Conicting objectives of energy development and water security in Africa

Africa's economic and population growth prospects are likely to increase energy and water demands. This quantitative study shows that pathways towards decarbonization of the energy sector in Africa may lead to higher water withdrawals and consumption than expected. By 2065, investments in low-carbon energy infrastructure increase annual withdrawals from 1% (2.0 o C) to 2% (1.5 o C) of total renewable water resources compared to 3% in the baseline scenario, despite lower nal energy demands in the mitigation scenarios. Water consumption, in comparison to the baseline, increases by 282% (2.0 o C) and 300% (1.5 o C) by 2065, due to the high water-intensity of the low-carbon energy system. To meet the 1.5 o C pathway, the energy sector requires higher water consumption overall and per unit of energy than other scenarios. These ndings demonstrate the crucial role of integrated energy planning and water resources management if Africa is to achieve climate-compatible growth.

List of African countries per power pool considered in the analysis (with the ISO 3166-1 alpha-2 28  Computational models provide insights into the continent's future energy transformation, energy-water nexus linkages and climate change 29 . Flörke et al. 30 conducted a global water use assessment using a water model in 177 countries for the period 1950-2010, demonstrating the relationship between water use and global-socioeconomic development. Lohrmann et al. 31 estimated water demand for power production on a global, regional and country-level with mitigation strategies for the period 2015-2050 considering a high temporal and spatial resolution. Previous studies on estimating water requirements in the energy sector have focused on either a continental [32][33][34][35] or regional 24,25,36,37 level. Several studies have examined the effects of climate change upon the power sector, such as supply disruption 25,38−40 , increase in electricity prices 41 and environmental impacts [42][43][44][45] .
This study builds on past work by providing a quantitative multi-annual analysis estimating energy requirements, water withdrawal and consumption, carbon dioxide emissions, and costs of covering the future needs of the energy system in Africa including trade links among the African nations.
The energy systems of forty-eight African countries, with a focus on the power sector, are modeled individually using an opensource cost-optimization modeling framework (OSeMOSYS) 46 . We model three different scenarios for the period 2015-2065 47 .
The reference scenario assumes no change from national renewable energy policies after 2017. In two mitigation scenarios (2.0ºC, 1.5ºC), the nal energy demands are lower due to energy e ciency measures and the annual emission levels are constrained to the emission pathways compatible with the 2.0ºC, 1.5ºC scenarios 48,49 .

Results
Our results show that countries currently under high-risk of water scarcity are expected to further increase their water use in the future due to fossil fuel consumption, leading to a vicious cycle. Under mitigation scenarios (2.0ºC, 1.5ºC), a reduction in demand in comparison to the reference scenario, coupled with the use of low-carbon technologies, leads to a decrease, but then an increase in water withdrawal and consumption, highlighting the role of nuclear, solar PV, hydropower and carbon capture with storage technologies (Supplementary Figure 2-7). Higher levels of electricity and gas trade are required to achieve lower emission limits, highlighting the role of transit countries and key net exporters. Nevertheless, it needs to be noted that higher levels of electricity exports in speci c countries lead to higher levels of water withdrawal and consumption.
Future energy sector water requirements in the African continent In the reference scenario, water withdrawals in Africa grew by almost eight times from 2015, reaching 159 billion cubic meters (bcm) in 2065. This increase corresponds to approximately 3% (2065) of the Total Renewable Water Resources (TRWR -5,290bcm in 2015) in the continent 27 , assuming no changes in precipitation patterns due to climate change. This growth is mainly due to the penetration of high water-intensive technologies (coal, oil) and the increase in the hydropower share. Water consumption also increases by four times (187bcm) by 2065.
In the reference scenario, in the Northern Africa power pool (NAPP) and Central African power pool (CAPP), there is an increase in water consumption despite the transformation of the future energy mix to a higher share of renewables and the use of less water-intensive thermal general technologies. This transformation is due to the rapid increase in energy demands. On the other hand, in the East African Power Pool (EAPP) and South African Power Pool (SAPP), the increase in water consumption is explained by the adoption of water-intensive technologies. NAPP, EAPP and the Western African power pool (WAPP) experience an increase in their water withdrawals by 2065 while SAPP and CAPP experience a decrease, 13% and 61%, respectively. However, all power pools experience an increase in their water consumption ranging from 116% (NAPP) to 1576% (CAPP). Notably, countries with considerable hydropower potential (e.g., Angola, Cameroon, DRC, Ethiopia, Nigeria, Zambia) experience increased evaporative losses in the future (Supplementary Figure 47-49).
In the mitigation scenarios, overall water withdrawal also rise (2.0 o C, 52bcm, TRWR 1%) (1.5 o C, 85bcm, TRWR 2%) but to a lesser extent compared to the reference scenario (2.0 o C, 67%), (1.5 o C, 47%) by 2065. Also, decarbonizing further, the energy sector leads to an increase in water consumption of 282% (2.0 o C) and 300% (1.5 o C) by 2065, in Africa ( Figure 2). These results support the message that pathways towards decarbonization of the energy sector in Africa may lead to higher water withdrawal and water consumption. This observation highlights the signi cant role of clean technologies with a low water footprint versus hydropower with integrated water resources management that secures water for other purposes (e.g., agriculture, municipal services). It is particularly relevant in NAPP and WAPP, which have set ambitious renewable energy targets. However, the projected nuclear investments in Egypt, Morocco, Chad, Guinea, Gabon, Uganda, Nigeria, Benin, Côte D Ivoire, Ghana, Senegal, Mali, increase water withdrawals in the above scenarios.
The role of inter-regional trades, national mega-projects and transit countries Fossil-fuel reserves, renewable potential and water resources are unevenly distributed in Africa. The results show how trade among African countries could in uence water resources management, decrease electricity generation costs and lower emission levels across the scenarios. The largest electricity net exporters by 2065 are Kenya, South Africa and Sudan, while the main net importers are Uganda, Burkina Faso and Mali. The high potential for renewables in the EAPP makes the region the largest net exporter of electricity. A notable nding of this study is the identi cation of some countries as transit-traders such as Egypt, Sudan, South Africa and Tanzania ( Figure 3). Indicatively, under the reference scenario, Egypt imports 659TWh (94%) of its cumulative electricity imports from Sudan. In parallel, Egypt exports approximately 1194TWh (96%) of its cumulative electricity exports (2015-2065) to Asia. In parallel, 64TWh or 15% of Sudan's total electricity exports are derived from imports of electricity generated in Ethiopia. Increased electricity trade enables optimized system operation and typically results in cheaper electricity costs and lower emission levels. However, for the exporting countries, there are consequences for their national water needs. Speci cally, the potential implementation of the Grand Inga project in the Democratic Republic of Congo, together with trade links, could increase the electricity exports to neighboring countries and displace part of their fossilfuel based generation. Zimbabwe is one of the countries which undergo a transformation from net exporter in the reference scenario to a net importer in the decarbonization scenarios to reduce its fossil fuel-based generation capacity, importing on average 14 TWh of electricity annually.
These results highlight the importance of an enhanced electricity trading scheme on the continent to reduce greenhouse gas emissions and system costs. Nevertheless, this could come at the expense of increasing the water consumption in the main electricity exporter countries, particularly Ethiopia, Guinea, Liberia, Sudan, South Africa, putting them at risk of water shortages. However, countries such as Ghana, which increase their electricity net imports in all scenarios, experience a concurrent decrease in their water requirements. In short, investment decisions in large hydro-electricity generation projects cannot be separated from water resource management and electricity trade and require regional coordination across countries, tailored to the local geopolitical and topological realities.
Also, speci c gas pipeline projects (e.g., West African, Trans-Saharan) could change certain countries (Algeria, Mozambique, Nigeria) to become energy hub exporters assisting their neighboring countries to transform their energy sector.
Transformation of the energy system The evolution of the energy mix of a core group of fossil fuel resource-rich countries play an essential role in African greenhouse gas emissions. For example, South Africa and Lesotho extract most of the continent's nal coal consumption. Nigeria and Egypt are large consumers of oil products. Algeria, South Africa, Nigeria, Egypt and the Democratic Republic of Congo will consume most of the continent's natural gas, in nal energy terms. The analysis of how the nal energy consumption evolves among the scenarios can be found in Supplementary Information.
In the reference scenario, the total African primary energy supply more than doubles compared to 2015, reaching 1853Mtoe by 2065. While the share of fossil fuels increases over the years (64%), renewables experience a gradual decrease, eventually reaching 36% of the total primary supply by 2065 ( Figure 4). Without a carbon constraint, coal, as the cheapest source of electricity, constitutes most of the continent's primary energy supply, followed by oil and biomass. Nuclear power disappears from the electricity supply system by 2065. The WAPP stands out as the most signi cant energy supplier (35%) in Africa, followed by EAPP (28%), SAPP (19%), NAPP (11%) and CAPP (7%) in 2065. WAPP is also the largest supplier (37%) of fossil fuels and renewables (32%) in the continent in 2065.
However, in the 2.0 o C and 1.5 o C scenarios, due to the relatively lower nal energy demand, African countries increase their 2065 total primary energy supply by only 50% and 31%, respectively, in comparison to 2015. Moreover, the primary supply of fossil fuels in those two scenarios declines dramatically throughout the years, reaching 27% and 8% by 2065. On the contrary, renewables increase by 64% and 72%, respectively, by 2065 ( Figure 4).
As natural gas reserves are scattered among nations, the role of natural gas trade through pipelines and LNG terminals is key to decarbonize the African energy system. Countries with signi cant natural gas reserves, such as Algeria, Nigeria, or Mozambique, increase their natural gas exports signi cantly to reduce the consumption of more polluting fossil fuels in the continent. In particular, Mozambique increases its gas exports to South Africa for replacing coal in the power sector with natural gas.
Under the mitigation scenarios, natural gas supply to Europe through the Northern African countries gradually declines, which is in line with Europe's aim to become a climate-neutral continent by 2050. The Western African power pool and speci cally Nigeria, will be the leading natural gas supplier in the African continent. Several coastal countries (Côte D Ivoire, Ghana, Morocco, Sudan, Senegal, Tunisia, Tanzania and South Africa) increase their LNG imports. This increase in imports leads to lower emission limits by replacing the above countries' fossil fuel capacity in the power sector as well as decreasing their water requirements.
Evolution of the electricity supply sector In the reference scenario, the overall generation capacity in Africa rises ten-fold from 181GW (2015) to 1863GW (2065 Our results show that to decarbonize the energy sector, the gas-red power generation technologies, along with the penetration of renewables, CCS and nuclear technologies, replace coal-based power generation. In the reference scenario, electricity generation in Africa increases from 64Mtoe (2015)  Developing strategies for the African continent should, therefore, prioritize sustainable technologies, demand-side management and set ambitious targets. This also applies to oil-producing countries of the continent (e.g., Algeria, Tunisia) since they are expected to pro t from electricity trading leading to savings on their total fuel costs by 48% (2.0 o C) / 46% (1.5 o C) and power system costs by 28% (2.0 o C) / 7% (2.0 o C) compared to the reference scenario, improving in parallel their water productivity by approximately 50%.

Discussion
The results offer insights into how the policy agenda on sustainability and economic growth could be strengthened at the global and continental level by considering the interdependency of energy and water sectors.
Previous studies have highlighted the role of low carbon electricity to meet climate change targets. Our results indicate that to meet the 1.5 o C pathway, the energy sector requires higher water consumption overall and per unit of nal energy, so proper water resource management is essential. The reason behind this is the penetration of nuclear and hydropower technologies.
Aside from increasing water consumption, hydropower creates new opportunities as it is an enabling infrastructure for effective water resources management. The additional services (water storage, ood risk mitigation) may counterbalance the associated withdrawals.
The results highlight the role of electricity and gas trading schemes among the African nations and how a trading scheme could help achieve more ambitious emission targets and lower system costs. Findings also reveal the future role of some countries as key net exporters and transit hubs. Besides, the possible implementation of mega-projects (Grand Inga Dam) could further boost trading and supply with low carbon electricity on the continent. In any case, integrated and transboundary water resource management is a sine qua non of Africa's sustainable development and needs to be harmonized with energy investments.
Findings show the different strategies between countries based on their indigenous energy resources while increasing electricity trade and substituting fossil-fuel-based power generation to reduce generation costs. The integrated model highlights the linkage between energy and water security. As nations invest in new power plants to increase electricity exports, this could come at the cost of their available water resources and even trigger geopolitical con icts. The increased ambition of African energy and climate targets would probably reduce energy exports to non-African countries, negatively affecting their energy security and limit potential trade options. This would, however, be outweighed by the signi cantly lower system costs of transforming Africa's energy sector following ambitious climate targets. LNG imports through coastal countries also play an essential part in decarbonizing the energy system by replacing coal-based power generation plants.
Also, countries such as Egypt, Algeria, Morocco, South Africa and Angola, are expected to construct signi cant thermal generation capacity on the coast, using seawater for cooling as the water is freely available. Thus the associated risks with climate change (e.g., sea-level rise) need to be considered.
If countries were to continue energy exports at the current level, the effects upon water withdrawals demonstrated in these results would be exacerbated. In the decarbonization scenarios, exports of primary fossil fuel resources decrease at a level commensurate with global action on climate change. Large exporting countries, such as Algeria, Nigeria, South Africa, need to consider the trade-offs between the revenues of energy exports with other countries, the local consumption that will boost economic growth and increase water requirements and energy security.
It is important to understand the linkages as well as the individual challenges of energy and water to develop effective strategies towards sustainable development, especially under climate change. Global-national-regional 50,51 policies, along with future adoption of technologies, declining costs of renewables, oil price uctuations and new business models, should be considered in an integrated manner to address the challenges indicated in the United Nations Sustainable Development Goals (SDGs) in particular SDG6, SDG7, SDG13 8 . Also, climate change and sustainable development governance should be better integrated to maximize the effectiveness of action in both domains 52 .
The continental-scale insights could inform the National Determined Contributions targets (to be reviewed in 2020) 50 , by demonstrating the broader African context of national greenhouse gas emission targets. National and governmental institutions, as well as universities involved in capacity building activities, could bene t from this open-source study since the provided datasets could strengthen the capacity for developing others and extending existing energy systems models.
There is scope for further work based on the results of this study. Combining agricultural and municipal water withdrawals 27 for each African nation with our results would show the impact of the scenarios upon levels of water-stress 26 .
Linking a water-systems model to the energy-system model of this study would provide insights into the resilience of the African continent in terms of water and energy under climate change. Hydrological modeling of each of the planned dams would provide more accurate quanti cation of dam productivity and impacts. Broader social and environmental effects of hydropower and nuclear, which are outside the scope of this study, should be further examined at a power pool or regional scale along with mitigation strategies to understand the implications of those technologies better. Battery storage for solar PV and pumped hydropower storage are only implicitly modeled due to the macroscopic nature of the study that focuses on Africa's urgent need for access to energy and water and computational constraints. Better data and spatial techniques could help identify and allocate the different cooling technologies to future thermal power plants. Also, incorporating countryspeci c reserve margins rather than using an average, the power sector projections would improve the representation of national energy systems. Finally, although included in the scenario analysis, some signi cant uncertainties around technology costs, fuel prices and future energy demands are not yet systematically explored, which would provide further evidence into the robustness of these results.

Methods
We developed a model to evaluate energy supply and water requirements to cover the energy needs of the African continent during the period 2015-2065. The model was developed using the open-source modeling system for long-term energy planning OSeMOSYS 46 . The objective function is to minimize total energy system costs, rather than, for example, co-optimize the energy and water sectors. Other energy resources were also included in the model except for adding the water analysis, and the dataset was updated based on the latest available information. The OSeMOSYS model developed to conduct the study "Energy projections for African countries" 47 , itself extended from the Electricity Model Base for Africa (TEMBA) 53 , was further extended, including exports for all fuels and water loss due to evaporation in hydropower plants. Furthermore, the latest available data on the energy system of Africa was also updated considering national energy policies. In the following subsections, the detailed methodology followed to develop the model is described.

Energy-water model development
The energy model was developed using the open-source cost-optimization energy system modeling framework (OSeMOSYS) 46 for medium to long term energy planning. OSeMOSYS is a bottom-up modelling framework that uses linear optimization techniques to satisfy an exogenously de ned energy demand. OSeMOSYS has been employed in the scienti c literature to provide insights on possible transformation pathways of large energy systems and their impacts on the economy, the society and the environment 54 . The open-source method of the study assists in overcoming the lack of transparency needed to address future long-term decarbonization trajectories 55 .
The electricity supply system, including a simple representation of other energy sources (heavy fuel oil, light fuel oil, gas, coal, biofuel & waste), of forty-eight (48) African countries was modeled individually and linked via electricity interconnectors and natural gas pipelines to shape the African model. Countries were then associated with the following regional power pools 56 : Central Africa (CAPP), Eastern Africa (EAPP), Northern Africa (NAPP), Southern Africa (SAPP) and Western Africa (WAPP) to identify trades. The North African Power Pool (NAPP), as it is reported here, is o cially called the Maghreb Electricity Committee, or COMELEC. The analysis provided results on a continental-regional-country level at an annual temporal resolution for the period 2015-2065. Country speci c hourly electricity generation pro les and seasonality of the electricity load were considered in the analysis.
Starting from the supply, fuel supply technologies were categorized per country in extraction and imports, associated with fuel prices distinguishing between inland and coastal, as well as conversion technologies (Liqui ed natural gas, gas regasi cation plant, re neries); electricity and natural gas interconnection projects among African countries, as well as with other continents; and country-speci c energy resources (renewable potential, fossil fuel reserves).
Moving to power generation, centralized and decentralized (off-grid) technologies were modeled, taking into consideration techno-economic parameters. New technologies such as carbon capture and storage (CCS) were also represented. Countryspeci c capacity factors (wind, solar) were introduced to capture better the granularity of electricity generation as well as transmission and distribution losses.
Lastly, at the end-user level, country-speci c nal energy demands (electricity, biofuel and waste, coal, gas, heavy fuel oil, light fuel oil) were captured and modi ed accordingly in each of the scenarios (Supplementary Table 1 -15). The nal energy demands and fuel exports (coal, crude oil, oil products, biofuels and waste) are exogenous parameters to the model. Emission factors were assigned per type of fuel, while the emission limits in decarbonization scenarios were placed on a regional and not country level.

Power plant data
Power plant data was obtained from the World Electric Power Plants database 13 and validated with online sources 57,58 and national master plans. In total, 5,300 power plants were examined, of which 3,646 power plants are thermal power plants. The power generation technologies are categorized into centralized and decentralized, per type of technology and separated in fossil fuel and nuclear, renewable and carbon capture and storage (Supplementary Table 39). Techno-economic parameters were assigned to the technologies (Supplementary Table 17 -20). In addition, the power plants are categorized as old (existing capacity until 2014), or new with e ciency and cost improvements.

Water demand
The Total annual Renewable Water Resources (TRWR) in Africa is the theoretical maximum annual volume of water resources available 9 , approximately 5,290 billion cubic meters 27 . Water withdrawal is the total volume removed from a water source such as a lake or river. In contrast, water consumption refers to the portion of the withdrawn water that is permanently removed and not returned to its source 11 .
To estimate the water withdrawal and consumption per type of technology and consequently the water demand by a country, we considered different cooling types: dry cooling (AIR), natural and mechanical draft tower (MDT/NDT), once-through cooling tower with freshwater (OTF) and once through cooling tower with salt water (OTS) per type of technology (power generation). Based on the identi ed cooling type, speci c water factors were assigned (Supplementary Table 22). The cooling type of the existing power plants was obtained from the World Electric Power Plants database 13 and completed with visual estimations based on satellite images using Google Earth. The new power plants were assumed to have the same cooling type as the majority of the existing power plants in the corresponding country for each fuel type. However, in cases where the available data for the current technologies were insu cient to identify their cooling type in the relevant countries, we used weighted averages. Water factors have also been applied to estimate water use for fuel extraction and processing (Supplementary Table 23). The following equations 59 were used to obtain water use of fuel production (1) and water withdrawal and consumption for power plant operation (2) for each country: Where WFP is the water use for fuel production and processing, WOP is the water use for power plant operation and n the number of power plants in each country.
Future water consumption allocated to hydro-electricity was also addressed based on a previous study that analyses the water loss due to evaporation in the biggest 158 hydropower reservoirs in Africa in 2016 59 . The corresponding hydropower plants in these reservoirs represent 95% of the total hydropower installed capacity in the continent. These country-speci c water factors were applied to the hydro-electricity projections for the different scenarios.
The water use for biomass extraction was not included in the study. Decarbonization scenarios To explore plausible future developments of the African energy sector, we developed three scenarios. The reference scenario assumes that there is no change from the national renewable energy policies after 2017. In the two mitigation scenarios (2.0ºC, 1.5ºC), the fuel demands of each African country are derived from the JRC GECO 2018 report 48 and disaggregated based on socio-economic factors (GDP 48 , population 2 ). In the reference scenario, the carbon dioxide emissions increase by approximately three times, compared with 2015 levels, reaching to 3,466 Mt of CO2 by 2065. The annual emission levels are constrained to the emission pathways compatible with the 2.0ºC, 1.5ºC scenarios 48,49 . Carbon removal technologies were also introduced in these two scenarios.
Under the 2.0 °C and 1.5 °C scenarios, we assume the implementation of energy e ciency policies and a signi cant reduction in fossil fuel consumption (Fig. 4). As a consequence, the total nal energy demand increases by only 79% and 49%, respectively, by 2065 over 2015 levels.
Projections for the energy transition of each country were calculated based on historical energy balances 60,61 , future GDP 48 and population 2 estimates and calibrated following the report of Keramidas et al. 48  The osemosys code used to develop the model for Africa can be found on a Github repository.

Competing interests
The authors declare no competing interests.

Additional information
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional a liations.