Reply to Comment on 'Egypt's water budget deficit and suggested mitigation policies for the Grand Ethiopian Renaissance Dam filling scenarios' by Kevin Wheeler et al'

Two technical misinterpretations of Heggy et al's (2021) results are the pillars of the comment by Wheeler et al, as stated in their abstract and summarized above:

• (a)  
  Wheeler et al consider the median 31 BCM yr⁻¹ of Egypt's total water budget deficit calculated in Heggy et al (2021) to be entirely the result of Nile flow impoundments by the GERD omitting the different components constituting the calculated value. This is stated by Wheeler et al on page 1 of their abstract:

  Their median estimate is that filling the GERD will result in a water deficit in Egypt of ∼31 billion m³/year.

  —Wheeler et al

  whereas Heggy et al (2021) assessed Egypt's median total water budget deficit (i.e. intrinsic + seepage losses + GERD impoundment) for different filling scenarios—using several peer-reviewed published GERD filling scenarios (available at the time of the preparation of the manuscript) from 2.5 to 29.6 years. Heggy et al (2021) clearly stated that 60% of the above deficit is attributed to intrinsic sources and only 32% + 8% is associated with GERD (filling + seepage) as shown in figure 2 and page 11 of Heggy et al (2021).

  If we account for all the components, Egypt may experience in the years of filling of the GERD a total annual water deficit from 24.7 to 36.5 BCM yr⁻¹ (distributed in percentage as: 32% due to GERD filling, 8% for seepage under GERD, and 60% of intrinsic water budget deficit), which corresponds to an average +35.5% of its current internal annual water demand.

  — Heggy et al (2021)

  It is clear that the main assumption of Wheeler et al is not consistent with what is published in Heggy et al (2021).
• (b)  
  Wheeler et al also erroneously consider the economic impacts in Heggy et al (2021) to be completely attributed to the GERD, as stated in their abstract:

  Their median estimate is that filling the GERD will result in a water deficit in Egypt of ∼31 billion m³/year. They estimate that under a rapid filling of the GERD over 3 years, the Egyptian economy would lose US $51 billion and 4.74 million jobs, such that in 2024, GDP per capita would be 6% lower than under a counterfactual without the GERD.

  — Wheeler et al, p 1

  whereas in Heggy et al (2021), Egypt's equivalent economic impacts and potential unemployment rates are calculated for the unmitigated total water budget deficit resulting from both intrinsic and GERD components. The unmitigated case is considered so as to show the magnitude of the water budget deficit if there is no mitigation due to socioeconomic instability, drought or the lack of joint AHD-GERD operation agreement. Heggy et al (2021, p 13) clearly state:

  in the years 2022, 2023 and 2024, our first order model suggests an equivalent agricultural GDP losses arising from the unmitigated total water budget deficit, for the total filling period, of ∼$51 billion for the 3 years filling scenario, ∼$28 billion for the 5 years filling scenario, ∼$17 billion for the 10 years filling scenario, and ∼$10 billion for the 21 year filling scenario.

  — Heggy et al (2021, p 13)

  According to the above, the statement in Wheeler et al is not consistent with the text in Heggy et al (2021), in which we calculated the equivalent economic impact of the unmitigated total water budget deficit from Egypt's intrinsic and GERD sources during the filling of the GERD. Wheeler et al mistakenly attributes the economic losses entirely to the filling of the GERD and ignores the fact that our results are for the unmitigated scenario. As the cost and feasibility associated with these mitigations remain widely debatable, in Heggy et al (2021), we opted to provide a feasibility index (section 3.2 and figure 3—Heggy et al 2021) to be a guide for decision makers when setting the mitigations plans for Egypt's total water budget deficit (including intrinsic + GERD) and not only from GERD.

While Wheeler et al states that:

In this response to Heggy et al (2021) we draw on high quality peer-reviewed literature and appropriate modeling methods related to our aim and approach to identify and analyze many flaws in their article.

— Wheeler et al, p 1

The call for using high quality peer-reviewed sources starts on a wrong note with two-thirds of the comment (from pages 2 to 7) based solely on a self-citation of Wheeler et al (2020) and the use of an inaccurate citation from a conference oral presentation by Deltares (2019) in the first section. For the second part (from pages 7 to 10), the comment is mainly composed of inappropriate comparisons exclusively made between the economic findings in Boehlert et al (2017), Kahsay et al (2017, 2019) and Heggy et al 2021) Boehlert et al (2017) and Kahsay et al (2017) are two book chapters that are outside of the rigorous peer review process claimed throughout the Wheeler et al's comment and constitute the main supporting arguments opposing our double-blind peer reviewed study as detailed further below:

• (a)  
  In the first section, and to support their first claim that our study provided a 'miscalculation of the water deficit' (pp 2–7), Wheeler et al cited two publications (Deltares 2019, Wheeler et al 2020) along with an updated simulation of the GERD filling (figures 1–3 in Wheeler et al) that is originally based on the simulation simulated in Wheeler et al (2020).First, we have numerous concerns regarding the validity of the simulation results modeled in Wheeler et al (2020). For instance, Wheeler et al (2020) simulation results indicated that the GERD filling under wet conditions would take 5 years, as shown in figure 3 in Wheeler et al (2020), while under average conditions (figure 4 in Wheeler et al 2020). i.e., with less water flow, the filling would take a shorter time within 4 years. The above contradiction suggests that there are inconsistencies and probably methodological flaws in the modeled data of Wheeler et al (2020). The scientific community and policy makers should carefully consider this inconsistency, when using the results of Wheeler et al (2020). For more details on the technical concerns on Wheeler et al (2020), see part 2 of this document, which contains our point-by-point reply (part 2. section B6).Second, the claim of using high-quality peer-reviewed material is not exemplified in Wheeler et al with the misuse of the results of Deltares (2019)—a conference paper presented at the Cairo Water Week 2019 to support their claim that there is no downstream water deficit under average conditions as stated in Wheeler et al:

  A major study conducted by the consulting firm Deltares, for the Egyptian Ministry of Water Resources and Irrigation (MWRI), agreed with this finding and stated that no shortages in water releases from the AHD would occur under average hydrological conditions (Deltares, 2019).

  — Wheeler et al, p 4

  However, when reviewing the presentation—which is available online (www.youtube.com/watch?v=CXd4povRFkM, minutes 19–21, and 28–29)—and when carefully examining the presented results of Deltares (2019), we find that the study results are not consistent with the claims in Wheeler et al's comment. According to the study of Deltares (2019), the annual water availability in Egypt under average conditions, following the establishment of GERD (including its filling and long time operation for 115 years), will suffer an average cumulative deficit equal to 47 BCM over its 115 years life span. The Deltares (2019) study also concludes that for short-term filling scenarios of 1–7 years, and after mitigation with the AHD, a water deficit from GERD impoundment ranging from 6.5 to 17 BCM can still occur under average conditions (Deltares (2019); minutes 19–21), which contradicts both Wheeler et al's comment and Wheeler et al (2020) regarding the no-water shortage occurrance at AHD under average flow conditions. Hence, the claims in Wheeler et al's comment are not supported by Deltares (2019). It is important to note that both Heggy et al (2021) and Deltares (2019) agreed on the occurrence of a measurable water budget deficit during the filling of GERD under average conditions.Moreover, it is worthy to note that according to Deltares (2019), which is cited as a reliable study in Wheeler et al, every shortage of 1 BCM results in a reduction of agricultural land equal to 294 000 feddan, reducing agricultural production values by 430 million US dollars, increasing imports by 150 million US dollars, and resulting in the loss of income for 290 000 people (Deltares (2019), minutes 21–22). The above linear approach aligns with our assumptions of the economic calculations in Heggy et al (2021) that are clearly stated as:

  Yearly water budget loss and equivalent land loss are approximated using relationships from Mazzoni et al (2018) and Hamada (2017) respectively, where the loss of each 5 BCM would result in total degradation of ∼0.42 million hectares (∼1.2 million feddan).

  — Heggy et al (2021, p 6)

  The above suggests that once again Wheeler et al misinterpret Heggy et al (2021) and misuse the results of Deltares (2019) to support their claims.Finally, a newly published UNDP report (UNDP and MPED 2021) on the impact of the GERD on Egypt's water budget deficit aligns with our results in Heggy et al (2021) and contradicts Wheeler et al's claims that there are no studies that have comparable outcomes to Heggy et al (2021). The UNDP and MPED (2021) report states that:

  It is expected that the process of filling the dam's reservoir will seriously affect the availability of water in Egypt and reduce the country's per capita share, thus affecting various economic activities, especially if Ethiopia fills the dam's reservoir in a manner that Egypt deems 'uncooperative'. For example, if the filling process takes five years, as Ethiopia has said it will, the cumulative water shortage rate at the Aswan High Dam will increase to 92 billion cubic meters, distributed over several years. Likewise, the water level in Lake Nasser will soon drop to 147 meters, making it impossible to compensate for the water lost.

  — UNDP, MPED (2021, p 220)

• (b)  
  The second part of the comment on the economic impacts (pages 7–10 of Wheeler et al's comment) is based on the erroneous assumption presented above that the entire deficit in Heggy et al (2021) is attributed entirely to the GERD's filling. This claim is neither true nor reproduced in the text or figures of Heggy et al (2021) as demonstrated multiple times above. As such, the references cited to support Wheeler et al's claims (e.g. Block and Strzepek 2010, Goor et al 2010, Dinar and Nigatu 2013, Arjoon et al 2014, Geressu and Harou 2015, Satti et al 2015, Nigatu and Dinar 2016, Osman et al (2016), Wu et al 2016, Boehlert et al 2017, Jeuland et al 2017, Kahsay et al 2017, 2019) are inappropriate for this purpose, as they assumed a marginal to no downstream water budget deficit during GERD mostly long filling period and considered several mitigation scenarios, which is different from the approach in Heggy et al (2021) where the deficit was assessed for several filling periods (short, medium and longueur ones) and assumed no mitigations. Siddig et al (2021) agrees with the above assessment that there are different models with different approaches that analyzed the impacts of the initial reservoir filling and steady-state operation of the GERD on Sudan and Egypt where different mitigation scenarios and economic assumptions yield different results.The cited studies by Wheeler et al focus on calculating the economic impact associated only with the GERD's filling and assume mitigation plans based on the CGE adjustment model (Siddig et al 2021). The CGE model is an adjustment model that derives a solution (mitigation plans) based on an agreement (e.g. a AHD–GERD water release agreement, crop types and irrigation efficiency), which in turn relies on consumption from the only national strategic water reserve that Egypt can use to mitigate future extreme drought periods that may affect the area (Abd Ellah 2020, El-Saadawy et al 2020, Abdelmohsen et al 2022). Moreover, the CGE model in the cited papers only adjusts the deficits caused by GERD (not considering the intrinsic deficits and seepage). On the other hand, Heggy et al (2021) calculated the economic impact on GDP arising from the unmitigated total water budget deficit (intrinsic = 60% + GERD = 40%) under different filling scenarios, not only from GERD comparing to the studies cited in Wheeler et al. Furthermore, the source of funding required to achieve the mitigation of such water budget deficit is not discussed in the CGE models as stated in Osman et al (2016):

  The scenario does not specify the source of funding for the simulated improvements in irrigation efficiency, e.g. government expenditures on research and development are not explicitly specified.

  — Osman et al (2016)

  Additionally, the studies cited in Wheeler et al only assessed the impacts of water deficit caused by the GERD considering Egyptian mitigation measures in their calculations and did not include Egypt's intrinsic deficit as the CGE model's baseline assumes that the economy starts from a stable or equilibrium position (Chief Economist Directorate, Scottish Government 2016). In contrast, our paper Heggy et al (2021) aimed to estimate the equivalent economic impacts of the total water budget deficit in Egypt including the intrinsic deficit and considering no mitigation efforts are taken. Heggy et al (2021) did not attempt to cost the mitigations as the reliable data necessary for such efforts are not publicly available for the authors at the time of writing the paper. At present, the projected costs for the mitigation plan as announced by the Egyptian authorities (the National Water Resources Plan 2037) attain approximately ∼50 billion US dollars (Adly et al 2021, Al-Jazeera 2021), close to the value of the 51 billion US dollars of potential impacts as estimated in Heggy et al (2021) for the total water deficits of Egypt under the GERD filling period. For the above reasons, we instead chose to present the feasibility index of different mitigation measures in a separate section '3.2. Feasibility index' Heggy et al (2021, p 11), to be further used as a guide for the decision makers when putting the mitigations plans.When overlooking these technical details in Heggy et al (2021), it is no wonder that our results can be perceived as exaggerated when compared to those that only consider the potential water budget deficit arising from GERD filling scenarios under idealistic conditions and their associated mitigations scenarios (CGE model). While the cited studies in Wheeler et al's focus on few suggested filling scenarios that can minimize downstream impacts, but they are still neither agreed upon nor yet implemented and they do not consider the rising intrinsic water budget deficit of Egypt's nor its technological readiness to address the total intrinsic deficit as shown in Mazzoni et al (2018) and Nikiel and Eltahir (2021)).

In addition to the misinterpretation of the results in Heggy et al (2021) and the inappropriate comparisons with other studies with of different objectives and approaches, Wheeler et al's comment support their key claims with inaccurate citations. For example:

• (a)  
  Wheeler et al take a statement from Mazzoni et al (2018) out of context to support their claim that the incorporation of the intrinsic water deficit in Egypt is inappropriate, as Mazzoni et al (2018) according to them only aimed to estimate the magnitude of groundwater depletion. Wheeler et al states that:

  However, such mass balance models are not designed to understand short-term interactions between hydrology and infrastructure. In fact, Mazzoni et al (2018) state that 'the objective of our model is limited to the estimate of the overall magnitude of the groundwater volume depletion and its timescale for each major transboundary aquifer as a consequence of the simultaneous aggregate withdrawal of one or more riparian countries on a regional scale.' Ignoring this stated purpose, Heggy et al (2021) apply the same model to quantify deficiencies in surface water availability.

  — Wheeler et al, p 7

  In the above, Wheeler et al misuse a sentence from the methodology section of Mazzoni et al (2018) that describes the model for groundwater depletion in North Africa and the Arabian Peninsula and present this sentence as the main aim of the cited study. However, the scope of Mazzoni et al (2018) clearly stated :

  we establish a water budget model that combines country-level demographic, macroeconomic, water supply and demand data in order to quantify the water deficit volumes per country and the groundwater depletions rates, from 2016 to 2050.

  — Mazzoni et al (2018)

  Hence the water budget model developed in Mazzoni et al (2018) is applicable to the specific case of Egypt to assess the increase of the existing intrinsic total water budget deficit during the filling of GERD as stated in Heggy et al (2021):

  To address this deficiency, we estimate that the median total annual water budget deficit for Egypt during the filling period, considering seepage into the fractured rocks below and around the GERD reservoir, as well as the intrinsic water deficit and assuming no possible mitigation efforts by Egyptian authorities, will be ∼31 BCM yr⁻¹.

  — Heggy et al (2021), abstract

  Therefore, it is central to our analysis in Heggy et al (2021) to incorporate the intrinsic water deficit volume, which is calculated using a rigorous numerical model published in a top tier journal (Global Environmental Change IF = 11, Cited 28 times) with an outcome consistent with other published materials regarding the intrinsic water deficit in Egypt (Omar and Moussa 2016 and Bekhit 2019).
• (b)  
  Wheeler et al claims that the analyses of Kahsay et al (2017, 2019) and Boehlert et al (2017) of the impacts of the 3 year filling scenario of the GERD on Egyptian GDP per capita yield a contradictory value that is two orders of magnitude smaller than our values in Heggy et al (2021). Wheeler et al state that:

  Kahsay et al (2015), Kahsay et al (2017) and Boehlert et al (2017) all performed analyses of the impacts of 3 year filling scenarios for the GERD on Egyptian GDP per capita. Those papers report a range of decreases from 0.01% to 0.1% in GDP per capita, depending on upstream hydrological conditions and uncertainty related to Sudanese management responses. The most risk averse decision-makers might look to the high end of this range. We believe that results would likely be similar using the information plotted in figure 3. Heggy et al (2021) have overestimated the economic losses to Egypt by nearly two orders of magnitude.

  — Wheeler et al, p 10

  When consulting Kahsay et al (2017, 2019) and Boehlert et al (2017), we found that they only considered the impacts of GERD (and do not consider the intrinsic deficit) based on the CGE adjustment model (again considering several mitigations) as clearly explained above Furthermore, did not consider the 3 year filling scenario for the GERD on Egyptian GDP per capita as stated in Wheeler et al. For instance, Kahsay et al (2019) stated that:

  Table 5 shows the potential impacts of the GERD impounding on the AHD in terms of hydropower generation and irrigation water supply in Egypt for the three possible climatic and hydrological regimes and the Sudanese water withdrawal options. If it occurs in a sequence of average or wet years, GERD impounding is planned to be realized in 6 years (2014–2019).

  — Kahsay et al (2019, p 19)

  And

  In case it occurs during a sequence of dry years, GERD impounding will require one more year and will be realized in 7 years (2014–2020).

  — Kahsay et al (2019, p 19)

  It is worth mentioning that Wheeler et al relied primarily on those three cited papers, i.e. Kahsay et al (2017, 2019) and Boehlert et al (2017) to support their claims that the economic findings in Heggy et al (2021) are two orders of magnitude higher than other published research.Finally, it is important to note, that Heggy et al (2021) did not portray a grim image for all GERD filling scenarios as portrayed in Wheeler et al, instead we clearly stated based on our analysis that:

  Longer-filling scenarios above 7 years will allow a more manageable deficit below 10 BCM yr⁻¹ that can be mitigated using existing resources and minimize the associated socio-economic impacts.

  — Heggy et al (2021)

  Wheeler et al comment only focused on the deficit outcome of the three years filling scenarios, the shortest one reported in Heggy et al (2021), which includes no mitigation of the deficit and compared its outcome to other studies which consider longer filling periods with several mitigations of the deficit, which misdirect the reader and incorrectly suggest that the results in Heggy et al (2021) are exaggerated.

We note that the focus of Heggy et al (2021) is on potential impacts on Egypt resulting from filling the GERD Reservoir rather than assessing the regional benefits that the GERD might provide if cooperative agreements for operating the dam are reached, or the economic implications to Ethiopia of delaying hydropower generation.

— Wheeler et al, p 2

A.1: Response: the above claim is inaccurate. In Heggy et al (2021), the focus, as clearly described in the paper's, abstract and text body, is specifically assessing the evolution of the existing water budget deficit in Egypt during the different filling scenarios of the GERD. An extensive analysis of the regional benefits or downsides of the GERD are beyond the scope of our paper. With that said, Heggy et al (2021) clearly summarized the benefits of GERD for Egypt and Sudan in two paragraphs in page 3 section 'introduction' and in page 15 section 'implications and mitigation strategies'.:

B.1: Response: We take note that the main concern of Wheeler et al is our assessment of the water deficit, which is later subdivided into three questions (addressed in the next three replies). It is important to note that the primary concern of Wheeler et al is based on misunderstanding of our methods and results as well as misinterpreting the supporting peer-reviewed references. Wheeler et al claim that Heggy et al (2021) did not clearly explain the method of calculating the water deficit and they stated that:

It is important to note that the published article does not clearly explain how the deficit predictions were calculated; we therefore requested access to the authors' analysis, a request to which they responded positively. On reviewing the spreadsheet analysis provided, it became clear that a model of the Nile River and reservoir systems was not used to make predictions of potential GERD-induced deficits. Rather, Heggy et al (2021) selectively used results from previous studies, including several conducted by authors of this response.

— Wheeler et al, p 3

While, in Heggy et al (2021), we clearly stated that:

In the present study, we evaluate the effects of the GERD on the short-term Egyptian availability of the Nile streamflow by considering different filling scenarios and strategies simulated in the most recent published literature Then, we assess Egypt's total water budget deficit arising from the GERD filling scenarios, seepage and the intrinsic one resulting from population growth. We evaluate the feasibilities of the suggest mitigation policies to address the resulting deficit and discuss the socioeconomic impacts considering each of the proposed filling scenarios for the GERD.

— Heggy et al (2021, p 4)

As stated above, our paper used existing published results of the deficits from several models of the Nile River and reservoir systems available at the time of preparation of the manuscript to assess the deficit under multiple scenarios as shown in table 1 and discussed thoroughly in page 4 and 5 of the original paper. Heggy et al (2021) did not attempt to model the altered Nile flow but rather performed statistical analysis of existing ones as summarized in table 1 and shown in figure 1 of Heggy et al (2021).

Moreover, Wheeler et al do not explain which values we selected and which ones we omitted at the time of acceptance of the manuscript, and how these would have changed the results in Heggy et al (2021).

B.2: Response: As explained above, our paper estimated the total water budget deficit for Egypt in BCM/year including its rising intrinsic deficit and the one that can arise from all GERD filling scenarios from 2.5 to 29.6 years considering the different published filling scenarios for the GERD (King and Block 2014, Zhang et al 2015, 2016, Wheeler et al 2016, Keith et al 2017, Liersch et al 2017, Donia and Negm 2018, Omran and Negm 2018). These published scenarios used in our analysis already considered the most critical element of the Nile systemt.

We also clearly illustrated the role of AHD in the mitigation of the impacts of the total water deficits of Egypt budget for instance in section 3.2 in Heggy et al (2021) in the paragraph:

Inspection of the feasibility index estimations (figure 3) indicates that among all the proposed solutions to address the water budget deficit, the most practical option (feasibility index >50%) for Egypt in the short term would be to temporarily reduce or even completely suspend the operations at the AHD to partially offset the lower streamflow during the filling of the GERD.

— Heggy et al (2021, p 11)

Wheeler et al claims that our analysis omitted the role of the AHD reservoir as stated:

Nonetheless, a key assumption in all of the authors' calculations is that the quantity of water retained by the GERD Reservoir during filling translates exactly into a deficit in supplies to water users downstream of the AHD in Egypt. This ignores the storage that exists in the AHD reservoir and all of the other natural and human modifications to the Nile flowing between the GERD and the AHD.

— Wheeler et al, p 2

They also stated that:

The deficits estimated by Heggy et al (2021), however, assume that this current unusually large amount of storage would not be used to maintain Egypt's water supply during filling of the GERD Reservoir.

— Wheeler et al, p 3

B3. Response: This is a key misunderstanding of our results. On the contrary, the normal role of the AHD in mitigating the impacts of water shortages in our study was clearly highlighted in our results and was identified as the most feasible solutions for mitigating the adverse impacts of the GERD filling (section 3.2: Feasibility index; and figure 3). We stated that:

...among all the proposed solutions to address the water budget deficit, the most practical option (feasibility index >50%) for Egypt in the short term would be to temporarily reduce or even completely suspend the operations at the AHD to partially offset the lower streamflow during the filling of the GERD.— Heggy et al (2021)

Our rationale to avoid including a specific volume of the water reserves in the AHD that can be used to mitigate the deficit from the GERD filling, is the large uncertainty surrounding the time scale of the filling scenarios and whether they are followed with dry period or not which remains a highly debated topic. It is important to note that the water reserve in the AHD reservoir is the only available strategic water reserve for Egypt during drought periods (Abd Ellah 2020). Hence using it to compensate the deficit from GERD filling must carefully consider all risks on longer time periods accounting for a succession of wet, average, and dry periods and not to downplay them by assessing them over short periods of 5–7 years as portrayed in Wheeler et al.

Moreover, the declaration of principles signed by Ethiopia, Egypt, and Sudan in 2015; in its 4th statement on the principle of fair and appropriate use, clearly states: 'To ensure fair and appropriate use, the three countries will take into consideration social and economic needs for the concerned Nile Basin countries' (Ahram 2015). Therefore, our methodology support a better implementation of the 4th statement of the declaration of principles between the Nile riparian countries, and thus help decision makers to be aware of the water needs for a better social and economic evaluation in Egypt.

B.4: Wheeler et al stated that:

The estimate of the median incorrectly assumes that various scenarios that Heggy et al (2021) selected from the literature are all equally likely, even though they come from studies that analyze different filling policies.

— Wheeler et al, p 3

B.4 Response: We did not assume that all scenarios are equally likely. In fact, we used the Probability Distribution Function (PDF) (figure 1 of Heggy et al 2021) to weigh the different scenarios according to their probability of occurrence in all adopted models. For example, the five years filling scenario, which were frequently suggested in the filling simulations, obtained a PDF value of 0.15. On the other hand, the other filling scenarios such as the three, fifteen- and twenty-years' scenarios were assigned much lower PDF values of less than 0.05 (figure 1 in Heggy et al 2021). Accordingly, the median estimation did not consider all scenarios as equally likely as mistakenly interpreted in Wheeler et al's above claim.

B.5: Wheeler et al claims that we neglected the current high levels of the AHD as stated that:

However, as -the AHD Reservoir is now nearly full with 110 BCM of active storage, the AHD Reservoir could fully compensate for the water required to completely fill the remaining space in the GERD Reservoir, even without considering other mitigating aspects that we describe below.

— Wheeler et al, p 3

B.5 Response: This claim ignores the hydrological and political complexity of the Nile River system and the unilateral filling operations. Wheeler et al considered the initial levels of the AHD at 179.1 m and their results show that the filling of the GERD would not have significant adverse impacts downstream on Egypt. Yet, neither their comment nor their analysis considered the following:

• The suggestion of using the current high level of the AHD would be reasonable if there is an agreement that guarantees cooperative filling and management scenarios of the GERD with clear framework and timetable of the filling period and the amount of water impoundment. An agreement that does not yet exist. Therefore, there is no assurance that the filling will be completed, while there are high levels of the AHD, given the extremely high annual evaporative (10% of total volume; Elsawwaf et al (2014) and seepage losses (i.e. 6 BCM during high stands; Sultan et al 2013, Abdelmohsen et al 2019, 2020) from the AHD as well as the difficulty to predict the drought and flooding events in the Nile River basin, which is characterized by 'widely divergent climate change forecasts; Wheeler et al (2020)'.
• For example, Ethiopian authorities originally planned to fill the GERD reservoir with 13.5 BCM in 2021 (Wheeler et al 2020) raising its level, by the completion of the second filling on 19 July 2021, to 595 m above mean sea level (amsl). This was not the case; the reservoir only attained a water level of 580 m amsl as indicated by the altimetry data (https://blueice.gsfc.nasa.gov/gwm/lake/1296), which is equivalent to a much lower water volume than planned by the Ethiopian authorities. Therefore, in case of unilateral filling of the GERD, it is unrealistic to depend on assumptions of initial AHD levels, where there is no strict plan to fill the GERD reservoir, while the AHD is still at high stands. This was the primary motivation of using different scenarios in our analysis in Heggy et al (2021). Furthermore, the high level of the AHD reservoir is not neglected in our study. A scenario of initial high level of the AHD (i.e. 180 m), which was used in Wheeler et al (2016), has been already incorporated in our analysis (see table 1 in Heggy et al 2021).
• Moreover, the AHD reservoir is designed to divert any excess of water above the level of 178 m (Eldardiry and Hossain 2021) into a free spillway towards Tushka depressions at a maximum daily discharge rate of 250 MCM d⁻¹ (Donia and Negm 2018). Given that the Tushka depressions occur at topographic levels below the AHD (i.e. 140–153 m above mean sea level), it is unrealistic to consider lifting the water back to the AHD. Therefore, any AHD water level above the 178 m level should be considered with caution and not taken for granted.

Wheeler et al introduce an updated model figures 1–3 (Wheeler et al, pp 3–6) and they suggest that this model challenges our deficit calculations, which is based on a statistical analysis from peer-reviewed published papers.

B.6 Response: We take note that all the claims as well as the new model (figures 1–3 in Wheeler et al) discussed in this section (pp 3–6) are mainly based on Wheeler et al (2020), as stated in Wheeler et al:

To contrast the resulting differences when using a rigorous stochastic approach, we have updated the model and methods of Wheeler et al (2018) and Wheeler et al (2020) to reflect the reservoir storage conditions that were observed in January 2021 and Ethiopia's current filling plan for the GERD.

— Wheeler et al, p 4

In addition to the fact that the comparison of Wheeler et al (2020) is inappropriate to our study as it assesses only the impact caused by the 5 year GERD filling scenario considering mitigation (assuming an AHD and GERD joint operation agreement) while our paper Heggy et al (2021) assessed the unmitigated total water budget of Egypt (intrinsic + seepage + GERD), we have several concerns on the simulation models of Wheeler et al (2020), which are summarized as follows:

• (a)  
  the simulation results indicated that the GERD filling under wet conditions (i.e. inflow > 90 BCM) will be completed in 5 years as shown in figure 3 in Wheeler et al (2020), while under near-average conditions (i.e. inflow of ∼86 BCM; figure 4 in Wheeler et al 2020) the filling takes only 4 years. This simple observation can adversely challenge the accuracy of their simulation and denotes a methodological flaw in their analysis and the selection of a single unrepresentative historical sequence in the simulation of each flow condition (i.e. wet, average and dry). It is important to note that this flaw may be the main reason that makes Wheeler et al (2020) not aligned with other published papers.
• (b)  
  The reason for the above discrepancy is the high interannual flow varibility which is omitted in Wheeler et al (2020). For instance, the inflow during year 1946, which is assumed to be the fourth year of the filling during the selected average flow period (1943–1952), witnessed an anomalous flood (∼106.7 BCM) that is more than 20 BCM above the average of the selected period (86.08 BCM). However, Wheeler et al (2020) claim that they avoid in their selection criteria anomalous events such as single-year floods or droughts during the initial filling years, as quoted below:

  This selection used both parametric and non-parametric criteria, including a statistical analysis of periods of flows along with hand selection to avoid anomalous events such as single-year floods or droughts.

  — Wheeler et al (2020, supplementary material)

  Hence, the selected average period (1943–1952) does not meet the criteria defined by Wheeler et al (2020) and accordingly the deficit estimate is erroneous. Furthermore, the estimate of the deficit under the above average period is not reproduced when considering the other near-average periods identified in their study. The latter suggest that Wheeler et al (2020) have methodological flaws that compromise its results comparison with Heggy et al (2021).
• (c)  
  The value of the AHD seepage, which was quantified as 6 BCM during high stands (Sultan et al 2013, Abdelmohsen et al 2019) and the GERD initial filling seepage, which according to Abuzaied (2021) pessimistic and optimistic scenarios ranges from 4 to 20 BCM, respectively, have not been clearly considered in Wheeler et al (2020). We believe that if these amounts of seepages are added to the simulation, they will change the results and show a decrease in the AHD storage different from that of the simulated outcomes in Wheeler et al (2020) and thus the results may align with other published results we used in Heggy et al (2021).
• (d)  
  Inflating the High Aswan Dam storage in an additional step to minimize the water deficit: We noted the use of an exaggerated values for the active and total storage of the AHD in the model as shown in table 1 of Wheeler et al (2020, supplementary data, p 16) where the paper used 148 BCM and 182 BCM to represent both active and total storage, respectively, which is not consistent with the widely accepted storage values of 87.2 BCM for the active storage and 160 BCM for the total one (Moussa 2012). These large differences call into question the output results from the simulation model of Wheeler et al (2020). It is important to note that even though Wheeler et al (2020) stated the proper values of the active and total storage (Wheeler et al 2020, supplementary data, p 3 (line 71–75)), they used the wrong ones in their model entries as stated in table 1 in Wheeler et al (2020) of the paper supplementary material.

Therefore, the claim of Wheeler et al that our analysis fails to meet the results in Wheeler et al (2020) is unreasonable considering the raised concerns on the methodological flaws in their analysis.

Wheeler et al claim that we used highly exaggerated seepage losses in our estimates of the water budget deficit (i.e. 15 BCM yr⁻¹) and they assumed, with no reliable or verifiable evidence, that we built our estimates on a newspaper article claim by Noureddin (2013). They stated that:

The high seepage rate [0.1%] scenario was included because Noureddin (2013) assumes that, due to rock formations, seepage losses may amount to 25% of the actual storage volume.

— Wheeler et al, p 6

B.7. Response: No citation to Noureddin (2013) exists in our paper. As for the value of 15 BCM per year for the seepage, Wheeler et al overlooked our detailed explanation of the data we used in our statistical analysis. In the method section, we clearly stated that:

A special emphasis is given to the water loss from the GERD due to seepage due to two reasons. First, the seepage losses have been largely underestimated in previous studies that dealt with evaluating the impact of the GERD and second, the seepage-related losses can reach annually up to 15 BCM compared to a much lesser annual evaporation related losses of only 3.8 BCM (Liersch et al 2017).

— Heggy et al (2021)

The above is a direct citation of a value estimated in a peer-review published article on the subject matter assuming the total inflow to range between 43.2 and 53.3 BCM and seepage rates of 1% and 32% (Liersch et al 2017). If Wheeler et al examined our results carefully, they would find that the 15 BCM seepage rate was considered to be an outlier and thus it was not involved in the paper's statistical analysis. In the results (section 3.1 and figure 2) we clearly stated that:

The decrease in streamflow for Egypt due to the seepage losses in the GERD, not considering the outliers, could range from a minimum of 43 MCM yr⁻¹ to a maximum of 10.4 BCM yr⁻¹ (median: 2.5, Q1: 0.185, and Q3: 4.9 BCM yr⁻¹)

— Heggy et al (2021)

and Heggy et al (2021) only used the median value of 2.5 BCM as clearly shown in figure 1 in our estimates to represent the annual seepage rate from the GERD. It is important to note here that the 2.5 BCM value can be considered as an optimistic lower-end estimate of the annual seepage rate when compared to the simulated optimistic (4 BCM) and pessimistic (20 BCM) scenarios of the initial filling seepage from the GERD at 74 BCM as recently published by AbuZeid (2021).

Wheeler et al claims that the incorporation of the mass balance model from (Mazzoni et al 2018) that was used to estimate the 'intrinsic' deficit is inappropriate.

B.8 Response: We responded earlier in details to this claim, see Part 1-point 3a).

Moreover, the aim of our study is to assess the total water budget deficit in Egypt, which involves the intrinsic water deficit as well as the water shortage associated with the GERD filling. Therefore, it is central to our analysis to incorporate these data, which is consistent with other published materials regarding the intrinsic water deficit in Egypt (Omar and Moussa 2016, APS 2018, Bekhit 2019).

They then add this figure to their estimates of the downstream impacts of the GERD filling (which, as we have argued above we believe to also be wrong) to arrive at a projected deficit of 'an average +35.5% of its current internal annual water demand'. However, this 'deficit' bears no relationship to the probability of water shortages occurring during the filling of the GERD Reservoir, which as we and others have calculated, is unlikely.

— Wheeler et al, p 7

B.9 Response: We clearly discriminate between the two figures throughout the results as well as the figures in our study (Heggy et al 2021). In figure 2 in Heggy et al (2021) we display individually each component of the water budget deficit in Egypt and the caption states that:

The left box (GERD) refers to the annual water deficit induced by water impounding in the GERD reservoir during the filling period (numbers are statistically derived from available scenarios in table 1. The middle box (Seepage) refers to the annual water deficit resulting from seepage losses in the GERD reservoir after Liersch et al (2017). The right box refers to the current intrinsic water deficit in Egypt after Mazzoni et al (2018).

— figure 2 caption, Heggy et al (2021)

C. Response: Wheeler et al raised concerns regarding the socioeconomic assessment of the impact of the GERD filling on Egypt. Their concern is mainly based on their assertion that we exaggerated the water deficit in Egypt and that it is totally attributed to GERD. They also suggest that the complexity of the Egyptian economy is not considered in estimating these socioeconomic impacts in Heggy et al (2021).

We demonstrated in our responses to questions 1 of Wheeler et al, that they misinterpreted the key findings of our study confusing the total unmitigated water budget deficit for Egypt to the null or minimal one solely caused by GERD under the five-year filling scenario after AHD mitigation as in Wheeler et al (2020). The above explain the discrepancies between the findings in Heggy et al (2021) and the references cited in Wheeler et al. In Heggy et al, we clearly mentioned that our study (2021) provides a simplified model to assess these impacts only in the agricultural sector that is translated into economic figures assuming no mitigations, to estimate the cost of the total water budget deficit on the economy. We stated in section 2.4. (Modeling the economic impact) that:

We use a simplified economic model to assess the impacts of the total water budget deficit under the different GERD's filling scenarios.

— Heggy et al (2021)

We also assigned a whole paragraph (35 lines) at the end of section 3.3 (GDP and unemployment projections) to discuss the model limitations. Finally, we did not claim that the model outputs provide a comprehensive assessment of the full economic impact of the total water deficit in Egypt for all economic sectors during the filling period of the GERD. On the contrary, we show that the intrinsic water budget deficit, combined with the filling of the GERD, can generate a water deficit equivalent to high losses of present cultivated area assuming no mitigation, as stated in our abstract and throughout the text for example:

In the years 2022, 2023 and 2024, our first order model suggests an equivalent agricultural GDP losses arising from the unmitigated total water budget deficit, for the total filling period, of ∼$51 billion for the 3 years filling scenario, ∼$28 billion for the 5 years filling scenario, ∼$17 billion for the 10 years filling scenario, and ∼$10 billion for the 21 year filling scenario.

— Heggy et al (2021, p 13)

C1: Question 2a: How did Heggy et al (2021) estimate the economic losses of water deficits to Egypt?

Wheeler et al claims that our estimations of economic losses associated with water deficits to loss of productive agricultural land and agricultural land area to agricultural GDP by a linear function and argue with this relationship between water deficits and agricultural GDP stating that:

The main problem with the assumption of a linear relationship between water deficits and agricultural GDP is that it does not represent how the Egyptian economy makes adjustments to reduce the economic consequences of deficits in inputs to production, in this case water. The value of water supply is best represented by its marginal product in its various uses, not by the average value of GDP from a sector that uses water as one of many inputs (Young et al 2014).

— Wheeler et al, p 8

C.1 response: The irrigation in the agriculture sector is the most important socio-economic sector in Egypt. There is about 9.176 million feddans of agriculture lands in Egypt that require about 63.51 BCM/year for irrigation and thus the agriculture sector consumes about 85% of the total demand for water (MWRI 2005, Abdelhaleem and Helal 2015, Deltares 2019). Therefore, the agricultural sector mainly consumes the bulk of the available water supplies (Badawy 2014) compared to industrial sector that only needs 5.4 BCM yr⁻¹ (Fanak 2018). For this reason, we translate that total water budget deficit into an equivalent (and not the actual) cost of total land degradation and agricultural employment loss. The same approach was adopted in Deltares (2019, minutes 21–22) study, which is cited in this comment by Wheeler et al as reliable study from the major consulting firm Deltares. Moreover, we carefully use the word equivalent throughout the text to show that this is a translation of the deficit into economic figures and is not the actual. Adding to the aforementioned reasons, we clearly justified the lack of data input from other sectors such as industry into the model, which was considered as a limitation of the model, and we attributed this deficiency to the lack of reliable input data. We stated that:

This model presents a first-order analysis providing a lower limit estimate of the equivalent economic impact of the total water budget deficit on the agricultural sector due to loss in crop productivity arising from land degradation from lack of irrigation) and not accounting for the consequences on the industrial sector that are difficult to assess due to the lack of published records.

— Heggy et al (2021)

C.2: Question 2b: Did Heggy et al (2021) appropriately apply the 'simple economic calculation' to estimate the economic losses to Egypt of filling the GERD Reservoir?

C.2.1: The unrealistic deficit

In this section, Wheeler et al concern is mainly based on their assertion that our estimation of the water deficits is wrong, as Wheeler et al stated that:

As we have demonstrated above, Heggy et al's (2021) estimate of median deficits relative to expected releases from the AHD Reservoir is wrong. As the economic loss is a function of the deficit, so too will the economic loss estimates be wrong.

— Wheeler et al, p 8

C.2.1 Response: We demonstrated in our above responses that Wheeler et al misinterpret our deficit results. Therefore, their assessment of the validity of our economic losses is compromised with this fact resulting in several inappropriate comparisons and questions made throughout the following questions as both are related as can be seen from their statement above.

C.2.2: Impacts on agricultural GDP and employment

Wheeler et al stated that the industry connected to agriculture (e.g. crop production, livestock production and fisheries) should be included in the model.

C.2.2 Response: The statement suggested by Wheeler et al is already existed in our text as we stated that:

Other service industries connected to industry and agriculture are likely to experience significant consequences because of decreased production capabilities. As such, the results on the equivalent decrease in the GDP per capita and the rise of unemployment should be considered as lower limits, the economic damage considering all the above factors will be higher.

— Heggy et al (2021, p 14)

As iterated several times in this reply and the original text, Heggy et al (2021) provided a first order simple economic impact based on the agricultural sector only as it is the main sector consuming water resources and due to the large uncertainties related to data availabilities for assessing the other sectors.

C.2.3: Results reported in the text are inaccurate and do not match Heggy's supporting materials.

Wheeler et al claims that our economic impacts number (∼$51 billion for the 3 years filling scenario, ∼$28 billion for the 5 years filling scenario, ∼$17 billion for the 10 years filling scenario, and ∼$10 billion for the 21 year filling scenario) do not match our calculations in the spreadsheet and figure 4.

C.2.3 Response: In our spreadsheet entitled 'Egy GERD Model AG book' provided to Dr Wheeler, the losses under the 3 years is $16.08 B for year1, $17.08 B for year 2 and $18.6 B in year 3 as shown in table 2 below. Summing the total of these three years is 50.74 B = ∼$51 B. Below is the screen capture of the spreadsheets and associated numbers., Wheeler et al, did not sum the numbers for the three years filling period.

C.3: Question 2 c: Did Heggy et al (2021) have to rely on such a simple economic calculation to estimate the economic losses to Egypt of filling the GERD?

Wheeler et al noted that our paper, Heggy et al (2021), did not mention or compare the economic results with several published papers based on the CGE model as they stated that:

Robinson, and Gelhar (1995), Lofgren et al (1996), Yates and Strzepek (1996), Osman et al (2016), Kahsay et al (2015) have all developed CGE models of Egypt that explicitly include water as an input to production activities. Again, none of these are referenced in Heggy et al (2021) or used for comparison to their results.

— Wheeler et al, p 9

C.3 Response: As we illustrated above in part 1, the CGE model is an adjustment model as the model derives a solution (mitigations plan based on an agreement) by finding a new set of prices and allocation of goods and factors such that the economy is in an equilibrium (Chief Economist Directorate, Scottish Government 2016). Also, the CGE model in the cited paper only adjust the deficits caused by GERD (not consider the intrinsic deficits and seepage) as the CGE baseline assumes that the economy starts from a stable or equilibrium position (Chief Economist Directorate, Scottish Government 2016).

The above comparison is inappropriate to our study Heggy et al (2021), as we did not assume any mitigation (because there is no any agreement yet). We simply cost the total water budget deficit (intrinsic + seepage + GERD) under this assumption as the cost and plans for mitigations remains largely unquantified, disputable with no available public data to properly perform this comparison. Hence the two models are incomparable in their purpose of use and assumptions.

Furthermore, as iterated above in section 3 of this reply that the cited papers in Wheeler et al only focused on the impacts of the GERD filling and do not include the intrinsic deficit and considers different mitigations scenarios for addressing potential GERD-caused water deficit. In the suggested references for comparison by Wheeler et al, Osman et al (2016) and Robinson et al (2008) and others, CGE models are presented as numerically solvable simulation aimed to provide suggest adjustment to address the deficit with measures as the AHD water release, crop types and irrigation efficiency, which either will consume the strategic water reserve storage that protect Egypt from future drought periods, or will require costly measures that their source of funding has yet to be defined as clearly stated in Osman et al (2016) when referring to CGE adjustment (mitigation) scenarios:

The scenario does not specify the source of funding for the simulated improvements in irrigation efficiency, e.g., government expenditures on research and development are not explicitly specified.

— Osman et al (2016)

Moreover, the scenarios based on the CGE models, have limitations as Osman et al (2016) stated that:

The scenarios do not explore the extent to which the Egyptian government can use the capacities of the Nile dams' reservoirs to adjust the seasonality of Nile water availability so as to influence structural changes in cropping decisions across seasons. In the longer term, the Egyptian government's ability to take such action will be limited to its control over the Aswan dams, so the option of collaboration with Ethiopia and Sudan during the period when the GERD's lake is filling may be fruitful. Similarly, the analysis does not take account of the extent to which the cost effectiveness of improvements in irrigation efficiencies may differ across crops and hence the potential benefits of targeting.

— Osman et al (2016)

To avoid all these discrepancies that are beyond the scope of Heggy et al (2021), we studied the impact of unmitigated total water budget deficits for Egypt during the different filling scenarios of the GERD and discussed the feasibility of the mitigations in a separate section 'feasibility index' according to their cost, time, and availability (For more details, see section 3 and introduction above) to be used as a guide for the decision makers to put the best mitigation plans.

For the above reasons, we did not discuss the above-mentioned literature as they are irrelevant to the scope of our study. The socioeconomic part of Heggy et al (2021) aimed at providing a first order costing of the size of the total water budget deficit for Egypt including its intrinsic deficit and the one caused by GERD, which should provide a first order estimate of the needed resources to address it. Our findings assess the cost of the total deficit to be equivalent to 51 B$, a value that is close to the 50–57 B$ recently announced by Egyptian authorities National Water Research Plan for 2037 (Adly et al 2021, Al-Jazeera 2021).

C.4: Wheeler et al claims that the CGE models are all fully demonstrated nonlinear relations which challenge our results as they stated that:

They also demonstrate substitution of fossil and renewable electricity to compensate for the losses of hydropower production from the AHD. These CGE models all demonstrate that the relationship between water releases from the AHD and Egyptian economic output (GDP) is highly nonlinear, not linear as Heggy et al assumed.

— Wheeler et al, p 9

C.4 Response: It is understandable that a full economic assessment cannot be made with one linear function but one segment of it can be assumed linear as a first order approximation as performed in Heggy et al (2021) relating the water and land irrigation constrains. In Osman et al (2016); cited by Wheeler et al to support their claims the CGE models not fully demonstrated by nonlinear relations the paper stated:

The effect of fiscal reforms, in the form of removing subsidies and taxes, were examined subject to physical supply constraints for both water and land (Robinson and Gehlhar 1995a). The first-order conditions for water and land constraints are given by linear cost functions with an explicit maxim and, which ensured that at least one of the two constraints is binding.

— Osman et al (2016)

Further Osman et al (2016) state:

Also, the models which CGE based on is using a linear programming problem for each sector.

— Osman et al (2016)

Additionally, Robinson et al (2008), also cited by Wheeler et al states:

'In effect, the model specifies the choice of land-water combinations by crop as a linear programming problem' and 'Given the solution to each sectoral linear programming problem, the model generates a land-water aggregate for each agricultural sector that is then assumed to enter a neoclassical production function along with labour and capital.'

— Robinson et al (2008)

It is hence appropriate as a first order approximation, as stated in the assumptions in Heggy et al (2021) to use a linear model to cost the impact on the agriculture sector that is the main consumer of water resources in Egypt (Abdelhaleem and Helal 2015, Deltares 2019). The same approach has been used in Deltares (2019) cited by Wheeler et al here in a reliable study.

C.5: Question 2d: If Heggy et al (2021) AHD used an appropriate economic model of the relationship between water and economic activity, would their conclusions have been different?

Wheeler et al claims that Heggy et al (2021) mistakenly calculate the economic impact of the three-year GERD filling scenario on Egypt and that the assessed 6% decrease in GDP per capita per year is nearly two orders of magnitude comparing to those published by Kahsay et al (2019), Kahsay et al (2017) and Boehlert et al (2017) as stated:

Kahsay et al (2015), Kahsay et al (2017) and Boehlert et al (2017) all performed analyses of the impacts of 3-year filling scenarios for the GERD on Egyptian GDP per capita. Those papers report a range of decreases from 0.01% to 0.1% in GDP per capita, depending on upstream hydrological conditions and uncertainty related to Sudanese management responses. The most risk averse decision-makers might look to the high end of this range. We believe that results would likely be similar using the information plotted in figure 3. Heggy et al (2021) have overestimated the economic losses to Egypt by nearly two orders of magnitude.

— Wheeler et al, p 10

C.5 Response: First, the cited papers mentioned by Wheeler et al are based on the CGE model, which as demonstrated above, are inappropriate to compare with Heggy et al (2021) results.

Second, it is important to note that the cited papers (Kahsay et al (2019), Kahsay et al (2017) and Boehlert et al (2017)) do not examine the impact of the 3 year filling scenario for the GERD on the Egyptian GDP per capita but rather they considered longer periods of 6 and 7 years contradicting the claims above in the comment. See our response to point 3b in part 1 for details.