According to the World Water Development Report 2018 [1
], due to economic developments, population growth, and changes in consumption patterns, global demand for water is rising at an annual rate of 1%, and the population suffering from water shortages is expected to reach nearly six billion by 2050. China will face severe water shortages as its per capita water resources is only one quarter of the world’s average level, and will be aggravated due to climate change, environmental pollution, and ecosystem damage [2
]. Therefore, the scarce water resources require proper management to maintain the sustainability of human societies.
Reallocation of physical water, such as the South-to-North Water Diversion project, is costly and may have negative impacts on ecosystems [3
]. The concepts of virtual water (VW) and water footprint (WF) provide a new approach to solve the water scarcity issues. VW refers to the total amount of freshwater required to produce a certain kind of product along their supply chains [4
]. This concept illustrates that regions with insufficient water resources can fulfill the consumption of water-intensive products by importing them rather than producing them locally so as to conserve local water resources. In this manner, fresh water embodied in the products is transferred from one region to another through inter-regional trade, which is termed VW flows [5
]. With the increasingly close inter-regional trade, the significance of studying the rule of VW flow to optimize the spatial and temporal allocation of water resources becomes prominent. Dalin et al. [6
] found that the number of trade connections and the volume of VW trade more than doubled in the past two decades. Feng et al. [7
] demonstrated that VW flows between regions redistributes water resources from water-rich areas to water-poor areas, and effectively reduced water stress within the Yellow River Basin.
The concept of WF is closely related to VW. WF is a consumption-based indicator that refers to the total amount of VW content consumed by a region [8
]. WF can be classified into three types, namely, blue WF (the amount of surface and underground water resources consumed in the product supply chain), green WF (water that is stored in the unsaturated soil and consumed by vegetation evapotranspiration), and gray WF (polluted water during production) [5
]. The green WF is only related to the agricultural sector. The grey WF, however, is difficult to calculate due to the complexity of contents and concentration of sewage and the lack of a unified quantitative method [10
]. In this study, we focus on blue WF (BWF), mainly because the incorporation of green WF may cause content inconsistency between the agricultural sector and other sectors [11
], and grey WF as an environmental indicator does not reflect the consumptive use of water [12
For a country/region, a variety of goods or services consumed within its territory are not only produced by using local water resources, but also by using water resources in other regions. Therefore, the WF of a country (region) is composed of two parts: Internal WF and external WF, with the former referring to the actual local and the latter non-local water resources used in practice [13
]. Distinguishing the internal WF from the external WF is helpful to reveal the source and destination of the VW flows and to explore the interdependence of water resources among regions [14
]. For instance, Zoumides et al. [15
] assessed the WF of crop production in Cyprus and found that this semi-arid nation largely relied on external water resources, which may increase the risk of food security. Steen-Olsen et al. [16
] analyzed the water displacement relationships among European Union countries, in which Spain and France are two primary exporters of embodied water.
The input-output model (IO) is an important method for the assessment of WF. Single-region input-output model (SRIO) uses input-output data of a single region to calculate the WF based on the estimated regional total water use coefficient [17
]. The extended form of SRIO, i.e., multi-regional input-output (MRIO) model, can describe the input-output relationships between different sectors among different regions, and hence is capable of tracing the resource and environmental impacts of consumption activities in one region to a specific production sector in another region via the inter-regional supply chain within the system [18
]. Additionally, MRIO can distinguish the internal and external WF so that one can explore the VW flow relationships between regions [20
]. MRIO has been widely used in the studies on inter-regional water flows [21
], PM2.5 emissions [22
], and carbon emission transfers [23
In recent years, many studies have explored the WF and VW flows across different scales, e.g., from nations [24
], river basins [26
], provinces [27
], and to cities [11
]. Tian et al. [28
] examined the temporal changes of China’s WF and their results suggest that China was a net exporter of VW during 1995–2009. Wang et al. [29
] found that China’s total WF decreased from 495.5 billion m3
in 1997 to 447.6 billion m3
in 2007. Zhang et al. [30
] compared the direct water consumption coefficient and the net VW outflows between different sectors of 30 provinces in China from 2002 to 2007, and explored the spatial and sectoral characteristics of China’s VW flows. The ongoing rapid development of China’s economy puts more pressure on China’s water demand [31
]. China’s GDP growth rate decreased from 14.2% in 2007 to 10.63% in 2010, while the development level varies greatly among regions [32
]. In this context, how do China’s WF and inter-regional VW flow relationships change? What are their driving forces? This study aims to answer these questions by examining the magnitudes and structures of inter-regional VW trade and provincial WF based on an MRIO analysis for the years, 2007 and 2010. Moreover, the assessed VW was combined with an integrated factor of WAVE+ (water accounting and vulnerability evaluation) [33
], thereby providing a basis for better management of water resources.
The results of MRIO analysis for both years, 2007 and 2010, suggest a consistent north-to-south and south-to-south net VW trade pattern in China. However, changes can be observed in the VW supplying regions by comparing the VW trade patterns for the years, 2007 and 2010, respectively. As shown in Figure 7
, the net outflows of VW decreased in Heilongjiang, Inner Mongolia, Hebei, and Gansu while they increased in Xinjiang and Ningxia in Northwestern China. Sichuan, Guizhou, Hubei, Anhui, Jiangxi, and Hainan, which are in southern China, were also found having an increased amount of net outflows of VW in 2010. For the main VW-consuming coastal provinces, namely Shandong, Jiangsu, Shanghai, Zhejiang, and Guangdong, a decrease in the net inflows was observed for these provinces in 2010. This result suggests a reduced dependence of these regions on imports of external VW. Contrastingly, there was an increase in the net inflows of VW for some inland provinces, such as Qinghai, Shaanxi, Shanxi, Henan, and Yunnan.
Moreover, according to the results of the VSW assessment, the amounts of VSW inflows were consistently greater than those of VSW outflows at the national level for both years, 2007 and 2010. The national total of net VSW inflows were 24.10 billion m3 and 21.16 billion m3, respectively, for the years, 2007 and 2010. These results indicate an increased pressure on the environmentally vulnerable provinces due to water consumption displacement. The vulnerable provinces with water deprivation issues are also the important regions of agricultural production. Therefore, the results of our analysis may raise special concern about the proper formulation of policies to alleviate the water deprivation issues in northern China.
To further investigate the influential factors of WF, a multivariate regression analysis was applied with the explanatory variables of per capita GDP, total amount of water resources, per capita amount of water resources, urban population, and rural population. The results show that at the 95% confidence level, per capita GDP, total amount of water resources, and per capita amount of water resources have significant relationships with regional WF, albeit to varying degrees, in both years of 2007 and 2010. Compared to 2007, in 2010, each region’s WF was also significantly affected by urban population (Table 3
Unbalanced economic development is a main factor that causes the regional differences in WF. China’s eastern coastal regions are its most economically advanced regions. These regions consistently rely heavily on large amounts of VW imports through inter-provincial trade of, for instance, crop and grain from those less developed regions. The economically underdeveloped regions primarily export agricultural products in the domestic trade system. Therefore, the per capita GDP of a region is significantly positively correlated to the size of its WF. Regions with affluent water resources are found to be less dependent on the consumption of VW products and services, as suggested by the significantly negative relationship between per capita amount of water resources and the amount of WF. Urban population was significantly negatively correlated to WF only in 2010. This result partially suggests the impacts of urbanization on improving the usage and efficiency of water resources. In the results for both years of 2007 and 2010, rural population does not show a significant relationship with WF.
Additionally, although we did not include a policy factor into the analysis due to data unavailability, some previous research has identified the importance of policy intervention to influence WF. For instance, Wang et al. [40
] found a downward trend of water use in Shandong from 1997 to 2002, which was largely owing to the implementation of water use quotas’ policy and the shut-down of backward-technology production lines. Moreover, climatic variability/change may cause substantial impacts on water resources in the future [41
]. Our results indicate that provinces in the regions of northeast and northwest China are primary exporters of water resources. However, empirical research has suggested a decreasing trend in precipitation in these regions mainly due to the climatic variability/change since the late 1980s, which may impose challenges on future water supply [42
]. These results further stress the importance of policy makings to mitigate the impacts of climatic change and better manage water resources both in local regions and the entire country.
The presented analysis has several limitations. First, the data we used cannot reflect the latest situation of inter-regional water flows, although the data were indeed published quite recently in 2012 and 2014. This drawback is mainly due to the complicated procedures for a set of well-established IO tables. In future study, we will try to apply more recent data if there is any. Second, the MRIO model applied only provides a snapshot of the overall WF pattern in China from the sectoral perspective and fails to reflect the detail impacts of individual products on WF [43
]. Additionally, the data we used only covered six economic sectors, which have a relatively coarse resolution. The aggregation of multiple sub-sectors (e.g., the agricultural sector) may introduce uncertainty to the modeling results [44
]. Third, the presented analysis focuses more on domestic inter-regional water flows while it does not account for those embodied in international trade between China and other countries in the world. We will address this issue in future work by incorporating complementary data of international trade.