Cooperative operation of the Grand Ethiopian Renaissance Dam reduces Nile riverine floods

The construction of the Grand Ethiopian Renaissance Dam (GERD) on the Nile River has triggered much debate between Ethiopia, Sudan, and Egypt on the dam's effects. Once completed, the GERD will be Africa's largest hydropower plant. This study analyzes the implications of cooperative long‐term operation of the GERD for Nile riverine flooding downstream of the dam. A daily river system model of the Eastern Nile is developed and used to examine how cooperative long‐term operation of the GERD would affect the occurrence of three flood alarm levels (alert, critical, and flooding) in Sudan downstream of the dam. A reconnaissance‐level flood inundation model is developed for the Nile within Khartoum State (Sudan's capital) to assess the GERD's impacts on the flood extent in the state based on simulated flows from the river system model. Assuming the GERD is operated to achieve a 90% power reliability and active upstream‐downstream data sharing between Ethiopia and Sudan on the dam's daily outflows, results show that the GERD would reduce the occurrence of the three alarm levels. Based on 34 simulated river flow sequences, the proportion of years with at least one flooding alarm day at Khartoum Gage declined from 37% without the GERD to 11% with the GERD. Seasonal coordination and planning between Ethiopia and Sudan are necessary to mitigate the remaining riverine flood hazard. Although cooperative long‐term operation of the GERD could play a positive role in reducing the riverine flood hazard in Sudan, the associated river flow alterations would adversely impact recession agriculture and the environment.

the Grand Ethiopian Renaissance Dam (GERD)-the largest hydropower plant in Africa and among the largest worldwide (Mulligan, van Soesbergen, & S aenz, 2020)-on the Blue Nile River for hydropower production, and the initial filling of the dam's reservoir started in July 2020. The GERD is located around 15 km upstream of the Ethiopia-Sudan border and is planned to have an installed power capacity of 5,150 MW and a maximum reservoir storage capacity of 74,000 Mm 3 .
The dam's storage capacity is around 1.5 times the average annual river flow at the dam location. Because the purpose of the GERD is electricity generation, its high storage capacity is expected to regulate the Blue Nile flow and potentially benefit Sudan in terms of irrigation water reliability and hydropower generation (Wheeler et al., 2016).
However, the dam is expected to pose adverse environmental impacts and a loss of recession agriculture in Sudan (Alrajoula, Al Zayed, Elagib, & Hamdi, 2016;. Moreover, the operation of the GERD would reduce Egypt's hydropower generation (Wheeler et al., 2016). Depending on how the GERD is operated during multi-year droughts, the dam could result in irrigation water deficits in Egypt (Wheeler, Jeuland, Hall, Zagona, & Whittington, 2020). Negotiations between Ethiopia, Sudan, and Egypt on the initial filling and long-term operation of the dam are ongoing since 2011.
Recently, the United States Government, the World Bank, and the African Union hosted and convened several rounds of negotiations between the three countries, but no agreement has been reached as of April 20, 2021.
Around 22-80% of the total flood-affected people in Sudan (from both flash and riverine floods) during the period 2012-2020 were located in these six states (Figure 1b;OCHA (2021)).
This study assesses the impacts of cooperative long-term operation of the GERD on the riverine flood hazard in Sudan and recommends measures for mitigating the associated challenges. The author developed and used a daily river system model of the Eastern Nile Basin and a reconnaissance-level riverine flood inundation model of Khartoum State (Sudan's capital) to examine the impacts of cooperative long-term operation of the GERD on the flood hazard of the Nile.
The river system model was calibrated and validated over the period from 1983 to 2017 and includes a detailed representation of the daily operations of the major water-related infrastructures in the river system. The flood inundation model of Khartoum State was based on satellite topographic data and was driven by simulated flows from the daily river system model.

| Eastern Nile River system model
In this study, a daily river system model was developed for the Eastern Nile Basin downstream of the GERD. A river system model is a numerical network representation of the water supplies and demands, infrastructures, and operating rules of a river. Figure 2 depicts a schematic of the model. The model includes 27 inflow nodes, 9 storage dams, 21 water withdrawal nodes, 13 stage-discharge gages, reservoir evaporation, and channel seepage from 13 river reaches. Table S1 in the supplementary material reports the main characteristics of the nine storage dams included in the model. The stage-discharge gages are populated with rating curves to translate river flows to river water levels. A rating curve is a relationship between the river discharge and the river water level of a specific river cross-section. The generated water levels were then utilized to calculate the number of days within each of three flood alarm levels used in Sudan by the Ministry of Irrigation and Water Resources: alert, critical, and flooding. These alarm levels are based on thresholds of river water levels. Table S2  The model is driven by hydrological inflows at nodes F 1-27 and the system operating rules. The observed flow records of the Nile at Eldiem (G1), Elgewisi (G3), Elhawata (G2), Malakal (G5), and Kubur (G9) gages were used as inflows for the nodes F1, F2, F3, F4, and F5, respectively. Data for the rest of the inflow nodes were obtained from hydrological models previously developed by , Basheer and Elagib (2018), and Basheer, Sulieman, and Ribbe (2021).
Average monthly evaporation coefficients were used to simulate evaporation from reservoirs. Channel seepage percentages and lag times were populated for the river reaches. Constant percentages were used to estimate the return flows of the Rahad (I1), the Gezira and Managil (I4), and the New Halfa (I12) irrigation schemes (5, 7, and 4%, respectively).
The model was created using RiverWare, a generalized river and reservoir simulation software developed by the University of Colorado Boulder (Zagona, Fulp, Shane, Magee, & Goranflo, 2001).
RiverWare can simulate several river system processes using a variety of methods and time steps. RiverWare enables the user to add the system operating rules using logical statements in the RiverWare policy language. This river system simulation software has been termed a "hydro-policy" tool by Wheeler, Robinson, and Bark (2018b) due to its flexibility in modeling water management policies.   (Manning, 1891).

| Flood inundation model
In GFT, the river stream is divided into segments, and a river crosssection is generated for each segment to construct a depth-discharge  Chow (1959). The flood inundation model was driven by daily simulated streamflow values generated using the river system model described in Section 2.1.

| Data sources
The observed river flows and reservoir water levels, water withdrawal targets, and dams' characteristics were obtained from Copernicus Global Land Service (2021)

| Simulation assumptions and scenarios
In this study, it was assumed that the long-term operation of the GERD starts with reservoir storage of 49,300 Mm 3 , following the outcomes of recent negotiations between Ethiopia, Sudan, and Egypt (Amde, 2020;Edrees, 2020). An energy-oriented operation was assumed for the GERD by targeting a power production of 1,600 MW (38.4 GWh/day) to achieve a 90% power reliability, as revealed by Wheeler et al. (2018a). A higher priority was given to keeping the reservoir water level between the minimum operating and full supply levels. It was assumed that Sudan's Roseires, Sennar, and Merowe dams (see Figure 2) are operated at their full supply levels and are allowed to drop only to meet the water or energy demands. Wheeler et al. (2016) and  found that this configuration eliminates the water supply shortages in Sudan. All other dams in Sudan were operated using their historical operating rules in both the baseline and with the GERD. Table S4  Cooperation on transboundary rivers can occur in various forms ranging from unilateral action to joint action (Sadoff & Grey, 2005). In this study, cooperation between Ethiopia and Sudan in the form of active upstream-downstream data sharing on the GERD's daily outflows (i.e., instant downstream knowledge at each simulated time step) was assumed in modeling the operation of the Roseires, the Sennar, and the Merowe dams.
To examine the impact of the GERD on the riverine flood hazard in Sudan, two scenarios were simulated: (a) a baseline scenario without the GERD in which the river system is operated following the historical rules, and (b) a scenario in which cooperative long-term operation of the GERD is introduced, and the operation of the Roseires, the Sennar, and the Merowe dams are modified as explained earlier. The two scenarios were examined across 34 river flow sequences (or traces) generated using the index-sequential method (Kendall & Dracup, 1991;Ouarda, Labadie, & Fontane, 1997). Each of the river flow sequences is 34-year long. The index-sequential method uses the historical record of river flows to generate river flow sequences assuming that every year in the record is a possible starting point. This method was used because it is non-parametric and preserves the serial and spatial correlations in the historical flow data.
However, it does not capture non-stationarity in the hydrologic system (e.g., climate change) and river flow patterns outside the historical record. The 34 hydrologic sequences were based on the daily flow data of the 27 inflow nodes of the river system model (Figure 2) for June 1983 to May 2017.

| RESULTS
Results reveal that cooperative operation of the GERD to achieve a 90% power reliability (see Section 2.5 for details) would change the pattern of river flows and water levels of the Blue Nile and the Main Nile in Sudan. Figure 3 shows box plots of the annual number of days in each of the three alarm categories used in Sudan (i.e., alert, critical, and flooding; see Table S2  It was found that cooperative operation of the GERD to achieve a 90% power reliability would reduce the riverine flood hazard in Sudan on the Blue Nile and the Main Nile. Nevertheless, the results suggest that the occurrence of floods remains possible, especially at Khartoum Gage, as indicated by a maximum of two flooding alarm days and 11% of the simulated years with at least one flooding alarm day (see Figure 3h). Figure 4 shows the daily simulated reservoir storage and outflow of the GERD for one of the 34 examined river flow sequences. During successions of high inflow years that coincide with high GERD water levels, the GERD would reach its full supply level and pass higher water volumes than the volumes required for generating the assumed 1,600 MW power target (or 38.4 GWh/day).

| DISCUSSION
The results show that high outflows from the GERD could occur during the dam's long-term operation, as depicted in  (Kron, 2005). Mitigating the impacts of high outflows from the GERD requires seasonal coordination and planning between Ethiopia and Sudan on dam operation in addition to active daily data sharing.
Seasonal coordination and planning and active data sharing would enable Sudan to gradually drawdown the Roseires, the Sennar, and the Merowe reservoirs (see Figure 2) to partly absorb the impact of high GERD outflows. Furthermore, awareness of the populations living along the Blue Nile and the Main Nile in Sudan should be raised against complacency, given the anticipated sporadic occurrence of riverine floods. In this study, the GERD was assumed to target 1,600 MW of power generation to achieve a 90% power generation reliability (Wheeler et al., 2018a). Adjusting the assumed operation of the GERD by increasing the power target would reduce intense daily downstream outflows to Sudan and, in turn, the flood hazard, but would cost a decline in the dam's power reliability and/or annual electricity generation, given the high interannual variability (see Figure S1 in the supplementary material) and low predictability of the Blue Nile flow.
The impact of cooperative operation of the GERD to achieve a 90% power reliability on flood occurrence in Khartoum would/should affect the operation of the Jebel Aulia Dam, which is located on the White Nile around 40 km upstream of the confluence of the Blue Nile and the White Nile (see Figure 1). The outlets of the Jebel Aulia Dam are usually closed when the water level at Khartoum Gage exceeds certain thresholds during the flood season of the Blue Nile in July-October (Basheer & Elagib, 2018;Sobeir, 1983)  Although floods are often associated with destruction and damage, they provide environmental and economic benefits. Generally, river flow regulation results in a loss of natural floodplains and negatively affects the flora and fauna that rely on the floodplains (Brismar, 2004;Cuny, 1991 (Mohammed, 2015).
Moreover, the hydrological alterations associated with the operation of the GERD would affect the water temperature, salinity, suspended nutrients, and oxygen content downstream of the dam and adversely impact aquatic life and biodiversity .

| CONCLUSIONS
This study showed that cooperative (i.e., active daily data sharing) energy-oriented long-term operation of the GERD would reduce Nile riverine floods in Sudan. Mitigating the remaining riverine flood hazard requires seasonal coordination and planning in addtion to active daily upstream-downstream data sharing between Ethiopia and Sudan. Climate change is expected to alter the amount and variability of the Nile streamflow (Siam & Eltahir, 2017), and cooperative operation of the GERD could contribute to reducing the riverine flood hazard of these anticipated changes.
Although cooperative energy-oriented operation of the GERD would reduce the riverine flood hazard in six states in Sudan, a large proportion of flood-affected people in the country would not benefit from reduced Nile riverine floods. Flash floods are the more common and frequent cause of flood-related losses and disruptions in the six GERD-affected states in Sudan (Mahmood, Elagib, Horn, & Saad, 2017;Mahmoud, Elagib, Gaese, & Heinrich, 2014;Zerboni et al., 2020)  health services, and public awareness of flood risks (Elagib, Gayoum Saad, Basheer, Rahma, & Gore, 2021;Horn & Elagib, 2018;Mahmood et al., 2017;Mahmoud et al., 2014).

ACKNOWLEDGMENTS
This work had been presented at the "Enhancing flood forecast and early warning in Eastern Nile Basin" forum organized by the Eastern Nile Technical Regional Office (ENTRO) from 16 to 17 August 2018.

CONFLICT OF INTEREST
The author reported no potential conflict of interest.

DATA AVAILABILITY STATEMENT
The data and model that support this study's findings are not publicly available due to third-party restrictions. The Digital Elevation Model (DEM) data used in developing the flood inundation model are freely accessible at: https://earthexplorer.usgs.gov/.