The Yellow River Basin is of the utmost importance for China in terms of food production, natural resources management, and socioeconomic development. Water withdrawals for agriculture, industry, and households in the past decade have seriously depleted environmental and ecological water requirements in the basin. This study presents a modeling scenario analysis of some water development strategies to harmonize water withdrawal demand and ecological water demand in the Yellow River Basin through water savings and interbasin water transfers. A global water and food analysis model including the Yellow River Basin as one of the modeling units is applied for the analysis. The model demonstrates that there is little hope of resolving the conflict between agriculture water demand and ecological water demand in the basin if the current water use practices continue. Trade-offs exist between irrigation water use and ecological water use, and these trade-offs will become more intense in future years with population growth, urbanization, and industrial development as well as growing food demand. Scenario analysis in this study concludes that increasing basin water use efficiency to 0.67 first and then supplementary water availability by interbasin water transfer through the South-North Water Transfer Project may provide a solution to water management of the Yellow River Basin in the next 25 years.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The Yellow River Basin plays a critical role in China's agricultural production, natural resource reservation, and socioeconomic development. The total cultivated area in the basin is 12.9 million hectares, about 13% of the total in China, but the basin holds only 3% of the country's water resources. The river also feeds over 50 cities along the basin with a population over 60 million and supplies significant amounts of water to chemical, oil, and mining industries in the middle and lower reaches of the basin. Figure 1 illustrates the location of the Yellow River Basin (YRB) and where the divisions between upper, middle, and lower reaches of the basin lie.
 Since the founding of the People's Republic of China, over 3100 large, middle, and small-sized reservoirs have been built in the basin, with a total storage up to 70 km3 in 1999 after the Xiaolangdi Reservoir was completed in the middle stream portion of the basin. The storage is greater than the long-term average annual runoff in the river (58 km3 based on hydrologic series during 1919–1975). Meanwhile, the expansion of irrigated area has been especially rapid in the YRB, rising from 0.8 million hectares in 1950 to 7.5 million hectares in 2000 (G. Li, Thoughts and considerations on long-term Yellow River development and management, http://www.yellowriver.gov.cn, accessed in December 2002). Irrigation growth, together with rapid growth of industrial and municipal water uses, has resulted in a dramatic increase of water withdrawal over the entire basin [Zhu et al., 2004]. Table 1 provides water withdrawal and consumption figures by sector and by source for 1998, 1999, and 2000. As can be seen, agriculture water uses cover over 80% of total water withdrawal and total water consumption in the basin. Another characteristic of water use in the basin is that the fraction of consumption (ratio of consumption to withdrawal) is higher than that in many other basins in the world, with an average in 1998–2000 as high as 75%. The world average is 46% (in 1995) [Rosegrant and Cai, 2003]. The high consumption in the YRB is due to the comparatively high percentage of agricultural water use, almost full use of return flow in the downstream irrigation districts through conjunctive use of canals and wells (Yellow River Conservation Commission (YRCC), Yellow River Water Resources Bulletins 1998–2000, http://www.yellowriver.gov.cn, accessed in November 2002), and considerable water loss (about half of the total consumption is categorized as nonbeneficial, see detailed discussion in section 3.2). Table 2 lists the total consumption (including water consumed during production process and nonprocess consumption such as river evaporation and use by nonagricultural vegetation), basin outflow to the ocean, and storage change in 1998, 1999, and 2000. The share of total consumption is as high as 77% of the total runoff, as an average figure of the 3 years, and outflow accounts only 13%. Total water withdrawal has been even greater than the amount of the total runoff, indicating that there is significant reuse of water. For example, the ratio of water withdrawal to total annual renewable water resources in 2000 was 1.10. This ratio, called the criticality ratio indicates a “very high water stress” according to Alcamo et al. .
Table 1. Water Uses by Sector by Source in the YRB in 1998, 1999, and 2000a
Depletion By Sector
Sources are Yellow River Conservation Commission (YRCC) Water Bulletins for 1998–2000.
Table 2. Water Balances in the YRB in 1998, 1999, and 2000a
Total Renewable Water (TRW), km3
Consumption (WC), km3
Basin Outflow (BO), km3
Storage Change, km3
Sources are YRCC Water Bulletins for 1998–2000.
 Similar to other regions in the world, in the YRB, human demand on natural resources has increased rapidly over the past few decades, while the aquatic and riparian ecosystems have felt the effects of the increases, resulting in loss of high-quality water, productive soil, and diverse ecological functions that are critical for plant and animal communities. Excess water withdrawal and consumption in the YRB has caused river desiccation. The most striking evidence of this desiccation stems from cutoffs experienced in the main channel downstream of the river during 1972–1998, which caused serious problems to water supply and ecosystems in the downstream area. During the period, flow in the main channel has been cut off in 21 years, accumulated to 1050 days. Moreover, the river had been cut off every year during 1990–1998 with both the duration of time and the distance from the river mouth increasing. In 1997, there was no flow out of the basin for 226 days, and the river dried up to Kaifeng, ∼600 km from the river mouth. Starting in 1999, the Chinese government began to strengthen the “Water Allocation Program,” which specifies water withdrawal quotas for provinces along the basin to remedy this problem. The status of the river has since been improved, and there has been no absolute flow cutoff in recent years. However, flow for ecological uses, especially for sediment flushing, is still far below optimal levels.
 Continued high-flow diversions would become self-defeating. In the YRB, sediment accumulation due to insufficient flow for sediment flushing makes the middle and downstream of the basin more vulnerable to flooding damage than ever. Wastewater discharge around the large cities along the river has polluted the river, particularly during the low-flow season partially because of lack of enough flow for pollutant dilution, which affects water availability and uses in the downstream reaches. Groundwater overdraft is found in the middle stream reaches and can likewise lead to the loss of an important water source for human uses in the future. Flow depletion in the delta area has already led to seawater intrusion and wetland recession in the delta area.
 In part because of the effects of the problems on the ecological environment, ecological restoration for sustainable water development has been given a high priority in the recently amended Chinese Water Law (2002). The law stipulates that ecological restoration must be given added weight in areas where ecosystems are seriously destructed. In the YRB, agriculture uses extract more water than any other sector, and they are primarily responsible for the depletion of flow needed for downstream sediment flushing and ecological preservation. Therefore saving agricultural water is now under greater pressure in order to reserve water to serve ecological and environmental functions.
 Other options have also been considered to eliminate the tension between water withdrawals and ecological water requirements, including interbasin transfer to supplement water sources in the YRB during the dry season. The South-North Water Transfer Project (SNWTP) will divert water from the Yangtze River in south China (see Figure 1 for the location of the Yangtze, whose annual runoff is about 15 times of the YR's) to north China [Berkoff, 2003]. Three diversion routes have been considered for the SNWTP, including the eastern, middle, and western routes. Currently, the eastern route is under construction, the middle route is approved for construction, and the decision for the western route is pending. The western route will divert water from the upper Yangtze tributaries to the upper Yellow River, providing an important supplement to the resources already in the YRB [Li, 2002; Berkoff, 2003].
 This study presents a modeling scenario analysis for water development strategies that could potentially be employed to harmonize water withdrawal demand and ecological water demand in the basin through water savings and interbasin water transfers. In the remaining sections we discuss the environmental and ecological water requirements (EEWR) in the YRB and then describe the modeling method and definition of the analytical scenarios based on data assessments of 1995 and projections of 1996–2025. Following that, results are presented, showing trade-offs between agricultural and ecological water uses and the impacts of water saving and possible interbasin water transfers.
2. Minimum EEWR in the YRB
 The major EEWR for the YRB is the preservation of flow, particularly for the downstream reaches of the basin, including flow to protect the wetland ecosystems, recharge groundwater in the costal area to prevent seawater intrusion, and flush sediment to the sea. The minimum level of EEWR refers to a threshold below which the quality of the environmental ecological systems may be degraded [Ni et al., 2002].
 According to estimates from the YRCC, the average flow rate at Lijing Station (close to the river mouth) should be 300 m3/s. This average rate entails a minimum flow rate of 50 m3/s and a total volume in the nonflooding season of 5 km3, which is ∼10% of the 8-month nonflooding period during the 11-year successive droughts from 1922 to 1932. Following some of professional assessments (e.g., H. Shangchi, YRCC, personal communication, 2002), the delta ecosystems were basically sustained during that period under such flow rate and volume, and therefore this flow rate and volume are considered to be the minimum sustainable EEWR. Ni et al.  and Zhu and Zhang  identified a similar threshold for the minimum water requirement during the nonflooding season.
 The flow requirement for sediment flushing is recognized as the largest EEWR in the YRB. The most challenging engineering aspect of managing the river is without doubt the control of the exceptionally high sediment load that the river carries to its lower reaches [Leung, 1996]. Through thousands of years, excessive sediment deposits have raised the riverbed several meters over the surrounding grounds; it is as much as 10 m above the city level of the ancient capital, Kaifeng. Over the last 50 years, sediment deposit in the downstream channel is estimated to be 10 billion tons, resulting in 2–6 m increase of the riverbed level [Li, 2002]. The accumulated sediment deposit has aggravated the threat of flooding. For example, in 1996, flow passing Huayuankou (a hydrologic station at the river section established as the beginning of the downstream) at a peak rate of 7600 m3/s made the river level rise to 94.7 m, which is 0.9 m higher than that caused by a flow rate of 22,300 m3/s in 1958 at the same station. Therefore ensuring flow enough to flush sediment is a problem unique to the YRB. It is estimated that sediment flushing requires 15 km3 of flow in the downstream channel during the flood season in which most of the sand is loaded into the river channel [Zhu and Zhang, 1999], while Ni et al.  argued that the minimum flow requirement for sediment transport during the flood season should be around 20 km3, 5 km3 higher than the previous estimate. In the analysis of this study we use 15 km3, which is currently under the approval of YRCC.
 Other EEWRs for the Yellow River include water to be stored within small catchments in the middle reaches with heavy and coarse-sediment yield to reduce sand load to the main channel, water diversion from the major tributaries or the main channel to recharge groundwater in regions where groundwater is over drafted, and adequate levels of streamflow to dilute wastewater discharge in the nonflooding season. The required levels for these EEWRs are relatively small compared to flow required for downstream wetland ecosystems and sediment flushing. According to Zhu and Zhang , flow blocked within the small catchments in the middle reaches to prevent soil erosion was 1.0 km3 in 1997. The total EEWR is then estimated as 21 km3 for 1 year, including 15 km3 for sediment flushing in the flood season, 5 km3 for the preservation of downstream ecosystems in the nonflood season, and 1 km3 blocked within the small catchments for the purpose of erosion control.
 The EEWRs for the YRB as discussed above are based on separate assessments in terms of various environmental and ecological requirements in the context of the basin. Arthington  provided a comparative evaluation of environmental flow assessment techniques and recommended “a holistic approach” to the assessment of environmental flows. A holistic approach is based on an all-inclusive concept of “riverine ecosystems.” It builds critical elements of the natural flow regime into the managed system, including the source area, river channel, riparian zone, floodplain, groundwater, wetlands, and estuary and features such as the preservation of species. Such an approach is promising if a comprehensive assessment of the environmental and ecological flows for the YRB is to be constructed.
 Moreover, until now, only objective measure of ecological flows has been used in current YRB management. That is to say, the measure of the flows is based on scientific studies or empirical judgments, while in many western countries, EEWRs are determined by a combination of legislative, regulative, and legal procedures tempered by social values and only partly predicated on scientifically justified criteria [Zhu et al., 2004]. This comprehensive approach will be valuable for the YRB considering the cultural history which has lasted thousands of years in the basin.
 In this study, we use the EEWR currently adopted in the water management of YRB as a base and deal with the trade-off between EEWR and agricultural production in the basin. The analysis is conducted in an integrated modeling framework, which implements the interactions between water management and agricultural production at the river basin scale. Some engineering and managerial measures are analyzed as possible solutions to resolve the conflict between EEWR and agricultural production in YRB.
3.1. An Integrated Water and Food Modeling Framework
 An integrated water and food model with the YRB as one of the 69 modeling units in the global scope is applied for this study [Rosegrant et al., 2002]. The global modeling framework combines an extension of the international model for policy analysis of agricultural commodities and trade [Rosegrant et al., 2001] with a newly developed water simulation model [Cai and Rosegrant, 2002] based on state-of-the-art global water databases and models, integrated basin management, field water management, and crop water modeling. The model attempts to project and analyze how water availability and demand would evolve over the next three decades (from a base year of 1995), taking into account the availability and variability in water resources, water supply infrastructure, and irrigation and nonagricultural water demands as well as the impact of alternative water policies and investments on water supply and demand.
 In the water module, water available for food production depends on climiate varaibles, water supply infrastructure, water quality, and socioeconomic and environmental policies, as illustrated by Figure 2. Water demands are simulated as functions of the year-by-year hydrologic fluctuations, irrigation development, growth of industrial and domestic water uses, and environmental and other flow requirements (committed flow). Committed environmental and ecological flows are treated as a predetermined hard constraint in water supply, and off-stream water supply for domestic, industrial, livestock, and irrigation sectors is determined through two steps. The first step determines the total water that could be depleted in each time period (month) for various off-stream uses, and the second determines water supply for different sectors. The model assumes domestic water demand is the first priority, industrial and livestock water demand is the second, and irrigation water supply is the residual claimant. Moreover, irrigation water supply is further allocated to different crops in the basin based on crop water requirements and profitability.
 In the food module, crop harvested areas and yields are calculated through crop-wise irrigated and rain-fed area and yield functions. These functions include water availability as a variable, through which the food module is connected with the global water simulation model. The combined water-food modeling framework provides a wide range of opportunity for analysis of water availability, food security, and environmental preservation at basin, country, and global scales. Many policy-related water variables are involved in this modeling framework, including potential irrigated area and cropping patterns, maximum allowed water withdrawal due to infrastructure capacity and environmental constraints, water use efficiency, water storage and interbasin transfer facility, rainfall harvest technology, allocation of agricultural and nonagricultural uses, and allocation of in-stream and off-stream uses. For the sake of exploring alternative futures, investment and management reform can influence the future paths of these variables, which influence food security at both national and global scales. The driving forces for scenario analysis are presented in Figure 3. A detailed description of the modeling method is given by Rosegrant et al. .
3.2. Terms for Water and Food Analysis in the Context of the YRB
 Three terms are introduced: Agricultural water use efficiency, irrigation water supply reliability, and irrigation crop production reliability. These terms are used in the definition of scenarios and the result are presented in the following contexts.
3.2.1. Agricultural Water Use Efficiency (AWUE)
 We use effective efficiency at the river basin scale [Keller et al., 1996] to represent irrigation water use efficiency, which is a ratio of beneficial irrigation water consumption (BIWC) to total irrigation water consumption (TIWC):
BIWC refers to actual crop evapotranspiration of all crops over 1 year in the entire basin, and TIWC is equal to BIWC plus nonbeneficial water consumption or depletion including evaporation loss, salt and other pollutant sinks, and economically nonrecoverable seepage. TIWC is assessed as 30.4 km3 in 1995 in the YRB, and BIWC is estimated as shown in Table 3 based on hydrologic parameters, irrigated areas and yields, and empirical relations and assumptions, which are illustrated in the notation of Table 3. The BIWC is estimated as 15.8 km3. From equation (1), AWUE is calculated as 0.52. It should be noted that the estimated AWUE represented as effective efficiency is higher than the efficiency value estimated by Chinese irrigation experts, ∼0.3–0.4, which is represented by classical irrigation efficiency [Qian and Zhang, 2001]. Although the classical efficiency concept is appropriate for irrigation system design and management [Doorenbos and Pruitt, 1977], it could lead to erroneous conclusions and serious mismanagement of scarce water resources if it is used for water accounting at a larger scale. This is because the classical approach ignores potential reuses of irrigation return flows [Keller et al., 1996]. The effective efficiency takes into account potential reuses of irrigation return flows. Reuses of irrigation return flows have been common in the YRB. Table 4 shows irrigation water withdrawals and return flows in different river reaches in 2000 assessed by YRCC. In the upper reaches (above Toudaoguai), return flow is as high as 35% of the withdrawal, which obviously provides a “source” for middle stream and downstream water withdrawal. In the downstream reaches (below Huayuankou), return flow cannot come back to the main channel because the riverbed is above the ground level. However, water is recycled and reused through the well-developed canal-well systems in the downstream areas (H. Li, YRCC, personal communication, 2002).
Table 3. Estimation of Beneficial Irrigation Water Consumption in the YRBa
ETC, potential crop evapotranspiration; Yr relative crop yield [Cai and Rosegrant, 1999]; ETA, actual crop evapotranspiration; PE, effective rainfall [Cai and Rosegrant, 1999]; IW, irrigation water requirement; AI, irrigated area; WI, irrigation water requirement; N/A, not applicable.
ETA = ETC [1 − (1 − Yr)/Ky], where Ky is the crop response coefficient to water stress [Doorenbos and Kassam, 1979].
IW = ETA − PE.
WI = AI × IW/105.
Other roots and tubers
Table 4. Year 2000 Withdrawal and Depletion Along YR Reachesa
Withdrawal, 109 m3
Depletion, 109 m3
Return Flow, 109 m3
Return Flow/Withdrawal, %
Source is 2000 YRCC Water Bulletin. LZ, Lanzhou; TDG, Tou-Dao-Guai; LM, Long-Men; SMX, San-Men-Xia; HYK, Huan-Yuan-Kou; LJ, Li-Jin.
3.2.2. Irrigation Water Supply Reliability (IWSR)
 IWSR is defined as the ratio of actual irrigation water supply (AIWS) to the potential irrigation water requirement (PIWR), with AIWS represented by TIWC:
and PIWR is estimated by
in which c is the index for crops, and st is the index for crop growth stages, kc is the crop coefficient, LR is the salt leaching factor, characterized by soil salinity and irrigation water salinity, and other variables have been defined in the notation of Table 3.
 With AWUE = 0.52, LR = 0.15, and data given in Table 4, PIWR is estimated as 34.5 km3, and by equation (2), IWSR of the YRB in 1995 is calculated as 0.88. For years 1996–2025, IWSR is an output from the model as shown in section 5.
3.2.3. Irrigated Crop Production Reliability (ICPR)
 ICPR is defined as a ratio of actual irrigated crop production (AICP) to the potential irrigated crop production (PICP):
In this study, we assess this item for cereal crops as a whole. AICP in the YRB is estimated to be 15.8 million tons in 1995, with actual yield of 3.6 metric tons per hectare and actual harvested area of 4.4 million hectare [Gunaratnam and Kutcher, 1997]. The potential yield is 4.0 metric tons per hectare, or 10% higher than the actual yield, and potential harvested area is 4.6 million hectares, or 4% higher than the actual harvested area [Cai and Rosegrant, 1999]. Thus PICP is estimated as 18.2 million tons. From equation (4), ICPR is calculated as 0.87. For the years 1996–2025, ICPR is an output from the model, as shown in section 5.
4. Scenario Definition
 Five scenarios are analyzed in this study. The first one is a business-as-usual scenario (BAU) which projects the likely water and food outcomes for a future trajectory based on the recent past, whereby current trends for water investments, water prices, and management are broadly maintained. Table 5 shows assessments in 1995 and projections in 2025 of key parameters under BAU. Since the downstream flow requirement for sediment flushing and ecosystem preservation is approximately equal to the basin outflow to the ocean, we use basin outflow to represent the ecological water use. Under BAU, there is no requirement set for the basin outflow, which basically reflects the practices during the last decade. That is to say, irrigation water withdrawals were driven by crop water demand under low water use efficiency and were not constrained by the downstream ecological flow requirement. The BAU assumes this state of affairs continues through next 25 years.
Table 5. Data Assessments (1995) and Projections (2025)a
 The second scenario (ecological scenario (ECO)) assigns higher priority to EEWR than agriculture, assuming the downstream EEWR will be satisfied as much as possible after 2001. That is, the basin outflow should not be less than 20 km3, as discussed in section 2. The two scenarios, BAU and ECO, are used to explore trade-offs between agricultural water demand and ecosystem water demand. The three scenarios defined below attempt to search possible measures to meet both agricultural and ecological water demand.
 1. ECO-AWUE is ECO with significant improvement in irrigation water use efficiency, assuming the effective efficiency in the basin will gradually increase to 0.76 by 2025 from 0.52 in 1995. According to the assessment of Rosegrant et al. , AWUE of 0.76 would be very high compared to other basins in the world. For example, the effective efficiency in California basins was estimated as 0.77 in 1995. These increases will be accomplished by substantial improvement of water demand management and large investments in advanced irrigation systems.
 2. ECO-SNWTP is ECO with source supplement from SNWTP, assuming that up to 4.0 km3 of water can be provided to the YRB from the SNWTP in the dry season by 2010 and 9.0 km3 can be provided by 2015. These are the amounts predicated to come with the first and second stage of the western route of the SNWTP, respectively (G. Li, Thoughts and considerations on long-term Yellow River development and management, http://www.yellowriver.gov.cn, accessed in December 2002).
 3. ECO-AWUE-SNWTP is ECO with a suitable improvement of irrigation water use efficiency and possible water supplement from the SNWTP, assuming AWUE increases to 0.67 by 2010 (and this value stays to 2025), followed by the same water supplement from SNWTP as specified under ECO-SNWTP.
 Under all of these scenarios the projected hydrologic regime between 1996 and 2025 is modeled based on data (including precipitation, evapotranspiration, and runoff) from the period between 1961 and 1990 from WaterGap 2.0, Kassel University, Kassel, Germany [Alcamo et al., 2000]. In this study, we use the global model but focus on the YRB only, and assumptions and projections for other units under all these scenarios follow business-as-usual conditions described by Rosegrant et al. .
5. Analysis and Results
 Modeling results from the five scenarios defined above are compared in terms of irrigation water supply reliability, irrigated crop production reliability, cereal production, and satisfaction of EEWR. First, we examine the trade-offs between irrigation water use and ecological water use, based on the outputs from BAU and ECO; then we discuss optional water development strategies that aim to match both agricultural water demand and ecological water demand in the next 25 years, based on outputs from ECO-AWUE, ECO-SNWTP, and ECO-AWUE-SNWTP.
 To show the trade-offs between irrigation water use and ecological water use, Figure 4a plots the IWSR under BAU and ECO, and Figure 4b plots the basin outflows for the period of 1995–2025. As can be seen, under BAU without control on the basin outflow, irrigation water demand will be almost completely met, except in a few years when there will be minor water shortages for irrigation. However, in most years (16 years out of 25 years during 2001–2025), basin outflow will be below the target level (20 km3), and in 6 years it will fall below 10 km3.
 Under ECO with higher priority on EEWR, a much different picture emerges. Basin outflows will meet usage requirements except for small deficits in a few years, while the IWSR will show a decline trend and drop to 0.5 in some years. Such drops in irrigation may not mean a disaster for the entire basin, but in some typically water scarce areas in the basin, substantial vulnerability with irrigation water supply may hit the irrigated crops.
 It should be noted that the hydrologic time series used in this analysis do not include the consecutive drought periods in the basin such as 1922–1933 and 1990–2002 [Chen, 2002]. With data from the consecutive drought periods the basin outflow will be reduced substantially, as observed in the 1990s. Including hydrologic series to represent more complete climate variability and climate change is beyond the scope of this paper and remain an important area for future research.
Figure 4c shows ICPR for cereal crops, which consume over 65% of total irrigation water in the YRB. Under BAU the value of ICPR is over 0.9; under ECO we can see more variability and a significant declining trend of the ICPR. The ICPR will be as low as 0.4 in some years after 2015, which means the total irrigated cereal production will decline by 60% compared to potential production. Moreover, although the IWSR is almost 1.0 in most years under BAU, indicating minor water shortages, ICPR shows a slight declining tread. This is because IWSR accounts for seasonal irrigation water supply reliability and ignores irrigation water supply in individual crop stages; however, the model accounts the effects of water shortage in individual crop growth stages in the calculation of irrigated crop production. The declining trend in ICPR implies a growing effect on crop production from water stress in individual crop growth stages due to growing nonirrigation water demand (Table 4).
Table 5 presents cereal production and basin outflow as an annual average of 2011–2015 and 2021–2025 under BAU and ECO. From BAU to ECO the ratio of the reduction in cereal production over the increase in basin outflow is calculated. During 2011–2015 each increase of 1000 m3 basin outflow will result in cereal production loss of 0.67 metric tons; during 2021–2025 the same increase of basin outflow will result in cereal production loss of 0.81 metric tons. The growing trade-off is due to the higher potential irrigated area, yield, and production in later years than those in earlier years; the same water stress will then result in larger production loss. Cereal production will be nearly one third lower in 2021–2025 under ECO compared to BAU (Table 6).
Table 6. Compare Irrigated Cereal Production and Basin Outflow Under Business-as-Usual Scenario (BAU) and Ecological Scenario (ECO)a
 The three scenarios with an improvement in water use efficiency, water supplement from the SNWTP, or both are discussed here. As shown in Figure 5a, under ECO-SNWTP the same reductions in IWSR will occur as under the ECO before the SWNTP water supplement is realized in 2010; some reductions may still occur even after the SWNTP objectives are realized after 2015. Under ECO-AWUE, although the reductions are smaller than those under ECO, they are still significant, especially in later years when potential irrigation water demand becomes larger. Under ECO-AWUE-SNWTP, however, similar IWSR values as those under BAU are achieved. In terms of basin outflows, ECO-AWUE and ECO-SNWTP will result in basin outflows equal to those under ECO, and ECO-AWUE-SNWTP will result in even higher basin outflows than ECO in later years, as shown in Figure 5b.
 Correspondingly, Figure 5c shows the ICPR under the same three scenarios. As can be seen, only ECO-AWUE-SNWTP can achieve the same irrigated food production reliability as achieved under BAU. Therefore ECO-AWUE-SNWTP will match the irrigation water demand as BAU and better meet the ecological water requirements than ECO. Cereal production and annual basin outflow as average values during 2011–2015 and 2021–2025 under the three ECO scenarios are presented in Table 7, which shows that ECO-AWUE-SNWTP is better than the other two in terms of both cereal production and basin outflow.
Table 7. Comparison of Irrigated Cereal Production and Basin Outflow Under ECO-WUE, ECO-SNWT, and ECO-WUE-SNWTa
 One might ask if the assumed improvement of AWUE is feasible. The question needs to be examined from both technological and institutional improvements. Different water conservation measures should be tailored to the special conditions of different regions. For irrigation districts in Ningxia and Neimeng (Ning-Meng) located in the upper reaches of the YRB, water requirement for salt leaching needs to be carefully checked to see if there is any potential to reduce water withdrawal; in some subregions of Neimeng, water is withdrawn to crop fields after crop harvesting to maintain soil moisture for the next crop season, which covers 40% of total water withdrawal in this region according to the estimation of the YRCC. Research is needed to examine if that amount of water is used efficiently. In downstream regions, combined canal and well systems have been efficient in enhancing water recycling, and these practices can be further explored, including increasing the capacity for large-scale conjunctive surface-groundwater uses and maintaining water quality requirements during water recycling. In terms of irrigation water saving technology, improvements in irrigation canal linings may be more important in the upper (Ningxia) and middle Yellow River because canal leakage raises groundwater levels along with salinity levels and because leakage increases evaporation. In the lower reaches of the basin, canal leakage may beneficially recharge groundwater. The potential benefits from canal lining improvements may be greatest near the main river channel because the groundwater table there is already high and any recharge from leakage may increase salinity levels. The cost effectiveness of these interventions must also be carefully assessed.
 Managerial improvements include the adoption of demand-based irrigation scheduling systems and improved equipment maintenance. Institutional improvements involve the establishment of effective water user associations and water rights, the creation of a better legal environment for water allocation, and the introduction of higher water prices [Rosegrant and Cai, 2003].
 In recent years the YRCC has been promoting water pricing as an economic incentive for irrigation system and management enhancement and water saving [YRCC, 2001]. Water prices for irrigation increased from 0.006 Yuan/m3 in 1988 to 0.04 Yuan/m3 in 2000 in the Ning-Meng Irrigation District, located in the upper reach of the basin, and they increased from 0.003 Yuan/m3 in 1988 to 0.03 Yuan/m3 in 2000 in the downstream irrigation districts [YRCC, 2001]. According to a field survey conducted by YRCC in 2001, there is some potential to increase agricultural water prices, which will reduce agricultural water use but will not affect agricultural profits significantly. Rosegrant and Cai  assessed water price elasticities for domestic, industrial, and agricultural water demand based on a review of relevant literature and compiled the results of many empirical studies. From the research they estimated price elasticities of demand for water in the relevant sectors mentioned above in China. The elasticities for the YRB are estimated as −0.11 for irrigated agriculture and livestock, −0.70 for the industrial sector, and for −0.45 for the domestic sector.
 Assuming water prices for agricultural water uses in the YRB will gradually increase and that by 2025 the prices will be 3 times higher than the current levels, irrigation water consumption in the basin will fall by 11%, or 2.4 km3, based on the model outputs. Farmers can respond to higher water prices not only by a direct reduction in water withdrawals and consumption but also by improving the efficiency of water use so that a greater portion of it is used beneficially for crop production [Varela-Ortega et al., 1998; Zilberman et al., 1997]. With the assumed higher water prices, if the basin water use efficiency increases to 0.73, beneficial water use will not be affected by the price increase in the basin. This value is close to the high BE (0.76) used in the scenario analysis. Therefore the assumed price increase and efficiency improvement may be acceptable in terms of sustaining agricultural production, while more water can be reserved for ecological purposes.
6. Conclusions and Discussions
 Water withdrawals in the past decade have seriously depleted ecological water requirements in the YRB. One may argue that this is mainly due to the consecutive droughts with declining precipitation and runoff during 1991–2002. During that period, annual average renewable water was estimated as only 47.6 km3 [Chen, 2002], 10.4 km3 below the long-term average renewable water 58.0 km3, based on hydrologic series 1919–1975. However, even with time series (1961–1990), which exclude the consecutive droughts, ecological flow will be depleted if the current water use practices continue, with agricultural water withdrawal driven by demand but not constrained by downstream ecological requirements. If this remains, there is little chance that both agriculture water demand and ecological water demand can be sustained. The system will be highly vulnerable, especially when successive droughts as during 1990–2000 and 1922–1932 occur.
 Strong trade-offs exist between irrigation water use and ecological water use. Population growth, urbanization and industrial development, and rising food demand in the YRB are likely to place even greater stress on already severe conditions in the YRB. The stress will make the trade-offs more intensive in future years. Substantial pressure exists in the basin to improve controls on water demands, enhance water saving, and consider other measures, including interbasin water transfer through the South-North Water Transfer Project (SNWTP). Scenario analysis in this study concludes that plausible improvement of irrigation water use efficiency by itself cannot completely resolve the problem. The SNWTP as it is currently planned cannot solve the problem by itself either. Increasing the AWUE to a feasible level first and then implementing the water supplement by SNWTP may provide a better solution, as shown from the modeling results. The SNWTP can generate significant benefits for agricultural production and ecological water flows in the Yellow River. A full assessment of the cost effectiveness of the SNWTP would require a complete cost accounting together with an estimate of the incremental economic benefits from domestic and industrial water use and an estimate of the benefits from agricultural and ecological water use.
 The key issue for saving water in the YRB is to reduce nonbeneficial water depletion, especially in the upper and middle reaches. Making particular improvements in water use efficiency in the YRB as well as in other basins will require site-specific analysis and implementation. AWUE depends on improvements in water-saving technologies such as advanced irrigation systems, conjunctive use of surface and groundwater, and precision agriculture; simultaneously, it will also depend on improvements in the institutions governing water allocation, water rights, and water quality. Great care must be taken in designing a water-pricing system for agriculture in the YRB, given that agriculture is the dominant water use for at least the near term. Direct water price increases are likely to be punitive to farmers because water plays such a large role in their cost of production. Better alternatives would be pricing schemes that pay farmers for reducing water use and water rights and water trading arrangements that provide farmers or water user associations with incentives to reduce wasteful water use.
 This study is based on empirical results and assumptions from other studies, such as the assessment of ecological water demand and irrigation water demand. These estimates will likely to be updated or improved in future research. A new methodology for determining appropriate ecological water demand might be used since the current method is solely based on scientific assessment or empirical judgments. Crop pattern changes and agricultural research will change the irrigation water demand. Moreover, flow regulation through the storage system may increase water availability and may affect the ecosystems in the basin. Future research with an integration of ecological, hydrological, and policy sciences on these possibilities is needed for further verification of the conclusions presented in this study.
 The work presented in this paper was partially funded by the Comprehensive Assessment project, International Water Management Institute, Colombo, Sri Lanka.