The ecological water diversion project in the Heihe River Basin is the first successful case in China in which the ecological systems in a river basin have been rescued. This project serves as a valuable example for the management of ecosystems in other inland river basins. This paper reviews the integrated studies of the water–ecosystem–economy relationship in the Heihe River Basin and concludes that sustainable development in inland river basins requires the basin to be considered as a whole, with the relationships between the upstream, midstream and downstream areas of the basin coordinated appropriately. Successful development in these basins will be reflected in an improved output per cubic meter of water and the implementation of integrated river basin management practices.
Inland river basins in northwestern China exhibit alternating high mountains and basins. The mountainous areas usually experience much more precipitation, snowmelt and glacier melt. These mountains, usually distributed in the upstream areas of river basins, are water production areas. The mid- and downstream areas of inland river basins, where precipitation is scarce, are water consumption areas. After runoff leaves the mountains and flows into a basin, it is the determining ecological factor for the inland river basin. Without water, the oases cannot survive and become desertified. However, too much water results in salinization. For thousands of years, humans have lived in inland river basins, where sufficient water has been available to support grazing and agriculture, and people have altered the landscape to support thriving communities. However, in recent decades, many of the inland river basins in China's northwestern arid region have experienced a common challenge. With the population booming and the rapid economic development in the up- and midstream areas of the river basins, the consumption of water increases dramatically and diminishes the water available for ecological processes. The terminal lakes dry up, sandstorms become more common and the Populus euphratica forests die, causing a series of severe ecological disasters . Similar to the Heihe River Basin (HRB), the competition for water between economy and ecosystem is also getting more intense in other inland river basins all over the world, e.g. the Aral Sea basin  and the Tarim River Basin  leading to a ‘Tragedy of the Commons’ problem . In all of these cases, the water, the ecosystem and the economy are closely interrelated. Therefore, the solution to this problem must involve the careful and rational use of the limited water resources in such a way that not only supports economic development but also sustains the health of the ecosystems .
The HRB is located in the middle part of the Hexi corridor in the arid region of northwestern China (Fig. 1). The HRB is the second largest inland river basin and is representative of all of the inland river basins. At the end of the 20th century, an ecological water diversion project (EWDP) was successfully implemented by the central government of China, and as a result, the severe deterioration of ecosystems in the downstream areas of the HRB has been greatly alleviated . Numerous studies on the water, atmosphere, ecology and anthropogenic activities in the HRB have also been conducted. These studies have been crucial in supporting the sustainable development of the inland river basins. This paper reviews the integrated studies of the water–ecosystem–economy in the HRB to serve as an example for strategies for the sustainable development of inland river basins.
Ecological research in the HRB has expanded from focusing on the ecology itself to exploring how to coordinate the relationships among water resource sustainability, ecosystem health and economic development by considering the basin as an integrated entity with water as the central determining factor (Fig. 2). The specific approaches involve ascertaining the quality and quantity of water resources and determining how to rationally allocate, efficiently use and sustainably exploit the available water resources. The key aims of these strategies are to improve the output per cubic meter of water and to implement integrated river basin management practices. Toward this end, eco-hydrological studies have been introduced, with a focus on improving the output per cubic meter of water , and eco-economic studies have been undertaken with a focus on coordinating the relationship between ecosystem health and economic development . Additionally, integrated studies of the water–ecosystem–economy relationship at the river basin scale have made encouraging progress .
HISTORICAL EVOLUTION OF THE WATER AND ECOLOGICAL ENVIRONMENT IN THE HRB
Historical deterioration of water and the ecological environment of the HRB
At the millennial scale, the climate in the mountainous areas in the upstream area of the HRB fluctuated between warm-wet and cold-dry periods . Starting 1320 years ago, dry, normal and wet years accounted for 42.8%, 20.5% and 36.7% of the total years, respectively. The pluvial periods in the Qilian Mountains occurred from 1390 to 1413, 1425 to 1450, 1530 to 1649 and 1792 to 1920 AD , and the longest wet period occurred from the mid-8th century to the mid-10th century. The longest dry period occurred from the early 19th century to the early 20th century. Currently, the area is experiencing a century-long wet period [9,11]. Because climate change has caused a transition from warm-dry to warm-wet conditions in northwestern China, rainfall might increase, but rising temperatures may weaken the influence of increased precipitation and runoff from the mountains .
In the midstream area of the HRB, natural oases have been gradually replaced by artificial oases, driven by changes in the water environment and human activities as well as the interactions between these factors over the past 2000 years. The expansion, recession and migration of oases are closely related to wet or dry periods, population growth or shrinkage and farmland reclamation or abandonment . The HRB experienced three large-scale agricultural developments in the Western Han Dynasty, Sui–Tang Dynasties and Ming–Qing Dynasties, all during wet or normal periods . A massive expansion of artificial oases occurred starting in the early Ming Dynasty, and the oasis area increased gradually from this time through the early Qing Dynasty, mid-Qing Dynasty and after the formation of the Republic of China. Desertification mainly occurred during the Northern and Southern Dynasties, Five Dynasties Period and Ming–Qing Dynasties [15,16].
Natural oases in the downstream area experienced some desertification due to reduced water from the middle reaches. This phenomenon is common in northwestern China's arid regions. East Juyan Lake (Sogo Nuur), West Juyan Lake (Gaxun Nuur) and the Juyanze Lake dried out several times and alternately served as the main terminal lake during different historical periods . The natural oases consequently migrated with the migration of rivers and terminal lakes. Oasis development and desertification alternated at century scale , and changes in river and lake environments exhibited great spatial and temporal heterogeneity [19,20].
Rapid deterioration of the water and ecological environment of the HRB since the 1950s
In the upstream area, deforestation, overgrazing and grassland reclamation had caused serious degradation of the vegetation since the 1950s. The carrying capacity of the rangeland and the soil-water conservation ability also had been substantially degraded. The glacier area in the HRB has decreased by 29.6% over the past 50 years ; for example, the Qiyi Glacier has been characterized by a negative mass balance trend since the 1990s and an increase in runoff . Additionally, runoff from the Heihe River has increased due to increasing summer precipitation and a warming climate in winter . In particular, the increase in annual precipitation has surpassed the increased water consumption caused by increases in temperature.
In the midstream area, the dynamics of the rivers and lakes have become dominated by artificial water regulation. The development of artificial oases occurred simultaneously with the disappearance of natural oases. The consumption of runoff from mountainous areas greatly increased in the midstream area because of rapid agricultural development. The farmlands, urban areas, water bodies and grasslands have all experienced dramatic changes. Agricultural development has caused the water use patterns, irrigation water reuse rate and seasonality of water resource availability in these areas to change dramatically. In addition, the water quality has continued to deteriorate because of the overuse of synthetic substances .
The amount of water entering the downstream area decreased significantly from the 1950s to 2000 due to a large expansion of irrigated farmlands in the midstream area. With the rivers drying up and groundwater tables falling, Juyan Lake dried up in the 1980s. The riparian vegetation evolved from riparian trees to shrubs and then to desert shrubland, and a large area of forest died .
Driving forces: human activities vs. climate change
In the mountainous upstream area, climate change has been the controlling factor both historically and at present. The zones of alpine tundra, snow and glacier above the elevation of 3600 m are major water production areas of the Heihe River, accounting for 80.2% of the total runoff out of the Qilian Mountains . Water from glacial melt only accounted for 3.6% of the runoff. In the past few decades, runoff has increased due to the increase in summer precipitation, and glaciers have retreated due to a rise in air temperature . There has been almost no runoff from the alpine forest region, and this region has even consumed stored soil water during dry years . Afforestation, farmland reclamation and reservoir construction have undoubtedly affected the process of runoff generation in the mountainous area.
Human activities have been the dominant factor for environmental change in the mid- and downstream areas, but climate change has also played a role. The three periods of major desertification in the Hexi Corridor in the last 2000 years corresponded to phases of cold and dry climatic conditions. In the mid-Qing Dynasty, the population density of the Hexi Corridor broke through the ‘critical threshold’ of population pressure in arid zones, and the water utilization rate exceeded 40%. This was the main reason for the rapid desertification in the past 200 years . In recent history, the increase in the volume of water utilized, caused by oasis expansion in the midstream area of the HRB, had led to serious water and ecological environmental degradation in the downstream area. The oases migrated from the lower reaches to the upper reaches, and the natural processes had gradually been replaced by artificial processes . The desert-oasis pattern had been reshaped. The dried Juyan Lake became a source of sandstorms . In summary, the water system and water environment in the mid- and downstream areas of the HRB had been completely altered by human activities, which obscured the impacts of climate change in recent decades .
ECOLOGICAL WATER DIVERSION IN THE MAIN STREAM OF THE HEIHE RIVER
Water reallocation in the main stream of the Heihe River
Ecosystem deterioration and other environmental problems in the HRB drew the attention of the authorities and scientists in 1980s. Since 1987, Lanzhou Institute of Desert Research, Chinese Academy of Sciences (CAS), organized several comprehensive expeditions to the HRB to investigate water, land and grassland resources . In July 1995, a joint research group constituted by experts from seven Ministries and Commissions conducted a field investigation on the environmental problems in the lower reaches of the Heihe River [1,32]. In November 1995, the Earth Sciences Division of CAS organized another investigation to the HRB and the Shiyang River Basin. Based on these investigations, consulting reports on urgently rescuing the Ejin oasis were submitted to the State Council, suggesting that an authoritative and uniform water resources allocation and management institution should be established, and integrated water resource management strategies should be applied in order to guarantee a yearly discharge to the Ejin Oasis sufficient for rescuing the ecosystem in the downstream areas of the HRB [1,32].
To achieve this goal, the Chinese government has invested 2.352 billion RMB Yuan since 2000 to the program for the recent comprehensive management of the HRB, in which the EWDP plays the key role. According to this program, the middle reaches should discharge 0.95 billion m3 of water to the lower reaches when the Yingluo Gorge, which is the outlet of the Heihe River from the mountains, discharges 1.58 billion m3 of water in normal years. The EWDP has been implemented since late 2000. As a result, water use in the midstream area of the HRB has been reduced and more water has been delivered to the lower reaches of the Heihe River. After 12 years of water reallocation, the spatio-temporal distribution of water resources has changed dramatically in the downstream area of the HRB.
Change in the ecological environment in the downstream area of the HRB since the implementation of the EWDP
Water resources have undergone substantial spatial and temporal changes since the implementation of the EWDP. The multi-year average runoff was 5.29 × 108 m3 in the Langxinshan section during 2000–2012, an increase of approximately 1.52 × 108 m3 compared to that of the period between 1990 and 1999. After being dried out for approximately 10 years, East Juyan Lake has been recharged with water from the Heihe River since 2002, receiving fresh water for approximately 10 years . Water recharging to East Juyan Lake has accumulated to 6.19 × 108 m3 during this period, and the maximum area covered by water is approximately 42.8 km2. The groundwater table has increased by 0.56 m on average in the downstream areas of the HRB .
Vegetation is used as an indicator of environmental change. After water reallocation, Populus euphratica forests recovered and the desert shrublands have been rehabilitated to some degree . Areas of forest and grassland have increased by 20.6% and by 12 km2, respectively . Furthermore, while the normalized difference vegetation index in the downstream area was declining during the period 1987–2000, an increasing trend has occurred in the core area of the Ejin Oasis since the water reallocation.
The soil moisture content at different depths has increased since the water reallocation. In Populus euphratica forests, the soil moisture content has increased from 0.06 m3 m−3 in 1998 to 0.17 m3 m−3 in 2010 at a depth of 80–120 cm. The soil-soluble salt content has increased variably in different vegetation communities. Compared to the original content, soil-solute salt content has increased approximately by 3.3-fold and 2-fold in Populus euphratica and herbaceous communities, respectively. In addition, there has been a conversion of soil type from slightly saline, with a solute content of 1.0–2.0%, dominated by HCO3− and Na+, to medium saline soil, with a content of approximately 2.0–4.0%, and intense saline soil, with a content of approximately 4.0–6.0%, dominated by SO42−, Cl− and Ca2+ .
The local climate in the downstream area has also improved following water reallocation. The frequency of sandstorms was 5.85 yr−1 on average from 1987 to 1999 and declined to 3.50 yr−1 on average during 2000 to 2009. Clearly, the EWDP has played a role in constraining extreme weather in the downstream area .
Impacts on the midstream area of the HRB since water reallocation
EWDP has both positive and negative impacts on water and the ecological environment in the midstream areas of the HRB. There has been a slightly negative effect on the aquatic environment in the middle reaches since the water reallocation program was implemented. The replenishment of groundwater resources by surface water has declined while the extraction of groundwater has increased, causing the depth of the water table in the middle reaches to gradually decline. However, the water supply in the midstream area has experienced almost no change . The amount of available water resources in the midstream area remains at approximately 25 × 108 m3. The rate between surface water and groundwater usages has remained stable in the Ganzhou and Liyuanhe irrigation districts. However, in the Linze and Gaotai irrigation districts, the exploitation of groundwater has increased dramatically. For example, groundwater exploitation in Linze and Gaotai counties increased by 505% and 40% from 2000 to 2008, respectively .
The diversity, heterogeneity and fragmentation of the landscape have increased at the landscape level since water reallocation, but the spatial connectedness and landscape dominance index have decreased. The types of patches at the landscape scale present a uniform distribution . The dynamics have changed in woodland and grassland areas since water allocation. The area covered by woodland has varied from a slight increasing trend to a decreasing trend, whereas the area covered by grassland has declined continuously, although the speed of decline has slowed since water reallocation .
Water reallocation has had positive impacts on air temperature, precipitation, water quality, the frequency of sandstorms, the soil environment, plants and animals in the terrestrial ecosystem and organisms in the aquatic ecosystem in midstream area. However, there are also negative influences on groundwater depth and the regional woodlands. Additionally, there have been both positive and negative impacts on the ecological landscape . Areas of desertification and salinization have decreased by approximately 2.0% and 28.0% after water reallocation, respectively. These trends illustrate that water reallocation has played a positive role in the ecological environment in the midstream areas of the HRB. In summary, no new ecological problems have resulted from the implementation of the EWDP .
There is no large reservoir in the upper reaches of the Heihe River; therefore, it is very difficult to control and regulate seasonal runoff and to determine the best time for ecological water diversion.
Water consumption has undergone great changes in the HRB since the EWDP was implemented. The allocation rate between ecological and economic water uses and the timing of water diversion are suggested to be adjusted.
Through more rational management of water resources and the implementation of water-saving practices, water consumption in the midstream areas has decreased. However, these changes have not completely ameliorated the problem, and there is still an imbalance between irrigation and ecological water consumption in the midstream areas of the HRB.
It is important for the limited water resources allocated for the downstream area to be utilized for ecosystem rehabilitation and not the illegal reclamation of new farmland.
ECO-HYDROLOGY RESEARCH IN THE HRB
Plant–water relationships and plant water use
In inland river basins, the main principle of water management is to regulate the eco-hydrologic processes to maintain an efficient and sustainable ecosystem. Plant–water relationships and plant water use are keys for eco-hydrological regulation. The plant–water relationship mainly refers to the effect of water on plant life, including the root uptake by plants, water transportation within the plant and evaporation into the atmosphere from the plant (i.e. transpiration) and the response and adaptation of plants to water stress . In desert zones in inland river basins, plants adapt to the extremely dry conditions through morphology adjustments in both carbon assimilation and metabolism . The different methods of assimilation and metabolism determine the differences in plant water use efficiency. According to the anatomic structure of assimilation tissue and the stable carbon isotope ratios (i.e. δ13C) in desert plants, the assimilating shoots in C4 plants have Kranz anatomy, whereas C3 plants do not; additionally, the δ13C values in C4 plants range between 14 and 15‰, whereas the corresponding values in C3 plants range from 25.8‰ to 28.1‰ . More than 40 woody C4 plants have been found in the desert area of HRB, including Haloxylon persicum, Calligonum mongolicum, Salsola arbuscula and other species . The desert plant population adapts to water stress with a unique dynamics that combines competition and collaboration. For example, the hydraulic lifting in deep rooting plants can lift deep soil water, which can then be utilized by shallow rooting plants . When C3 plants and C4 plants grow together, the C3 plants’ photosynthetic rate and water use efficiency improve, and the corresponding functions of the C4 plants decline .
Interaction between ecological and hydrological processes
Eco-hydrological processes involve many interactions among water, soil, plants, atmosphere and other factors. In the vertical direction, these processes are reflected by energy and water transport in the soil–plant–atmosphere continuum, manifesting in the exchange of water vapor and CO2 flux driven by near-surface atmospheric forcing. In the horizontal direction, the processes are reflected in the transmission of energy and water at individual, population, community and ecosystem (i.e. landscape) scales . Differences in radiation components and water transport and its ecological effects are caused by landscape heterogeneity.
Vegetation in alpine meadow ecosystems can regulate the radiation and energy budget, resulting in changes in the hydrological processes in frozen soil and increased mountain runoff . Climate change leads to an upward migration of the upper limit of the forest and steppe belts [53,54], resulting in increased vegetation productivity and therefore causing an increase in evapotranspiration and a decrease in soil moisture and runoff . In the desert-oasis region in the midstream area, water-saving practices, an increase in water use efficiency and the expansion of the oasis have caused an increase in groundwater extraction and a decrease in groundwater recharge . In the downstream area of the HRB, the change in the groundwater table depends on the amount of water from the upper and middle reaches. Maintaining flood processes is not only beneficial to groundwater recharge but also plays an essential role in the reproduction of riparian forests. The shallow underground water table has increased since the implementation of EWDP, but studies of the groundwater recharge range, flow path and flow rate are still limited . The desert vegetation in rain-fed areas is distributed in a patchy pattern, resulting from the interaction of soil, water, landforms and sandstorms. In particular, soil-water and groundwater heterogeneity caused by microgeomorphology results in the patchy pattern of vegetation . Most desert plants absorb soil water from rainfall, whereas a few desert riparian plants absorb soil water converted from groundwater . The desert plants maintain their growth and exhibit a unique absorption strategy to optimize their use of the limited precipitation [60,61].
Water consumption and water requirements in the oasis ecosystem
Accurate estimations of water consumption in the oasis ecosystem form the basis for allocating the water resources reasonably, improving the water use efficiency in the oasis and determining the ecological water requirement for maintaining oasis stability. In view of the water shortage in inland river basins, the water requirement in the oasis ecosystem includes water to sustain riparian forest, wetland and farmland shelterbelts as well as water to flush salt from farmland. Water requirements at the different scales of individual, population (i.e. plot), community (i.e. irrigation district) and oasis have been investigated through long-term in situ monitoring and remote-sensing technologies [60,62]. The ratio of maize transpiration to evaporation was found to be approximately 3:1 . The water consumption of a typical irrigation field in the midstream areas of the HRB is 7500–9600 m3 ha−1, in which the ecological water consumption accounts for approximately 25%. This amount of water is required to maintain the lower limit of metastability in the oasis based on a water and heat balance analysis [60,64]. Therefore, the average amount of water should not be less than 7500 m3 ha−1, and the ecological water consumption should not be less than 25% to maintain the stability of the oasis.
ECOLOGICAL ECONOMIC RESEARCH IN THE HRB
Ecological economics has been introduced as a new research direction in the integrated study of the HRB. This discipline mainly focuses on studying the relationships between human and natural factors . This section reviews the research on the ecological economics of the changing relationship between the above two factors in the HRB.
Integrated accounting of economic development and ecosystem services
Economic value is a key factor for coupling ecosystems with economic systems. To measure the value of ecosystem services, Xu et al. assessed the economic losses caused by environmental degradation , the economic value of restoring an ecosystem  and the weighted benefits and costs of alternative ecosystem management strategies , including monetary and non-monetary benefits and costs. The contingent valuation method assessment results showed that, by taking into account the ecosystem discount rate, the aggregated benefit of restoring the ecosystems in Ejin Banner was about 1637 million RMB Yuan, accounting for 24.19% of the GDP in Zhangye City in 2001. In terms of the construction of national economic accounting systems, Chen established a sub-accounting system, embedding both water and land resources by combining environmental loss and ecosystem services into the national accounting system . The results showed that the annual depletion cost of excessive water use was 197.9 million RMB Yuan, accounting for 9.39% of traditional net-capital formation in Zhangye district in 2000. On this basis, Ma et al. also built the material flow accounts in Zhangye City , midstream area of the HRB. The results showed that the ecological rucksack was as high as 86.5% of the total material requirement, manifesting that the exploitation of natural resources was in low efficiency and economic development was affected by natural resources significantly. In terms of ecosystem management, ecological compensation is the main method we used to couple ecosystem services and economic accounting. Song et al. simulated the relationship between the ecosystem services and the payments by using a hydrological model coupled with benefit–cost analysis and addressed the problem of ‘whom and how many should be paid’ in the up- and midstream areas of the HRB .
Coupling research on ecosystems and economic systems
Mass, energy and information are the three main media that connect ecosystems and the economic systems. Eco-economically speaking, the coupling studies on ecosystems and economic systems have mainly focused on these three media.
In terms of material media, the coupling studies have been carried out on the ecological footprint, water footprint and ecosystem health evaluation, which are based on the evaluation of water for ecological purposes. Earlier this century, many empirical studies using an ecological footprint and virtual water methods were carried out in the HRB when these methods were introduced [72–74]. Xu et al. found defects in the concept of the ecological footprint. They therefore proposed the sustainability evaluation equation, ImPACTS, which takes into account social adaptability . Xu et al. expanded upon and provided a new interpretation of the virtual water strategy. They believed that the key to implementing the virtual water strategy is to find concrete ways to develop secondary and tertiary industries, i.e. deriving new positive feedback from secondary and tertiary industries . Guo evaluated the health status of the HRB ecosystem based on ecological water requirements .
The studies on another coupling medium, energy, focused mainly on emergy and exergy analysis. Applying the emergy analysis method, Du et al. analyzed the sustainability of agricultural eco-economic systems in Zhangye City . Applying exergy analysis, Liu analyzed the development of sustainable economic systems . The results showed that in 2005, there were about 1732 PJ energy and raw material resources flowing into the industrial sector of Gansu Province, but only 39.5% of them reaching end-use. The efficiency was therefore very low. It was suggested that the positive resource-saving technologies could significantly improve the efficiency.
Research on the third coupling medium, information, focused on information economics, health evaluations of the economic system and the social economic cycle of water resources. Deng et al. showed how to reduce the information rent issue in ecological compensation. They used simulated auctions to explore the optimal bidding model under different risk preferences of participants for ecological compensation and determined ways to reduce the information rent . Huang et al. used an ascendency formula inferred from the perspective of information theory to evaluate the sustainability of the economic system in Gansu Province . The results showed that the utilization of water resources of Ganzhou District was still in its infancy and was not sustainable. The economic development was mainly driven by increasing the amount of water use rather than optimizing the structure of water utilization.
Human impacts on water systems
Like natural factors, human factors have a broad impact on water systems. The identification of human factors is similar to the discovery of new resources, and the analysis of the impact of human factors on water systems is also similar to analyzing the usage of new resources. Together, these strategies offer the potential to derive new relationships from the system. To identify and quantify the human factors, Xu et al. pioneered a study using the IPAT equation to identify some of the key human factors according to a hierarchical system decomposition, including factors such as cultural type, social capital, institution and others . Liu  and Liang et al.  used geo-statistical methods to spatialize the population and GDP in the HRB. To examine the impact of human factors on water systems, Zhong et al. analyzed the impact of different types of cultural awareness in the HRB on the nature of cognitive and water resource management . Li et al. analyzed the impact of social capital on the performance of water management in the HRB and pointed out that it had a more significant positive correlation . Zhou  and Zhang et al.  analyzed the impact of changes in the water system on water management performance. Research on water systems’ ability to adapt to human factors has become a new research direction . For instance, Tang and Xu  applied the contingent valuation method to analyze the farmers’ bearing capacity of water price. The results showed that the highest price the farmers could bear was 1027.5 RMB Yuan ha−1 yr−1. Compared with the actual water price, the price of irrigation water in Zhangye had already been beyond the farmers’ bearing capacity. In addition, water market trade has been proved an important mechanism in solving water competition conflicts and improving water use efficiency. Long et al.  introduced a two-part allocation method of water resource property, and stimulated the efficiency of water property trade based on the game theory, with a case study in Zhangye. The results showed that the water property trade would result in a potential maximum benefit of 1.34 × 108 RMB Yuan yr−1.
The development of a win–win strategy for the environment and economy in the HRB
All theories originate from practice and, in turn, guide practice. Combined with local development practices, research has been carried out on a number of development strategies aimed at finding a win–win situation for the ecosystem and economy in the HRB.
To summarize the transformative development strategy of Zhangye City, development should be based on reconstructing the ecological environment, developing an eco-friendly city and developing new opportunities for ecotourism . This strategy provides an economic benefit from ecological protection and achieves a win–win situation for the ecological environment and the economy. Herman Daly described ecological economic problems in terms of scale, equity and efficiency, and he stressed that independent policies are needed to solve these three problems. The development strategy in Zhangye is an expansion of and supplement to the ecological economic theory described above. Cheng et al.  described a planning strategy that avoids the worst ecological and economic states when proposing a happiness-oriented water resources management planning strategy. They also proposed countermeasures to avoid falling into development traps by identifying potential development traps and evaluating these traps in terms of scale, equity and efficiency.
INTEGRATED STUDY OF THE WATER–ECOSYSTEM–ECONOMY IN THE HRB
Necessities of the integrated study
Land-surface systems are giant, open, complex systems . System science and complexity science advocate holism but, practically, are mostly dominated by reductionism . Integrated studies can, perhaps, bridge the gap between holism and reductionism. Through integrated study, it is possible to efficiently ‘achieve innovations by rearranging and expanding tried-and-true strategies’.1 Thus, the concept that ‘the whole is greater than the sum of its parts’ need not be empty words but can be applied in scientific practice.
Generally speaking, land-surface system science is thought to involve a number of different scientific subjects and is a typical interdisciplinary science for which three types of integration are necessary: technology integration, knowledge integration and team integration .
A river basin is the most appropriate unit for testing integrated study methods in land-surface system science because a watershed is a microcosm as well as a basic unit of the earth system. Hydrologically, a river basin is a relatively closed drainage system which is hierarchically composed of different levels of watersheds; ecologically, a watershed is probably a basic spatial unit for terrestrial ecosystem dynamics; and socio-economically, economic development carried out using a river basin as the basic unit has been a trend . Realizing the importance of watersheds in hydrological science and even for the earth system science, the United States National Research Council has conducted a number of watershed studies since 1991 and has so far published Opportunities in the Hydrologic Science, Watershed Research in the U.S. Geological Survey, New Strategies for America's Watersheds, River Science at the U.S. Geological Survey and Challenges and Opportunities in the Hydrologic Sciences [98–102]. Particularly, the National Science Foundation, USA, established a Science and Technology Center on ‘Sustainability of semi-arid hydrology and riparian areas (SAHRA)’ to develop multi-disciplinary understanding of complex interactions between physical, biological, economic and human factors in semi-arid river basins and to support sustainable development by building multi-scale integrated basin-scale models . The strategy was not only producing advances in basic science but also providing support to decision making . In recent years, a number of terrestrial research programs in many countries have used watersheds as the basic unit for research and observation, including the Critical Zone Observatory (CZO), the Consortium of Universities for the Advancement of Hydrologic Science, Inc. (CUAHSI), the Terrestrial Environmental Observations (TERENO)  and the Danish Hydrological Observatory (HOBE) .
In China, the HRB has long served as a testbed for the integrated study of the water–ecosystem–economy relationship [107,108]. The integrated study in the HRB aims to explore research methods for land-surface system science and to refine land-surface system science theories. At the same time, this integrated study aims to develop a water resource management decision support system (DSS) based on scientific models. The purpose of this DSS is to provide a powerful supporting tool for sustainable development in inland river basins.
Progress on the development of the 3M integration platform
A 3M (i.e. monitoring, modeling and data manipulation) integration platform has been developed to support the integrated study in the HRB.
The HRB is the area where eco-hydrological modeling research has been most intensively conducted in China on the watershed scale. Many existing models have been used in the HRB [109,110], and several new models have also been developed . However, most of the newly developed models are still single-subject models. The development of integrated models has gained rapid momentum in recent years.
Model integration for the upstream area of the HRB focuses on the development of a distributed hydrological model, which can solve the problem of runoff prediction in mountainous area and investigate the complex interactions among atmosphere, plant, soil, frozen soil and snow. Wang et al. developed an integrated model by coupling the distributed hydrological model, i.e. the geomorphology-based hydrological model (GBHM) with the Simple Biosphere 2 (SiB2) model, which contains a parameterization scheme of frozen soil hydrothermal processes . Zhang et al. further coupled the simultaneous heat and water model, which contains a more advanced frozen soil hydrological process module, with the GBHM and developed a distributed hydrological model that was suitable for the alpine area . Through verification studies in the experimental watersheds in the upstream areas of the HRB, Zhang et al. confirmed that the model was suitable for simulating frozen soil and snowmelt runoff processes in cold-region watersheds.
Model integration in the midstream and downstream areas of the HRB focused, however, on coupling of the surface water, groundwater and ecological models. Wang et al. developed 3D saturated and unsaturated zones for the water flow model, AquiferFlow, as a general modeling tool . Tian et al. further realized the coupling between the land-surface process model, SiB2 and AquiferFlow, and therefore greatly improved the modeling ability to simulate evapotranspiration and the interaction between surface water and groundwater . Using the aforementioned model, Tian et al. simulated the hydrological cycle of the midstream areas of the HRB, and the results showed that the coupling model could be used to realistically and comprehensively simulate the water cycle and the energy balance in arid regions . Zhou et al. coupled the groundwater model, MODFLOW, the soil-water model, Hydrus-1D, and the crop growth model, WOFOST [116–118]. Li et al. further coupled the land-surface process and the stomatal photosynthesis modules with the integrated model to establish a farmland eco-hydrological model . This integrated eco-hydrological model was used to simulate the transpiration and evaporation of the crops during their growth cycle as well as the crop yield under different irrigation regimens and environmental and climatic conditions. Additionally, this model can be used for prediction and decision-making support. For instance, compared to the current irrigation schedule, the irrigation schedule optimized by the integrated model could save 27.27% of irrigation water, indicating that the coupled eco-hydrological model could be used to analyze the interaction between the ecosystem and hydrological cycle as well as to provide guidance for agricultural water saving.
Data are a major component of the information infrastructure of the watershed. Data are the foundation for analysis, the fuel for the model and the basis for decision making. The development of watershed science depends on the collection and integration of all types of data. The progress of data integration in the HRB can be divided into two components.
First, there is database integration. Based on ‘Digital Heihe’ , various types of data for the HRB were reorganized and standardized . A data center was established for data sharing. Currently, there are as many as 500 data sets incorporating a total of 3.4 TB of data available online at the HRB data center. More than 4000 Chinese and international users have used the data service. The amount of data provision has reached 42 TB . An automatic data harvesting and assembly system has been developed for collecting field observation data with the support of a high-performance relational database to effectively improve the efficiency of applications such as massive data query, on-line visualization and on-demand downloading. This system also supports automatic control of data quality, on-line analysis of data and on-line model application .
Second, there is scientific data integration. The purpose of the scientific data integration is to convert multi-source and multi-resolution data into data sets that can be directly used for model development, validation and improvement . The major developments include the following: (1) statistical and dynamical downscaling methods were developed, and the data sets of near-surface meteorological elements with a 1–5 km spatial resolution and hourly temporal resolution for the HRB from 2000 to 2012 were produced [125–127]; and (2) land cover, plant functional type and soil texture maps suitable for hydrological and ecological models for the HRB were produced. Using the evidence-reasoning method, Ran et al. produced a land cover map based on the International Geosphere-Biosphere Program classification system by integrating a number of thematic maps (land use/cover, plant functional type, wetland, glacier, etc.) . Gao et al. produced a land cover map based on the US Geological Survey classification system by integrating the vegetation classification data from the vegetation map of China into the land-use map of the HRB . Using the fuzzy logic inference method, Lu et al. produced a soil texture map for the HRB by integrating nearly 200 profiles of observed data for soil texture and a number of high-accuracy thematic maps. These model parameter sets have effectively supported the integrated study in the HRB.
Using the watershed as the unit to establish an integrated watershed observing system has been an important trend in integrated eco-hydrologic studies in the past 10 years. Thus far, a relatively comprehensive watershed observing system has been established in the HRB. The main branches of this observing system are located at the Heihe Upstream Watershed Ecology-Hydrology Experimental Research Station, the Linze Inland River Basin Comprehensive Research Station in the midstream areas of the HRB, the Alashan Desert Eco-hydrology Experimental Research Station in the downstream area of the HRB and the Heihe Remote Sensing Experimental Research Station (for distributed observation on the river basin scale). In addition, several field observation experiments have been conducted in the HRB, including the HRB Field Experiment (HEIFE) (1990–1992), the Jinta Experiment (2004), the Watershed Allied Telemetry Experimental Research (WATER) (2007–2010) and the Heihe Watershed Allied Telemetry Experimental Research (HiWATER) (2012).
Among the experiments, an important result of WATER has been the generation of some multi-scale, high-quality comprehensive data sets, which have greatly supported the development, improvement and validation of a series of ecological, hydrological and quantitative remote-sensing models. The goal of a breakthrough for solving the ‘data bottleneck’ problem has been achieved [130,131].
HiWATER was initiated in 2012. This project has established a world-class hydrological and meteorological observation network, a flux measurement matrix and an eco-hydrological wireless sensor network. A set of super high-resolution airborne remote-sensing data has also been obtained. In addition, there has been important progress with regard to the scaling research . Furthermore, the automatic acquisition, transmission, quality control and remote control of the observational data have been realized through the use of wireless sensor network technology . The observation and information systems have been highly integrated, which will provide a solid foundation for establishing a research platform that integrates observation, data management, model simulation, scenario analysis and decision-making support to foster 21st-century watershed science in China.
Another important aspect of model integration is the application and development of modeling environment, which is a computer software platform that supports the efficient development of integrated models, the convenient coupling of existing models or modules, model management, data pre-processing, parameter calibration and visualization. The study on the modeling environment in the HRB has followed two directions: using internationally recognized, mature modeling environments for model coupling to solve the key scientific problems in the HRB and developing new modeling environments after taking into consideration the deficiencies of existing modeling environments in terms of their flexibility.
A variety of applications used internationally recognized modeling environments. For example, with the support of the Spatial Modeling Environment, a distributed runoff model for the mountainous areas in the HRB was formulated [134,135], and by using the first-generation Modular Modeling System, the precipitation-runoff models were modified for the forecasting of runoff in the upstream areas of the HRB with improved functions of snowmelt and frozen soil processes . Additionally, aiming to solve the irrigation water-optimization problem in the midstream areas of the HRB, an agricultural eco-hydrological model for the midstream areas of the HRB was developed using the new-generation Object Modeling System 3, by coupling the groundwater model, MODFLOW, the hydrological model for unsaturated zones, HYDRUS, and the crop growth model, WOFOST . Based on the Cold Regions Hydrological Model, the frozen soil and snow-melting modules were successfully introduced into the traditional hydrological model . The new model was used in the Binggou Watershed, upstream of the HRB. The results revealed that estimating snow sublimation loss was vital for accurate snowmelt estimation in this region.
Nan et al.  developed a new modeling environment for hydrological and surface process studies. The focus of this modeling environment is to realize the reusability and expansibility of the modules using an efficient and flexible mechanism for module control and data transmission. A number of case studies for model integration have been carried out based on this modeling environment. For instance, the hydrological model TOPMODEL was coupled with the evapotranspiration module from the land-surface process model Noah so that the impact of vegetation on water balance was more reasonably simulated in the modified TOPMODEL. Using the aforementioned modeling environment, Liu et al.  simulated the ‘who should be compensated, by how much and how’ issue from the ecological compensation perspective in the upstream areas of the HRB by using the newly developed modeling environment. Additionally, Liu et al. accurately simulated the spatial changes of land use in Zhangye City, midstream of the HRB by combining the general equilibrium models with the land use model CLUE-S by considering the demand change on land use.
Decision support system
One of the ultimate goals of the integrated study in the HRB is to establish some DSSs based on integrated watershed models for the sustainable utilization of water resources and other natural and social economic resources (Fig. 3). Thus far, two types of DSS have been developed: research-oriented and application-oriented systems.
The design objective of the research-oriented DSS is to establish spatially explicit DSSs by integrating comprehensive scientific models for the ‘water–soil–air–plant–human’ continuum as the framework and incorporating expert knowledge of the sustainable development of water and soil resources in the HRB. This system should also be capable of designing different scenarios and carrying out scenario analysis. To date, the development of the framework for a DSS that integrates multiple hydrological models and geographic information system functions and supports multi-disciplinary model coupling has been completed. Technically, a breakthrough has also been made that solves the key problem to handle the inconsistencies of different temporal and spatial scales of models . One of its applications is to evaluate ecosystem service payments in the mountainous area of the HRB by integrating the Soil and Water Assessment Tool (SWAT) and the minimum data method in the uniform DSS framework and to establish various scenarios to evaluate the effects of payment price changes on reducing grazing area .
Ge et al. [143–145] developed an application-oriented DSS for water resources management, and analyzed the relationship between water resources management and effects from climate change and human activities under three scenarios over the coming 15 years (2015–2030) in the midstream areas of the HRB. The hydrological and climatic scenario supposed that river runoff and precipitation were calculated using the SWAT model and the CCSR/NIES model, respectively, based on greenhouse emission scenario (A2) from the IPCC report. Hypotheses of population scenario alternatively assumed a constant population of the year 2010 and a population growing with the average growth rate of the period 2000 to 2010. Assumptions on farmland included a constant farmland area of the year 2010 and a growing farmland area increasing at the average annual growth rate of the period 2000 to 2010. Results from three hybrid scenarios indicated that (1) increasing planted area would lead to more vulnerability in dealing with extreme hydrological events and (2) more groundwater would be pumped. Some recommendations were proposed for decision makers and managers to improve water resources management, which comprised that (1) uncertain climate change should be considered in future water resources management and planning; (2) planted area should be reduced by approximately 60 000 ha in order to increase the capacity of responding to extreme hydro-meteorological events and to decrease the risk related to lack of water; (3) groundwater abstraction should be effectively controlled; and (4) water use efficiency should be improved along with increased water productivity.
The HRB, a typical inland river basin, has been used as a testbed to carry out integrated studies. We reviewed the water problems, ecological and eco-economic researches as well as integrated studies of the water–ecosystem–economy relationship in the HRB. We suggest to coordinate the relationships among water resource sustainability, ecosystem health and economic development and balance the relationships involving the upstream, midstream and downstream areas of the basin by considering the basin as an integrated entity.
The downstream areas of the HRB experienced severe ecosystem deterioration during the second half of the 20th century. From 2000, based on scientific research previously carried out in the HRB, the ecological water diversion project was successfully implemented and the severe deterioration of ecosystems in the downstream areas of the HRB has been greatly alleviated. However, the water conflicts between economy and ecosystem and among different regions of the HRB still exist, and how to manage the water and land resources to achieve a sustainable river basin development is still a grand challenge. For this goal, further in-depth integrated studies are required.
An integrated platform, which incorporates monitoring, modeling and data manipulation, has been developed to support the integrated study in the HRB. Integrated models have been used for understanding complex interactions within the ‘water–soil–air–plant–human’ continuum and some DSSs have been preliminarily developed and used in small-scale water management, such as irrigation water allocation. Future work is to develop a fully integrated water–ecosystem–economy model and a spatial explicit DSS that takes integrated scientific models as its framework for supporting sustainable development of the river basin.
We believe that the experiences in the HRB are useful for sustaining both the ecological health and socio-economic development of other inland river basins.
We thank the four anonymous reviewers and the editor for their insightful comments.
This work was supported by the National Natural Science Foundation of China (91125001).