Understanding and managing new risks on the Nile with the Grand Ethiopian Renaissance Dam

The paper addresses how the Grand Ethiopian Renaissance Dam (GERD) and Egypt’s High Aswan Dam (HAD) will jointly affect flows, storage, and water security in the transboundary Nile under the absence of a comprehensive operations agreement. The Nile’s hydrology is highly variable, with major contributions from the Blue Nile (Ethiopia), White Nile, and Atbara, and historical average annual flow at Aswan of ~86.5 bcm. The 1959 Egypt–Sudan agreement allocated flows downstream but was not recognized by Ethiopia. Ethiopia built the GERD (installed capacity 5150 MW; reservoir total storage 74 bcm, active 59 bcm) primarily for hydropower, which will alter downstream flow timing. The research goal is to clarify risks and opportunities for Egypt, Sudan, and Ethiopia across three eras: (1) the reservoir filling period, (2) a post-filling ‘new normal,’ and (3) a severe multi-year drought, to inform negotiations and risk management.

The study builds on decades of Nile hydrology research and multiple analyses of GERD–HAD operations that evaluate tradeoffs, cooperation benefits, and hydro-economic outcomes (e.g., Arjoon et al., Block & Strzepek, Digna et al., Geressu & Harou, Jeuland & Whittington, Kahsay et al., Mulat & Moges, Strzepek et al., Wheeler et al., Nigatu & Dinar, Sangiorgio & Guariso). Prior work highlights the potential for cooperative operations to reduce risks and increase benefits, as well as the uncertainties introduced by climate variability and change. This paper synthesizes and extends these insights into policy-relevant narratives spanning filling, typical operations, and severe drought.

The analysis uses the Eastern Nile RiverWare Model (ENRM), a rule-based water resources systems model implemented on the RiverWare platform, simulating operations from 2020–2060. Operating rules are encoded as prioritized logical statements for releases to meet objectives including irrigation and M&I demands, hydropower generation, seasonal targets, minimum flows, flood management, and shortage policies. Key assumptions: Era 1 (filling) uses a preliminary proposal from the National Independent Scientific Research Group (NISRG, 2018–2019). Year 1 retains 4.9 bcm to reach 565 masl (two low-head turbines); surplus passes via temporary spillway. Year 2 retains 13.5 bcm to reach 595 masl to test additional turbines. Thereafter, Ethiopia releases at least 35 bcm per year during filling, with retention during flood months (Jul–Sep) and more even releases across the water year, continuing to fill to FSL 640 masl (peak flood) after reaching 625 masl in the dry season. Egypt aims to release 55.5 bcm annually from HAD unless storage triggers Drought Management Policy (DMP) reductions of 5%, 10%, and 15% when storage falls below 60, 55, and 50 bcm, respectively (first reducing Toskha pumping, then outlet releases). Sudan withdrawals are assumed at 16.7 bcm from current diversion points, with reservoir evaporation varying with levels. Ethiopia irrigation withdrawals are 0.45 bcm (Finchaa) during filling, plus implicit Lake Tana uses via calibrated inflow–outflow relationships. Consumptive uses are assumed not to expand immediately and do not imply endorsement of water rights. Eras 2–3 (post-filling): GERD targets reliable hydropower of 1600 MW whenever possible (approx. 90% maximum reliable rate). GERD pool is lowered to 625 masl each June for flood planning. Future Ethiopian withdrawals increase by 1.0 bcm (Upper Beles, Anger, Arjo tributaries), totaling 1.45 bcm beyond Lake Tana irrigation. Sudan’s future diversions remain 16.7 bcm exclusive of reservoir evaporation; disagreements on accounting for Sudanese reservoir evaporation under the 1959 agreement are noted. Egypt continues to target 55.5 bcm releases from HAD, subject to DMP. Hydrologic conditions: The model uses naturalized flows at Aswan (1900–2018) constructed from 41 gauges (Deltares) and Dongola records adjusted for Sudanese uses (2003–2018). Representative historical sequences are selected for average, high, and low flows to illustrate outcomes in the three eras, including the notorious 1970s–1980s multi-year drought. The study discusses the rationale for using historical traces (transparency, reduced perceived bias, and narrative clarity) while acknowledging climate change and possible unprecedented sequences.

- Filling era (Era 1): HAD reservoir levels could drop to levels not seen in recent decades during GERD filling, yet the risk of significant shortages in Egypt is relatively low due to existing HAD storage and Egypt’s DMP. Under an extremely dry 10-year sequence (similar to 1978–1987) extended to 20 years: even without GERD, a cumulative Egyptian shortage of 35 bcm occurs, with HAD nearing its minimum operating level (147 masl). With GERD filling, in the first 6 years HAD storage falls to 47 bcm while GERD stores 46 bcm; deficits in Egypt occur by year 5 (two years earlier than without GERD). The GERD adds 46 bcm to Egypt’s cumulative deficit over the 20 years, for a combined total of 81 bcm during the filling plus drought period. - New normal (Era 2): After GERD reaches FSL and begins normal operations, intra-annual downstream flow variability decreases due to turbine releases matched to power generation. GERD net evaporation losses average ~1.7 bcm/year, partially offset by ~1.1 bcm/year reduced evaporation at HAD as the system equilibrates. In a representative average 20-year sequence (1934–1953), GERD maintains steady releases and never reaches its minimum operating level (18.4 bcm at 595 masl); HAD remains above 60 bcm and Egypt’s DMP is never triggered. HAD storage is typically lower with GERD than without due to evaporation at GERD and higher Sudanese reservoir levels under smoothed flows, but Egypt can almost always meet the 55.5 bcm target under normal/wet conditions. Sudan benefits from smoothed Blue Nile flows: reduced floods and sediment, increased hydropower generation, and more summer irrigation water (e.g., Gezira). - Entering a severe multi-year drought (Era 3, onset using 1972–1987): As drought begins, both GERD and HAD draw down. GERD releases for power initially bolster Nile flows, reducing Egypt’s cumulative shortages from 42 to 27 bcm up to the drought peak (a 15 bcm reduction) compared to a no-GERD case. However, both reservoirs can become nearly or fully depleted at drought nadir. - Post-drought recovery (1988–2001): Simultaneous refilling is challenging. If GERD continues to generate 1600 MW whenever possible, it refills slowly and operates near minimum level while HAD recovers. Egypt’s cumulative shortage during recovery is 21 bcm with GERD versus 14 bcm without GERD. Ethiopia’s operational choice matters: prioritizing immediate power generation mitigates Egypt’s shortages; prioritizing rapid GERD refill exacerbates them. - Cross-cutting insights: Perceptions of risk matter; visibly falling HAD levels during filling or drought can be misattributed to GERD operations alone, potentially fueling mistrust and a ‘water panic.’ Cooperative planning, data sharing, and agreed rules are essential to minimize harmful impacts during rare multi-year droughts. Ethiopia gains substantial hydropower benefits (~16 TWh/year). Sudan gains from stabilized flows. Egypt may experience slightly lower HAD levels and reduced HAD hydropower, but generally maintains releases under average conditions.

The findings show that GERD–HAD interactions create distinct risk profiles across three eras. During filling, existing HAD storage and Egypt’s DMP limit shortages, but reservoir level declines can appear alarming and be misinterpreted, especially during concurrent low tributary inflows. In typical post-filling years, GERD’s regulation stabilizes flows, yielding net benefits to Ethiopia (reliable hydropower) and Sudan (reduced floods/sediment, more stable irrigation supply), with minimal impact on Egypt’s ability to release 55.5 bcm, though HAD levels trend lower. In severe multi-year droughts, coordinated management becomes critical: GERD’s early releases can reduce Egypt’s shortages, but prolonged drought risks near-depletion of both reservoirs and difficult allocation choices. Public perception and behavior are pivotal; the prospect of a ‘water panic’ emphasizes the need for transparent data, pre-agreed drought management rules, and proactive communication. Overall, the results underscore that cooperation, agreed operating policies (including minimum releases or firm power contracts), and contingency plans are necessary to manage transboundary risks effectively.

The study synthesizes model-based narratives for three eras—filling, new normal, and severe drought—to clarify how GERD will reshape Eastern Nile hydrology and risk. Key contributions include quantifying potential shortages and storage dynamics under extreme sequences, identifying benefits to Ethiopia and Sudan from regulated flows, and highlighting Egypt’s vulnerabilities primarily during drought onset and recovery. The authors conclude that: (1) an agreement on GERD filling rules is urgently needed; (2) basin-wide, transparent drought management and recovery policies should be established before a crisis occurs, specifying how reduced flows are shared when both reservoirs are low and balancing power generation with consumptive use; (3) coordinated communication strategies are essential to maintain public confidence and avoid a water panic. Future work should refine cooperative operating strategies, incorporate broader climate-informed flow ensembles, improve accounting of upstream uses and reservoir evaporation, and evaluate economic and social implications of alternative drought contingency plans.

- Hydrology is based on historical naturalized flow sequences (1900–2018). While transparent and relatable, this approach may not capture unprecedented droughts/floods under climate change or all possible flow sequences. - Key water use and operations assumptions (e.g., Sudanese withdrawals at 16.7 bcm, Ethiopian future withdrawals of 1.45 bcm, GERD target 1600 MW, Egypt’s DMP thresholds) may differ from future policies and actual practices. The accounting of Sudanese reservoir evaporation under the 1959 agreement is contested, adding uncertainty to allocations. - Model simplifies aspects of intra-annual operations and does not explicitly model all upstream irrigation developments (e.g., Lake Tana irrigation is implicit via calibration). - Data and model code are available only upon request due to restrictions; lack of open data may limit independent replication. - Behavioral and political dynamics (e.g., public perception, ‘water panic’) are discussed qualitatively and not modeled quantitatively.