China is a country of agriculture, and agricultural production consumes a great deal of water. In this paper, we quantify the provincial food production water footprint (WF) in China during 1997–2011, and then analyze its change trend by the method LMDI (Logarithmic Mean Divisia Index). The results indicate the following. (1) China's food production WF increased during 1997–2011 as a whole. The food production WFs at the provincial level are obviously different. (2) The main reason for the change of WF of food production in China related to the virtual water content and total food production. As for the changes of food production WFs for each province, they were not always in accordance with the total food production. For example, in Guizhou, Qinghai, Sichuan, and other provinces, the food production WFs grew while total food production declined, thus indicating strong negative decoupling. Thus, it is necessary to take the measure of agricultural products' transportation ‘green channel’ to promote the development of domestic food trade and virtual water trade.

China is a large agricultural country with a long history, whose agricultural production requires great consumption of water resources. According to China's national census of water1 over the past years, agricultural water has accounted for about 70% of the total production water supply. Agricultural water, to a great degree, has been used for the production of rice, corn, wheat, beans, potato, and other food crops, for which it is necessary to study the water resources required for the production of food crops in China. In order to calculate the water resources consumed in agricultural production, Hoekstra & Hung (2002) put forward the concept of water footprint (WF). As an indicator of freshwater use, the WF takes both direct and indirect water use of a consumer or producer into consideration. WF can be subdivided into green, blue, and gray WF. The green water refers to soil moisture. The blue water refers to the surface and ground water resources along the supply chain of a product. The gray water refers to the volume of freshwater required to assimilate the load of pollutants based on natural background concentrations and existing ambient water quality standards (Hoekstra et al., 2011).

There are many studies about the production or consumption WF calculation of food and other agricultural products. The calculation methodology of WF can be subdivided into bottom-up and top-down methods (Zhang et al., 2017). (1) Bottom-up method: Production tree method is a typical bottom-up method. The calculation of production tree method is complicated, and is mainly applied for calculating WF of crops and animal products. Based on the software Crop-water, this method, first of all, calculates crops' virtual water with meteorological data or data of their plating areas, output, and so on (Allen et al., 1998; Hoekstra & Hung, 2002; Hoekstra et al., 2011). Then, production WF of each crop can be calculated by multiplying each crop's virtual water with its production. Similarly, consumption or trade WF of each crop can be calculated by multiplying its virtual water with its consumption and trade volume. For animal products, WF of daily water used for drinking, cleaning, and fodder during animals' feeding is taken into account. Then, their virtual water can be calculated by combining WF with each value factor and scaling factor. Last, production, consumption, and trade WF can be calculated by multiplying the virtual water with their production, consumption, and trade volume. Ever since this method was put forward, many researchers have empirically studied WF of crops and animal products in the world (Chapagain et al., 2006; Chapagain & Hoekstra, 2007, 2011; Mekonnen & Hoekstra, 2010; Zheng et al., 2012; Gerbens-Leenes et al., 2013; Shrestha et al., 2013; Vanham & Bidoglio, 2013; Zoumides et al., 2014; Cao et al., 2014; Fulton et al., 2014; Wang et al., 2015; Su et al., 2015; Chouchane et al., 2015). (2) Top-down method: Input–output method is a typical top-down method. It, first of all, calculates direct water consumption coefficient with water consumption in different sectors (in general, it refers to blue water consumption) and output data. Then, the total water consumption coefficient is calculated by multiplying it with Leontief inverse matrix. Finally, production, consumption, and trade WF can be calculated by multiplying those total water consumption coefficients with their output, consumption, and trade volume. Some researchers have only considered the blue WF, and use input–output method to assess the WFs of agriculture, industry, and service sectors (Zhao et al., 2009; Feng et al., 2011; Zhang & Anadon, 2014; Li & Chen, 2014; Huang et al., 2015; Deng et al., 2016). China covers a vast territory, and food products vary greatly in different regions, along with the soil and climatic conditions. As a result, it leads to greatly differentiated virtual water contents (VWCs). Therefore, this study carries out research on WF accounting at the provincial level. In addition, the WF calculated by input–output method is only related to the macro-industrial sectors, taking no account of rice, corn, wheat, beans, potato, or other specific food crops, and only the calculation results of blue WF are taken into account. Therefore, in this paper, to take both green and blue WF into account, the production tree method is adopted to calculate the WFs in the food production process of Chinese provinces.

Although there have been many studies on the accounting of the WF, the factors driving the changes of WFs, and the proportions of various driving factors are further studied, in which process the choice of factor decomposition method is involved. At present, there are mainly two factor decomposition methods, namely SDA (Structural Decomposition Analysis) and IDA (Index Decomposition Analysis), while LMDI (Logarithmic Mean Divisia Index) is the most common method. (1) SDA: Duarte et al. (2014) utilized the SDA methodology to study the change trend of the WF in Spain during 1860–2011, while Xie (2014) presented a decomposition analysis of changes in China's energy usage during 1992–2010. Wang et al. (2016) provided detailed insight into how diverse policy evolution affected the WF in China from 1997 to 2007 through input–output analysis and SDA. (2) IDA: Through the employment of the LMDI method, Jeong & Kim (2013) assessed the change trend of greenhouse gas emissions in South Korea during 1991–2009, and Voigt et al. (2014) applied the LMDI to decompose the energy intensity changes of 40 major economies. Guo et al. (2014) studied water consumption-related chemical oxygen demand emission in Chinese industrial sectors with LMDI method. Xu et al. (2015) used the LMDI method to quantitatively analyze the driving factors for changes in WF from 1978 to 2012 in Beijing. There are n! different approaches to decompose n factors in the method of SDA (Dietzenbacher & Los, 1998), while LMDI method satisfies the uniqueness of decomposition. Therefore, in this paper, the LMDI method has been adopted to decompose the change trend of China's food production.

In order to analyze the relationship between the change of WF and food output, this paper further analyzes the decoupling. Originally proposed by OECD (2001), the concept of decoupling analysis aimed to investigate the link between ‘environmental bads’ and ‘economic goods’. Then, Tapio (2005) developed the theory of decoupling analysis, based on which research was conducted on the relationship between economic development and carbon emissions in the European Union. Since then, decoupling analysis has been applied to a large number of empirical studies. For example, Zhang & Wang (2013) studied the decoupling between CO2 emission and economic growth in Jiangsu Province, China during 1995–2009. Zhang & Yang (2014) researched decoupling agricultural water consumption and environmental impact from crop production. Song (2014) found a decoupling cultivated land loss by construction occupation from economic growth in Beijing. In the same year, after investigation on the changes in rural register population and rural settlement area, Song & Liu (2014) analyzed the decoupling between them. Furthermore, Tang et al. (2014) applied decoupling indicators to analyze tourism-related CO2 emissions in China during 1990–2012.

The main innovations in this paper are listed as follows. (1) With the province as a research unit, in this paper, the WF of food production in China during 1997–2011 is calculated. (2) Through the adoption of the LMDI methodology, the decomposition and analysis of the changes in WF of food production at the national and at the provincial level are conducted. (3) The decoupling between changes of food production WF and food output in China is analyzed. The rest of the paper is organized as follows: below the research methodology of the model employed in this study is presented, while the next section discusses the adopted data. This is followed by the results and discussion, and finally the conclusions and suggestions are presented.

Decomposition of the changes in WF

The LMDI method can satisfy the uniqueness of the decomposition. Therefore, in this paper, the LMDI method is adopted to decompose the change trend of China's food production WF. Based on the calculation formula of the WF (Hoekstra & Hung, 2005; Duarte et al., 2014), the WF of each food crop is equal to the VWC of the crop multiplied by the crop output. Then, we add them up by the number of provinces and crop species, and we can get:
formula
(1)
where indicates the WF of grain production in China for the year t, represents the VWC of the ith food species in Chinese provinces t for the year t, and is the food production output in Chinese provinces t for the year t. Provided , , where is the food production structure in Chinese provinces t for the year t, indicates the proportion of food production by each province, means the total food production output in Chinese provinces t for the year t, refers to the total food production output in China's provinces for the year t, then we get:
formula
(2)
We further obtain the derivative of Equation (2) as follows:
formula
(3)
Provided , we can get:
formula
(4)
Then, Equation (2) was integrated by t to get Equation (5):
formula
(5)
In order to solve Equation (5), add weight function as follows:
formula
(6)
Through the combination with weight function, we can solve Equation (5):
formula
(7)
where represents the change of WF caused by that of VWC and is the change of WF caused by that of food production structure, is the change of WF caused by that of the proportion of food production in each province, and means the change of the WF caused by that of the total food output in China.
Similar to the derivation of Equation (7), the decomposition formula of WF change trend for each province can be obtained with the following equations:
formula
(8)
where is the change of the WF caused by that of VWC in the province r, is the change of the WF caused by that of food production structure in the province r and is the change of the WF caused by that of the total grain output in the province r.
Similar to the derivation of Equation (7), provided , the decomposition formula of the WF change trend for rice, corn, wheat, beans, potato, and another five food crop categories can be obtained as follows:
formula
(9)
where indicates the change of WF caused by that of the VWC of crop i, represents WF change caused by the proportion change of the crops i produced by each province in the nation, and refers to the WF change caused by the total production change of crops i in the nation.

Decoupling WF change from change in total food production output

When it comes to food production, capital, labor, land, and other production factors are required. However, the food productions per unit area are different, while the WF changes are not consistent with the total food production output changes in each province. Accordingly, following the method of decoupling analysis proposed in the literature such as that of OECD (2001) and Tapio (2005), the decoupling indicators used to measure the decoupling of WF changes from the total food production output changes are defined as follows:
formula
(10)
where refers to the value of WF change in the province r and is the value of total food production output in the province r; represents the original value of WF in the province r, and indicates the original value of total food production output in the province r. Then, with Equation (8) substituted into the calculation of Equation (10), we can get Equation (11) as follows:
formula
(11)

Therein, is the decoupling indicator which aims to measure the decoupling of the total food production output changes from WF changes caused by VWC changes, is the decoupling indicator used to measure the decoupling of the total food production output changes from WF changes caused by food production structure changes, and is the decoupling indicator employed to measure the decoupling of the total food production output changes from WF changes caused by total food production output change in the province r.

According to the research of Tapio (2005), decoupling indicators can be divided into the following eight categories, as illustrated in Table 1.

Table 1.

The division of decoupling indicators.

Type
Weak decoupling >0 >0 (0,0.8] 
Expansive coupling >0 >0 (0.8,1.2] 
Expansive negative decoupling >0 >0 (1.2, + ∞) 
Strong decoupling <0 >0 (−∞,0] 
Strong negative decoupling >0 <0 (−∞,0] 
Weak negative decoupling <0 <0 (0,0.8] 
Recessive coupling <0 <0 (0.8,1.2] 
Recessive decoupling <0 <0 (1.2, + ∞) 
Type
Weak decoupling >0 >0 (0,0.8] 
Expansive coupling >0 >0 (0.8,1.2] 
Expansive negative decoupling >0 >0 (1.2, + ∞) 
Strong decoupling <0 >0 (−∞,0] 
Strong negative decoupling >0 <0 (−∞,0] 
Weak negative decoupling <0 <0 (0,0.8] 
Recessive coupling <0 <0 (0.8,1.2] 
Recessive decoupling <0 <0 (1.2, + ∞) 

This paper studies the WFs in the production of rice, corn, wheat, beans, potato, and another five different agricultural products, in which the data during the time span of 1997–2011 for 30 provinces in China (limited by statistical data available, Tibet and Hong Kong, Macao, Taiwan regions are not included) have been employed for analysis. Similarly to Duarte et al. (2014), this paper calculates VWC of various kinds of food crops, in which the green water and blue water usage, instead of the gray water usage2 is taken into consideration. Accordingly, except for any specific explanation, WF in this paper refers to green WF and blue WF, but not gray WF. The methodology for calculating the VWC of various grain crops can be consulted from Allen et al. (1998) and the climate data have been acquired from the China meteorological data sharing service system3, with the software used in the calculation being Cropwat. As for the production output and planting area data of rice, corn, wheat, beans, potato and other food crops, they are collected from China's Rural Statistical Yearbook (The Division for Rural Social Economy Investigation of the National Bureau of Statistics, 1998–2012).

Calculation results of food production WF in China

National level

According to Equation (1), this paper calculates the WF of food production in China (WFs relating to rice, corn, wheat, beans and potato, as well as another five different agricultural products). The calculation results are shown in Figure 1.

Fig. 1.

China food production WF during 1997–2011 (unit: 1 billion m3).

Fig. 1.

China food production WF during 1997–2011 (unit: 1 billion m3).

Close modal

The following is shown in Figure 1. (1) For the sum of green and blue WF. During 1997–2003, the WF of food production in China shows a decreasing tendency (decreased from 625.53 billion m3 in 1997 to 511.32 billion m3 in 2003). However, in 1999, there is a relative rise compared with the situation in 1998 (it was 606.30 billion m3 in 1999 and 602.77 billion m3 in 1998). During the period 2003–2011, the WF of China's grain production shows a rising tendency (rose from 511.32 billion m3 in 2003 to 637.29 billion m3 in 2011). Nevertheless, in 2010, it has relatively declined compared with the situation in 2009, which is the same for the condition in 2008 compared with that of 2007 (it was 611.31 billion m3 in 2007, 597.40 billion m3 in 2008, 654.19 billion m3 in 2009, and 624.15 billion m3 in 2010). The main reason for this phenomenon lies in the improvement of water saving technology in the food production process; meanwhile, the VWC of all kinds of food crops (can be understood as the water consumption of per unit mass food production) shows a downward trend, which will lead to the decline of food production WF in China. On the other hand, the improvement of food production technology has led to the increase of food production output, which thus results in the rise of China's food production WF. Furthermore, analyses of the reasons for the above are shown in the section ‘Trend decomposition of the food production WF in China’. (2) The comparison between green and blue WF. When compared with blue WF during 1997–2011 in China, green WF is as much as 42%–55% larger, which shows that production of China's food crops relies mostly on natural rainfall. The storage from rivers and lakes is used for supplying the insufficiency of natural rainfall. In addition, the change trend of green and blue WF remains almost the same due to the effect of changes in food crops' output.

It needs to be pointed out that, in this study, WF related to accounting and change trend analysis of WF in provincial and crops' aspect. The decoupling analysis of change in food production WF and food output refers to the sum of green and blue WF.

Provincial level

The WFs of food production at provincial level during 1997–2011 are shown in Figure 2(a)2(c).

Fig. 2.

China's provincial WFs of food production during 1997–2011 (unit: 1 billion m3): (a) larger group, (b) medium group, (c) fewer group.

Fig. 2.

China's provincial WFs of food production during 1997–2011 (unit: 1 billion m3): (a) larger group, (b) medium group, (c) fewer group.

Close modal

The following is shown in Figure 2(a)2(c). (1) During 1997–2011, the change of food production WF in China appears to be more frequent. (2) During 1997–2011, those provinces with larger food production WFs are Henan province, Hebei province, Shandong province, and Heilongjiang province, due to their large food production. Moreover, during 1997–2011, those provinces with fewer food production WFs are Qinghai Province, Beijing city, and Shanghai City because of their lower food production volume. Also, the analyses of the reasons for the above are shown in the section ‘Trend decomposition of the food production WF in China’.

Crop level

The WFs of food crops such as rice, corn, wheat, beans, and potato are shown in Figure 3.

Fig. 3.

WFs of five kinds of food crops production in China during 1997–2011 (unit: billion m3).

Fig. 3.

WFs of five kinds of food crops production in China during 1997–2011 (unit: billion m3).

Close modal

From Figure 3, it can be seen that the productions of rice, wheat, and corn have relatively larger WFs during 1997–2011, while the productions of beans and potato have relatively fewer WFs. During 1997–2003, the WFs of rice, wheat, and corn productions generally have a decreasing tendency, but there is a general rise during 2003–2011. However, when it comes to the WFs relating to rice, beans, and potato production, the gap between the amount in 1997 and that of 2011 is smaller (in 1997, it was 198.06 billion m3, 37.65 billion m3, 48.70 billion m3, respectively, while it was 199.53 billion m3, 35.11 billion m3, 44.45 billion m3, respectively, in 2011).

Spatial distribution maps of WF

In this paper, taking 1997 and 2011 for example, we illustrate the WF spatial distribution in Chinese provinces, as shown in Figures 4 and 5.

Fig. 4.

The Chinese spatial distribution of WFs in 1997 (unit: 1 billion m3).

Fig. 4.

The Chinese spatial distribution of WFs in 1997 (unit: 1 billion m3).

Close modal
Fig. 5.

The Chinese spatial distribution of WFs in 2011 (unit: 1 billion m3).

Fig. 5.

The Chinese spatial distribution of WFs in 2011 (unit: 1 billion m3).

Close modal

As can be seen from Figures 4 and 5, in 2011, the spatial distribution of WF in China shows little change. The WFs of Qinghai, Ningxia, Shanxi, Beijing, Tianjin, Shanghai, Hainan, and other provinces are still less than 5 billion m3, while the amounts in such provinces as Heilongjiang, Henan, Hebei, and Shandong are still greater than 35 billion m3.

Trend decomposition of the food production WF in China

National level

According to Equation (7), the results of the changes in food production WFs in China, through factor decomposition, are shown in Table 2.

Table 2.

Factor decomposition of changes in food production WF in China (1997–2011) (unit: 1 billion m3, %).

YearIncrement/Proportion
1997–1998 Increment (billion m3−38.79 −9.52 2.43 23.12 −22.76 
Proportion (%) 170.39 41.82 −10.66 −101.55 100.00 
1998–1999 Increment (billion m35.03 2.61 −2.10 −2.02 3.53 
Proportion (%) 142.70 74.03 −59.44 −57.29 100.00 
1999–2000 Increment (billion m346.30 0.35 0.52 −57.52 −10.35 
Proportion (%) −447.25 −3.41 −5.03 555.68 100.00 
2000–2001 Increment (billion m37.24 −3.99 2.10 −11.66 −6.30 
Proportion (%) −114.83 63.27 −33.32 184.88 100.00 
2001–2002 Increment (billion m3−39.34 −0.59 1.58 4.55 −33.79 
 Proportion (%) 116.41 1.74 −4.68 −13.47 100.00 
2002–2003 Increment (billion m3−12.25 −0.15 −0.11 −32.02 −44.53 
Proportion (%) 27.51 0.33 0.26 71.90 100.00 
2003–2004 Increment (billion m30.20 −3.02 1.22 48.54 46.94 
Proportion (%) 0.42 −6.43 2.59 103.42 100.00 
2004–2005 Increment (billion m3−5.73 −2.91 1.14 17.52 10.02 
Proportion (%) −57.17 −29.10 11.40 174.87 100.00 
2005–2006 Increment (billion m3−3.18 −0.57 3.01 18.23 17.48 
Proportion (%) −18.20 −3.26 17.19 104.27 100.00 
2006–2007 Increment (billion m325.35 −3.73 −0.93 4.85 25.55 
Proportion (%) 99.23 −14.60 −3.63 19.00 100.00 
2007–2008 Increment (billion m3−47.88 0.64 1.13 32.20 −13.91 
Proportion (%) 344.32 −4.58 −8.14 −231.60 100.00 
2008–2009 Increment (billion m347.20 −3.53 −0.14 4.26 47.79 
Proportion (%) 98.77 −7.38 −0.30 8.91 100.00 
2009–2010 Increment (billion m3−35.69 −3.44 0.79 17.29 −21.04 
Proportion (%) 169.63 16.33 −3.76 −82.20 100.00 
2010–2011 Increment (billion m3−16.18 1.35 −0.81 28.78 13.14 
Proportion (%) −123.11 10.24 −6.18 219.05 100.00 
1997–2011 Increment (billion m3−67.70 −26.50 9.83 96.14 11.76 
Proportion (%) −575.70 −225.32 83.55 817.47 100.00 
YearIncrement/Proportion
1997–1998 Increment (billion m3−38.79 −9.52 2.43 23.12 −22.76 
Proportion (%) 170.39 41.82 −10.66 −101.55 100.00 
1998–1999 Increment (billion m35.03 2.61 −2.10 −2.02 3.53 
Proportion (%) 142.70 74.03 −59.44 −57.29 100.00 
1999–2000 Increment (billion m346.30 0.35 0.52 −57.52 −10.35 
Proportion (%) −447.25 −3.41 −5.03 555.68 100.00 
2000–2001 Increment (billion m37.24 −3.99 2.10 −11.66 −6.30 
Proportion (%) −114.83 63.27 −33.32 184.88 100.00 
2001–2002 Increment (billion m3−39.34 −0.59 1.58 4.55 −33.79 
 Proportion (%) 116.41 1.74 −4.68 −13.47 100.00 
2002–2003 Increment (billion m3−12.25 −0.15 −0.11 −32.02 −44.53 
Proportion (%) 27.51 0.33 0.26 71.90 100.00 
2003–2004 Increment (billion m30.20 −3.02 1.22 48.54 46.94 
Proportion (%) 0.42 −6.43 2.59 103.42 100.00 
2004–2005 Increment (billion m3−5.73 −2.91 1.14 17.52 10.02 
Proportion (%) −57.17 −29.10 11.40 174.87 100.00 
2005–2006 Increment (billion m3−3.18 −0.57 3.01 18.23 17.48 
Proportion (%) −18.20 −3.26 17.19 104.27 100.00 
2006–2007 Increment (billion m325.35 −3.73 −0.93 4.85 25.55 
Proportion (%) 99.23 −14.60 −3.63 19.00 100.00 
2007–2008 Increment (billion m3−47.88 0.64 1.13 32.20 −13.91 
Proportion (%) 344.32 −4.58 −8.14 −231.60 100.00 
2008–2009 Increment (billion m347.20 −3.53 −0.14 4.26 47.79 
Proportion (%) 98.77 −7.38 −0.30 8.91 100.00 
2009–2010 Increment (billion m3−35.69 −3.44 0.79 17.29 −21.04 
Proportion (%) 169.63 16.33 −3.76 −82.20 100.00 
2010–2011 Increment (billion m3−16.18 1.35 −0.81 28.78 13.14 
Proportion (%) −123.11 10.24 −6.18 219.05 100.00 
1997–2011 Increment (billion m3−67.70 −26.50 9.83 96.14 11.76 
Proportion (%) −575.70 −225.32 83.55 817.47 100.00 

Note: The last column of the increment in Table 1 is the sum of the previous four columns, and the proportion is the ratio of the corresponding increment.

Table 2 shows the following. (1) Factor decomposition of the annual changes in WFs. The change of WF of food production is mainly caused by the change of VWC and total food production output . Changes in the production structure of food and the proportion of the provinces' production in the nation have exerted little influence on the WF, which can be easily explained according to Equation (1). Among these years, WF was changed mainly by the VWC in 1997–1998, 1998–1999, 2001–2002, 2006–2007, 2007–2008, and 2009–2010. WF was changed mainly by the proportion of provinces' production in 1999–2000, 2002–2003, 2003–2004, 2004–2005, 2005–2006, 2008–2009, and 2010–2011. (2) The factor decomposition of the changes of WF during 1997–2011. The variation of total food production output is a leading cause of WF change of food production , while the VWC serves as another cause, followed by the change of food production structure and change of the provincial production proportions .

Provincial level

According to Equation (8), through the use of factor decomposition, the results of the change in WF of food production in each province of China are shown in Table 3 (lists the decomposition results for the whole time period).

Table 3.

Factor decomposition of changes in food production WF at provincial level (1997–2011) (unit: 1 billion m3, %).

ProvinceIncrement/Proportion
Anhui Increment (billion m3−6.15 −0.47 3.81 −2.81 
Proportion (%) 218.45 16.83 −135.29 100.00 
Bejing Increment (billion m30.17 −0.15 −1.92 −1.91 
Proportion (%) −8.69 7.80 100.88 100.00 
Chongqing Increment (billion m3−3.03 −0.24 −0.79 −4.06 
Proportion (%) 74.66 5.93 19.41 100.00 
Fujian Increment (billion m3−1.36 −0.08 −3.46 −4.90 
Proportion (%) 27.79 1.66 70.55 100.00 
Gansu Increment (billion m3−2.32 0.26 5.98 3.92 
Proportion (%) −59.06 6.58 152.48 100.00 
Guangdong Increment (billion m33.67 −0.03 −6.44 −2.79 
Proportion (%) −131.29 1.02 230.27 100.00 
Guangxi Increment (billion m3−0.07 −0.09 −1.57 −1.73 
Proportion (%) 4.12 5.46 90.42 100.00 
Guizhou Increment (billion m33.60 −0.31 −2.13 1.17 
Proportion (%) 308.87 −26.27 −182.60 100.00 
Hainan Increment (billion m3−0.52 −0.01 −0.62 −1.15 
Proportion (%) 45.44 0.44 54.12 100.00 
Hebei Increment (billion m3−11.01 −2.28 6.38 −6.90 
Proportion (%) 159.40 32.97 −92.37 100.00 
Henan Increment (billion m3−8.26 −2.38 25.42 14.78 
Proportion (%) −55.87 −16.08 171.94 100.00 
Heilongjiang Increment (billion m3−5.98 −8.23 31.93 17.72 
Proportion (%) −33.75 −46.45 180.20 100.00 
Hubei Increment (billion m3−4.36 −0.27 −2.32 −6.95 
Proportion (%) 62.79 3.84 33.38 100.00 
Hunan Increment (billion m32.61 −0.43 1.75 3.93 
Proportion (%) 66.49 −10.96 44.47 100.00 
Inner Mongolia Increment (billion m3−5.61 −0.73 12.28 5.94 
Proportion (%) −94.50 −12.28 206.78 100.00 
Jilin Increment (billion m3−5.34 −0.59 12.88 6.95 
Proportion (%) −76.77 −8.53 185.30 100.00 
Jiangsu Increment (billion m3−2.00 −0.29 −1.46 −3.76 
Proportion (%) 53.29 7.75 38.96 100.00 
Jiangxi Increment (billion m33.32 −0.32 4.17 7.17 
Proportion (%) 46.36 −4.49 58.13 100.00 
Liaoning Increment (billion m3−6.00 −1.67 7.46 −0.21 
Proportion (%) 2,838.86 789.62 −3,528.48 100.00 
Ningxia Increment (billion m30.39 0.02 1.28 1.69 
Proportion (%) 22.89 1.19 75.92 100.00 
Qinghai Increment (billion m30.13 −0.02 −0.11 0.0015 
Proportion (%) 8,925.55 −1,295.15 −7,530.40 100.00 
Shangdong Increment (billion m3−11.65 −3.27 7.22 −7.70 
Proportion (%) 151.28 42.42 −93.70 100.00 
Shanxi Increment (billion m3−0.42 −5.17 5.85 0.27 
Proportion (%) −154.24 −1,914.23 2,168.48 100.00 
Shannxi Increment (billion m3−6.12 1.01 1.74 −3.37 
Proportion (%) 181.48 −29.97 −51.51 100.00 
Shanghai Increment (billion m3−0.05 0.02 −0.75 −0.78 
Proportion (%) 6.43 −2.79 96.36 100.00 
Sichuan Increment (billion m31.58 −0.41 −0.87 0.30 
Proportion (%) 522.56 −135.91 −286.65 100.00 
Tianjian Increment (billion m30.19 −0.14 −0.80 −0.75 
Proportion (%) −25.87 18.61 107.26 100.00 
Xinjiang Increment (billion m3−1.85 0.17 3.54 1.86 
Proportion (%) −99.78 8.95 190.83 100.00 
Yunnan Increment (billion m3−0.39 −0.50 4.69 3.79 
Proportion (%) −10.39 −13.23 123.62 100.00 
Zhejiang Increment (billion m3−0.88 0.09 −7.16 −7.95 
Proportion (%) 11.09 −1.19 90.10 100.00 
ProvinceIncrement/Proportion
Anhui Increment (billion m3−6.15 −0.47 3.81 −2.81 
Proportion (%) 218.45 16.83 −135.29 100.00 
Bejing Increment (billion m30.17 −0.15 −1.92 −1.91 
Proportion (%) −8.69 7.80 100.88 100.00 
Chongqing Increment (billion m3−3.03 −0.24 −0.79 −4.06 
Proportion (%) 74.66 5.93 19.41 100.00 
Fujian Increment (billion m3−1.36 −0.08 −3.46 −4.90 
Proportion (%) 27.79 1.66 70.55 100.00 
Gansu Increment (billion m3−2.32 0.26 5.98 3.92 
Proportion (%) −59.06 6.58 152.48 100.00 
Guangdong Increment (billion m33.67 −0.03 −6.44 −2.79 
Proportion (%) −131.29 1.02 230.27 100.00 
Guangxi Increment (billion m3−0.07 −0.09 −1.57 −1.73 
Proportion (%) 4.12 5.46 90.42 100.00 
Guizhou Increment (billion m33.60 −0.31 −2.13 1.17 
Proportion (%) 308.87 −26.27 −182.60 100.00 
Hainan Increment (billion m3−0.52 −0.01 −0.62 −1.15 
Proportion (%) 45.44 0.44 54.12 100.00 
Hebei Increment (billion m3−11.01 −2.28 6.38 −6.90 
Proportion (%) 159.40 32.97 −92.37 100.00 
Henan Increment (billion m3−8.26 −2.38 25.42 14.78 
Proportion (%) −55.87 −16.08 171.94 100.00 
Heilongjiang Increment (billion m3−5.98 −8.23 31.93 17.72 
Proportion (%) −33.75 −46.45 180.20 100.00 
Hubei Increment (billion m3−4.36 −0.27 −2.32 −6.95 
Proportion (%) 62.79 3.84 33.38 100.00 
Hunan Increment (billion m32.61 −0.43 1.75 3.93 
Proportion (%) 66.49 −10.96 44.47 100.00 
Inner Mongolia Increment (billion m3−5.61 −0.73 12.28 5.94 
Proportion (%) −94.50 −12.28 206.78 100.00 
Jilin Increment (billion m3−5.34 −0.59 12.88 6.95 
Proportion (%) −76.77 −8.53 185.30 100.00 
Jiangsu Increment (billion m3−2.00 −0.29 −1.46 −3.76 
Proportion (%) 53.29 7.75 38.96 100.00 
Jiangxi Increment (billion m33.32 −0.32 4.17 7.17 
Proportion (%) 46.36 −4.49 58.13 100.00 
Liaoning Increment (billion m3−6.00 −1.67 7.46 −0.21 
Proportion (%) 2,838.86 789.62 −3,528.48 100.00 
Ningxia Increment (billion m30.39 0.02 1.28 1.69 
Proportion (%) 22.89 1.19 75.92 100.00 
Qinghai Increment (billion m30.13 −0.02 −0.11 0.0015 
Proportion (%) 8,925.55 −1,295.15 −7,530.40 100.00 
Shangdong Increment (billion m3−11.65 −3.27 7.22 −7.70 
Proportion (%) 151.28 42.42 −93.70 100.00 
Shanxi Increment (billion m3−0.42 −5.17 5.85 0.27 
Proportion (%) −154.24 −1,914.23 2,168.48 100.00 
Shannxi Increment (billion m3−6.12 1.01 1.74 −3.37 
Proportion (%) 181.48 −29.97 −51.51 100.00 
Shanghai Increment (billion m3−0.05 0.02 −0.75 −0.78 
Proportion (%) 6.43 −2.79 96.36 100.00 
Sichuan Increment (billion m31.58 −0.41 −0.87 0.30 
Proportion (%) 522.56 −135.91 −286.65 100.00 
Tianjian Increment (billion m30.19 −0.14 −0.80 −0.75 
Proportion (%) −25.87 18.61 107.26 100.00 
Xinjiang Increment (billion m3−1.85 0.17 3.54 1.86 
Proportion (%) −99.78 8.95 190.83 100.00 
Yunnan Increment (billion m3−0.39 −0.50 4.69 3.79 
Proportion (%) −10.39 −13.23 123.62 100.00 
Zhejiang Increment (billion m3−0.88 0.09 −7.16 −7.95 
Proportion (%) 11.09 −1.19 90.10 100.00 

Note: The last column of the increment in Table 3 is the sum of the previous four columns, and the proportion is the ratio of the corresponding increment.

We can see the following from Table 3. (1) In province r, the change in WF of food production is mainly caused by the change of VWC and total food production output , while changes in the production structure of food would bring about fewer changes in the WF of food production. The provinces where the change of VWC could affect most are Anhui, Beijing, Chongqing, Fujian, Guangdong, Guangxi, Hainan, Hebei, Hubei, Jiangsu, Liaoning, Shandong, Shannxi, Shanghai, Tianjin, Zhejiang, etc., while for the other 19 provinces, they would be mostly influenced by the changes in the production structure of food. (2) During 1997–2001, WF of food production in the whole country gained an increase, but it experienced a decrease in the provinces of Anhui, Beijing, Chongqing, Fujian, Guangdong, Guangxi, Hainan, Hebei, Hubei, Jiangsu, Liaoning, Shandong, Shannxi, Shanghai, Tianjin, and Zhejiang. However, there was also an increase in another 15 provinces. (3) During 1997–2011, the changes in WF of food production in Heilongjiang, Henan, Zhejiang, etc., are relatively larger4, and were 17.72 billion m3, 14.78 billion m3, −7.95 billion m3, respectively. As for their counterparts in Shanxi, Liaoning, Qinghai, etc., they are relatively smaller, and were 0.27 billion m3, −0.21 billion m3, and 0.0015 billion m3, respectively.

Corps level

According to Equation (9), with the adoption of factor decomposition method, the paper analyzes the changes in the production WF of five crops in China, namely rice, corn, wheat, beans, and potato. The results are shown in Table 4 (only the decomposition results of the whole time are listed).

Table 4.

Decomposition analysis of production WF of the five crops during 1997–2011 (unit: 1 billion m3, %).

ProvinceIncrement/Proportion
Rice Increment (billion m3−0.57 0.75 1.29 1.47 
Proportion (%) −38.78 50.72 87.92 100.00 
Corn Increment (billion m3−42.27 2.95 87.69 48.37 
Proportion (%) −87.39 6.10 181.30 100.00 
Wheat Increment (billion m3−10.55 −4.39 −16.36 −31.30 
Proportion (%) 33.71 14.02 52.26 100.00 
Beans Increment (billion m3−4.31 0.88 0.90 −2.53 
Proportion (%) 170.31 −34.66 −35.63 100.00 
Potato Increment (billion m3−10.00 4.22 1.53 −4.25 
Proportion (%) 235.18 −99.18 −36.01 100.00 
ProvinceIncrement/Proportion
Rice Increment (billion m3−0.57 0.75 1.29 1.47 
Proportion (%) −38.78 50.72 87.92 100.00 
Corn Increment (billion m3−42.27 2.95 87.69 48.37 
Proportion (%) −87.39 6.10 181.30 100.00 
Wheat Increment (billion m3−10.55 −4.39 −16.36 −31.30 
Proportion (%) 33.71 14.02 52.26 100.00 
Beans Increment (billion m3−4.31 0.88 0.90 −2.53 
Proportion (%) 170.31 −34.66 −35.63 100.00 
Potato Increment (billion m3−10.00 4.22 1.53 −4.25 
Proportion (%) 235.18 −99.18 −36.01 100.00 

Note: Increment of the last column in Table 3 is the sum of the first four columns and the ratio is the ratio of the corresponding increment.

The following can be seen from Table 4. (1) Compared with the condition in 1997, the WFs of rice and corn production increased in 2011, but WFs of wheat, beans, and potato production in 2011 experienced an decrease. (2) In 1997–2011, for rice, corn, and wheat, the changes in WF are mainly caused by the changes in total food production output . For beans and potato, the counterparts were mainly caused by the changes in VWC .

To summarize these results in Tables 24, it can be noticed that the changes in the WF (, , ) caused by the VWC in most cases are negative, which thus indicates a downward trend in the VWC of the five crops (rice, corn, wheat, beans, and potato) in China. Based on the calculation method of virtual water and water foot print raised by Hoekstra et al. (2011), the increment of the food production output per unit area, as well as the promotion of agricultural water-saving irrigation technology can help reduce the crop VWC and food production WF.

Decoupling China's provincial WF changes from changes in total food production outputs during 1997–2011

In this paper, the year 1997 is taken as the initial period, while the year 2011 is taken as the last period, so as to research the decoupling of WF changes from the total food production output changes. According to Equations (10) and (11), the decoupling indicators can be obtained, as shown in Table 5.

Table 5.

The decoupling indicators between the changes in food production WFs and the changes in food production outputs in China during 1997–2011.

ProvinceDr
Anhui −6.15 −0.47 3.81 −2.81 3.54 −1.50 −0.12 0.93 −0.69 
Bejing 0.17 −0.15 −1.92 −1.91 −1.15 −0.10 0.09 1.14 1.13 
Chongqing −3.03 −0.24 −0.79 −4.06 −0.27 9.56 0.76 2.48 12.8 
Fujian −1.36 −0.08 −3.46 −4.90 −2.81 0.36 0.02 0.92 1.30 
Gansu −2.32 0.26 5.98 3.92 3.61 −0.34 0.04 0.87 0.57 
Guangdong 3.67 −0.03 −6.44 −2.79 −5.32 −0.67 0.01 1.17 0.51 
Guangxi −0.07 −0.09 −1.57 −1.73 −1.15 0.05 0.06 1.02 1.12 
Guizhou 3.60 −0.31 −2.13 1.17 −1.55 −2.09 0.18 1.23 −0.68 
Hainan −0.52 −0.01 −0.62 −1.15 −0.30 1.00 0.01 1.20 2.21 
Hebei −11.01 −2.28 6.38 −6.90 4.81 −1.21 −0.25 0.70 −0.76 
Henan −8.26 −2.38 25.42 14.78 16.72 −0.28 −0.08 0.86 0.50 
Heilongjiang −5.98 −8.23 31.93 17.72 25.29 −0.18 −0.24 0.94 0.52 
Hubei −4.36 −0.27 −2.32 −6.95 −2.36 1.81 0.11 0.96 2.88 
Hunan 2.61 −0.43 1.75 3.93 1.44 1.85 −0.30 1.24 2.78 
Inner Mongolia −5.61 −0.73 12.28 5.94 9.67 −0.36 −0.05 0.79 0.38 
Jilin −5.34 −0.59 12.88 6.95 13.69 −0.36 −0.04 0.87 0.47 
Jiangsu −2.00 −0.29 −1.46 −3.76 −1.97 1.12 0.16 0.82 2.11 
Jiangxi 3.32 −0.32 4.17 7.17 2.87 1.11 −0.11 1.39 2.40 
Liaoning −6.00 −1.67 7.46 −0.21 7.84 −0.54 −0.15 0.67 −0.02 
Ningxia 0.39 0.02 1.28 1.69 1.10 0.32 0.02 1.06 1.40 
Qinghai 0.13 −0.02 −0.11 0.00 −0.08 −1.71 0.25 1.45 −0.02 
Shangdong −11.65 −3.27 7.22 −7.70 6.08 −1.07 −0.30 0.66 −0.71 
Shanxi −0.42 −5.17 5.85 0.27 3.43 −0.05 −0.64 0.73 0.03 
Shannxi −6.12 1.01 1.74 −3.37 1.54 −2.41 0.40 0.68 −1.33 
Shanghai −0.05 0.02 −0.75 −0.78 −0.90 0.06 −0.03 0.97 1.01 
Sichuan 1.58 −0.41 −0.87 0.30 −1.20 −1.53 0.40 0.84 −0.29 
Tianjian 0.19 −0.14 −0.80 −0.75 −0.38 −0.30 0.22 1.26 1.17 
Xinjiang −1.85 0.17 3.54 1.86 4.04 −0.46 0.04 0.87 0.46 
Yunnan −0.39 −0.50 4.69 3.79 4.47 −0.08 −0.10 0.95 0.77 
Zhejiang −0.88 0.09 −7.16 −7.95 −6.71 0.13 −0.01 1.04 1.15 
ProvinceDr
Anhui −6.15 −0.47 3.81 −2.81 3.54 −1.50 −0.12 0.93 −0.69 
Bejing 0.17 −0.15 −1.92 −1.91 −1.15 −0.10 0.09 1.14 1.13 
Chongqing −3.03 −0.24 −0.79 −4.06 −0.27 9.56 0.76 2.48 12.8 
Fujian −1.36 −0.08 −3.46 −4.90 −2.81 0.36 0.02 0.92 1.30 
Gansu −2.32 0.26 5.98 3.92 3.61 −0.34 0.04 0.87 0.57 
Guangdong 3.67 −0.03 −6.44 −2.79 −5.32 −0.67 0.01 1.17 0.51 
Guangxi −0.07 −0.09 −1.57 −1.73 −1.15 0.05 0.06 1.02 1.12 
Guizhou 3.60 −0.31 −2.13 1.17 −1.55 −2.09 0.18 1.23 −0.68 
Hainan −0.52 −0.01 −0.62 −1.15 −0.30 1.00 0.01 1.20 2.21 
Hebei −11.01 −2.28 6.38 −6.90 4.81 −1.21 −0.25 0.70 −0.76 
Henan −8.26 −2.38 25.42 14.78 16.72 −0.28 −0.08 0.86 0.50 
Heilongjiang −5.98 −8.23 31.93 17.72 25.29 −0.18 −0.24 0.94 0.52 
Hubei −4.36 −0.27 −2.32 −6.95 −2.36 1.81 0.11 0.96 2.88 
Hunan 2.61 −0.43 1.75 3.93 1.44 1.85 −0.30 1.24 2.78 
Inner Mongolia −5.61 −0.73 12.28 5.94 9.67 −0.36 −0.05 0.79 0.38 
Jilin −5.34 −0.59 12.88 6.95 13.69 −0.36 −0.04 0.87 0.47 
Jiangsu −2.00 −0.29 −1.46 −3.76 −1.97 1.12 0.16 0.82 2.11 
Jiangxi 3.32 −0.32 4.17 7.17 2.87 1.11 −0.11 1.39 2.40 
Liaoning −6.00 −1.67 7.46 −0.21 7.84 −0.54 −0.15 0.67 −0.02 
Ningxia 0.39 0.02 1.28 1.69 1.10 0.32 0.02 1.06 1.40 
Qinghai 0.13 −0.02 −0.11 0.00 −0.08 −1.71 0.25 1.45 −0.02 
Shangdong −11.65 −3.27 7.22 −7.70 6.08 −1.07 −0.30 0.66 −0.71 
Shanxi −0.42 −5.17 5.85 0.27 3.43 −0.05 −0.64 0.73 0.03 
Shannxi −6.12 1.01 1.74 −3.37 1.54 −2.41 0.40 0.68 −1.33 
Shanghai −0.05 0.02 −0.75 −0.78 −0.90 0.06 −0.03 0.97 1.01 
Sichuan 1.58 −0.41 −0.87 0.30 −1.20 −1.53 0.40 0.84 −0.29 
Tianjian 0.19 −0.14 −0.80 −0.75 −0.38 −0.30 0.22 1.26 1.17 
Xinjiang −1.85 0.17 3.54 1.86 4.04 −0.46 0.04 0.87 0.46 
Yunnan −0.39 −0.50 4.69 3.79 4.47 −0.08 −0.10 0.95 0.77 
Zhejiang −0.88 0.09 −7.16 −7.95 −6.71 0.13 −0.01 1.04 1.15 

Note: The unit of , , , and is 1 billion m3, and the unit of is 1 billion kg; in the case of ignoring the rounding error, the sum of the three decomposition of decoupling indicator is equal to the decoupling indicator (e.g. −0.69 = −1.50–0.12 + 0.93).

Through the combination of the results in Table 5 with the classification rules in Table 1, the following can be seen. (1) Judging by , and , the provinces belonging to weak decoupling include Gansu, Henan, Heilongjiang, Inner Mongolia, Jilin, Shanxi, Xinjiang, and Yunnan, with no province belonging to expansive coupling; moreover, the provinces belonging to expansive negative decoupling are Hunan, Jiangxi, and Ningxia while Anhui, Hebei, Liaoning, Shandong, and Shannxi are provinces belonging to strong decoupling. The provinces belonging to strong negative decoupling are Guizhou, Qinghai, Sichuan; the province belonging to weak negative decoupling is Guangdong; the provinces belonging to recessive coupling are Beijing, Guangxi, Shanghai, Tianjin, and Zhejiang, and the provinces belonging to recessive decoupling are Chongqing, Fujian, Hainan, Hubei, and Jiangsu. It is known that food production calls for capital, labor, land, and other productive factors, and also the food production outputs per unit area are not the same, which thus means that the changes in WFs are not consistent with changes in total food production outputs. (2) Judging by decomposition results, there are for every province, which is due to the positive or negative sign of (changes in WFs caused by changes in total food production output in province r) being consistent with that of (changes in total food production output in province r). Additionally, and can either be greater or less than zero, for the reason that there are differences among (changes in the WFs caused by the changes in VWCs in province r), (changes in WFs caused by changes in the structure of food production in province r), and (changes in food production output in province r) in the positive or negative sign.

Recently, the calculation methodology of WF has been studied in many studies (Hoekstra et al., 2011), and empirical analysis of WF for some crops' production in some specific areas is available (Chapagain et al., 2006; Chapagain & Hoekstra, 2007, 2011; Mekonnen & Hoekstra, 2010; Shrestha et al., 2013; Wang et al., 2015). Compared with above-mentioned literature, this study is dedicated to accounting and change trend analysis of food production WF in China from the aspects of different countries, provinces, and crops. The comparison of production WF in different provinces or crops shows that there is huge difference. For example, those provinces maintaining larger food production WF are Henan, Hebei, Shandong, and Heilongjiang during 1997–2011. In contrast, Qinghai, Beijing, and Shanghai maintain fewer. As for crops during 1997–2001, those maintaining larger WF are rice, wheat, and corn, etc. In contrast, such crops as beans and potato maintain fewer. Input–output method is utilized for calculating WF in some literature, aside from production tree method (Zhao et al., 2009; Feng et al., 2011; Zhang & Anadon, 2014; Li & Chen, 2014; Huang et al., 2015; Deng et al., 2016). However, the WF calculated by input–output method is only related to the macro-industrial sectors, taking no account of rice, corn, wheat, beans, potato, or other specific food crops, and only the calculation result of blue WF is taken into account. Therefore, in this paper, the production tree method is adopted to calculate the WFs in the food production process of Chinese provinces.

Change trend of WF for China or other countries has been analyzed in some literature by the method of SDA (Duarte et al., 2014; Deng et al., 2016; Wang et al., 2016) and LMDI (Xu et al., 2015), but there is still a lack of any study decomposing change trend factors of China's WF from the aspect of provinces and crops at the same time. Taking decoupling analysis of energy and environment field as reference (Tapio, 2005), this study has also further analyzed the relation between changes of food production WF and food output, showing that food output can increase when food production WF decreases in provinces like Anhui, Hebei, Liaoning, Shangdong, and Shannxi.

In this paper, based on related data of China's Rural Statistical Yearbook and China meteorological data sharing service system, we calculate the provincial WFs of food production in China during 1997–2011. Meanwhile, the LMDI method is utilized to produce factor decomposition analysis on changes in food production WFs at the national level, provincial level, and crop level, with relevant results drawn as follows. (1) During 1997–2011, the WFs of China's food production are increasing in general, but a partial time period is likely to decrease. In addition, there is a huge difference in China's provincial WFs of food production. (2) Compared with 1997, WFs of corn and rice production in 2011 gained an increase. However, there appears to be a decline in WFs of wheat, beans, and potato production in 2011. Changes in VWCs and food production output are the main cause for the changes in the WFs of China's food production. (3) The decoupling analysis of changes between food production WF and food output shows that those changes might not be synchronized.

According to the conclusions of this study, the following policy implications can be obtained. (1) Due to the big differences of WFs of different provinces in China, in some places (like Shanghai, Beijing, and Tianjin), it is difficult to meet local demands by using the local water resources or food production. Thus, the agricultural product transportation policy named ‘the green channel’5 (free charge of agricultural products transport vehicles on the highway) and other preferential policies need to be applied to promote domestic grain trade and virtual water trade. (2) For the reason that the changes in VWC and total food production output are the main causes of the changes in WF of food production in China, water saving measures need to be taken, such as sprinkler irrigation and drip irrigation, so as to improve the efficiency of agricultural water use. As well, the improvement of agricultural production technology is also required to promote the increase of food production output, which could eventually satisfy people's demand for water and food. (3) There are other factors, like capital, labor, and land, which can affect food production, and the food production outputs per unit area are not the same. Hence, the changes in WF and the changes in food production outputs are not consistent in different provinces, thus indicating the existence of strong negative decoupling. In some provinces (such as Guizhou, Qinghai, and Sichuan), the VWC increased; however, the food production outputs experienced no increase. Therefore, from the point of view of saving water resources, it is necessary for these provinces to introduce VWC from other provinces.

It needs to be pointed out that this study has some innovation in research content and method on the basis of some related literature. (1) It is dedicated to systematical accounting for food production WF from provincial and crops' aspects during 1997–2011. (2) Analysis of change trend in WF has been made by the method of LMDI. (3) The decoupling analysis of changes between food production WF and food output has been made among provinces in China. However, this paper takes green and blue WF into consideration, excluding gray WF.

This study was funded by National Natural Science Young Foundation of China (grant number 71704070), Ministry of Education for the Humanities and Social Sciences Research Young Fund on the West and Borderland Project (grant number 17XJC790002), Gansu Provincial Higher Education Research Project (grant number 2017B-41), Lanzhou University of Finance and Economics Silk Road Economic Research Institute (grant number JYYZ201603), Lanzhou University of Finance and Economics' scientific research projects (grant number Lzufe201601), and Shanghai University of Finance and Economics through postgraduate innovation fund projects (grant number CXJJ-2014-411). The authors declare that they have no conflict of interest.

2

The gray WF cannot be obtained by Equation (1) directly, so this paper only considered the green WF and blue WF.

4

The larger changes in food production WFs refer to larger absolute values, while the fewer changes in their counterparts refer to fewer absolute values.

5

For detailed content about the policy ‘green channel’ for agricultural products’ transportation, please refer to the website of Central Government of China: http://www.gov.cn/zwgk/2010-01/05/content_1503574.htm.

Allen
R. G.
,
Pereira
L. S.
,
Raes
D.
&
Smith
M.
, (
1998
).
Crop Evapotranspiration – Guidelines for Computing Crop Water Requirements
.
FAO Irrigation and Drainage Paper 56
,
FAO
,
Rome
.
Cao
X.
,
Wu
P.
,
Wang
Y.
&
Zhao
X.
, (
2014
).
WF of grain product in irrigated farmland of China
.
Water Resources Management
28
(
8
),
2213
2227
.
Chapagain
A. K.
&
Hoekstra
A. Y.
, (
2007
).
The WF of coffee and tea consumption in the Netherlands
.
Ecological Economics
64
(
1
),
109
118
.
Chapagain
A. K.
&
Hoekstra
A. Y.
, (
2011
).
The blue, green and grey WF of rice from production and consumption perspectives
.
Ecological Economics
70
(
4
),
749
758
.
Chouchane
H.
,
Hoekstra
A. Y.
,
Krol
M. S.
&
Mekonnen
M. M.
, (
2015
).
The WF of Tunisia from an economic perspective
.
Ecological Indicators
52
,
311
319
.
Dietzenbacher
E.
&
Los
B.
, (
1998
).
Structural decomposition techniques: sense and sensitivity
.
Economic Systems Research
10
(
4
),
307
324
.
Duarte
R.
,
Pinilla
V.
&
Serrano
A.
, (
2014
).
The WF of the Spanish agricultural sector: 1860–2010
.
Ecological Economic
108
,
200
207
.
Feng
K.
,
Chapagain
A.
,
Suh
S.
,
Pfister
S.
&
Hubacek
K.
, (
2011
).
Comparison of bottom-up and top-down approaches to calculating the WFs of nations
.
Economic Systems Research
23
(
4
),
371
385
.
Fulton
J.
,
Cooley
H.
&
Gleick
P. H.
, (
2014
).
WF outcomes and policy relevance change with scale considered: evidence from California
.
Water Resources Management
28
(
11
),
3637
3649
.
Gerbens-Leenes
P. W.
,
Mekonnen
M. M.
&
Hoekstra
A. Y.
, (
2013
).
The WF of poultry, pork and beef: a comparative study in different countries and production systems
.
Water Resources & Industry
1
,
25
36
.
Hoekstra
A. Y.
&
Hung
P. Q.
, (
2002
).
Virtual Water Trade: A Quantification of Virtual Water Flows Between Nations in Relation to International Crop Trade
.
Value of Water Research Report Series 11
.
IHE
,
Delft
,
The Netherlands
.
Hoekstra
A. Y.
,
Chapagain
A. K.
,
Aldaga
M.
&
Mekonnen
M. M.
, (
2011
).
The WF Assessment Manual: Setting the Global Standard
.
Earthscan
,
Abingdon
,
UK
.
Huang
K.
,
Wang
Z.
,
Yu
Y.
&
Yang
S.
, (
2015
).
Assessing the environmental impact of the WF in Beijing, China
.
Water Policy
17
(
5
),
557
560
.
Mekonnen
M. M.
&
Hoekstra
A. Y.
, (
2010
).
A global and high-resolution assessment of the green, blue and grey WF of wheat
.
Hydrology and Earth System Sciences
14
,
1259
1276
.
Organization for Economic Co-operation and Development (OECD)
(
2001
).
Sustainable Development: Indicators to Measure Decoupling of Environmental Pressure From Economic Growth
.
OECD
,
Paris
.
Shrestha
S.
,
Pandey
V. P.
,
Chanamai
C.
&
Ghosh
D. K.
, (
2013
).
Green, blue and grey WFs of primary crops production in Nepal
.
Water Resources Management
27
(
15
),
5223
5243
.
Su
M. H.
,
Huang
C. H.
,
Li
W. Y.
,
Tso
C. T.
&
Lur
H. S.
, (
2015
).
WF analysis of bioethanol energy crops in Taiwan
.
Journal of Cleaner Production
88
,
132
138
.
Tang
Z.
,
Shang
J.
,
Shi
C.
,
Liu
Z.
&
Bi
K.
, (
2014
).
Decoupling indicators of CO2 emissions from the tourism industry in China: 1990–2012
.
Ecological Indicators
46
,
390
397
.
The Division for Rural Social Economy Investigation of the National Bureau of Statistics
(
1998–2012
).
Chinese Rural Statistical Yearbook
.
Chinese Statistics Press
,
Beijing
.
Vanham
D.
&
Bidoglio
G.
, (
2013
).
A review on the indicator WF for the EU28
.
Ecological Indicators
26
,
61
75
.
Voigt
S.
,
Cian
E.
,
Schymura
M.
&
Verdolini
E.
, (
2014
).
Energy intensity developments in 40 major economies: structural change or technology improvement?
Energy Economics
41
,
47
62
.
Wang
Y. B.
,
Wu
P. T.
,
Engel
B. A.
&
Sun
S. K.
, (
2015
).
Comparison of volumetric and stress-weighted WF of grain products in China
.
Ecological Indicators
48
,
324
333
.
Wang
X.
,
Huang
K.
,
Yu
Y.
,
Hu
T.
&
Xu
Y.
, (
2016
).
An input–output structural decomposition analysis of changes in sectoral WF in China
.
Ecological Indicators
69
,
26
34
.
Zhang
Y.
,
Huang
K.
,
Yu
Y.
&
Yang
B.
, (
2017
).
Mapping of water footprint research: a bibliometric analysis during 2006–2015
.
Journal of Cleaner Production
149
,
70
79
.
Zhao
X.
,
Chen
B.
&
Yang
Z. F.
, (
2009
).
National WF in an input–output framework – a case study of China 2002
.
Ecological Modelling
220
(
2
),
245
253
.
Zheng
Z.
,
Liu
J.
,
Koeneman
P. H.
,
Zarate
E.
&
Hoekstra
A. Y.
, (
2012
).
Assessing WF at river basin level: a case study for the Heihe River Basin in northwest China
.
Hydrology and Earth System Sciences
16
(
8
),
2771
2781
.
Zoumides
C.
,
Bruggeman
A.
,
Hadjikakou
M.
&
Zachariadis
T.
, (
2014
).
Policy-relevant indicators for semi-arid nations: the WF of crop production and supply utilization of Cyprus
.
Ecological Indicators
43
,
205
214
.