and a water board that supplies bulk water to the authority. The institutional framework therefore includes the management relationships between these entities.

There are various local level instruments for developing responses or adaptation strategies to changes projected for the future in South Africa, including the Water Services Development Plans, Provincial Spatial Development Plans (ECPSDP 2010) and Integrated Development Plans (BCM 2012) developed by the local government agencies in collaboration with stakeholders. In addition, the National Water Resources Strategy-2 (DWA 2013), the National Climate Change Response Strategy (RSA 2010; http://www.climateresponse.co.za/home) and Water Resources Reconciliation strategies are developed by the DWA with national, regional and local inputs. One must not forget that any planning instrument must be informed by the most up-to-date and reliable information that is generated from the analysis of the available historical observed data, and the application of models or estimation methods to fill in gaps in the observed data. All of the information generated by prediction models for developing responses to future situations is necessarily uncertain (we cannot predict the future with certainty). The dominant issue is the need to recognise the uncertainties in future predictions and ensure that they are part of any planning instruments such that the links can be made between uncertain information and decision-making risk (Lempert & Collins 2007; Beven 2011; Beven & Alcock 2012). The IPCC stresses that ‘it is important that uncertainty over future climate change risks not become a barrier to climate change risk reduction actions’ and thus, ‘in cases where climate change uncertainties remain high, countries may choose to increase or build on their capacity to cope with uncertainty, rather than risk maladaptation from use of ambiguous impact studies or no action’ (IPCC 2012, p. 351).

With this in mind, detailed results from this study and various adaptation measures, some that fall under good governance, have been presented to the water service institutions in the Amatole system, to contextualise the expected future uncertainty in water availability and to use and to promote adaptive planning (Hughes et al. 2013). These adaptation measures include integrated land and water management, building resilience, maintenance (an essential part of good governance), monitoring (essential to reduce the uncertainty in the predictions), water literacy and training, and dialogue with agencies at local, regional, national and international level (essential to learn from and to share experiences and knowledge) (details in Hughes et al. 2013). One of the primary recommendations for future water resources management in any system, which was echoed by various stakeholders involved in the project (DWA, Amatola Water Board, Water Research Commission and various scientists), is the importance of integrated management and monitoring across various water management agencies in the catchment to reduce the uncertainty in future predictions. Collaboration is critical to moving forward to meet the three principles of the South African National Water Act (no. 36 of 1998), those of equity, sustainability and efficiency.

ACKNOWLEDGEMENTS

This research was funded by the South African Water Research Commission (project no. K5/2018) and was conducted in collaboration with the Amatola Water Board and other steering committee members, who are thanked for their input. The reservoir data were kindly provided by Mr Cobus Ferreira of the Department of Water Affairs (DWA). Downscaled climate data were provided by the Climate Systems Analysis Group (CSAG) based at the University of Cape Town, South Africa.

REFERENCES

REFERENCES
Adger
W. N.
Agrawala
S.
Mirza
M. M. Q.
Conde
C.
O'Brien
K.
Pulhin
J.
Pulwarty
R.
Smit
B.
Takahashi
K.
2007
Assessment of adaptation practices, options, constraints and capacity
. In:
Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
(
Parry
M. L.
Canziani
O. F.
Palutikof
J. P.
van der Linden
P. J.
Hanson
C. E.
, eds).
Cambridge University Press
,
Cambridge
,
UK
, pp.
717
743
.
Arranz
R.
McCartney
M.
2007
Application of the Water Evaluation and Planning (WEAP) Model to Assess Future Water Demands and Resources in the Olifants Catchment, South Africa
.
IWMI Working Paper 116
,
report by the International Water Management Institute
,
Colombo
,
Sri Lanka
.
BCM
2002
Buffalo City Municipality Water Services Development Plan
.
Buffalo City Municipality
,
South Africa
.
BCM
2012
Buffalo City Metropolitan Municipality Integrated Development Plan 2012/13 Review.
Buffalo City Municipality
,
South Africa
. ).
Bharati
L.
Smakhtin
V. U.
Anand
B. K.
2009
Modeling water supply and demand scenarios: the Godavari–Krishna inter-basin transfer, India
.
Water Policy
11
(
Suppl. 1
),
140
153
.
Chavez-Jimenez
A.
Lama
B.
Garrote
L.
Martin-Carrasco
F.
Sordo-Ward
A.
Mediero
L.
2013
Characterisation of the sensitivity of water resources systems to climate change
.
Water Resour. Manage.
27
,
4237
4258
.
Christensen
J. H.
Hewitson
B.
Busuioc
A.
Chen
A.
Gao
X.
Held
I.
Jones
R.
Kolli
R. K.
Kwon
W.-T.
Laprise
R.
Magaña Rueda
V.
Mearns
L.
Menéndez
C. G.
Räisänen
J.
Rinke
A.
Sarr
A.
Whetton
P.
2007
Regional climate projections
. In:
Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
(
Solomon
S.
Qin
D.
Manning
M.
Chen
Z.
Marquis
M.
Averyt
K. B.
Tignor
M.
Miller
H. L.
, eds).
Cambridge University Press
,
Cambridge, NY
,
USA
.
Cutter
S.
Osman-Elasha
B.
Campbell
J.
Cheong
S.-M.
McCormick
S.
Pulwarty
R.
Supratid
S.
Ziervogel
G.
2012
Managing the risks from climate extremes at the local level
. In:
Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation
(
Field
C. B.
Barros
V.
Stocker
T. F.
Qin
D.
Dokken
D. J.
Ebi
K. L.
Mastrandrea
M. D.
Mach
K. J.
Plattner
G.-K.
Allen
S. K.
Tignor
M.
Midgley
P. M.
, eds).
A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change (IPCC)
.
Cambridge University Press
,
Cambridge, NY
,
USA
, pp.
291
338
.
DEA
2011
South Africa's Second National Communication under the United Nations Framework Convention on Climate Change
.
Department of Environmental Affairs
,
Pretoria
,
Republic of South Africa
. ).
DWA
2013
National Water Resource Strategy
.
2nd edn.
Department of Water Affairs and Forestry
,
Pretoria
,
South Africa
.
DWAF
2004a
National Water Resource Strategy
.
Department of Water Affairs and Forestry
,
Pretoria
,
South Africa
.
DWAF
2004b
Mzimvubu to Keiskamma Water Management Area: Amatole – Kei Internal Strategic Perspective
.
DWAF Report No. P WMA 12/000/00/0404
.
Prepared by FST Consulting Engineers (Pty) Ltd in association with Tlou & Matji and Umvoto Africa, on behalf of the Department of Water Affairs and Forestry Directorate, National Water Resource Planning
.
DWAF
2008
Development of a Reconciliation Strategy for the Amatole Bulk Water Supply System
.
Final Report
.
Two Volumes
.
Prepared by SSI Engineers and Environmental Consultants on behalf of the Department of Water Affairs and Forestry Director, National Water Resource
,
Pretoria
,
South Africa
.
ECPSDP
2010
Eastern Cape Provincial Spatial Development Plan
.
Tshani Consulting C.C.
,
East London
,
South Africa
.
http://www.psdp.ecprov.gov.za (accessed February 2011
).
Hewitson
B.
Crane
R.
1996
Climate downscaling: techniques and application
.
Climate Res.
7
,
85
95
.
Hughes
D. A.
Kapangaziwiri
E.
Mallroy
S. J. L.
Wagener
T.
Smithers
J.
2011
Incorporating Uncertainty in Water Resources Simulation and Assessment Tools in South Africa
.
WRC Report No. 1838/1/11
.
Water Research Commission
,
Pretoria
,
South Africa
.
Hughes
D. A.
Mantel
S. K.
Slaughter
A.
2013
Informing the Responses of Water Service Delivery Institutions to Climate and Development Changes: A Case Study in the Amatole Region, Eastern Cape
.
Water Research Commission Report 1843/1/11
,
Pretoria
,
South Africa
.
Intergovernmental Panel on Climate Change
2000
Emission Scenarios: A Special Report of IPCC Working Group III
(
Nakićenović
N.
Swart
R.
, eds).
Cambridge University Press
,
Cambridge
,
UK
, p.
608
. ).
Intergovernmental Panel on Climate Change
2012
Managing the risks of extreme events and disasters to advance climate change adaptation
. In:
A Special report of Working Groups I and II of the Intergovernmental Panel of Climate Change
(
Field
C. B.
Barros
V.
Stocker
T. F.
Dahe
Q.
Dokken
D. J.
Ebi
K. L.
Mastrandrea
M. D.
Mach
K. J.
Plattner
G.-K.
Allen
S. K.
Tignor
M.
Midgley
P. M.
, eds).
Cambridge University Press
,
Cambridge, NY
,
USA
,
582
pp.
Intergovernmental Panel on Climate Change
2013
Climate change 2013: the physical science basis
. In:
Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
(
Stocker
T. F.
Quin
D.
Plattner
G.-K.
Tignor
M.
Allen
S. K.
Boschung
J.
Nauels
A.
Xia
Y.
Bex
V.
Midgley
P. M.
, eds).
Cambridge University Press
,
Cambridge, NY
,
USA
,
1535
pp.
Jackson
R. B.
Carpenter
S. R.
Dahm
C. N.
McKnight
D. M.
Naiman
R. J.
Postel
S. L.
Running
S. W.
2001
Water in a changing world
.
Ecol. Appl.
11
,
1027
1045
.
Mehta
V. K.
Rheinheimer
D. E.
Yates
D.
Purkey
D. R.
Viers
J. H.
Young
C. A.
Mount
J. F.
2011
Potential impacts on hydrology and hydropower production under climate warming of the Sierra Nevada
.
J. Water Climate Change
2
,
29
43
.
Middleton
B. J.
Bailey
A. K.
2008
Water Resources of South Africa, 2005 Study (WR2005)
.
WRC Report No. TT381/08
.
Water Research Commission
,
Pretoria
,
South Africa
.
Pappenberger
F.
Beven
K. J.
2006
Ignorance is bliss: or seven reasons not to use uncertainty analysis
.
Water Resour. Res.
42
(
5
),
W05302
.
Parry
M. L.
Canziani
O. F.
Palutikof
J. P.
2007
Technical summary
. In:
Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
(
Parry
M. L.
Canziani
O. F.
Palutikof
J. P.
van der Linden
P. J.
Hanson
C. E.
, eds).
Cambridge University Press
,
Cambridge
,
UK
, pp.
23
78
.
RSA
2010
National Climate Change Response Green Paper
.
Department of Environmental Affairs
,
Pretoria
,
the Government of the Republic of South Africa
, p.
38
.
Sieber
J.
Purkey
D.
2007
Water Evaluation and Planning System. User Guide for WEAP21
.
Stockholm Environment Institute
,
Massachusetts
,
USA
.
http://www.weap21.org/ (accessed 15 October 2013).
Tadross
M.
Davis
C.
Engelbrecht
F.
Joubert
A.
van Garderen
E. A.
2011
Regional scenarios of future climate change over southern Africa
. In:
Climate Risk and Vulnerability: A Handbook for Southern Africa
(
Davis
C. L.
, ed.).
Council for Scientific and Industrial Research
,
Pretoria
,
South Africa
,
Chapter 3
, pp.
28
50
.
UWP
2012a
Amatole System Reconciliation Strategy: System Integration/Optimisation Workshop
.
Workshop held on 21 June 2012, East London Golf Course
.
UWP Consulting
,
East London
,
South Africa
.
UWP
2012b
Amatole System Reconciliation Strategy: Screening of Surface Water Supply Augmentation Options Workshop
.
Workshop held on 26 March 2012, East London Golf Course
.
UWP Consulting
,
East London
,
South Africa
.
van Wageningen
A.
du Plessis
J. A.
2007
Are rainfall intensities changing, could climate change be blamed and what could be the impact for hydrologists?
Water SA
33
(
4
),
571
574
.
Walmsley
R. D.
Walmsley
J. J.
Silberbauer
M.
1999
Sustainability of Freshwater Systems and Resources
. In:
National State of the Environment Report – South Africa
.
Department of Environmental Affairs and Tourism
,
Republic of South Africa
. ).
Willows
R.
Connell
R.
(eds)
2003
Climate Adaptation: Risk, Uncertainty and Decision Making
.
United Kingdom Climate Impacts Programme (UKCIP) Technical Report
.
UKCIP
,
Oxford
,
UK
. ).
Yates
D.
Purkey
D.
Sieber
J.
Huber-Lee
A.
Galbraith
H.
2005
WEAP21 – A demand, priority, and preference-driven water planning model
.
Water Int.
30
(
5
),
501
512
.
Yates
D.
Purkey
D.
Sieber
J.
Huber-Lee
A.
Galbraith
H.
West
J.
Herrod-Julius
S.
Young
C.
Joyce
B.
Rayei
M.
2009
Climate driven water resources model of the Sacramento Basin, California
.
J. Water Resour. Plann. Manage.
September/October
,
303
313
.
Yates
D. N.
Lavin
F. V.
Purkey
D. P.
Guerrero
S.
Hanemann
M.
Sieber
J.
2013
Using economic and other performance measures to evaluate a municipal drought plan
.
Water Policy
15
,
648
668
.