Water, Energy and Carbon Tradeoffs of Groundwater Irrigation-Based Food Production: Case Studies from Fergana Valley, Central Asia

: In arid environments, water shortages due to over-allocation of river ﬂow are often com-pensated by lift irrigation or pumping groundwater. In such environments, farmers using pumped irrigation can deploy on-farm energy-efﬁcient and water-saving technologies; however, pumping water requiring extra energy is associated with carbon emissions. This study explores how to increase crop production using pumped irrigation with minimal energy and carbon emissions. The purpose of this research is twofold: ﬁrst, to examine on-farm energy consumption and carbon emissions in gravity and groundwater irrigation systems; and second, to explore system-level alternatives of power generation and water management for food production based on the results from the farm-level analysis. This study employs a novel system-level approach for addressing water, energy, and carbon tradeoffs under pumped irrigation using groundwater. These tradeoffs are assessed at farm and system levels. On-farm level estimates showed that farm-level interventions were insufﬁcient to produce mutual gains. According to the results of the system-level evaluation, system-level interventions for water and energy conservation, the use of renewable energy to pump water for irrigation, and river basin scale cooperation are all required to maintain crop production while reducing energy consumption and carbon emissions. This study examines the sustainability of pumped irrigation using a system-level approach through linking water and energy use at the on-farm level to the appropriate off-farm level mobilization of natural resources and carbon emissions to supply and energy for on-farm crop production. The impact of water cooperation at the basin/system level on the sustainability of crop production is assessed at the on-farm level in groundwater Thus, it facilitates the analysis of between and in groundwater The research question is: how to reduce The hypothesis is that The of compare


Introduction
Increased agricultural production is critical to ensuring food security for the world's growing population. This requires more agricultural farm inputs, including water and energy, and carries associated soil, water, and atmosphere pollution risks due to increasingly poor-quality return flows and greenhouse gas (GHG) emissions. Over time, the return flows and emissions may also accelerate global changes in the biosphere, negatively impacting the environment and depletion of non-renewable fossil energy resources. Consequently, agricultural development, aimed at obtaining short-term benefits, increases the risk of failures in the future for the environment and human life [1].
The unsustainability of agricultural development has become a much-researched subject [1][2][3]. One challenge for the intensification of food production in arid regions is the vulnerability and over-allocation of river flow, which causes farmers to shift to pumped irrigation, using available water sources, including groundwater [4][5][6][7][8][9]. The pumped irrigation ensures the reliability of water supply according to crop water requirements; however, This study examines the sustainability of pumped irrigation using a system-level approach through linking water and energy use at the on-farm level to the appropriate off-farm level mobilization of natural resources and carbon emissions to supply water and energy for on-farm crop production. The impact of water cooperation at the basin/system level on the sustainability of crop production is assessed at the on-farm level in groundwater irrigation areas. Thus, it facilitates the analysis of tradeoffs between water, energy, and carbon in groundwater irrigation areas. The research question is: how to reduce energy use and carbon emissions in crop production using pumped irrigation. The hypothesis is that combined analyses of water, energy, and carbon tradeoffs from field to system-level can result in sustainable crop production in groundwater irrigation areas. The objectives of this study are (1) to compare on-farm level energy use and related carbon emissions in gravity and groundwater irrigation systems; and (2) to examine alternatives of electricity generation and use, irrigation, and food production at a system-level based on these farm-level findings. The novelty of this study lies in the system-level approach to addressing water, energy, and carbon tradeoffs for pumped irrigation, including lift and groundwater irrigation.

Challenges in Food Production Due to Water-Energy-Carbon Nexus
The Aral Sea Basin, spread across the southern part of Central Asia, has a fast-growing population. The population is expected to rise from 57 million in 2020 to 83 million in 2050, making food crop production a priority for farming in irrigated lands where over 90% of the region's agricultural products are produced. Population growth will increase the demand for more water withdrawal. At the same time, the flow of the main rivers in the region is projected to decrease due to climate change and melting glaciers upstream. Thus, by 2050, the annual unmet demand in the Amudarya and Syrdarya River Basins is projected to increase to 50% and 35% of the total water demand, respectively [31]. Considering that over 40% of the irrigated land downstream of the rivers is under lift irrigation, food production in the region is projected to become highly dependent on the availability of water, the energy to lift water, and a sustainable environment, which also requires water for its functioning. Under these conditions, the interlinkages between water, energy, and carbon have become critical considerations for sustainable food production and an environmentally friendly strategy. Food production in the basin depends on river flow, which is transboundary. Much of the flow of the Syrdarya (over 73%) and Amudarya (over 88%) rivers originate in upstream countries-Kyrgyzstan, Tajikistan, and Afghanistan, while productive farmlands requiring irrigation are located primarily in downstream countries-Uzbekistan, Turkmenistan, and Southern Kazakhstan.
Reaping the benefits of such an environment requires close cooperation between riparian states in the upstream hydropower development and food crop production downstream of the rivers [32]. There is high demand for such collaboration, as 95% of the hydropower potential is undeveloped in the upstream of the Amudarya in Tajikistan and 87% in the upstream of the Syrdarya in Kyrgyzstan [33]. Increasing hydropower generation upstream in the future could boost the electricity market in the region; however, it might reduce the quantity of water available for agriculture in the summer months. While downstream crop producers are concerned about upstream hydropower development, hydropower is very important for the upstream economy and potential for the region's future [34].
At the beginning of the 1990s, the previously united regional power supply system collapsed. The downstream states preferred access to water than hydropower from upstream and, facing energy shortages, focused on internal thermal power development using fossil energy resources. The electricity supply from upstream hydropower stations was replaced with local thermal power stations. This strategy causes rapid depletion of fossil resources, including natural gas, in Uzbekistan, where reserves are available for only 20-30 years. In addition, the shift from hydropower to thermal power increases carbon dioxide emissions, contributing to climate warming and melting glaciers upstream. Thus, focusing on thermal power in the downstream states causes depletion of fossil energy resources and increases Sustainability 2022, 14, 1451 4 of 24 CO 2 emissions, contributing to glacier melting and subsequent depletion of river flow upstream. This dilemma of water and energy tradeoffs and its impact on the agriculture and environment of the region is still not well studied [35,36].
The government of Uzbekistan, foreseeing an increase in water scarcity in agriculture, has prioritized the widespread adoption of drip irrigation. The area under drip irrigation is planned to increase from 44,000 ha to 600,000 ha by 2030, with reduced irrigation demand by 1500 million m 3 of water [37], which could be reallocated to another watershort irrigated area or allowed to flow to the Aral Sea. This may cause a gradual lowering of the groundwater table in many systems where irrigation water losses are the primary source of groundwater recharge. This may produce several gains, such as reducing unproductive evaporation from the groundwater table and creating free capacities in the aquifers, which can be used to store floodwater. The river has excessive flows in high water years. The shortcoming is that lowering the water table would make developing groundwater less attractive for farmers due to the increased energy demand for pumping. Current water savings may be achieved at the cost of future groundwater development. In such cases, groundwater depletion can be arrested by managed aquifer recharge activities in the winter seasons, for example, by using hydropower releases from the upstream reservoirs [38]. Irrigation in CA is already energy-intensive. Currently, irrigated agriculture is solely based on river flow, which is delivered to farm gates by a multi-level system of canals, often receiving water through inter-basin water transfers. This strategy of reallocating water resources from one basin to another, developed in the past (during the Soviet time with no borders between the riparian countries), is widely applied in the region and has resulted in low water-use efficiency of 0.3-0.4; water losses from the canals and the agricultural fields fill aquifers and raise groundwater levels.
Inter-basin water transfers have often been achieved by lifting water from one basin to another, requiring energy to pump water. Consequently, the electricity to pump irrigation water amounts to 12% of the total power generated in Uzbekistan. Thus, low water-use efficiency and high energy consumption make agricultural water management a critical starting point for sustaining national food production.
Developing farmer-driven groundwater irrigation under such conditions can improve water and energy efficiencies. Despite the government's water policy prioritizing water-saving by introducing drip irrigation or advancing furrow irrigation, groundwater development for agricultural purposes remains essential, especially for irrigated land in piedmont areas and river valleys, often under lift irrigation. One of the advantages of groundwater development over lift or small pump irrigation is that farmers using groundwater decide when to irrigate and how much water to apply.
Farmers cultivate grapes outside of canal command areas where groundwater is 20-40 m deep. They acquire water from lift irrigation schemes, if available, or pump groundwater. Lift irrigation schemes pump water from a river to a height of 50-200 m and more to the highest point. Water is transported by lift canals, usually long-distance by gravity, and then distributed between farmers for irrigation. That is why groundwater irrigation is less energy-intensive than lift irrigation, with an average value of 0.2-0.4 kWh/m 3 versus 0.3-0.6 kWh/m 3 , respectively, indicating the regions where groundwater irrigation may produce water and energy savings. When the lift irrigation infrastructure is aged, the difference in the energy intensity between lift irrigation and the groundwater becomes higher. There are also risks of increasing salinization and depletion of groundwater, especially in drought years. Thus, the issues are complex and require system-level analyses of power generation, irrigation, food production, and the environment.

Study Area
This study was carried out in the Fergana Valley, where the Syrdarya River begins, by the confluence of two tributaries, the Naryn and Karadarya. These two main tributaries and small rivers, flowing into the valley from the south and north, are the primary water sources for irrigation. At the same time, losses from the riverbeds, irrigation canals, and agricultural fields recharge groundwater. Borisov [39] estimated that the total storage of groundwater, including renewable and non-renewable, in the Fergana Valley exceeds 100 km 3 ; despite this significant potential, groundwater is being developed and applied for irrigation to cover a shortage of canal water only. Two representative farms growing grapes using gravity and groundwater irrigation were selected for comparative analysis (Figure 1).

Study Area
This study was carried out in the Fergana Valley, where the Syrdarya River begins, by the confluence of two tributaries, the Naryn and Karadarya. These two main tributaries and small rivers, flowing into the valley from the south and north, are the primary water sources for irrigation. At the same time, losses from the riverbeds, irrigation canals, and agricultural fields recharge groundwater. Borisov [39] estimated that the total storage of groundwater, including renewable and non-renewable, in the Fergana Valley exceeds 100 km 3 ; despite this significant potential, groundwater is being developed and applied for irrigation to cover a shortage of canal water only. Two representative farms growing grapes using gravity and groundwater irrigation were selected for comparative analysis (Figure 1). The two farms selected for this study have different practices for growing grapes. Farm 1, situated in the western part of the valley in Soghd Province of Tajikistan, has GPS coordinates of 40°08′40″N and 69°38′47″E. The farm has a total land area of 630 ha, of which 86 ha is allocated to grape cultivation. The farmer's vineyards are grown in rows using trellis on sandy loam soils, underlain by gravel deposits, with gravity furrow irrigation applied from the Khodjabakirgan River. Farm 2 is located in the Fergana Province of Uzbekistan with GPS coordinates of 40°28′40″N and 71°35′36″E. In the early 1990s, the farmer received land from the government through a long-term lease. The 10-ha area is under groundwater/well irrigation, with 6 ha under grapes, while the remaining are young orchards. The farm's soil is sandy loam underlain by gravel 50 cm below the ground surface. There is approximately 500 ha of land under the same production technology of grape cultivation in the same area.
These two sites represent two different farming systems. Farm 1 has an extensive land area, is relatively well-supplied with machinery, and applies gravity irrigation. Farm 2 is a family farm in a smaller area, relies extensively on manual labor, uses a mini tractor for soil tillage, and practices groundwater irrigation. While Farm 1 applies row planting of vineyards and gravity irrigation by furrows made along the rows, Farm 2 uses intensive crop cultivation practices-grape vines at the height of 2 m cover the entire area. Such a canopy dramatically increases solar energy harvest, while plant roots occupy the whole inter-row space, up to a depth of 50 cm. This canopy shape and root distribution allow plants to use solar energy, water, and nutrients effectively. The farmer applies a low rate, high-frequency irrigation, using groundwater along the short furrows with a width of 60 cm, thereby moistening the entire inter-row space shaded by leaves. Due to the indicated differences, including the farming practices used in grape production, the form of The two farms selected for this study have different practices for growing grapes. Farm 1, situated in the western part of the valley in Soghd Province of Tajikistan, has GPS coordinates of 40 • 08 40 N and 69 • 38 47 E. The farm has a total land area of 630 ha, of which 86 ha is allocated to grape cultivation. The farmer's vineyards are grown in rows using trellis on sandy loam soils, underlain by gravel deposits, with gravity furrow irrigation applied from the Khodjabakirgan River. Farm 2 is located in the Fergana Province of Uzbekistan with GPS coordinates of 40 • 28 40 N and 71 • 35 36 E. In the early 1990s, the farmer received land from the government through a long-term lease. The 10-ha area is under groundwater/well irrigation, with 6 ha under grapes, while the remaining are young orchards. The farm's soil is sandy loam underlain by gravel 50 cm below the ground surface. There is approximately 500 ha of land under the same production technology of grape cultivation in the same area.
These two sites represent two different farming systems. Farm 1 has an extensive land area, is relatively well-supplied with machinery, and applies gravity irrigation. Farm 2 is a family farm in a smaller area, relies extensively on manual labor, uses a mini tractor for soil tillage, and practices groundwater irrigation. While Farm 1 applies row planting of vineyards and gravity irrigation by furrows made along the rows, Farm 2 uses intensive crop cultivation practices-grape vines at the height of 2 m cover the entire area. Such a canopy dramatically increases solar energy harvest, while plant roots occupy the whole inter-row space, up to a depth of 50 cm. This canopy shape and root distribution allow plants to effectively use solar energy, water, and nutrients. The farmer applies a low rate, high-frequency irrigation, using groundwater along the short furrows with a width of 60 cm, thereby moistening the entire inter-row space shaded by leaves. Due to the indicated differences, including the farming practices used in grape production, the form of ownership, management, and irrigation techniques, the yield levels achieved by Farm 2 are significantly higher than those of Farm 1.

Data and Methods
This study used a two-step approach to analyze water-energy-carbon tradeoffs. First, we carried out field studies at the farm level to estimate the linkages between the shift from gravity to groundwater irrigation and the changes in energy consumption and GHG emissions. Since farmers gradually move from cotton and wheat to high-value crops, orchards, and vegetables, we selected farms growing grapes, an essential source of revenue for farmers from the Fergana Valley and the region. We then evaluated system-level alternatives to make farming practices more sustainable, including intensifying crop production, introducing water-and energy-saving technologies, and shifting to renewable energy sources.

Farm Inputs
To assess the gains and shortcomings of introducing groundwater irrigation, farming practices and inputs, including irrigation water, were monitored during 2013-2014. All farming practices and inputs were recorded daily. Daily climate data for this period were obtained from the Fergana weather station. Field-level water monitoring included: (1) measuring irrigation applications by using Cipolletti and Thomson weirs, at field level and furrow level, accordingly; (2) Cipolletti weirs were also applied for measuring tailend water discharges that occurred during irrigations; and (3) measuring soil moisture content for depths of 0-0.15, 0.15-0.30, 0.30-0.50 and 0.50-0.75 cm using soil sampling and air-drying methods. Phenology observations included measuring soil crop cover and plant height in major plant development phases.
The evapotranspiration from the grape fields was estimated using the water balance method using formula as follows: where ET-evapotranspiration from a grape field; I-total irrigation applied; P-effective precipitation, Fl-field losses of irrigation water, ∆SM-change of soil moisture content in the plant root zone. Soil moisture content was determined before and at the end of the crop-growing season as well as before and after irrigation applications. Deep percolation losses were estimated as the difference between the volume of water applied for irrigation and the amount of irrigation water accumulated in the plant root zone. Table 1 shows the inputs applied with the emission potential for 2013-2014 for both farms. Table 1 indicates the differences in farming inputs at Farm 1 and Farm 2. Intensive farming practices in grape cultivation, applied at Farm 2, require 3.7 times more labor than in Farm 1; soil tillage practices are applied using a mini tractor of 15 horsepower in Farm 2 and a tractor of 80 horsepower in Farm 1. These practices, monitored at the farms, are as follows: (a) soil tillage of interrow spacing and making furrows in spring; (b) soil cultivation after each irrigation at Farm 1 and four times per season and reshaping furrows at Farm 2; (c) applying herbicides and insecticides three times per year in the grape-growing season. Accordingly, the total diesel fuel consumption amounted to 56 l/ha and 60 l/ha at Farm 1 and Farm 2. These figures are within the range given for grape cultivation by Karimi and Moghaddam [41] and Osman and Engindeniz [42].
Furthermore, the irrigation application technique, another critical input in productivity, differed in the farms under consideration. Farmers applied eight irrigations per cropping season at Farm 1 and 15 irrigations at Farm 2. Although the recommended irrigation schedules were similar because of the same soil and climate conditions, the actual irrigation schedules differed ( Figure 2).   Figure 2 shows that the total irrigation application was higher at Farm 1, and the amount of water applied to Farm 2 was less than the recommended volume. The recommended irrigation applications were calculated based on climate conditions, soil properties, groundwater table depths, and the relevant irrigation efficiency factors [43]. In both farms, farmers themselves, based on the status of plants, soil moisture depletion, and availability of irrigation water, decided when to irrigate and how much water to apply. The prolonged intervals between irrigation applications were in practice, especially from the second half of June until the end of July when grape bushes did not receive irrigation  [40]. Data of the recommended irrigation schedule (green line) is adapted from Domullodjonov (1988) [43]. Figure 2 shows that the total irrigation application was higher at Farm 1, and the amount of water applied to Farm 2 was less than the recommended volume. The recommended irrigation applications were calculated based on climate conditions, soil properties, groundwater table depths, and the relevant irrigation efficiency factors [43]. In both farms, farmers themselves, based on the status of plants, soil moisture depletion, and availability of irrigation water, decided when to irrigate and how much water to apply. The prolonged intervals between irrigation applications were in practice, especially from the second half of June until the end of July when grape bushes did not receive irrigation water for 47 days. Such interruptions in water supply occurred due to the unregulated flow of the Khodjabakirgan River. In contrast, groundwater use for irrigation allowed timely irrigation at Farm 2.

Energy Use and Greenhouse Gas Emissions
The study made the following assumptions: • Water, energy, and carbon tradeoffs and conclusions based on grape production apply to other crops.

•
The ages of the vineyards were 25 years and 8 years for Farms 1 and 2, respectively. The study assumed that vineyards in this age range do not have different productivity potential if properly maintained farming inputs [44]. Thus, comparative evaluations can be undertaken.

•
The analysis applied the straight-line method of depreciation of agricultural machinery. • According to Siyal et al. [45], energy inputs for surface water delivery by canal systems in the lower Indus basin in southern Pakistan average between 3 and 4 KJ/m 3, and carbon emissions are 0.22-0.30 g/m 3 . These values are insignificant compared to on-farm energy consumption and carbon emissions in the gravity and groundwater irrigation systems. Considering the low energy inputs found in similar systems, energy inputs for water delivery by canal systems in the study areas were excluded from further analyses.
Water and energy use and GHG (CO 2 -equivalent) emissions were the basis for assessing tradeoffs between water, energy use, and carbon emissions. Farm inputs and operations used in grape production contribute to CO 2 emissions. The amount of carbon dioxide equivalent (i.e., CO 2 -e) depends on the type of GHG emitted from the farming inputs and respective conversion factors given in Table 2. Energy and CO 2 emissions were from electricity and diesel, agrochemicals, and agricultural machinery production.

Energy Requirement and CO 2 Emissions for Groundwater Pumping
Electricity consumption for pumping water is monitored using the meter. In scenario analyses, the energy required for groundwater pumping was calculated using Equations (2) and (3) borrowed from Robertson [59] and Shah [4], respectively: where Energy is the energy demand (kWh), Volume is the volume of groundwater lifted per annum; H is the total dynamic head (Equation (3)); gwd is the initial groundwater depth; drawdown is 3 m, and friction losses are 20%; T&Dlos is transmission and distribution losses; and η is the efficiency of the pump set (%), including the pump, the driver, and the motor.
In Uzbekistan, 89% of the electricity generated is from thermal power plants (TPPs), and 11% is from hydropower power plants (HPP). Meanwhile, 91% of the electricity produced at the TPPs is from natural gas, 2% from black oil, and 7% from coal. According to UzbekEnergo, the transmission and distribution losses of electricity in Uzbekistan are 11%, while pump set efficiency is 30%. Natural-gas-based electricity TPPs have an average efficiency of 28%, and coal-based TPPs have 25%.
The estimation of CO 2 emissions is based on the amount of electricity generated at the TPPs to support irrigation water pumping. The rate of natural gas consumption in the overall production of electricity for Uzbekistan is 0.276 m 3 /kWh for conventional power plants and 0.188 m 3 /kWh for new plants. The carbon dioxide emissions per unit of electricity from power plants ranged between 0.542 kg of CO 2 -e for traditional oncethrough systems and 0.368 kg of CO 2 -e for circulation systems. The Fergana Valley, where more than 10 million people live, covers a deficit in electricity supply from Kyrgyzstan, where electricity generation is assumed to be produced entirely by HPP. The estimations of CO 2 emissions from electricity generation by HPP are based on Scherer and Pfister [60], where CO 2 emissions are correlated to the relevant reservoir/electricity generation ratio area. Emissions of CO 2 are determined using this correlation separately for upstream, midstream, and downstream hydropower.

Energy and CO 2 Emissions from Agrochemicals
The agrochemicals considered in this study include fertilizers and chemicals used for controlling pests and weeds. The study derived the history of agrochemical use from monitoring activities and farm records. Farm 2 deployed ammonium nitrate (NH 4 NO 3 ) with 35% N content, whereas Farm 1 applied urea with 46% N content. For the production, packaging, delivery, and application of chemical fertilizers, emissions were assessed following the emission factor given in Table 2.

Energy Use and CO 2 Emissions from Agricultural Machinery Production
According to Pimentel [56], approximately 83.8 mega joules (MJ) of energy is used to produce a kilogram of agricultural machinery. Consequently, on average, it takes 9.6 kg of CO 2 -e to produce a kilogram of agricultural machinery [56,57]. It is worth noting that there could be differences in farm machinery and manufacturing technologies between countries. However, due to data constraints, this study uses the values from Pimentel [56] and Maraseni et al. [57].
Water-energy-carbon tradeoffs were analyzed using indicators as follows: (a) Total energy inputs per hectare of irrigated land; (b) Non-renewable fuel resources depleted; (c) Total CO 2 emissions per hectare of irrigated land; (d) Crops produced per kg of CO 2 emissions, that is, the ratio of the yield of grapes to total CO 2 emissions [61]; (e) Carbon dioxide emissions per MJ of input, that is, the ratio of total CO 2 emissions to the energy input; (f) Water productivity [61], calculated as follows: where WP = water productivity (kg/m 3 ), Y = yield of grapes (kg/ha), and ET = crop evapotranspiration (m 3 /ha); (g) Energy efficiency (Ee), estimated using the formula: where E out = energy output (MJ/ha), and E in = energy input (MJ/ha), (h) Energy productivity, calculated as follows: where EP = energy productivity (kg/kWh), and P is the total energy input for producing grapes (kWh/ha).

Farm-Level Analysis of Water, Energy, and Carbon Tradeoffs
Irrigation water-associated energy inputs and carbon emissions: Farm 1 and Farm 2 use different irrigation strategies. Farmers practicing gravity irrigation applied eight irrigations varying from 1201 m 3 /ha early spring to 985-1123 m 3 /ha, each during the summer season at Farm 1. Intervals between irrigation were prolonged, especially in the summer season, causing a mismatch between water supply and crop water requirements. The farmer, employing groundwater irrigation, provided fifteen low-rate high-frequency irrigations varying from 389 to 685 m 3 /ha, each at Farm 2. Intervals between irrigations varied from one month in spring to one week in July and the first half of August; such frequent irrigation allowed for maintaining soil moisture content at the optimal level. Water supplied for irrigation is spent almost entirely for crop transpiration thanks to covering the entire area by grape vines at the height of 2 m and low-rate high-frequency irrigation by short furrows. Table 3 summarizes water-associated energy inputs reflecting the differences in irrigation strategies. Water-related energy inputs are low under gravity irrigation and significantly increase for groundwater irrigation, causing high carbon emissions. Since water is free at Farm 1, the farmers have insufficient incentives to adopt water-saving technologies. In opposite, at Farm 2, the farmer is induced to reduce irrigation applications and use water more efficiently due to the high cost of electricity to pump water. The electricity consumption to operate the pump causes high carbon emissions.
Water balance studies found crop evapotranspiration at 5740 m 3 /ha at Farm 1 and 7496 m 3 /ha at Farm 2. Despite the minor difference between the irrigation applications at Farm 1 and Farm 2, there was a high difference in water productivity. This was because (1) according to coarse estimates using the crop cover change, over 83% of the water applied at Farm 2 and only 47% of the water applied at Farm 1 are depleted productively for crop transpiration; (2) timely irrigation according to crop water requirements at Farm 2, which were difficult to follow at Farm 1 acquiring water from the unregulated river, and; (3) intensive farming practices at Farm 2, including better management of soil nutrients, better use of labor and fertilizers, resulted in optimal crop canopy and high efficiency in harvesting solar energy. * Energy consumption to generate electricity is at off-farm level, that is why, further, electricity inputs to pump water were cut from the farm level and moved to the system level.
The data presented in Table 3 shows that irrigation water supply becomes very energy and carbon emission intensive under groundwater irrigation, which puts farmers under pressure to apply comprehensive farming practices to produce high yields of grapes to recover made investments in water. The intensive farming technology at Farm 2 resulted in the fast growth of grapes bushes, which required manual labor to cut a large number of leaves and branches; moreover, the grape harvest was labor-intensive. There were hidden electricity savings to pump water thanks to the highly efficient irrigation technology allowing the farmer to save energy against high water losses under gravity irrigation. GHG emissions due to groundwater pumping amounted to 0.51 kg CO 2 -e/m 3 .
Agricultural inputs energy and carbon emissions: Table 4 summarizes agricultural inputs, energy, and carbon emissions. Notes: Energy inputs are given for field-level only; the energy of applied agrochemicals is higher at Farm 2 mainly due to the increased application rate of ammonium nitrate given in Table 1. Table 4 showed that Farm 2 employing groundwater irrigation practices more labor inputs and agrochemicals, which required short-term investment. Farmers were ready to install deep wells, securing water for irrigation. However, smallscale farming and lack of capital did not allow them to invest in access to groundwater and agricultural machinery. Timely irrigation, intensive labor use, and agrochemicals permitted farmers to make farming practices comprehensive. However, it increased energy inputs and associated carbon emissions.

Data presented in
High grapes yield vs. high energy inputs and carbon emissions: Table 5 presents estimated farm-level environmental indicators of grape production under gravity and groundwater irrigation. Kg of CO 2 -e/MJ of energy input 0.04 0.05 * 944 grapes bushes at Farm 1 and 714 at Farm 2, ** Water productivity is given per m 3 of water depleted. Notes: Energy consumed is given for field-level only; the energy of the agrochemicals used on Farm 2 is higher than on Farm 1, mainly due to the increased application rate of ammonium nitrate. Table 5 also shows tradeoffs between higher yields and higher environmental impacts. The shift from gravity to groundwater irrigation and improved farming practices increased water-use efficiency. Despite similar application rates, groundwater irrigation allowed timely and efficient water application. Farmers receiving desired water applied intensive farming practices and invested more inputs, such as fertilizers, resulting in a three-timehigher grape yield at Farm 2, 18,600 kg/ha, compared to 6740 kg/ha on Farm 1. Introducing intensive farming and timely groundwater irrigation at low rates and high frequencies tripled the yield and water productivity. It also increased energy efficiency and energy productivity by 8% each. This was because the grapevines cover the entire area at the height of 2 m increased solar energy harvest. At the same time, plant roots occupying the inter-row space up to a depth of 50 cm allowed plants to use water and nutrients effectively.
The major disadvantage of the intervention was the significant increase in the energy inputs and CO 2 emissions per hectare of cropland and per m 3 of groundwater applied. The current 2.75 times increase in crop production is traded for future issues associated with 2.6 times more accumulation of GHGs in the atmosphere.
On-farm level estimates showed that farm-level interventions dealing with the waterenergy-carbon nexus were insufficient to produce mutual gains and disclose potential tradeoffs. Increasing food crop production and improving water productivity was balanced by growing energy inputs and CO 2 emissions, indicating a need for a wider, system-level consideration of the water-energy-carbon tradeoffs.

System-Level Analyses of Water-Energy-Carbon Nexus
There are different, case-specific, potential cases to improve the sustainability of agricultural production [62,63]. This study considers several alternative cases of grape production. These alternative scenarios are as follows.
Case 1: Gravity irrigation and conventional farming practices. Case 2: Pump irrigation using groundwater and conventional farming practices: a TPP is a source of electricity to pump groundwater. Case 3: Pump irrigation using groundwater, intensive farming practices, and water-saving irrigation: a TPP is a source of electricity to pump groundwater. Case 4: Using energy-efficient pumps to lift groundwater, intensive farming practices, and water-saving irrigation: a TPP is a source of electricity. Case 5: Case 4 plus supplying energy to pump groundwater from the upstream HPP. Case 6: Case 4 plus supplying energy to pump water from the midstream HPPs. Case 7: Case 4 plus supplying energy to pump water from the HPPs, located downstream of the main river. Case 8: Using a solar pump, intensive farming practices, and water-saving irrigation.
System-level analyses in this study include on-farm crop production and relevant water management and power generation and delivery system. System-level options assume that the impact of farm-level decisions on the system level and the opposite, system-level choices on the field level have to be considered to improve resources, water, energy, and productivity as well as reduce carbon dioxide emissions. For example, a transition to intensive farming practices, energy-efficient pumps, or water-efficient irrigation technologies, which is often farmer-driven, should consider a system-level impact, as this is where the decision would be made to switch from fossil fuels to renewable energy sources or improve the efficiency of electricity generation and distribution. For example, cases 3 and 8 may have similar results at the on-farm level. However, there is still a difference at the system level on CO 2 emissions. Alternatively, some cases may contribute to better resource productivity at the on-farm level. However, they may increase fossil non-renewable resources depletion at the off-farm level and increase carbon dioxide emissions. It is essential to clarify at the system level whether there are high productivities of water and energy, fossil resources still remain, and carbon dioxide emissions minimized. That is why linking on-farm alternatives with the system-level options is essential in one case. Analyzing such cases contributes to formulating future visions for developing agriculture, water, and energy sectors.
The first case represents Farm 1 producing grapes using conventional farming and gravity irrigation. The second case analyzes the shift from gravity to groundwater irrigation using thermal power-based electricity but no changes in farming practices. The third case represents Farm 2, producing grapes using groundwater, intensive farming practices, and water-saving furrow irrigation. The fourth case evaluates the adoption of energy-efficient pumps and improves power generation efficiency. The fifth to seventh cases consider shifting powering pumps from TPP to HPP. The eighth case considers shifting to solar energy. Cases 5, 6, and 7 analyze the gains and shortcomings of obtaining electricity from HPPs located upstream, midstream, and downstream of the rivers, respectively. They differ based on the relationship between electricity generation and evaporation losses from relevant reservoirs; evaporation losses were estimated using data available for HPPs located in the Syrdarya and Amudarya River basins, while CO 2 emissions from appropriate reservoirs were calculated using the relations given by IHA [64]. The emission rate for CO 2 was 48 gCO 2 -eq/kWh for solar PV (utility) [64].
The assessment of introducing drip irrigation to grape production at Farm 2 showed that replacing low-rate high-frequency furrow irrigation with drip irrigation may produce only minor water savings. This is due to the combination of intensive vineyard cultivation technology with furrow irrigation when: (a) plant roots cover the entire soil space between the vineyard rows up to 0.5 m below the ground surface, causing no significant water deep percolation losses; and (b) the vineyard leaves cover the entire surface area at the height of 2 m, resulting in irrigation water depletion mainly in the form of transpiration. The irrigation water losses from surface runoff, on average 10%-15% of the total water applied, are collected by the tailwater recovery system and directed to an adjacent field for irrigation use [40]. These estimates show that the farmer used the available water efficiently. For this reason, drip irrigation was excluded from further analysis.
The comparison of the cases included estimating the environmental indicators at the farm and system levels, including on-farm and off-farm. For example, in the case of the HPP energy supply, the on-farm level estimates include irrigation water and other farming inputs. In contrast, at the system level, additional water depletions have evaporation losses from a relevant reservoir and water delivery losses if losses form depletions. In the case of TPP energy supply, on-farm level estimates include energy inputs related to water and other farming inputs. In contrast, at the system level, additional energy inputs include the depletion of fossil fuel resources. The system-level inputs considered include on-farm-and off-farm-level water and energy inputs to pump groundwater; off-farm energy depletions associated with other farming inputs were beyond the scope of this study.
The comparison of the cases was undertaken using resource productivity and GHG emissions; however, there may be other case-specific factors and conditions beyond this study's scope. Figure 3 shows the difference between the energy input estimates for alternative cases at the on-farm and system levels.  . Energy inputs at (a) on-farm level and (b) system level for alternative cases of water and energy use. Depleted energy includes fossil fuel resources for TPP; renewable energy includes water evaporation losses for HPP and water for cooling solar panels. On-farm level estimates include the energy of water, farming inputs, and electricity to pump water. System-level estimates include energy for pumping water, system-level water consumption and fossil fuel resource depletion, and on-farm-level energy of farming inputs. The analysis at the farm level showed minor differences between the cases with pumping water (Figure 3a). While case 1, with gravity irrigation and conventional farming practices, indicates the lowest energy consumption per area, the shift to groundwater irrigation in case 2 and the introduction of intensive farming practices in case 3 increased energy inputs to a maximum. Other cases show similar energy inputs per ha, indicating . Energy inputs at (a) on-farm level and (b) system level for alternative cases of water and energy use. Depleted energy includes fossil fuel resources for TPP; renewable energy includes water evaporation losses for HPP and water for cooling solar panels. On-farm level estimates include the energy of water, farming inputs, and electricity to pump water. System-level estimates include energy for pumping water, system-level water consumption and fossil fuel resource depletion, and on-farm-level energy of farming inputs. The analysis at the farm level showed minor differences between the cases with pumping water (Figure 3a). While case 1, with gravity irrigation and conventional farming practices, indicates the lowest energy consumption per area, the shift to groundwater irrigation in case 2 and the introduction of intensive farming practices in case 3 increased energy inputs to a maximum. Other cases show similar energy inputs per ha, indicating no significant differences, except case 7, which uses hydropower downstream to pump groundwater.

Energy Inputs as Affected by Alternative Cases
The picture is different in Figure 3b. The system-level analysis considers fossil fuel consumption for electricity generation, its transportation and distribution, and the energy loss associated with water evaporation from reservoirs. The research shows high variations between cases with different energy sources for pumping water. In cases with TPP power supply, even in case 4 with improved water and energy efficiency, energy consumption is 1.5-1.9 times higher than those with HPP or solar power supply. In the cases with HPP power supply, the energy consumption in cases 5 and 6 is almost similar, while the energy consumption increases significantly in case 7. This is because of the excessive energy depletion associated with water delivery losses, recharging saline brackish water aquifers, and extra evaporation losses from the relevant reservoirs. Based on the system-level analysis, energy inputs are minimal for cases 5, 6, and 8, characterized by groundwater irrigation, improved water, and energy use, and renewable energy from the upstream (case 5) or midstream (case 6) HPPs or solar energy. System-level estimates show significant differences in energy inputs for crop production using thermal power and renewable power supply for groundwater irrigation. Figure 4 compares the total energy inputs, considering farm-level inputs against energy inputs at the system level. Figure 4 clearly shows that the worst cases characterized by high energy inputs occur when electricity is generated at TPPs for water pumping at farms using conventional (case 2) or intensified farming practices, including water-saving irrigation (case 3). The same crop production can be obtained with much lower energy inputs if electricity is produced using renewable energy, hydropower, or solar power (cases 5, 6, and 8, respectively). loss associated with water evaporation from reservoirs. The research shows high variations between cases with different energy sources for pumping water. In cases with TPP power supply, even in case 4 with improved water and energy efficiency, energy consumption is 1.5-1.9 times higher than those with HPP or solar power supply. In the cases with HPP power supply, the energy consumption in cases 5 and 6 is almost similar, while the energy consumption increases significantly in case 7. This is because of the excessive energy depletion associated with water delivery losses, recharging saline brackish water aquifers, and extra evaporation losses from the relevant reservoirs. Based on the systemlevel analysis, energy inputs are minimal for cases 5, 6, and 8, characterized by groundwater irrigation, improved water and energy use, and renewable energy from the upstream (case 5) or midstream (case 6) HPPs or solar energy. System-level estimates show significant differences in energy inputs for crop production using thermal power and renewable power supply for groundwater irrigation. Figure 4 compares the total energy inputs, considering farm-level inputs against energy inputs at the system level. Figure 4 clearly shows that the worst cases characterized by high energy inputs occur when electricity is generated at TPPs for water pumping at farms using conventional (case 2) or intensified farming practices, including water-saving irrigation (case 3). The same crop production can be obtained with much lower energy inputs if electricity is produced using renewable energy, hydropower, or solar power (cases 5, 6, and 8, respectively). Using power supply from the upstream HPP (case 5), the midstream HPP (case 6), or solar energy (case 8) reduces the energy intensity by 65%, and from the downstream HPP (case 7) by 53%, as compared to using thermal power (case 2). These data indicate no justification for replacing gravity irrigation with groundwater irrigation if electricity is supplied by thermal power. Figure 5 shows carbon dioxide emissions associated with crop production under different water and energy inputs cases. Using power supply from the upstream HPP (case 5), the midstream HPP (case 6), or solar energy (case 8) reduces the energy intensity by 65%, and from the downstream HPP (case 7) by 53%, as compared to using thermal power (case 2). These data indicate no justification for replacing gravity irrigation with groundwater irrigation if electricity is supplied by thermal power.  On-farm level estimates show that introducing intensive farming practices doubled CO2 emissions; however, there are no differences between the cases with groundwater irrigation using different sources of power supply (Figure 5a). The on-farm level assessments were insufficient to disclose water, energy, and carbon tradeoffs in the studied cases.

Carbon Dioxide Emissions as Affected by Alternative Cases
The system-level estimates of CO2 emissions clearly show the differences between the cases (Figure 5b). In the cases with TPP power supply for pumping water, conventional farming practices (case 2) resulted in 8.2 times higher CO2 emissions than under gravity irrigation. Intensifying farming and improving irrigation practices (case 3) increased CO2 emissions by 11%. Improving pumping efficiency and power generation efficiency (case 4) reduces CO2 emissions by 36% compared to case 2. Advancing TPP energy On-farm level estimates show that introducing intensive farming practices doubled CO 2 emissions; however, there are no differences between the cases with groundwater irrigation using different sources of power supply (Figure 5a). The on-farm level assessments were insufficient to disclose water, energy, and carbon tradeoffs in the studied cases.
The system-level estimates of CO 2 emissions clearly show the differences between the cases (Figure 5b). In the cases with TPP power supply for pumping water, conventional farming practices (case 2) resulted in 8.2 times higher CO 2 emissions than under gravity irrigation. Intensifying farming and improving irrigation practices (case 3) increased CO 2 emissions by 11%. Improving pumping efficiency and power generation efficiency (case 4) reduces CO 2 emissions by 36% compared to case 2. Advancing TPP energy generation (case 4) reduced the emission rate; however, the case remains uncompetitive against cases with renewable energy supply (cases 5-8).
The shift from thermal energy to hydropower energy reduced CO 2 -e. The energy supply of groundwater irrigation from the upstream HPPs (case 5) and midstream HPPs (case 6) had the lowest CO 2 emissions at 1173 and 1225 kg of CO 2 -e/ha, respectively. These rates are comparable to emissions when solar power was applied (1200 kg of CO 2 -e/ha) while using the downstream HPP power supply, CO 2 emission rates were 1655 kg of CO 2 -e/ha. The system-level assessments disclosed the benefits of applying renewable energy to the low-carbon transformation of food production on irrigated land. Table 6 shows the main environmental indicators of the studied cases.
The data presented in Table 6 indicate that increasing crop production and obtaining gains in water, energy, and carbon, including the highest crop production per carbon dioxide emissions, are found for cases 5, 6, and 8. No case with a TPP power supply was suitable for groundwater irrigation, even using water-and energy-saving technologies. These cases have high water productivity, but increased CO 2 emissions characterize low energy efficiency and productivity. The water productivity indicator alone was not enough to conclude on the sustainability of the crop production because the WP value did show no differences between the cases with non-renewable and renewable energy sources characterized by significantly different carbon emissions. In contrast, the crop yield ratio to the carbon emissions directly reflected the impact of the case on the use of energy sources, carbon emissions, and crop yields. The energy efficiency and the energy productivity indicators indirectly reflected differences in carbon emissions. The energy efficiency indicator did not consider the changes in crops yields. The worst-case for crop production development was the case of groundwater irrigation using conventional irrigation and farming practices; the second-worst case was the shift to groundwater irrigation using intensive farming practices with no improvements in irrigation practices. Case 4, using TPP-based electricity, had higher water productivity than conventional farming with gravity irrigation (case 1). However, the depletion of non-renewable resources and relatively high carbon emissions make these cases not applicable for the future.
Groundwater irrigation using downstream HPP (Case 7) is characterized by lower water and energy productivity and crop yields per kg of CO 2 emissions than the other HPP cases due to increased evaporation losses and CO 2 emissions from the relevant reservoir off-farm water losses to saline groundwater. Case 8, with groundwater irrigation using a solar pump, showed the highest water and energy productivity and the crop yield per kg of CO 2 emissions.
From an environmental viewpoint, the current strategy of gravity irrigation (case 1) showed many advantages. There are no fossil fuel inputs for power generation to lift water, and the energy inputs and CO 2 emissions are low. However, practicing gravity irrigation depends on the availability of water according to crop water requirements; conventional farming practices often applied along with gravity irrigation produce low crop yields and are characterized by low water and energy productivity. The case of replacing gravity irrigation with a low-head micro-irrigation system, which requires power to pump water into the system, was beyond the scope of this study.

Discussion
Water productivity is considered one of the main indicators of basin sustainability [65,66]. Brauman et al. [65] highlighted that improved crop water productivity increases water sustainability. Cai et al. [66] gave different examples, including when increasing a basin's water productivity was at the expense of agricultural sustainability. This study results suggest that in the environment, when carbon emissions become the main factor of environmental and agricultural unsustainability, WP indicator is not enough alone to decide whether the crop production is sustainable because, despite it being an important indicator of crop return on consumed water, it does not reflect differences of alternative cases in carbon emissions. From this point of view, the crop yield response to kg of carbon emissions indicator can be applied along with water productivity to compare alternative cases of agricultural production using pumped irrigation. To obtain a complete picture of environmental sustainability, it is also essential to consider the third indicator-the depletion of non-renewable fuel resources. The scale of analyses of the indicators is vital.
Analyzing the water-energy-carbon tradeoff only at the farm level [22,23] may mask the gains and shortcomings of interventions to improve water and energy use and hidden environmental impacts. On-farm level analyses show the effect of farm-level inputs and measures to improve their efficiencies on crop yield and the increasing energy inputs and corresponding CO 2 emissions. The energy source-hydro, solar, or thermal-makes no difference from a farm perspective. However, the energy source makes a big difference at the system level. A system-level analysis, including on-farm crop production and linked off-farm water and energy sources depletions, is necessary to find possible solutions that may produce gains in crop production, resource reduction, including water and energy, improved water, and energy productivity, and reduced CO 2 emissions.
The system-level analysis indicates that the energy consumption for crop production is high when the electricity supply for water pumping comes from a TPP. Fossil fuel resources become the main component of energy inputs, making thermal power less compatible with renewable energy resources. The results of this study suggest that it is not sustainable to expand thermal power because of the high depletion rate of fossil fuel resources, while improving and advancing existing thermal power plants may reduce CO 2 emissions and contribute to reducing the depletion of fossil resources. Besides introducing intensive farming practices and water and energy-saving technologies, improving pumping and energy generation efficiencies may bring about resource reduction and environmental benefits [27]. However, pumped irrigation using thermal power cannot deal with waterenergy-carbon tradeoffs. It reduces energy efficiency and productivity, accelerates fossil fuel resource depletion, and increases CO 2 emissions. Under conditions of increasing long-term temperature, due to climate change and depleting fossil fuel resources, the gradual shift to environmentally friendly renewable energy resources is a better solution than meeting the growing needs by expanding all available resources, including thermal resources.
For assessing the sustainability of water management, a basin's level is an appropriate scale [67][68][69]. Analyzing trade-offs for water, energy, and carbon may require linking basin water management to corresponding power generation alternatives located outside the basin. For example, inter-basin water transfers from the Amudarya River basin are critical for sustainable water supply in the lower reaches of the Zerafshan River and the Kashkadarya River. However, the reliable operation of pumping stations that pump water depends on power plants located outside of their basins. In such cases, trade-off analysis requires linking the basin water management analysis with energy supply alternatives.
The estimates show that hydropower-based pumped irrigation has advantages over thermal-power-based irrigation: fewer energy inputs in crop production and less carbon intensity at the system level. These gains can be achieved by accessing upstream and midstream hydropower resources. Many studies consider the practice of obtaining hydropower electricity in summer and trading for fossil natural gas in winter as a benefit-sharing approach for transboundary river basins with upstream hydropower and downstream agriculture [32]. This study indicates that the joint development and use of upstream hydropower may reduce the depletion of fossil fuel resources and lead to more sustainable agriculture based on pumped irrigation. The upstream hydropower development should consider water for food, energy, and the environment. This case can be pioneered in the case of the Zerafshan River upstream of Tajikistan and downstream to Uzbekistan-there is an agreement for two HPPs to be installed by the downstream country in the river's upstream.
Similar advantages are inherent in the development of midstream hydropower. This approach means developing small midstream hydropower in the Fergana Valley, Tashkent, Surkhandarya, and Kashkadarya regions in Uzbekistan or the northern part of Tajikistan. This proposal coincides with the ongoing construction of new small HPPs on small rivers and canals in many regions of Uzbekistan and Kazakhstan. Small HPP installation does not require extra water depletion because the downstream reservoirs can capture hydropower releases; there are minor power transportation and distribution losses and CO 2 emissions. HPPs on small rivers and widespread canals can produce energy from upstream hydropower releases and supply power for developing groundwater and wide adoption of micro-irrigation.
The study showed the shortcomings of installing hydropower downstream, which may cause high water losses due to evaporation in the hotter desert zone.
There are also environmental and social considerations in prioritizing hydropower while infringing the needs of other uses, such as the environment and agriculture [70][71][72][73]. Therefore, hydropower development must ensure environmental flows, with high importance for river ecosystem services. In the case of hydropower releases from reservoirs in winter, agriculture may have irrigation water shortages in summer. This is where further tradeoff analysis is required to minimize system-level losses. Mitigation measures may include the development of sufficient capacities for re-regulating upstream hydropower releases in the midstream, focusing on preserving river ecosystem services downstream, shifting from canal irrigation to groundwater, and using a managed aquifer recharge technique. Hydropower development must contribute to diversifying benefits from water rather than prioritizing a single water use. This requires close cooperation of riparian states in energy production and water management to enhance food production.
Solar power was confirmed as the best option for water-energy-carbon tradeoffs. Groundwater development using solar power requires minimum resources, water, energy, inputs, high resource efficiency, and productivity. It produces minimum CO 2 emissionsthe emissions were 51% less than water pumping using electricity from TPPs. This is higher than the emission reduction value found by De Vlugt [74] for groundwater pumping for irrigation in India.
Developing solar power could be an essential contribution to sustaining the growing demand for energy supply and ensuring win-win outcomes. While there are plans to install six 100 MW solar power plants by 2030 in Uzbekistan and similar developments in neighboring countries, they can be further expanded to cover the growing energy demand.

Conclusions
Amid the growing demand for food and increasing pressure on limited, vulnerable river flow, the shift to pumped irrigation using groundwater has addressed water scarcity in agriculture. This adaptation strategy creates a base for increasing crop production by introducing intensive farming practices, water-saving irrigation, and energy-saving technologies. However, in this case, improved yields of crops and water productivity may be traded for high carbon inputs and CO 2 emissions. This study results suggest that in the environment, when carbon emissions become the main factor of environmental and agricultural unsustainability, WP indicator is not enough alone to decide on the sustainability of crop production because, despite it being an important indicator of crop return on consumed water, it does not reflect differences of alternative cases in carbon emissions. Under climate change induced by anthropogenic carbon emissions, a combination of three indicators, including crop yields per kg of carbon emissions, water productivity, and deple-tion of non-renewable fuel resources, can be used to clarify whether crop production using groundwater irrigation is environmentally sustainable.
The study revealed that only a farm-level analysis of water-energy-carbon tradeoffs might mask the gains and shortcomings of interventions to improve water and energy use and hidden environmental impacts. Instead, a system-level that includes a combination of farm-level and off-farm-level analysis of the tradeoffs is necessary to find crop production cases, which are environmentally sustainable.
The study found that, in Fergana Valley, the shift to groundwater irrigation using thermal power supply makes crop production unsustainable-groundwater supply is based on rapidly depleting fossil energy resources, with crop yields per kg of CO 2 emissions two times less than when electricity for groundwater pumping comes from the upstream or the midstream hydropower. The worst cases, characterized by high carbon emissions, occur when electricity is generated at thermal plants that pump water to farms using conventional or intensified farming practices.
The study shows that hydropower-based pumped irrigation has advantages over thermal-power-based irrigation. Hydropower can ensure timely irrigation and increase resource efficiency and productivity simultaneously by improving power availability, producing much lower CO 2 emissions. With the same water productivity, grape yields (kg/ha) per kg of CO 2 -e/ha were 15.2-15.9 using electricity from hydropower against 7.7 kilograms of grapes/kg of CO 2 -e using thermal plants. Many previous studies have considered the practice of obtaining hydropower electricity in summer traded for fossil natural gas in winter as a benefit-sharing approach for transboundary river basins with upstream hydropower and downstream agriculture. However, the present study suggests that this approach does not sustain power generation in the basin while continuously utilizing fossil carbon reserves. Instead, joining the development and use of upstream hydropower by riparian states reduces the depletion of fossil fuel resources, making agriculture more sustainable based on pumped irrigation. Hydropower development must ensure environmental flows and consider food, energy, social, and environmental tradeoffs to bring basin-scale benefits. This requires close cooperation between riparian states on water management and energy generation to enhance food production.
Midstream hydropower can serve to re-regulate upstream hydropower releases. The installation of small HPPs does not require extra water depletion because the downstream reservoirs can capture hydropower releases; there are minor power transportation and distribution losses and CO 2 emissions. Installing power stations on small rivers and widespread canals can form a basis for developing groundwater and wide adoption of micro-irrigation. Thus, hydropower development must contribute to diversifying benefits from water rather than prioritizing a single water use. The study indicates that downstream hydropower is less attractive because it causes extra water depletion and CO 2 emissions from the related reservoirs.
The results of the study highlight that solar power is the best case to deal with water-energy-carbon tradeoffs. Pumped irrigation using solar power showed similar to hydropower grape yields per kg of carbon emissions, water, and energy productivities; In contrast, hydropower may reallocate water from food production and increase water shortages for agriculture solar power requiring minimum water resources, which has no environmental impacts.
The three pillars for addressing water, energy, and carbon tradeoffs for sustainable food production are system-level interventions for water and energy saving, use of renewable energy to pump water for irrigation, and river basin scale cooperation. These pillars can lead to low carbon transformation of pumped irrigation-based food production in Central Asia and elsewhere. Further studies are required to analyze other cases, link cropping systems, water saving irrigation strategies, and alternative energy sources, and make future projections to achieve environmental sustainability in agricultural production. A special focus in future studies is to be given to a scale of water management sustainability analyses under the water, energy, and carbon tradeoffs at different levels-field, system, and basin.
Funding: This research received no external funding.