Investigating Climate Change Effects on Evapotranspiration and Groundwater Recharge of the Nile Delta Aquifer, Egypt

: Climate change (CC) directly affects crops’ growth stages or level of maturity, solar radiation, humidity, temperature, and wind speed, and thus crop evapotranspiration ( ET c ). Increased crop ET c shifts the fraction of discharge from groundwater aquifers, while long-term shifts in discharge can change the groundwater level and, subsequently, aquifer storage. The long-term effect of CC on the groundwater ﬂow under different values of ET c was assessed for the Nile Delta aquifer (NDA) in Egypt. To quantify such impacts, numerical modeling using MODFLOW was set up to simulate the groundwater ﬂow and differences in groundwater levels in the long term in the years 2030, 2050, and 2070. The model was initially calibrated against the hydraulic conductivity of the aquifer layers of the groundwater levels in the year 2008 from 60 observation wells throughout the study area. Then, it was validated with the current groundwater levels using an independent set of data (23 points), obtaining a very good agreement between the calculated and observed heads. The results showed that the combination of solar radiation, vapor pressure deﬁcit, and humidity (H) are the best variables for predicting ET c in Nile Delta zones (north, middle, and south). ET c among the whole Nile Delta will increase by 11.2, 15.0, and 19.0% for the years 2030, 2050, and 2070, respectively. Zone budget analysis revealed that the increase of ET c will decrease the inﬂow and the groundwater head difference (GWHD). Recharge of the aquifer will be decreased by 19.74, 27.16, and 36.84% in 2030, 2050, and 2070, respectively. The GWHD will record 0.95 m, 1.05 m, and 1.40 m in 2030, 2050, and 2070, respectively when considering the increase of ET c . This reduction will lead to a slight decline in the storage of the Nile Delta groundwater aquifer. Our ﬁndings support the decision of the designers and the policymakers to guarantee a long-term sustainable management plan of the groundwater for the NDA and deltas with similar climate conditions.


Introduction
Water resources are one of the most important natural resources worldwide [1,2]; sustainable management plans are essential for ensuring the long-term availability of these sources, especially in arid and semi-arid regions [3][4][5][6].The agriculture sector is the highest water consumer.Therefore, successful irrigation management practices should consider the rational use of water alongside the maximum water use efficiency, high yield of crops, and return profit [7][8][9].The source of irrigation water can be surface water or groundwater.The quality of irrigation water should be considered when looking for sustainable agricultural practices [10].Knowing exactly how much water is needed to fulfill crop requirements helps growers to utilize the irrigation water more efficiently with minimal deterioration Water 2023, 15, 572 2 of 16 to water sources [11].Thus, the accurate estimation of crop evapotranspiration (ET c ) is crucial for the operation and designing of irrigation systems, irrigation scheduling, and entire management plans of irrigation water [12].
Evapotranspiration is a process governed by the energy and heat exchanges at the land surface, which is controlled by the amount of available energy on the ground [13].Additionally, it reflects the crop water requirements during the growing season and how much water is required for plants [14].The estimation of spatially distributed crop water consumption is challenging but important to determine water balance at different scales to promote the efficient management of water resources [15].Additionally, ET c from irrigated agriculture (e.g., the case of the Nile Delta of Egypt) is an important concern in arid and semi-arid regions, where it has a substantial impact on groundwater resource depletion and water management plans [16,17].
Moreover, ET c employs a key role in the watershed budget analysis [18], and a wide range of techniques have been developed to precisely estimate its spatial and temporal variability at site-specific locations.Moreover, it is generally considered in the hydrological modeling of surface water [19] by integrating empirical formulas, such as the Penman-Monteith equation [20], based on climate variables including temperature, solar radiation, wind speed, and humidity.The estimation of ET c is most important in the hydrological modeling of surface water, where it is not directly included in groundwater modeling [21].ET c is neglected in groundwater modeling when aquifers are formed by coarse-grained sediments where the capillary rise is very low [22].For leaky aquifers such as the NDA, ET c must be considered by subtracting from precipitation and irrigation water to calculate the input recharge to the groundwater flow models [23], especially when the water table is relatively shallow [24].
There are several ways to calculate the actual ET c , but surface energy balance methods [25] and surface water balance methods [25,26] are mainly used.Rana and Katerji [27] provided a summary of 10 methods for measuring and calculating actual ET c .The approaches were classified as hydrological, micrometeorological, and plant-physiological, depending on their use.Every approach has advantages and disadvantages.Xu and Chen [28] showed that a weighted lysimeter, for example, can offer precise information on water balance; nevertheless, it is difficult to estimate the ET c across large regions for an extended length of time.As a result, actual ET c is typically calculated using less sophisticated physically based or empirical methods.According to Linacre [29], actual ET c may be estimated using a link between accessible soil water content and evaporation rate.Actual ET c was estimated using two methods: the first method is the ET c model developed by Han and Tian [30], and the second method is the soil water balance model, in which the actual ET c is estimated by the fraction of potential ET c ; this fraction increases as soil water content increases.
Egypt faces several serious risks from CC, and the Nile Delta (ND) is now seriously suffering from these risks, such as rising sea level, saltwater intrusion, salinization, higher temperatures, and land subsidence [31].The impacts of these effects on surface water and groundwater should be carefully addressed and the adaptation scenarios for mitigation must be prepared [32,33].The agriculture sector has been negatively influenced by higher temperatures for lower crop productivity and the loss of some agricultural lands in the ND [34], where the productivity of crops was estimated to decrease by 1 to 17%.Moreover, climatic changes have economic and environmental impacts which should be considered to ensure food security [35].
The groundwater recharge in the NDA is mainly sourced from irrigation water, inflow from the two branches of the Nile River, Damietta and Rosetta, and rainfall along the northern coast [36].Groundwater recharge depends on the rainfall duration and intensity, evapotranspiration rates, infiltration rates, soil moisture, aquifer permeability, and groundwater table depths [37].Moreover, ET c and soil moisture depletion depend on climatic parameters such as temperature, relative humidity, wind speed, and sunshine hours.The climatic parameters should be considered for estimating the ET c and soil moisture depletion pattern to investigate the influence of CC on groundwater recharge [38].While flood irrigation is traditionally practiced in the ND of Egypt and the return flow from irrigation water is the main component of groundwater recharge for the NDA, accurate predicting of ET c is essential to quantify how much water will recharge the aquifer in the long term when the same irrigation method is still applied.
To date, there has been no accurate estimation of actual ET c where assumptions and uncertainty of CC patterns exist.Based on the hydrological cycle, Milly and Dunne [39] determined that the actual ET c over the Mississippi River basin increased between 1949 and 1997 due to higher precipitation and increasing water usage.Golubev et al. [40] discovered that the actual ET c increased in certain relatively dry sections of southern Russia and Ohio over the warm seasons of 1950-1990 using large weighted lysimeters, whereas actual evaporation decreased in two wetter taiga areas.Similarly, Linacre [29] observed that decreased pan evaporation did not always imply declining actual ET c .Trends for potential ET c during the last 50 years in the ND have been established recently [41][42][43][44], but a predictive trend of actual ET c has not been included in the ND aquifer modeling.Thus, our novel approach is to incorporate "near to accurate" predictive values of ET c into the numerical simulation model of the groundwater in the ND of Egypt.The objective of this study is therefore to investigate the effect of climate change on annual actual ET c throughout three zones of the ND (north, middle, and south) based on the historical data from 1958 to 2020 for predicting groundwater flow and differences in groundwater heads, and aquifer storage for the years 2030, 2050, and 2070.Our findings support the sustainable management plans of groundwater in the ND of Egypt for guaranteeing the sustainable use of water resources.

The Nile Delta Study Area
The Nile Delta (ND) of Egypt is located between the latitudes 30 • 00 and 31 • 45 N and longitudes 29 • 30 and 32 • 30 E (Figure 1a).This area is bounded by the Mediterranean Sea in the north and Ismailia and Nubaria canals in the east and west, respectively.The direction of the ground surface slopes is from the south to the north (the Mediterranean Sea) [45].The average range of the daily temperature is between 17 • C to 20 • C [46], while the average rainfall is between 250 mm year −1 in the north to 200 mm year −1 in the south and middle parts of the ND [47].Additionally, the evaporation rates range between 7 mm day −1 in Upper Egypt to about 4 mm day −1 on the Northern Mediterranean coast [48].Figure 1b shows the hydrological cross-section from south to north along the ND aquifer where the hydraulic conductivity is in m day −1 .Figure 1c shows the land use and land cover for the ND in 2015 which was used as a base map for assigning ET c in the model simulation.Only the temporal variation of ET c was considered in the simulation, while the spatial variation of ET c was not included where the land use/land (LU/LC) cover in the year 2015 (LU/LC) was assumed to be almost the same during the simulation period.This is because a precise prediction of LU/LC maps requires detecting and analyzing the changes in LU/LC satellite high-resolution images over an adequate period, which was not our main objective.The Nile Delta aquifer (NDA) is a semi-confined aquifer that is covered by a clay cap.The thickness of clay ranges from 5 m in the south to 20 m in the middle, reaching 50 m in the North of the ND [51][52][53].Said [54] illustrated that the stratigraphic of the quaternary sediments is divided into two units: the first unit is the Holocene deposits, formed by silty and sandy clay, and the second is the Pleistocene deposits, which represent the main aquifer.The average thickness of the NDA is about 200 m on the south to 900 m on the north.The hydraulic conductivity of the NDA ranges between 50 m day −1 to 120 m day −1 [55].The aquifer transmissivity ranges between 2000 m 2 day −1 to 3000 m 2 day −1 , and the effective and total porosity ranges between 12% to 40% [56].According to Morsy [57] and Abd-Elhamid et al. [58], the average depth of the groundwater levels ranges from 3 m to 5 m, while the groundwater flow direction is from the southeast to the northwest (Figure 2).The NDA recharge rates, groundwater heads, flow directions, thickness of clay cap, and hydraulic conductivities are controlled and managed by the distribution and the flow of the surface water system [59,60].This aquifer is recharged by infiltration from rainfall, surface water, and excess irrigation water.The total groundwater abstractions from the NDA are 85% of the groundwater resources in Egypt [61][62][63].The Nile Delta aquifer (NDA) is a semi-confined aquifer that is covered by a clay cap.The thickness of clay ranges from 5 m in the south to 20 m in the middle, reaching 50 m in the North of the ND [51][52][53].Said [54] illustrated that the stratigraphic of the quaternary sediments is divided into two units: the first unit is the Holocene deposits, formed by silty and sandy clay, and the second is the Pleistocene deposits, which represent the main aquifer.The average thickness of the NDA is about 200 m on the south to 900 m on the north.The hydraulic conductivity of the NDA ranges between 50 m day −1 to 120 m day −1 [55].The aquifer transmissivity ranges between 2000 m 2 day −1 to 3000 m 2 day −1 , and the effective and total porosity ranges between 12% to 40% [56].According to Morsy [57] and Abd-Elhamid et al. [58], the average depth of the groundwater levels ranges from 3 m to 5 m, while the groundwater flow direction is from the southeast to the northwest (Figure 2).
The NDA recharge rates, groundwater heads, flow directions, thickness of clay cap, and hydraulic conductivities are controlled and managed by the distribution and the flow of the surface water system [59,60].This aquifer is recharged by infiltration from rainfall, surface water, and excess irrigation water.The total groundwater abstractions from the NDA are 85% of the groundwater resources in Egypt [61][62][63].

Crop Consumptions
Crop water requirements are the major component of the water balance in the ND.Thus, updating the crop requirements is essential and important when considering all agroclimatic zones, climatic-dependent ETc, and crop coefficients [65].Figure 3 shows the yearly averaged values of the actual ETc (in cm) for the three zones of the ND, according to the finding of "The National Commission on Water Requirements" using the Penman-Monteith equation based on the monthly meteorological data.Actual ETc was calculated based on the crop water requirements for maize and wheat, which were selected as two major crops for this study area.This is because wheat is the most important food security crop and is the major winter cereal grain crop, while maize is the second most important grain crop according to the Food and Agricultural Organization [66].Additionally, Egypt is currently cultivating the largest area of wheat in its history at 3.6 million acres, which is equivalent to a third of Egypt's agricultural area.Maize is grown and harvested principally in summer from May to October, while wheat is grown in the winter season.ETc increased from 1958 to 1964 and decreased from 1964 to 1981.Additionally, it increased in the year 1982 and returned to decreasing from 1982 to 1999, while it increased in the year 2000.From 2000 to 2007, it decreased, returning to an increase from 2007 to 2020.This cycling behavior of the ETc can be observed over the years studied (Figure 3), which is attributed to the variation of these climate variables and climate change.

Crop Consumptions
Crop water requirements are the major component of the water balance in the ND.Thus, updating the crop requirements is essential and important when considering all agroclimatic zones, climatic-dependent ET c , and crop coefficients [65].Figure 3 shows the yearly averaged values of the actual ET c (in cm) for the three zones of the ND, according to the finding of "The National Commission on Water Requirements" using the Penman-Monteith equation based on the monthly meteorological data.Actual ET c was calculated based on the crop water requirements for maize and wheat, which were selected as two major crops for this study area.This is because wheat is the most important food security crop and is the major winter cereal grain crop, while maize is the second most important grain crop according to the Food and Agricultural Organization [66].Additionally, Egypt is currently cultivating the largest area of wheat in its history at 3.6 million acres, which is equivalent to a third of Egypt's agricultural area.Maize is grown and harvested principally in summer from May to October, while wheat is grown in the winter season.

Crop Consumptions
Crop water requirements are the major component of the water balance in the ND.Thus, updating the crop requirements is essential and important when considering all agroclimatic zones, climatic-dependent ETc, and crop coefficients [65].Figure 3 shows the yearly averaged values of the actual ETc (in cm) for the three zones of the ND, according to the finding of "The National Commission on Water Requirements" using the Penman-Monteith equation based on the monthly meteorological data.Actual ETc was calculated based on the crop water requirements for maize and wheat, which were selected as two major crops for this study area.This is because wheat is the most important food security crop and is the major winter cereal grain crop, while maize is the second most important grain crop according to the Food and Agricultural Organization [66].Additionally, Egypt is currently cultivating the largest area of wheat in its history at 3.6 million acres, which is equivalent to a third of Egypt's agricultural area.Maize is grown and harvested principally in summer from May to October, while wheat is grown in the winter season.ETc increased from 1958 to 1964 and decreased from 1964 to 1981.Additionally, it increased in the year 1982 and returned to decreasing from 1982 to 1999, while it increased in the year 2000.From 2000 to 2007, it decreased, returning to an increase from 2007 to 2020.This cycling behavior of the ETc can be observed over the years studied (Figure 3), which is attributed to the variation of these climate variables and climate change.

Data Collection
Environmental Systems Research Institute (ESRI) Arc GIS software 10.2.2 was used to extract the monthly minimum temperature (T min ), mean temperature (T mean ), maximum temperature (T max ), wind speed (WS), solar radiation (SR), humidity (H), and precipitation (rainfall, P) from the open access data, with a geographical resolution of around 4 km 2 .Temperature data were collected from the Climatic Research Unit (CRU) Time-Series (Ts) 4.0 in Network Common Data Format (NetCDF) format at 0.5 • grid for the global land surface [67].The Japanese Reanalysis (JRA-55) was used to collect the solar radiation, wind speed, and vapor pressure deficit data in NetCDF format as well [68].The JRA-55 provided complete spatio-temporal data covering the period from 1958 to the present [69].Additionally, World Weather Online (https://www.worldweatheronline.com/eg.aspx(accessed on 1 August 2022) was used to collect the humidity data.These datasets have been used in previous research of modeling the long-term dynamics of crop ET c and the impact of CC on the water footprint in the ND using an artificial neural network [69], using deep learning [70], using a deep neural network [71], and using Gaussian process regression [72].

Model Description
The groundwater modeling system (GMS) was applied and used to develop the required 3D simulation model in the current study area.The MODFLOW is a modular, three-dimensional finite-difference groundwater flow model.The governing flow equation based on water balance [73] is written as follows: where K x , K y , and K z are the hydraulic conductivities along the x, y, and z directions (LT −1 ), respectively, h is the piezometric head (L), q s is the volumetric flux per unit volume at sources or sinks in the porous medium (T −1 ), S s is the specific storage of the porous medium (L −1 ), and t is time (T). Figure 4a shows the model grid and boundary conditions used in the simulation, while Figure 4b,c show a vertical and horizontal cross-section along the conceptual model.Environmental Systems Research Institute (ESRI) Arc GIS software 10.2.2 was used to extract the monthly minimum temperature (Tmin), mean temperature (Tmean), maximum temperature (Tmax), wind speed (WS), solar radiation (SR), humidity (H), and precipitation (rainfall, P) from the open access data, with a geographical resolution of around 4 km 2 .Temperature data were collected from the Climatic Research Unit (CRU) Time-Series (Ts) 4.0 in Network Common Data Format (NetCDF) format at 0.5° grid for the global land surface [67].The Japanese Reanalysis (JRA-55) was used to collect the solar radiation, wind speed, and vapor pressure deficit data in NetCDF format as well [68].The JRA-55 provided complete spatio-temporal data covering the period from 1958 to the present [69].Additionally, World Weather Online (https://www.worldweatheronline.com/eg.aspx(accessed on 1 August 2022) was used to collect the humidity data.These datasets have been used in previous research of modeling the long-term dynamics of crop ETc and the impact of CC on the water footprint in the ND using an artificial neural network [69], using deep learning [70], using a deep neural network [71], and using Gaussian process regression [72].

Model Description
The groundwater modeling system (GMS) was applied and used to develop the required 3D simulation model in the current study area.The MODFLOW is a modular, three-dimensional finite-difference groundwater flow model.The governing flow equation based on water balance [73] is written as follows: where Kx, Ky, and Kz are the hydraulic conductivities along the x, y, and z directions (LT −1 ), respectively, h is the piezometric head (L), qs is the volumetric flux per unit volume at sources or sinks in the porous medium (T −1 ), Ss is the specific storage of the porous medium (L −1 ), and t is time (T). Figure 4a shows the model grid and boundary conditions used in the simulation, while Figure 4b,c show a vertical and horizontal cross-section along the conceptual model.

Model Calibration
The groundwater flow model was calibrated using 60 field observation wells during the year 2008 [57].The model was calibrated for steady-state conditions using the initial hydraulic parameters presented in Table 1 by trial and error until simulated groundwater levels matched those measured in the 60 observation wells.After model calibration, the groundwater flow model was validated with an independent set of data (23 points).The residuals between the simulated and the observed groundwater heads ranged from −0.834 to 0.01 m.The absolute mean residual was 0.277 m, while the root mean squared error (RMSE) was 0.354 m, with a normalization RMSE of 2.72% (Figure 5a).This relative agreement of parameters suggests a good correlation between simulated groundwater levels (Figure 5b).The locations of the observation wells used for model calibration and validation are shown in Figure 5c.The results of the groundwater levels in the NDA ranged from 14 m to 0 m above mean sea level (MSL) (Figure 5d); additionally, the groundwater flowed from the South to the North.

Climate Trend Analysis
Fundamentally, trend analysis is a strategy for understanding how and why things have changed-or will change-over time.One thing to keep in mind while seeking to comprehend trend analysis is the vast range of disciplinary contexts in which it is addressed.This makes it more difficult to describe universally, but for the sake of clarity, it may be defined here as a technique for analysis that collects data and then seeks to uncover patterns, or trends, within that data to explain or anticipate behaviors.A statistical study of trends inside time-series datasets would allow for direct conclusions on existing trends or patterns within the datasets, as well as future projections (e.g., population growth or decline).These fundamental ideas have been used in a variety of statistical trend studies in the past and continue to be relevant in scenario-building exercises and future research.The statistical methods employed in such studies vary, and there are no "automatic" trend analysis procedures that can supply all of the answers, such as reference evapotranspiration [74,75], evaporation [39], and actual ET c [76].Therefore, trend analysis is imperfect in certain ways, but it can help analysts to recognize patterns and trends that would not be feasible otherwise.Many statistical approaches, such as the Bayesian process, Spearman's Rho test, Mann-Kendall test, and Sen's slope estimator, have been developed to find trends within time series.In this study, to detect actual ET c trends, one parametric approach (linear regression) was used.

Case Scenarios
The MODFLOW simulated the groundwater flow and predicted the water heads for three projected periods, including the years 2030, 2050, and 2070, considering the impact of changing ET c as a discharge component of the out-flow budget of the groundwater aquifer.These simulation procedures allow a long-term assessment of the NDA behavior and the storage potential for the increasing pattern of temperature.

Climate Trend Analysis
Fundamentally, trend analysis is a strategy for understanding how and why things have changed-or will change-over time.One thing to keep in mind while seeking to comprehend trend analysis is the vast range of disciplinary contexts in which it is addressed.This makes it more difficult to describe universally, but for the sake of clarity, it may be defined here as a technique for analysis that collects data and then seeks to uncover patterns, or trends, within that data to explain or anticipate behaviors.A statistical study of trends inside time-series datasets would allow for direct conclusions on existing trends or patterns within the datasets, as well as future projections (e.g., population growth or decline).These fundamental ideas have been used in a variety of statistical trend studies in the past and continue to be relevant in scenario-building exercises and future research.The statistical methods employed in such studies vary, and there are no "automatic" trend analysis procedures that can supply all of the answers, such as reference evapotranspiration [74,75], evaporation [39], and actual ETc [76].Therefore, trend analysis is imperfect in certain ways, but it can help analysts to recognize patterns and trends that would not be feasible otherwise.Many statistical approaches, such as the Bayesian process, Spearman's Rho test, Mann-Kendall test, and Sen's slope estimator, have been developed to find trends within time series.In this study, to detect actual ETc trends, one parametric approach (linear regression) was used.

Case Scenarios
The MODFLOW simulated the groundwater flow and predicted the water heads for three projected periods, including the years 2030, 2050, and 2070, considering the impact of changing ETc as a discharge component of the out-flow budget of the groundwater aquifer.These simulation procedures allow a long-term assessment of the NDA behavior and the storage potential for the increasing pattern of temperature.

Trend Analysis of Actual Evapotranspiration in the Nile Delta
Results of the trend analysis for the annual actual ET c series over 1958-2020 are shown in Figure 6a-c for the three zones in the study area.The mean, median, and mode of the historical data from 1958 to 2020 for the north zone of the ND were 72.13, 69.6, and 67.6, respectively.They recorded 37.28, 36, and 36 for the middle zone, and 27.06, 26.7, and 35 for the southern zone of the ND.The analysis showed increases in annual values of actual ET c in the three zones.Increased actual ET c was due to increasing values of climate parameters.The combination of solar radiation, vapor pressure deficit, and humidity (H) were found to be the best variables for predicting ET c .The increasing percentages were 16.26, 8.76, and 12.31% for the northern zone, middle zone, and southern zone of the ND, respectively.
Furthermore, Figure 7 shows the future linear projections for actual ET c in comparison with historical time series.In the North zone of the ND, the actual ET c will reach 78.80, 82.05, and 85.30 cm year −1 in the years 2030, 2050, and 2070, respectively, compared with 72.13 cm year −1 .This represents an increase of 9.25, 12.59, and 16.05% in crop ET c in 2030, 2050, and 2070, respectively.ET c at the middle zone of the ND will reach 40.87, 42.62, and 44.37 cm year −1 compared with 37.28 cm year −1 , with an increase of 9.63, 13.07, and 16.65%, respectively, for the years 2030, 2050, and 2070, respectively.In addition, the actual ET c for the southern zone will increase to 32.10, 34.56, and 37.03 cm year −1 compared with 27.06 cm year −1 which equals 18.65, 23.38, and 28.84% increase that the ET c values between 1958 and 2020.The average value of actual ET c among the whole ND will reach 50.60, 53.08, and 55.57cm year −1 compared with 45.49 cm year −1 , with an increase of 11.2, 15.0, and Water 2023, 15, 572 9 of 16 19.0%for the years 2030, 2050, and 2070, respectively.These values of ET c were used as the mean ET c rate and applied to the whole NDA when running MODFLOW to simulate the groundwater modeling, and in order to determine the decline in groundwater levels by the years 2030, 2050, and 2070.
Water 2023, 15, x FOR PEER REVIEW 9 of 16 ETc in the three zones.Increased actual ETc was due to increasing values of climate parameters.The combination of solar radiation, vapor pressure deficit, and humidity (H) were found to be the best variables for predicting ETc.The increasing percentages were 16.26, 8.76, and 12.31% for the northern zone, middle zone, and southern zone of the ND, respectively.Furthermore, Figure 7 shows the future linear projections for actual ETc in comparison with historical time series.In the North zone of the ND, the actual ETc will reach 78.80, 82.05, and 85. 30

Hydrological Water Balance of Nile Delta Aquifer
The calibrated model was applied to simulate the future changes of ETc and its i on the groundwater levels in the ND for the years 2030, 2050, and 2070.Table 2 pr the zone budget analysis for the three predefined predictive periods in which the co heads inflow was increased to 1,073,700, 1,102,500, and 1,208,900 m 3 day −1 compare 840,030 m 3 day −1 at base case (2008); additionally, the aquifer recharge was decrea 4,815,900, 4,692,300, and 4,271,700 m 3 day −1 compared with 840,030 m 3 day −1 at bas while the canal leakage increased to 985,600, 1,014,400 and 1,116,100 m 3 day −1 com with 732,910 m 3 day −1 for the base case.

Hydrological Water Balance of Nile Delta Aquifer
The calibrated model was applied to simulate the future changes of ET c and its impact on the groundwater levels in the ND for the years 2030, 2050, and 2070.Table 2 presents the zone budget analysis for the three predefined predictive periods in which the constant heads inflow was increased to 1,073,700, 1,102,500, and 1,208,900 m 3 day −1 compared with 840,030 m 3 day −1 at base case (2008); additionally, the aquifer recharge was decreased to 4,815,900, 4,692,300, and 4,271,700 m 3 day −1 compared with 840,030 m 3 day −1 at base case, while the canal leakage increased to 985,600, 1,014,400 and 1,116,100 m 3 day −1 compared with 732,910 m 3 day −1 for the base case.The total inflow/outflow decreased by 9.2%, 10.10%, and 12.90% for the years 2030, 2050, and 2070, respectively.Table 3 summarizes these percentages of increase (+)/decrease (−) of each boundary parameter for the years 2030, 2050, and 2070 compared to the current situation of the aquifer (base case).On the other hand, the constant heads in flow were increased by 27.8, 31.2, and 43.9 for 2030, 2050, and 2070, respectively, compared to the base case.The flow to the aquifer was decreased by 19.7, 21.8, and 28.3 for 2030, 2050, and 2070, respectively, compared to the base case, while canal leakage was increased by 34.5, 38.4, and 52.3 for 2030, 2050, and 2070, respectively.The outflow to drains decreased by 27.9, 30.6, and 39.4 for 2030, 2050, and 2070, respectively, while outflow to wells remained constant where the same discharges were assigned for the model to strictly limit the discharge from the aquifer.The total inflow/outflow decreased by 9.2%, 10.10%, and 12.90% for the years 2030, 2050, and 2070, respectively.Table 3 summarizes these percentages of increase (+)/decrease (−) of each boundary parameter for the years 2030, 2050, and 2070 compared to the current situation of the aquifer (base case).

Quantification of Evapotranspiration-Aquifer Interaction
The relation between ETc and the flow to the NDA is presented in Figure 9.The results indicated that the future increase in ETc to 50.60 mm, 51.20 mm, and 53.70 mm year −1 , compared with 44.15 mm year −1 by an increase of 19.70%, 21.80%, and 28.80%, led to a

Quantification of Evapotranspiration-Aquifer Interaction
The relation between ET c and the flow to the NDA is presented in Figure 9.The results indicated that the future increase in ET c to 50.60 mm, 51.20 mm, and 53.70 mm year −1 , compared with 44.15 mm year −1 by an increase of 19.70%, 21.80%, and 28.80%, led to a decrease in the aquifer recharge to 4,815,900, 4,692,300, and 4,271,700 m 3 day −1 compared with 6,000,400 m 3 day −1 in the base case.The GWHD reached 0.95 m, 1.05 m, and 1.40 m while considering the increase of ET c in years 2030, 2050, and 2070, respectively.This indicated that future ET c should be considered in water resources management planning to mitigate the reduction of groundwater levels and any potential reduction in aquifer storage while monitoring abstraction by wells for limiting discharges from the NDA.Thus, a sustainable operation of the ND aquifer can be achieved.

Quantification of Evapotranspiration-Aquifer Interaction
The relation between ETc and the flow to the NDA is presented in Figure 9.The results indicated that the future increase in ETc to 50.60 mm, 51.20 mm, and 53.70 mm year −1 , compared with 44.15 mm year −1 by an increase of 19.70%, 21.80%, and 28.80%, led to a decrease in the aquifer recharge to 4,815,900, 4,692,300, and 4,271,700 m 3 day −1 compared with 6,000,400 m 3 day −1 in the base case.The GWHD reached 0.95 m, 1.05 m, and 1.40 m while considering the increase of ETc in years 2030, 2050, and 2070, respectively.This indicated that future ETc should be considered in water resources management planning to mitigate the reduction of groundwater levels and any potential reduction in aquifer storage while monitoring abstraction by wells for limiting discharges from the NDA.Thus, a sustainable operation of the ND aquifer can be achieved.

Conclusions
Groundwater management of the Nile Delta aquifer (NDA) in Egypt was assessed by considering climate change's effect on actual evapotranspiration, ET c (water requirements of a typical pattern).Firstly, historical data over 62 years from 1958 to 2020 were acquired, and the study area was categorized into three zones: the North (along the Mediterranean coast), the Middle, and the South.The conceptual model of MODFLOW was utilized to identify the current case of the aquifer and calculate the budget analysis after model calibration and validation in steady and transient states.Then, the visual MODFLOW was run to predict the aquifer behavior and calculate the different boundary parameters at three predictive dates: until 2030, until 2050, and until 2070, including the change of the annual actual ET c in the North, Middle, and South zones of the Nile Delta (ND).In addition, zone budget analysis was conducted for the total inflow and outflow at the end of the three predictive periods.
Actual evapotranspiration (ET c ) in ND zones correlated with the combination of solar radiation, vapor pressure deficit, and humidity (H), which were found to be the best variables for predicting ET c .The increasing solar radiation and vapor pressure deficit convert a considerable amount of liquid water into water vapor and water demands, while increasing humidity tends to reduce transpiration.Zone budget analysis revealed that the increase of ET c will decrease the inflow and the outflow by 9.2, 10.1, and 12.9% in 2030, 2050, and 2070, respectively.Based on the model simulation results, the groundwater head difference (GWHD) reached 0.95, 1.05, and 1.40 m while considering the increase of ET c in 2030, 2050, and 2070, respectively.Our results are helpful for addressing the behavior of the ND groundwater aquifer for ensuring a sustainable management plan considering the climate effect on crop water requirements.However, the combination of the spatial and temporal changes of ET c is an interesting topic to be studied in further research.

Figure 1 .
Figure 1.(a) Key map of the Nile Delta.(b) Hydrogeological cross-section from south to north along the Nile Delta aquifer [49].(c) The major five land use/land cover classes in the Nile Delta in 2015 [50].

Figure 1 .
Figure 1.(a) Key map of the Nile Delta.(b) Hydrogeological cross-section from south to north along the Nile Delta aquifer [49].(c) The major five land use/land cover classes in the Nile Delta in 2015 [50].

Figure 3 .
Figure 3. Historical data between 1958 and 2020 for the average ETc (cm) for the three zones of the Nile delta in the previous 62 years.
ET c increased from 1958 to 1964 and decreased from 1964 to 1981.Additionally, it increased in the year 1982 and returned to decreasing from 1982 to 1999, while it increased in the year 2000.From 2000 to 2007, it decreased, returning to an increase from 2007 to 2020.This cycling behavior of the ET c can be observed over the years studied (Figure3), which is attributed to the variation of these climate variables and climate change.

Figure 3 .
Figure 3. Historical data between 1958 and 2020 for the average ETc (cm) for the three zones of the Nile delta in the previous 62 years.

Figure 3 .
Figure 3. Historical data between 1958 and 2020 for the average ET c (cm) for the three zones of the Nile delta in the previous 62 years.

Figure 4 .
Figure 4. (a) Model grid and boundary conditions; (b) hydrogeological vertical cross-section; (c) hydrogeological horizontal cross-section of the underlying aquifer assigned in the simulation.

Figure 5 .
Figure 5. Observed vs. calculated groundwater heads (m) in steady-state condition for (a) model calibration; (b) model validation (solid line represents the 45 o line while the red and blue dotted lines represent the error bounds of ± 5% and ± 10%, respectively; (c) location of the observation wells used for model calibration and validation; (d) groundwater head distributions for the Nile Delta Quaternary aquifer (contour lines are in black while arrows represent velocity vectors).

Figure 5 .
Figure 5. Observed vs. calculated groundwater heads (m) in steady-state condition for (a) model calibration; (b) model validation (solid line represents the 45 o line while the red and blue dotted lines represent the error bounds of ± 5% and ± 10%, respectively; (c) location of the observation wells used for model calibration and validation; (d) groundwater head distributions for the Nile Delta Quaternary aquifer (contour lines are in black while arrows represent velocity vectors).

Figure 6 .
Figure 6.Trend analysis of actual evapotranspiration for historical data from 1958 to 2020, and the linear regression for the predictive values until 2070 for the three zones of the Nile Delta: (a) North, (b) Middle, and (c) South.

Figure 6 .
Figure 6.Trend analysis of actual evapotranspiration for historical data from 1958 to 2020, and the linear regression for the predictive values until 2070 for the three zones of the Nile Delta: (a) North, (b) Middle, and (c) South.

Figure 7 .
Figure 7. Future projection of actual ET compared with the historical period.

Figure 7 .
Figure 7. Future projection of actual ET compared with the historical period.

Figure 8a -
Figure 8a-c present the distribution of groundwater heads difference in the ND for the three predictive periods until the years 2030, 2050, and 2070.The results showed that increasing the ET will decrease the aquifer recharge and increase the groundwater head difference (GWHD).

Figure
Figure 8a-c present the distribution of groundwater heads difference in the ND for the three predictive periods until the years 2030, 2050, and 2070.The results showed that increasing the ET will decrease the aquifer recharge and increase the groundwater head difference (GWHD).

Figure 8 .
Figure 8. Distribution of groundwater head differences (in meters) for the Nile Delta aquifer for the three predictive periods: (a) until 2030, (b) until 2050, and (c) until 2070.

Figure 8 .
Figure 8. Distribution of groundwater head differences (in meters) for the Nile Delta aquifer for the three predictive periods: (a) until 2030, (b) until 2050, and (c) until 2070.

Figure 8 .
Figure 8. Distribution of groundwater head differences (in meters) for the Nile Delta aquifer for the three predictive periods: (a) until 2030, (b) until 2050, and (c) until 2070.

Figure 9 .
Figure 9.Comparison between the predicted average ETc in 2030, 2050, and 2070, and the corresponding reduction in aquifer recharge (in Mm 3 day −1 ).

Figure 9 .
Figure 9.Comparison between the predicted average ET c in 2030, 2050, and 2070, and the corresponding reduction in aquifer recharge (in Mm 3 day −1 ).

Table 1 .
Initial and calibrated hydraulic parameters of the two geologic units of the Nile Delta aquifer.
27.06 cm year which equals18.65,23.38,and 28.84% increase that the ETc values be 1958 and 2020.The average value of actual ETc among the whole ND will reach 53.08, and 55.57cm year −1 compared with 45.49 cm year −1 , with an increase of 11.2 and 19.0% for the years 2030, 2050, and 2070, respectively.These values of ETc wer as the mean ETc rate and applied to the whole NDA when running MODFLOW to late the groundwater modeling, and in order to determine the decline in ground levels by the years 2030, 2050, and 2070.

Table 2 .
Zone budget analysis for the base case (at 2020), 2030, 2050, and 2070 (values in m 3 d

Table 3 .
Percentage of difference (increase (+), or decrease (−)) for each simulation scenario (until the years 2030, 2050, and 2070) compared to the base case (current situation of the aquifer).

Table 3 .
Percentage of difference (increase (+),or decrease (−)) for each simulation scenario (until the years 2030, 2050, and 2070) compared to the base case (current situation of the aquifer).