The Temporal Evolution of Physical Water Consumption and Virtual Water Flow in Beijing, China

Highlights

• Water shortage has become the main bottleneck restricting the sustainable development of Beijing.
• The temporal evolution of physical water consumption and virtual water flow in Beijing is analyzed.
• The total physical water consumption of various economic departments in Beijing shows an overall decreasing trend.
• There are significant differences in virtual water flow patterns of different departments.
• Inter-regional joint production will be the key to solve the problem of uneven spatial and temporal distribution of water resources.

Abstract

With the rapid development of the socio-economic system and the close connection of inter-regional trade, the physical water consumption in production and the virtual water flow associated with inter-regional trade are both have a significant impact on local water systems, especially in megacities. Beijing is the political, economic and cultural center of China, which is a megacity that has severe water scarcity. To evaluate the status-quo of local water consumption and propose the countermeasures, this study quantitatively analyzed the evolution trend of physical water consumption and the virtual water flow in Beijing. The results show that the total physical water consumption in Beijing decreased from 2.43 billion m³ (2002) to 1.98 billion m³ (2017), while the net virtual water input increased from 1.76 billion m³ (2002) to 3.09 billion m³ (2017), which was mainly embedded in agricultural and industrial products. This study also reveals the equal importance of physical water and virtual water in ensuring the regional water security and sustainable economic development. In view of poor water resource endowment, Beijing should conduct the coupled management of physical water and virtual water to alleviate the local water shortage, i.e., to receive more virtual water embedded in agricultural and industrial products, and allocate the limited local water resources to domestic use and high-benefit sectors.

1. Introduction

With the continuous population growth and the accelerated socio-economic development, regional water security has become an important part for the sustainable development [1,2]. Beijing is one of the most water-deficient cities in China, where water shortage has become the main bottleneck restricting the sustainable development of the economy and society [3,4]. To guarantee water security, the South-to-North Water Diversion Project has provided Beijing with more than 880 million m³ water annually since 2014, accounting for more than 22% of its local water supply [5]. Furthermore, virtual water importing also has played significant role in ensuring Beijing’s water security. For example, from 2000 to 2016, the virtual water input embedded in food increased from 3.55 billion m³ to 16.76 billion m³, and the virtual water input embedded in energy products rose from 52.76 million m³ to 137.47 million m³ [6]. Currently, the external water consumption in Beijing has exceeded 50% of the total regional water consumption [7]. The overall water demand presents a dual dependence on physical water and virtual water, and the degree of dependence is increasing. Therefore, the physical water and virtual water need to be comprehensively considered for the water resources management and water security guarantee [8].

In recent years, inter-basin physical water transfer and virtual water trade have been regarded as two major methods that can effectively alleviate the regional water resources pressure. The implementation of the South-to-North Water Diversion Project [9], the Great Lakes Basin Water Diversion Project [10], the Central Valley Project [11], and other inter-basin water diversion projects are successful stories of physical water security guarantees [12]. With the rapid development of a socio-economic system and the close connection of interregional trade, the virtual water flow associated with interregional trade is having a great effect on the local water system. The concept of virtual water was put forward by Professor Tony Allan [13]. On the basis of the concept of virtual water, Hoekstra proposed the concept of water footprint [14]. Most of the research is to calculate the water footprint at different levels [15]. Allan et al. demonstrated that the virtual water trade was an extension of comparative advantage theory [16]. Existing studies have proved that virtual water trade can alleviate the unbalanced spatial distribution of water resources [17,18,19], and solve the problems of water security [20,21,22], food security [23,24,25,26] and energy security [27,28,29]. At present, the most extensive method for virtual water research is the input-output method [30,31,32,33], Zhang et al. made a multi-regional input-output analysis of domestic virtual water trade and provincial water footprint in China [34]. For the coupling of physical water and virtual water, Gao et al. put forward the cognitive framework of the interaction between the physical and virtual water and the strategies for sustainable coupling management [35]. Zhao et al. analyzed the physical and virtual water transfers for regional water stress alleviation in China [36]. For cities such as Beijing, both physical water and virtual water transfer are aimed to strengthen water management and ensure adequate water supply. One of the current challenges is how to efficiently and appropriately combine physical water and virtual water to relieve water resource pressures [37].

However, most of the previous studies implemented the water management measures from just a single perspective of physical water or virtual water, but paid less attention to combine the physical water and virtual water of each department under the sub-industry. Based on this situation, this paper analyzes the evolution of the water consumption structure and water supply structure in 2002, 2007, 2012 and 2017 in Beijing. Furthermore, the study systematically calculates the physical water consumption and virtual water flow of various research departments in Beijing and their impact on water resources pressure. Furthermore, this study also proposes a series of countermeasures for the local water scarcity response in Beijing. The main contents of this study are summarized as follows: (1) Calculating the physical water consumption of various departments in Beijing and analyzing the changes of water supply structure; (2) Quantifying the water consumption coefficient and virtual water flux of each department by constructing input-output tables; and (3) Analyzing the impact of various industrial water consumption and virtual water flow on local water resources pressure and proposing some countermeasures.

2. Material and Method

2.1. Study Area

Beijing is located in the northern part of the North China Plain, with a total area of 16,410 km² (39.4°–41.6° N, 115.7°–117.4° E). Beijing belongs to typical semi-arid monsoon climate, which is a famous water-deficient area in China. The annual precipitation is 595 mm, 80% of which is concentrated in the period of June to September. The total amount of water resources in Beijing is only 3.55 billion m³, of which 40.3% is surface water and 59.7% is groundwater [38]. It is a densely populated, highly developed and the most water-deficient city in China, whose per capita water resources is only 137 m³, equivalent to 6.8% of the national average and 1.5% of the global average [39] (Figure 1).

Due to the rapid population growth and economic development, the utilization rate of water resources in Beijing has reached 112%, which places tremendous pressure on the local water resources system [40]. In order to meet water demand, Beijing attaches great importance to the use of unconventional water. In 2017, the recycled water consumption reached 1.05 billion m³, accounting for 27% of the total water supply. In addition, the externally transferred water supply was 882 million m³, accounting for 22% of the total water supply [5]. Meanwhile, virtual water consumption related to food increased from 3.55 billion m³ (2000) to 16.76 billion m³ (2016) [6]. Based on the above analysis, it shows that Beijing is a typical water-receiving area of both physical water and virtual water, and the virtual water is also playing a more and more important role for regional water security in the coming years.

2.2. Data

This study mainly used the physical water consumption data of different industries and input-output tables to reveal the temporal evolution of physical water use and virtual water flow in Beijing. The detailed data sources are presented

Table 1. Data Sources.

Type	Data Sources
Agricultural water consumption data	Water Resources Bulletin [5]
Economic input-output table	Beijing Municipal Bureau of Statistics [41]
Per capita domestic water consumption data	China Water Resources Bulletin [42]
Rural and urban populations	China Census yearbook [43]
Industrial departments water consumption data	China Economic Census Yearbook [44]
Total industrial water consumption data	Beijing Water Resources Bulletin [5]
Domestic water consumption data	Beijing Water Resources Bulletin [5]
Water supply sources data	Beijing Water Resources Bulletin [5]

below.

2.3. Method

2.3.1. Calculation of Physical Water Consumption

Unlike agricultural water consumption data, which can be collected directly from the Water Resources Bulletin, the water consumption data of various industrial departments and the service industry are not available and need to be calculated as follows.

Due to the lack of statistics on water consumption of various industrial departments, these data are calculated by the water consumption of above-scale industrial departments and the total water consumption of all industrial departments [45]. The equation can be expressed as:

$$W_i=W_{ai}\times \frac{W_{tit}}{W_{ait}}$$

(1)

where $W_i$ is the total water consumption of the industrial department i in Beijing; $W_{ai}$ is the water consumption of the industrial department i above the scale in Beijing; $W_{ait}$ is the total water consumption of all industrial departments above the designated scale; $W_{tit}$ is the total water consumption of all industrial departments.

The water consumption of the service industry is calculated based on domestic water consumption, which includes rural domestic water consumption and urban domestic water consumption. The urban domestic water consumption consists of urban residential domestic water consumption and urban public water consumption (mainly utilized for services). Therefore, the water consumption of the service industry can be expressed as:

$$W_S=W_L-W_{LVP}\times N_V-W_{LTP}\times N_T$$

(2)

where $W_S$ is the physical water consumption of the service industry in Beijing; $W_L$ is Beijing’s total domestic water consumption; $W_{LVP}$ is the per capita domestic water consumption of rural residents; $N_V$ is the number of rural populations in Beijing; $W_{LTP}$ is the per capita domestic water consumption of urban residents; $N_T$ is the number of urban populations in Beijing.

2.3.2. Calculation of Virtual Water Flow (Input-Output Method)

This study calculated the water consumption coefficient and virtual net water output in four typical years of nine representative industries including agriculture, the mining industry, the textile and garment industry, the paper making and printing industry, the petrochemical industry, the metal smelting and product industry, the electricity production and supply industry, other industries in Beijing. We then analyzed the virtual water flow flux of these nine departments.

The input-output model is at the core of the above calculation. In this study, the water resources input-output model is constructed by adding the direct water consumption of various departments into the economic input-output table as a separate module, as shown in

Table 2. The basic form of the water resources input-output model.

Industry		Intermediate Use				Final Use				Import	Total Output
		Sector 1	…	Sector n	Subtotal	Consumption	Capital Formation	Export	Subtotal
Intermediate input	Sector 1	$X_{11}$	…	$X_{1n}$	${\displaystyle \sum }_{j=1}^n{X}_{1j}$	$C_1$	$f_1$	$e_1$	$y_1$	$i_1$	$X_1$
	…	…	…	…	…	…	…	…	…	…	…
	Sector n	$X_{n1}$	…	$X_{nn}$	${\displaystyle \sum }_{j=1}^n{X}_{nj}$	$C_n$	$f_n$	$e_n$	$y_n$	$i_n$	$X_n$
	Subtotal	${\displaystyle \sum }_{i=1}^n{X}_{i1}$	…	${\displaystyle \sum }_{i=1}^n{X}_{in}$		${\displaystyle {\displaystyle \sum }_{i=1}^n}{C}_i$	${\displaystyle {\displaystyle \sum }_{i=1}^n}{f}_i$	${\displaystyle {\displaystyle \sum }_{i=1}^n}{e}_i$	${\displaystyle {\displaystyle \sum }_{i=1}^n}{y}_i$	${\displaystyle {\displaystyle \sum }_{i=1}^n}{i}_i$	${\displaystyle {\displaystyle \sum }_{i=1}^n}{X}_i$
Value Added		$v_1$	…	$v_n$	${\displaystyle {\displaystyle \sum }_{j=1}^n}{v}_j$
Total input		$X_1$	…	$X_n$	${\displaystyle {\displaystyle \sum }_{j=1}^n}{X}_j$
Water consumption		$W_1$	…	$W_n$	${\displaystyle {\displaystyle \sum }_{j=1}^n}{W}_j$

.

The direct water consumption coefficient refers to the amount of direct water consumption per unit goods or services, which can reflect the utilization of water resources in various departments. It is equal to the ratio of water consumption to total output in the production process of each department:

$$k_j=\frac{W_j}{X_j}$$

(3)

where $k_j$ is the direct water consumption coefficient of sector j; $W_j$ is the total water consumption of sector j; $X_j$ is the total output of sector j. $K=\left[k_j\right]$ is the row vector of the direct water consumption coefficient.

The complete water consumption coefficient, referring to the amount of water consumed (directly and indirectly) by the production of unit final product, is equal to the direct water consumption coefficient times the Leontief inverse matrix. The calculation formula is as follows

$$Q=K{\left(I-A\right)}^{-1}$$

(4)

$$a_{ij}=\frac{x_{ij}}{X_j}$$

(5)

where $Q$ is the row vector of the complete water consumption coefficient; $I$ is the n-order unit matrix; $a_{ij}$ is the direct consumption coefficient, which refers to the direct input from department i to department j in the production of per unit product; $A=\left[a_{ij}\right]$, is the direct consumption coefficient matrix; $x_{ij}$ is the intermediate input provided by department i to department j.

The matrix of virtual water net output can be calculated by the following formula:

$$VWF=Q \left(EM-IM\right)$$

(6)

where $VWF$ represents the matrix of virtual water net output; $EM=\left[e_i\right]$ is the export matrix; $IM=\left[i_i\right]$ is the import matrix.

2.4. Water Stress Index (WSI) and Assumed Water Stress Index (WSI*)

This study introduced the water stress index (WSI) and the assumed water stress index (WSI*) to analyze the impact of virtual water flow on regional water resources. The WSI refers to the ratio of the actual amount of regional water consumed to the amount of regional water available, which can directly reflect the degree of regional water shortage. For another, WSI* represents the water stress index under a hypothetical scenario, i.e., the water pressure generated by local water consumption to meet the water demand without external physical water or virtual water input. The difference between WSI* and WSI represents the contribution of physical water and virtual water input in terms of mitigating water stress.

$$WSI=\frac{WU-P{W}_{net,im}-W_R}{W_a}$$

(7)

$$WS{I}^{*}=\frac{WU-W_R+V{W}_{net,im}}{W_a}$$

(8)

$$WS{I}^{*}-WSI=\frac{P{W}_{net,im}+V{W}_{net,im}}{W_a}$$

(9)

where $WU$ represents the total water consumption of the given departments; $W_R$ refers to the consumption of recycled water; $P{W}_{net,im}$ is the consumption of physical water imported from external areas; $V{W}_{net,im}$ is the external virtual water consumption; and $W_a$ is the available water resources. The WSI is divided into four levels: no stress (WSI < 0.2), moderate stress (0.2–0.4), severe stress (0.4–1.0) and extreme water stress (WSI > 1.0).

3. Results

In this section, we mainly clarify the characteristics of physical water consumption and virtual water flow pattern in Beijing. Based on this, the paper further analyzed the changes of water stress index under the combined influence of local physical and virtual water.

3.1. The Characteristics of Physical Water Consumption

To clearly show the characteristics of physical water supply in Beijing from 2001 to 2018, Figure 2 presents the evolution of different types of water sources. As shown in Figure 2, after an initial decrease from 2001 to 2002, the physical water supply had increased steadily to 3.93 billion m³ in 2018.

As shown in Figure 2a, the proportion of surface water and groundwater in the water supply had continued to decrease. Reclaimed water joined Beijing’s water supply structure in 2003, and its supply had increased from 0.21 billion m³ (2003) to 1.08 billion m³ (2018). The transferred water of the South-to-North Water Diversion joined the water supply structure in 2008, and its proportion overall raising. With the supply of reclaimed water and water transferred from South-to-North Water Diversion Project, the surface water supply had decreased from 1.17 billion m³ (2001) to 0.3 billion m³ (2018) and the groundwater supply had decreased from 2.72 billion m³ (2001) to 1.63 billion m³ (2018), which had significantly alleviated local water shortage.

In the Figure 2b, the proportion of water supply to various industries in Beijing had changed dramatically in the past 20 years. The proportion of domestic and ecological water supply had increased from 12% and 0.3% in 2001 to 18.4% and 13.4% in 2018, respectively. On the contrary, the proportion of industrial and agricultural water supply had shown a downward trend, in which the proportion of agricultural water supply decreased most significantly (about 30%), while the proportion of industrial water supply had dropped from 9.2% to 3.3%. Generally, the physical water supply had gradually shifted from the primary industry and the secondary industry to the tertiary industry.

From Figure 3, it can be found that the agricultural water consumption accounted for the largest share (exceeding 40%) of total water consumption in 2002. However, with the economic and social development of Beijing, the agricultural water consumption had continuously decreased from 1.55 billion m³ (2002) to 0.51 billion m³ (2017), and its proportion in total water consumption had correspondingly decreased from 40.2% to 25.8%. The reduction of agricultural water consumption had led to insufficient food production and supply in Beijing, which caused the inability to meet growing local demand and increased the import dependence. Simultaneously, the water consumption for energy production also showed a downward trend. In 2002, the total water consumption of petrochemical industry and electricity production and supply industry was 0.36 billion m³, and this figure had decreased to 0.17 billion m³ in 2017, which is only 30% of agricultural water consumption, accounting for 8.4% of the total regional water consumption. On the contrary, the service industry water consumption had increased from 0.70 billion m³ (2002) to 1.12 billion m³ (2017), and its proportion of total water consumption had risen from 23.33% in 2002 to 56.51% in 2017. Since 2007, the water consumption of service industry had accounted for the largest part of water consumption in Beijing. Due to the high economic benefits per unit of its water consumption, the rising physical water consumption in the service industry can accelerate the development of the local economy. To summarize, the overall water consumption of these 9 departments had decreased from 2.99 billion m³ (2002) to 1.98 billion m³ (2017).

3.2. The Virtual Water Flow Pattern

The coefficient of each research department had steadily decreased (Figure 4), which shows that with technological innovation and development, the amount of water required by various research departments to obtain per unit output value had constantly decreased. Furthermore, the water consumption coefficient of agriculture (from 0.13 to 0.02) had always been the largest, while that of the service industry (from 0.06 to 0.01) was the smallest, which was less than half of that of agriculture. For energy production, the water consumption coefficient of the petrochemical industry decreased from 0.01 (2002) to 0.002 (2017), with a decrease of 80%. Meanwhile, the coefficient of electricity production and the supply industry changed from 0.03 (2002) to 0.002 (2017), which is 7% of the initial level.

The virtual water net output of agriculture in Beijing had been negative invariably, indicating that the total amount of imported agricultural products was far more than that of exported agricultural products (Figure 5). The net inflow of agricultural virtual water had increased from 0.83 billion m³ (2002) to 2.62 billion m³ (2007) and then decreased to 2.09 billion m³ (2017), which shows an overall fluctuating upward trend. The virtual water net output of agriculture in the four typical years was the smallest (i.e., the net input was the largest), which means that agriculture had the highest dependence on the supply of external resources. Agricultural products are water-intensive products with low value. Therefore, importing agricultural products from outside can save more local water to create higher-value products, which is conducive to economic development.

For energy production, the petrochemical industry is different from the electricity production and supply industry. The virtual water net output of the petrochemical industry had decreased from 0.02 billion m³ (2002) to −0.47 billion m³ (2017), the main way that Beijing obtained petrochemical products had changed from internal production to external import. While that of electricity production and the supply industry had increased from −0.49 billion m³ (2002) to 0.09 billion m³ (2012), then decreased to 0.03 billion m³ (2017), changing overall from a negative value to a positive value, indicating that the department’s dependence on external input was decreased. In addition, the textile and clothing industry, the mining industry, and the paper making and printing industry were less dependent on external input than the metal smelting industry and the petrochemical industry.

With the development of economy and society, the industrial structure had been adjusted. The industries with high water consumption and low efficiency were gradually replaced by the emerging industries with low water consumption and high efficiency. The net output of virtual water in Beijing’s service industry had increased from 0.73 billion m³ (2002) to 1.58 billion m³ (2007) and then decreased to 1.28 billion m³ (2017), which shows an overall fluctuating upward trend. And in the four typical years, the service industry was the department that had the highest net output of virtual water. A large amount of virtual water outflow will also increase the water pressure, but the value produced by its unit water consumption and outflow is much higher than that of the primary industry and secondary industry.

3.3. Comprehensive Impact of Physical Water Consumption and Virtual Water Flow

To respond to the strictest water resources system in China (in which the total water consumption control target of each province is formulated) and facilitate the analysis of the regulation and utilization of water resources, this study adopted the total water consumption control target of Beijing as the available water volume. Based on the analysis of the two dimensions of physical water consumption and virtual water flow (Figure 6), it can be found that Beijing can barely balance the utilization of water resources only under the co-maintenance of physical water and virtual water. The total WSI* was always greater than 1, which indicates that without the input of external physical water and virtual water, local water resources cannot meet local water demand. This was caused by excessive water consumption in agricultural and industrial production. Among them, the WSI* of agriculture and industry were over half of the total WSI*, and the average value of agricultural reached 0.61, while that of industry was 0.64. On the contrary, the WSI* of the service industry was negative, and decreased from −0.006 (2002) to −0.195 (2017), which shows that if the physical and virtual water flow was not considered, the service industry will alleviate local water resources pressure and be advantageous to solving water security.

According to the four levels of water stress degree, Beijing’s water stress degree was “severe stress” in 2002 (WSI = 0.61), “moderate stress” in 2007 (WSI = 0.39), “no stress” in 2012 (WSI = 0.18), and no stress in 2017 (WSI = 0.02), means that the shortage of local water resources had been prodigious relieved. However, the WSI* in the same year was much higher than WSI. The increase of the difference between WSI and WSI* reflects that the improvement degree of external physical water and virtual water input on the water shortage in Beijing had been increasing. The external water not only plays an irreplaceable role in alleviating local water shortage, but also promotes local economic development and industrial optimization.

4. Discussion

4.1. Comparison Analysis with Previous Studies

The main uncertainty of this study comes from the water consumption data of industrial departments. Due to the limitation of data, this study introduced the method proposed by [45] to allocate the total amount of industrial water, so as to obtain the water consumption data of various industrial departments. Because the consumption pattern and industrial structure will change with the development of the local economy, there is uncertainty in using the data of 2008 for allocation.

To further analyze the uncertainty of the calculation results, this paper compares the results between our study and others. According to the results, Beijing is a virtual water input area, which is consistent with [46,47,48]. Furthermore, Han et al. quantified the momentum of virtual water flow in Beijing, and the results showed that the virtual water inflow has increased to 3.2 billion m³ (2017) [49], which is within ±5% of our estimate. Yang et al. analyzed the water consumption structure in Beijing and found that industry had a great impact on water consumption in Beijing, while agriculture has a small impact [50]. This is also consistent with our calculation. In addition, the conclusion that agriculture was the main input department of virtual water appeared in our study and Liu et al. [48]. Lastly, Zhang et al. calculated the WSI in Beijing, and they found that the WSI decreased from 0.6 (2002) to 0.05 (2017) [51], which is within ±10% of our results. From the comparison results, the results of this study are accurate, scientific and reliable, which can provide scientific and technological support for high-quality development in the study area.

4.2. Inter-Regional Joint Production as the Virtual Water Strategy

Compared with physical water, virtual water has many advantages, such as convenient transportation and low freight [52,53,54]. The mobility and transportation convenience of virtual water can be easily used to effectively cope with regional water scarcity [55]. For example, in order to alleviate the water shortage in food production, Japan imported abundant food from the United States and Canada, and the corresponding virtual water input of food was 8.76 billion m³ [56]. Particularly in the current double circulation system of the domestic and foreign economy, the virtual water flow will be exacerbated with the intensification of the commodity trade, and the inter-regional joint production as the virtual water strategy will be the key to solve the problem of the uneven distribution of water resources in time and space [57].

Inter-regional joint production is a kind of mutual economic production method, which emphasizes a win-win cooperation and mutually beneficial production relationship [58,59,60]. The principle of this method is to produce water-intensive products in certain areas where the water resources are abundant, and to produce high value-added products in areas with relatively poor water resources by importing water-intensive products to achieve greater economic benefits. For example, Beijing can transfer its high-water consumption industries to other provinces where the water resources are abundant [61,62], and the products produced in other provinces can be returned to Beijing through inter-regional trade to achieve a win-win situation [63,64]. The inter-regional joint production will be the best way for Beijing to solve the problem of its own water shortage and excessive pressure on water resources. Meanwhile, the method also has sustainability and security stability, which can promote the common sustainable development of different provinces at a larger scale.

4.3. How to Alleviate the Water Shortage in Beijing

Based on above analysis, the following suggestions are proposed to significantly alleviate the water shortage in Beijing:

(1) The first choice is to improve the water-saving technology and optimize the industrial structure based on water carrying capacity [65]. Beijing should vigorously develop agricultural water-saving technologies and change the crop planting structure aiming at lower water consumption. For industry and service, water use efficiency should be continually improved, and the industrial structure should also be optimally adjusted to match the local water carrying capacity.

(2) To increase the water resources availability and search for new water sources is an urgent mission [66]. Generally, Beijing could increase the availability of water supplies in different ways, such as the South-to-North Water Diversion Project, the utilization of reclaimed water, and seawater desalination. According to the estimation, the new increased water supply can reach 2.5 billion m³ in 2025, which could efficiently alleviate the problem of water scarcity in Beijing [67].

(3) The updating of water management practices is an important support [68]. Beijing should strictly enforce water resources management. It is urgent to implement multi-regional water resources governance to expand the existing regional water resources system, which can make the water resources system more stable. From the perspective of a virtual water strategy, the government should encourage markets to purchase food and other water-intensive industrial products from water-rich regions to reduce the use of local water resources, and the saved water could be used to produce more high value-added commodities.

5. Conclusions

This study assessed the physical water consumption and virtual water flow of nine research departments in Beijing. On this basis, this study also analyzed the water resources pressure induced by physical water consumption and regional virtual water flow. The main conclusions can be summarized as follows:

(1) The total physical water consumption of various industries in Beijing shows an overall downward trend (except for ecological environment water and residential water), while the temporal evolution characteristics of physical water consumption of various water sources and economic departments are different. In terms of different water sources, the consumption of reclaimed water and water transferred from South-to-North Water Diversion Project gradually increases, while the consumption of surface water and groundwater continuously decreases. For the studied nine economic areas, the total water consumption of these presents a downward trend. Among them, agricultural water consumption decreased most obviously, from 1.55 billion m³ (2002) to 0.51 billion m³ (2017). Furthermore, the water consumption of various industrial departments also declined, but not as dramatically as that of agriculture. However, the water consumption of the service industry presented an upward trend.

(2) Beijing is a typical area of net virtual water input, but there are significant differences in virtual water flow patterns of different economic departments. The net virtual water input had increased from 1.75 billion m³ (2002) to 3.08 billion m³ (2017). Among them, the inflow of agricultural virtual water was the largest, accounting for 47.43% (2002) and 67.86% (2017) of the total net virtual water inflow, respectively. Furthermore, most of the industrial departments (except for the petrochemical industry and the electricity production and supply industry, which experienced the change of virtual water flow direction) had virtual water imported. On the contrary, the service industry had exported virtual water, and its output showed an increasing trend, from 0.73 billion m³ (2002) to 1.28 billion m³ (2017).

(3) Without the external physical water and virtual water, the local water resources cannot meet the increasing water demand. In the four typical years, the total WSI* was always greater than 1. Moreover, the difference between WSI and WSI* had increased from 0.44 (2002) to 0.99 (2017), indicating that the improvement degree of external physical water and virtual water input on the water shortage in Beijing had increased significantly. Therefore, to ensure water security and sustainable development, Beijing should conduct the coupled management of physical water and virtual water.