Journal of Cleaner Production 279 (2021) 123670

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Non-negligible regional differences in the driving forces of crop-

related water footprint and virtual water flows: A case study for the
Beijing-Tianjin-Hebei region
Meng Li a, 1, Zenghui Xu a, 1, Shan Jiang b, La Zhuo c, d, e, *, Xuerui Gao c, d, e, Yong Zhao b, **,
Yilin Liu a, Wei Wang a, Jiming Jin a, f, Pute Wu c, d, e
a
College of Water Resources and Architectural Engineering, Northwest A&F University, Yangling, 712100, China
b
China Institute of Water Resources and Hydropower Research, Beijing, 100000, China
c
Institute of Soil and Water Conservation, Northwest A&F University, Yangling, 712100, China
d
Institute of Soil and Water Conservation, Chinese Academy of Sciences and Ministry of Water Resources, Yangling, 712100, China
e
National Engineering Research Centre for Water Saving Irrigation at Yangling, Yangling, 712100, China
f
Department of Watershed Sciences, Utah State University, Logan, UT, 84322, USA

a r t i c l e i n f o a b s t r a c t

Article history: Water scarcity is a significant risk for meeting increasing food demand around the world. The importance
Received 10 March 2020 of identifying the driving forces behind water consumption in agriculture and relative virtual water (VW)
Received in revised form flows has been widely reported in order to provide practical advice for regional sustainable agricultural
30 July 2020
water resource management. However, the differences of regional driving forces behind both water
Accepted 7 August 2020
Available online 15 August 2020
consumption and VW flows were largely ignored. To fill the crucial gap, taking nine major crops grown in
the Beijing-Tianjin-Hebei (BTH) region in China over 2000e2013 as the study case, we investigated the
Handling editor. Bin Chen regional differences in socio-economic driving forces on both the estimated water footprint (WF) in crop
production and relative inter-city VW flows for each crop per year. Results show that although there is
Keywords: little change in total WFs in crop production (~43.3 billion m3/y on annual average), the WF per unit mass
Driving forces of crop decreased and the crop structure in the total WFs changed greatly. The BTH region was a VW
Regional differences importer with net VW import of 11.7 billion m3/y by 2013 but the roles of each cities were different.
Crop water footprint Inter-city VW was predominantly exported from the southern regions. The per capita GDP was the main
Virtual water flows
positive driver of both total WFs and relative VW flows. Whereas economic productivity and con-
Beijing-Tianjin-Hebei region
sumption capacity were inhibiting factors for the WFs and VW flows, respectively. The levels of total crop
WFs in agricultural cities were more sensitive to the effects of the main driving factors. The intensity of
driving factors behind the inter-regional crop-related VW flows was shown to be directly related to
regional role as an importer or exporter. Aiming for a balance between water sustainability, food security
and economic developments, the current analysis suggests characteristic-based agriculture in terms of
regional differences in water consumption and local roles in relative inter-regional VW flows. Especially
for agricultural cities which are VW exporters, reducing water consumption per unit agricultural product
and increasing economic production value per drop of water used in all sectors are equally important.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction

Limited water resource is a key constraint on food security,

stable socio-economic development and ecological health
* Corresponding author. Institute of Soil and Water Conservation, Northwest A&F regarding the UN’s Sustainable Development Goals (Hoekstra and
University, Yangling, 712100, China. Wiedmann, 2014; Ja €germeyr et al., 2017; UN, 2015). Since the
** Corresponding author. onset of the Anthropocene, population growth and food demand
E-mail addresses: zhuola@nwafu.edu.cn (L. Zhuo), zhaoyong@iwhr.com
increase, which, along with increasingly diverse and water-
(Y. Zhao).
1
The authors contribute equally. intensive consumption patterns, has resulted in that the 41% of

https://doi.org/10.1016/j.jclepro.2020.123670
0959-6526/© 2020 Elsevier Ltd. All rights reserved.
2 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670

Table 1
Inventory of representative studies on driving forces analysis of water footprints (WFs) and virtual water (VW) flows.

Research Methodology Dependent Study scale Spatial Consideration of spatial-differences in driving

variable unit forces

Tamea et al. (2014) Commonality analysis VW flows Globe nation No

Wang et al. (2016) STIRPAT model VW flows Globe nation No
Qian et al. (2019) LMDI method VW flows China and the Belt and Road nation No
Countries
Su et al. (2020) LMDI method WFs Lithuanian nation No
Tian et al. (2016) IPAT equation WFs China nation No
Zhao et al. (2014) Extended STIRPAT model WFs China nation No
Zhao and Chen LMDI method WFs China nation No
(2014)
Zhang et al. (2012) SDA model WFs Beijing nation No
Xu et al. (2015) LMDI method WFs Beijing nation No
Sun (2019) SDA model WFs BTH region nation No
Cai et al. (2019) SDA model VW flows China province Yes
Zhao et al. (2019) Comparative advantage VW flows China province Yes
theory
Xiong et al. (2020) SDA model and linear VW flows China province Yes
regression
Current study LMDI method WFs & VW flows BTH region city Yes

Fig. 1. Location of the Beijing-Tianjin-Hebei region.

global water used for irrigation is in expense of environmental flow
requirements (Ja€germeyr et al., 2017). Virtual water (VW) flows via
food trades play a crucial role in satisfying food demands and even
alleviating water stress, especially for water scarce countries in the
Middle East (D’Odorico and Rulli, 2013). Water footprints (WFs)
(Hoekstra, 2003) measure the quantity of water resources required
for all the products and services consumed within a certain period.
WFs can be divided into three colors. Green WF refers to the con-
sumption of green water, i.e. rainwater insofar as it does not
become run-off. Blue WFs indicate consumption of blue water, i.e.
surface and ground water. Grey WFs measures the level of water
pollution as the volume of water required to assimilate the load of
pollutants given natural background concentrations and existing
ambient water quality standards (Hoekstra et al., 2011). Currently,
nearly a quarter of the global WF in agriculture is related to food
trade (Hoekstra and Mekonnen, 2012). However, existing interna-
tional or intra-national food trade networks are driven by either
wealth (Wang et al., 2016), population, geographical distance
(Tamea et al., 2014), land productivity (Zhao et al., 2019) or policies
(Zhuo et al., 2016a), yet water resource endowments contribute to
remote water scarcity in many parts of the world (Liu et al., 2018).
Hoekstra and Mekonnen (2016) found that 37% of the United
Kingdom’s blue WFs related to crop consumption directly related to
blue water scarcity in six other countries, including Indonesia,
Ghana, India, Brazil, Spain and Argentina. Within China, Zhao et al.
(2018) reported that the food exports of Xinjiang province
contributed to 66% of the total water loss through inter-provincial
food transfers to areas with more efficient water use. Under-
standing the socio-economic drivers of development related to
both the WFs of local food production and relative inter-regional
VW flows is crucial for constructing smart water management
and food trade regulations, which apparently can better balance
increasing food demands and sustainable water consumption.
Common methods for analyzing driving forces include the IPAT
equation which is proposed to simply state that environmental
impact (I) is the product of population (P), affluence (A), and
technology (T) (Ehrlich and Holdren, 1971) (now the stochastic
impact by regression on population, affluence and tech-
nology(STIRPAT) (Dietz and Rosa, 1994)), the structural decompo-
sition analysis method (SDA) (Leontief et al., 1972) and
Logarithmic-Mean Divisia Index (LMDI) method (Ang et al., 1998).
The SDA and SITRPAT method have their own limitations. The SDA
method is based on the input-output model, which needs the
inputeoutput coefficients and final demand per sector (Hoekstra
and van den Bergh, 2003). The data is difficult to acquire and the
decomposition is not unique. The decomposition of STIRPAT
 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670
Fig. 2. The spatial distribution of multi-year average (a) evapotranspiration and (b) precipitation in the Beijing-Tianjin-Hebei region.
method exists residual term. Comparing with the above decom-
position analyses, the LMDI methodology has the great advantage
of perfect decomposition whereby no unexplained residual term
and being consistent in aggregation and satisfying additivity (Ang,
2004). Many studies have been published on the driving forces of
regional WFs and VW flows, some at the global level and others
focusing on China (Table 1). However, the latest studies focused on
only WFs or VW flows. There remains no comprehensive analysis of
the drivers of both WFs and VW flows, which could evaluate the
water consumption from the perspective of production and con-
sumption to provide practical policy proposals for each city. It has
been widely shown and proved that there are significant intra-
national spatial heterogeneities in the magnitude, structure, and
developing trends in both WFs in crop production and corre-
sponding VW flows (Chouchane et al., 2018; Dalin et al., 2014; Dang
et al., 2015; Katyaini and Barua, 2017; Zhuo et al., 2014). For
instance, Dalin et al. (2014) showed large variations in agricultural
water-use efficiency among Chinese provinces. Such as the WF of
crop in Hainan province was about 3.3 times as large as in Tibet
province. The main VW exporters and importers are also changed
that China’s domestic food trade can save green water but lose blue
water. Katyaini and Barua (2017) found that the VW flows were
from highly water-scarce states in North to highly water-scarce
states in West and South of India. Chouchane et al. (2018) indi-
cated that the WF in Tunisia is largest in the north and the largest
share of blue WF in the total WF is found in the centre and south of
the country. Identification of the hot spot and its socio-economic
drivers in space and time of low efficient WFs and water-scarcity-
driven node in VW network are of greater practical significance
for local sustainable water and environmental management rather
than showing a brief big driver for a large scale boundary. Appar-
ently, analyzing inter-regional differences in the driving force for
WF and VW flows are crucial for managing more precisely the
resulted water issues from the WF and VW flows for certain places.
Crop-related VW flow assessment is defined by crop trade
network and WF of producing per unit mass of the traded crops.
Without the data availability on intra-national crop trade matrix,
particularly for developing countries like China, the multiregional
3
input-output model (Zhao et al., 2015), proximity principle-based
method (Ma et al., 2006) and linear optimization model (Dalin
et al., 2014) are existing approaches of simulating intra-national
trade flows. The multiregional input-output model is able to pro-
vide reliable and comprehensive system boundary for measuring
whole industry supply chain effects, but is failed to show details in
agricultural products and constitute information over years due to
lack of required information (Feng et al., 2011). With capability of
distinguishing individual agricultural products in successive annual
balance of supply and demand at multiple spatial scales, the
proximity principle-based method and linear optimization model
are widely used for agricultural related assessments. In consider-
ation of effects from cost in transportation, the linear optimization
model has kept the wider acceptance (e.g. Wang et al., 2019; Zhuo
et al., 2019; Gao et al., 2020).
The Beijing-Tianjin-Hebei (BTH) region, including thirteen cities
(Fig. 1), has a developed economy but scarce water resources (Zhao
et al., 2017a). The population and GDP in the BTH region account for
8% and 11% of China’s total, respectively. Influenced by the climate
change and human activities, the available water resource has
shown a significant decline, from 28 to 29 billion m3 in the late 19th
century to 14e15 billion m3 in the early 21st century (NBSC, 2018).
The per capita water resources in the region are less than 300 m3/
cap per year (CNKI, 2018). As shown in Fig. 2, the distribution of
water resources is extremely uneven among cities in the BTH,
which accompanied by highly unbalanced economic development
and urbanization levels. In the year 2015, the urbanization rate of
Beijing was 87%, while that of Hebei province was only 33% (CNKI,
2018). As a major agricultural province, crop production in Hebei
province accounts for 89.4% of the total in the BTH region. Thus, the
agricultural water consumption in Hebei is 84.5% of the region
(CNKI, 2018).
The current study, taking the BTH region over 2000e2013 as a
case study, aims to identify the regional differences in driving
forces behind both crop-related WFs and the relative inter-city VW
flows. Green and blue WF measures the actual green (rainwater)
and blue water (surface and ground water) consumption, also
called “consumptive WF”, which are contributors to the water
4 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670

scarcity issues (Hoekstra, 2013). Whereas grey WF, also called
“degradative WF”, which is indexed by a water quantity equivalent
value for indicating the water pollution level (i.e. water required to
diminish certain water pollutants), not an actual volume of water
appropartion (Hoekstra, 2013). The current study focus on water
quantity consumption and relative VW flows, aiming to investigate
the socio-economic drivers behind. Therefore, the WF in the cur-
rent study means consumptive WF. First, we estimated the green
and blue WFs in crop production for nine types of crop (wheat,
maize, rice, millet, potato, soybean, cotton, groundnuts, and vege-
tables) per city per year over the study period. The considered crops
contributed to 94% and 92% of the crop production and crop har-
vested area, respectively, of the BTH in 2013. Second, we estimated
(mm/d) is the precipitation on day t; IRR[t] (mm/d) refers to the net
irrigation depth on day t; CR[t] (mm/d) is the capillary rise of the soil
horizon due to groundwater; ET[t] (mm/d) is the actual crop
evapotranspiration, separated into soil evaporation (E) and crop
transpiration (Tr); RO[t] (mm/d) is the surface runoff; and DP[t] (mm/
d) is deep percolation.
At the beginning of crop planting, the soil water content is green
water, whereas precipitation and irrigation in the growing season
are the source of green and blue WFs, respectively. Therefore, by
tracking the proportion of daily rainfall and the contribution of
irrigation water to each element of soil moisture balance, we can
separate the daily green and blue parts of ET (mm/d) (Chukalla
et al., 2015; Zhuo et al., 2016b):

8 PR½t Sg½t1
>
>
> Sg½t ¼ Sg½t1 þ ðPR½t þ IRR½t  RO½t Þ  ðPR þ IRR Þ  ðDP½t þ ET½t Þ  S
>
< ½t ½t ½t1
(2)
>
>
>
> IRR½t Sb½t1
: Sb½t ¼ Sb½t1 þ ðPR½t þ IRR½t  RO½t Þ   ðDP½t þ ET½t Þ 
ðPR½t þ IRR½t Þ S½t1

the annual inter-city VW flows via the trade of each crop in each
year. Finally, we quantified the inter-regional driving forces of WFs where Sg[t] (mm/d) and Sb[t] (mm/d) are the green and blue soil
and VW flows using the LMDI method, considering both social and water content on day t, respectively. The remaining parameters are
economic drivers. Improving upon the existing literature, this is the the same as above.
first time in the current study that the spatial differences in For irrigated crops, we consider three different irrigation tech-
regional driving forces in both crop-related WFs and inter-regional niques: furrow, sprinkler, and micro-irrigation, by setting different
VW flows have been considered comprehensively on a sub-national surface soil wetting rates (80%, 100%, and 40%, respectively)
scale. (Chukalla et al., 2015; Raes et al., 2017). By creating different
standardized input files, we obtained the WFs under different
2. Methods and data irrigation methods at each station. Then, we calculate WFs more
accurately using the weighted average:
Firstly, we estimated the green and blue WFs in crop production !
X3 Xgp
of each city in BTH regions by the AquaCrop model based on station 10  ETb;i½t  wi
i¼1 t¼1
simulations. Secondly, the linear optimization model, which is a (3)
mathematical model solved by constraints and objective functions, Fb ¼ X3
ðYi  wi Þ
was applied to estimate the crop-related VW flows within BTH i¼1
region. Finally, the driving forces of WFs and VW flows were !
investigated through the LMDI methodology. X3 Xgp
i¼1
10  t¼1
ETg;i½t  wi
(4)
2.1. Estimating green and blue water footprints in crop production WFg ¼ X3
i¼1
ðYi  wi Þ
The green and blue WFs of growing per unit mass of a crop (m3/
t) are equal to the accumulation of daily green and blue evapo- where i indicates the type of irrigation technique; ETg,i[t] and ETb,i[t]
transpiration (m3/ha), respectively, over the crop growth period (mm/d) are the green and blue evapotranspiration under irrigation
divided by the crop yield (t/ha) (Mekonnen and Hoekstra, 2011). To mode i on day t; Yi (t/ha) is the yield level under irrigation mode i;
estimate the daily green and blue evapotranspiration and crop and wi is the percentage of irrigation method i in the total irrigated
yield, we used the FAO’s water-driven crop water productivity area. The wi is estimated as the same as the percentage of irrigated
model AquaCrop version 6.0 (Hsiao et al., 2009; Raes et al., 2009; area under irrigation method i in total irrigated area in the province
Steduto et al., 2009). AquaCrop model simulates crop transpiration where the station located in.
(Tr) and soil evaporation (E) separately. The daily Tr is used to
derive the daily biomass gain via the normalized biomass water 2.2. Estimating the crop-related inter-city virtual water trade
productivity of the crop (Steduto et al., 2009). Compared to other network
crop growth models, AquaCrop has a significantly smaller number
of parameters and better balances between simplicity, accuracy and The crop-related inter-city VW flow (m3/y) equals to the WF per
robustness (Steduto et al., 2007). AquaCrop model is based on the unit of crop (m3/t) in the exporting city multiplied by the crop trade
dynamic water balance in the rootzone: volume (t/y).

S½t ¼ S½t1 þ PR½t þ IRR½t þ CR½t  ET½t  RO½t  DP½t (1) VWFi;j;x ¼ WFi;x ,Ti;j;x (5)

Where S[t] (mm/d) is the soil water content at the end of day t; PR[t] where i, j and x represent the exporter, importer, and crop trade
 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670 5

volumes, respectively; VWFi,j,x is the VW amount caused by trade

volumes from i to j of crop x; WFi,x is the WF per unit of crop x in city
i; and Ti,j,x refers to the trade volumes and contains both interna-
tional and domestic trade volumes.
In this study, we estimated the crop trade volume by applying a
linear optimization model (Dalin et al., 2014), in which the opti-
mization goal is minimal trade costs, and use the relationship be-
tween supply and demand as the constraint condition. First, we
estimated the provincial crop trade balances. The per capita con-
sumption level per crop per year was assumed the same to the
national average level according to the FAO food balance sheets
(FAO, 2018) across the BTH region. The food balance sheet consists
of information on country level utilizations for food, feed, seed as Fig. 3. The water footprints per unit of cereals (wheat, rice, maize, millet, potato) and
well as waste per crop per year. The same assumptions have been cash crops (soybean, groundnut, cotton, vegetables).
applied in Ma et al. (2006) and Zhuo et al. (2016a), given that there
is no data available on intra-differences in terms of consumption
level for specific crops. Here, we used the inter-province trade unit cost of transport, which is modeled the same as above; Sm,x (in
volumes as an example to introduce the specific optimization t/y) is the available volumes of crop x in city m for export; Dn,x (in t/
conditions. We assumed that crop production of each province is y) is the total demand of city n for crop x.
given priority to meet their own needs; i.e., the province starts to The VW imports from outside the Beijing-Tianjin-Hebei region
export crops when the production of each province is greater than (including domestic and international region) were calculated by
the demand. We used four harbor provinces (Tianjin, Liaoning, multiplying the WF per unit of crop by the volume of crop trade.
Shandong, and Guangdong) as a hub for international trade when The domestic volume of crop trade is modeled by the linear opti-
estimating provincial crop trade volumes, with the capital of each mization model. The WF per unit of crop of China is calculated by
province as the link for transport with other provinces. The linear the FAST Track method (Tuninetti et al., 2017a) by using the data for
optimization formulas are as follows: the year 2008 in Zhuo et al. (2016a).

0 1 uWFð2008Þ  YðTÞ
8 X
i¼31;j¼31 uWFðTÞ ¼ (8)
>
> @ Ti;j;x  ci;j;x A Yð2008Þ
> min Cx ¼
>
>
>
> i¼1;j¼1
>
> where uWF(T) and uWF(2008) (m3/t) represent the WF per unit of
>
>
>
< X
j¼31
> crop of period T and 2008; Y(T) and Y(2008) (t/ha) mean the yield of
Ti;j;x  Si;x i ¼ 1…31 period T and 2008. The WF per unit of crop of the world is the
(6)
> j¼1
>
>
> weighted average of the WF and trade volume of countries which
>
> i¼31
X
>
> have trade relations with China.
>
> Ti;j;x ¼ Dj;x j ¼ 1…31
>
>
>
: i¼1
2.3. LMDI methodology
Ti;j;x  0
In this study, the LMDI method was used to analyze the driving
where Cx (in CNY) means the total transport cost of trade of crop x;
forces behind the changes in the total WFs of crop production and
Ti,j,x (in t/y) is the crop trade volumes from i to j of crop x; ci,j,x (in
in the inter-city VW flows within the BTH region. Logarithmic-
CNY/t) refers to the unit cost of transport, which is estimated by the
Mean Divisia Index (LMDI) method is developed by Ang et al.
“OD cost matrix analysis tool” in ArcGIS network analysis (see
(1998) which is a Divisia index using a logarithmic mean weight
Figure S1 and Table S1); Si,x (in t/y) is the available volumes of crop x
function. It has been used in many research fields for identify
in province i for export; Dj,x (in t/y) is the total demand of province j
driving factors of changes in space and time of environment ap-
for crop x.
propriations by human activities such as energy (e.g. Liu et al.,
Then we downscaled the trade volumes to inter-city crop trade
2019a; Tan and Lin, 2018; Wang and Feng, 2018), carbon emis-
volumes. The inter-province trade data could be the essential data
sions (e.g. Ma et al., 2018; Mousavi et al., 2017; Wang and Jiang,
for the inter-city trade. Here, we take the example of inter-city
2019; Zhao et al., 2016), WFs (e.g. Su et al., 2020; Xu et al., 2015;
trade volumes. The detailed linear optimization formulas are as
Zhao and Chen, 2014; Zhao et al., 2017b) and VW flows (Qian et al.,
follows:
2019).
0 1 Regarding the driving forces of crop WFs, based on the LMDI
8 X
m¼13;n¼13
>
> @
min Cx ¼ Tm;n;x  cm;n;x A method, we constructed 5 indexes including WF per unit of crop,
>
>
>
> structure of production, unit GDP productivity, per capita GDP, and
>
>
m¼1;n¼1
>
> population.
>
< X
> n¼13
Tm;n;x  Sm;x m ¼ 1…13 X X WFi Pdi Pd GDP X
(7)
>
> n¼1 WF ¼ WFi ¼ P¼ UWFi Psi GEP
>
> Pdi Pd GDP P
>
> X
m¼13 i i i
>
>
>
> Tm;n;x ¼ Dn;x n ¼ 1…13 (9)
>
>
: m¼1
Tm;n;x  0 where WF (m3/y) is the total WFs of crops in each city; i refers to the
type of crop; WFi refers to the total WFs of crop i in each city; Pdi (t/
where m, n represent 13 cities in the BTH region; Cx (in CNY) means y) is the yield of each crop in each city; Pd (t/y) is the total yield of
the total transport cost of trade of crop x; Tm,n,x (in t/y) is the crop each city; GDP (billion CNY) is the gross domestic product of each
trade volumes from m to n of crop x; cm,n,x (in CNY/t) refers to the city; P is the total population of each city; UWFi ¼ WFi/Pdi
6 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670

represents the unit water productivity; Psi ¼ Pdi/Pd denotes the

22%

20%
28%
32%
29%
18%
51%
65%
33%

23%
structure of production; G ¼ Pd/GDP refers to the unit GDP pro-

2%

9%

2%
13%
RC
ductivity; and E ¼ GDP/P refers to the degree of economic devel-
Total
opment. We chose the additive decomposition method because the
Vegetables

107
WF

81
48
47
57
77
80
69
62
72
68
64
90
65
results of additive decomposition and multiplicative decomposi-
tion using the LMDI method are the same. The decomposition
Blue

WF

36
26
15
17
14
23
29
31
18
23
20
14
34
21
formula is as follows:

127%
32%
10%
10%
35%

18%
42%
36%

27%
57%
18%
20%
DWF ¼ WF T  WF 0 ¼ DWFUWFi þ DWFPsi þ DWFG
RC

0
0
2046
1756
1878
1680
1579
1625
1670
2020

1968
1830
1959
1797
Total

þDWFE þ DWFP
WF

(10)

0
0
Cotton

Blue

538
476
375
401
359
249
449
548

453
310
504
420
in which △WF refers to the change of total WFs between period T
WF

0
0 and 0; WFT and WF0are the total WFs of periods T and 0, respec-
102% tively. The calculation formula for each part of the change is as
46%
76%

52%
13%
16%
59%
30%

33%
67%
9%

28
RC

follows:
0
0

!
Groundnuts

1699
1575
1267
1012
1436
1307
1379
1245

1495
1801
1049
1248
Total

WF

WFit  WFi0 UWFit

DWFUWFi
0
0

¼      ln (11)
ln WFit  ln WFi0 UWFi0
Blue

649
587
394
428
444
300
526
511

510
430
479
448
WF

0
0
The blue and green water footprints (WFs) of crops (m3/t) in the Beijing-Tianjin-Hebei region in 2013 and the relative changes (RC) from 2000 to 2013.

!
29%

34%
49%
34%
33%
55%
75%

54%
28%

WFit  WFi0 Pt
RC

DWFPsi    ln si
0
0

0
0
0

¼   (12)
t
ln WFi  ln WFi 0 Psi0
Total

131
121

112
106
140
110

124
WF

87
87
0
0

0
0
0
Potato

!
Blue

WF

WFit  WFi0 Gti

74
46
50
34
71
62
67
58

61
0
0

0
0
0

DWFG ¼      ln (13)
ln WFit  ln WFi0 G0i
45%
41%

24%
25%
58%
45%
20%
30%
30%
22%
1%
8%
5%

4%
RC

!
2966
4517
3170
1683
2053
1869
1847
1982
1850
1838
2624
2237
2053
2204
Total

WFit  WFi0 Et
WF

DWFE ¼      ln i
Soybean

(14)
ln WFit  ln WFi0 Ei0
1096
Blue

629

612
358
415
202
538
474
304
355
525
288
506
441
WF

!
11%
99%

19%
42%
25%
28%
27%
16%
36%
29%
18%
19%
17%

WFit  WFi0 Pt
94%

DWFP ¼ 
RC

    ln i (15)
ln WFit  ln WFi0 Pi0
2287

2111
1272
1452
1200
1029
1370
1595
1454
1390
2461
1241
1330
Total

WF

When evaluating the driving forces of VW flows, this study

Millet

Blue

465

320
240
232
167
209
300
264
277
243
319
251
230

considered the driving forces of VW import using consumption and

WF

the following five indexes: intension of VW import, crop con-

14%
49%
45%
36%
23%

58%

40%

sumption structure, unit GDP consumption ability, per capita GDP,

4%
RC

4%

4%
0

0
0
0

and population. It is because that the VW export of the same city is

driven by the same factors in the importing regions. The driving
Total

853
732
941
522
695
818

864
768
530

581
WF

forces of WF can also represent that of VW export as the VW export

0

0
0
0

is more related to the regional crop production and water

Blue
Rice

276
276
358
183
216
162

311
260
188

203
WF

consumption.
0

0
0
0
41%
42%

40%
49%
24%
33%
15%
35%
42%
30%
41%
21%
26%

X X VWIi Ci C GDP X
1%
RC

VWI ¼ VWIi ¼ P¼ Vi Csi Gc EP

Ci C GDP P
i i i
1053
Total

826
999
661
759
896
677
802
807

939
917
803
882
816
WF

(16)
Mazie

Blue

109
146

101
112

152
130
115
116
116

129
106
WF

73

49

62

where VWI is the VW import of each city, including VW import and

export; VWIi is the amount of VW import of the crop i in each city;
11%

17%
20%
28%

34%
15%

16%
5%

5%

5%

8%

Ci is the consumption of the ith crop in each city; C represents the

RC

2%

0
0

total consumption; Vi ¼ VWIi/Ci refers to the intension of VW flows,

1026

1000
Total

reflecting the dependence on external water resources; Csi ¼ Ci/C

892
582
787
668
686
733
771

809
834
766
WF

0
0

represents the crop consumption structure; and Gc ¼ C/GDP refers

Wheat

Blue

to the unit GDP consumption ability, indicating the crop con-

411
144

148
122

150
203

211
105
191
150
WF

80

85

0
0

sumption level in a region’s economic activities. The remaining

Qinhuangdao

parameters are the same as above.

Shijiazhuang

Zhangjiakou

Cangzhou
Tangshan

Hengshui
Langfang
Chengde
Baoding
Handan
Xingtai
Tianjin
Beijing

DVWI ¼ VWIT  VWI0 ¼ DVWIVi þ DVWICsi þ DVWIGc

Table 2

BTH
City
 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670 7

Fig. 4. The total water footprints in crop production by cities in the Beijing-Tianjin-Hebei region for the year of 2000 (left) and 2013 (right).

Fig. 5. The contribution of each crop in the virtual water flows of the Beijing-Tianjin-
Hebei region.
Fig. 7. The virtual water flows among cities for the year of 2000 (left) and 2013 (right)
in the Beijing-Tianjin-Hebei region (Unit: 106 m3/y). The regions with one color inside
the circle are net importers, and the others are net exporters. (For interpretation of the
references to color in this figure legend, the reader is referred to the Web version of
this article.)

Fig. 6. The net virtual water import of each city for the year of 2000 (left) and 2013
(right) in the Beijing-Tianjin-Hebei region.
Fig. 8. The blue virtual water flows among cities for the year of 2000 (left) and 2013
(right) in the Beijing-Tianjin-Hebei region (Unit:106 m3/y). (For interpretation of the
references to color in this figure legend, the reader is referred to the Web version of
þDVWIE þ DVWIP (17) this article.)

in which △VWI refers to the change of VW import between period

T and 0; VWIT and VWI0 are the VW import of period T and 0, !
respectively. Similarly, the variation of indicator factors is calcu- VWIit  VWIi0 Gtc
lated using the following formula:
DVWIGc ¼      ln (20)
ln VWIit  ln VWIi0 G0c

!
VWIti  VWI0i Vti !
DVWIVi ¼     ln VWIit  VWIi0 Et
t
(18) DVWIE ¼     ln i
ln VWIi  ln VWIi 0 V0i  (21)
ln VWIit  ln VWIi0 Ei0

! !
VWIit  VWIi0 Ct VWIit  VWIi0 Pt
VWICsi ¼      ln si (19) DVWIP ¼      ln i (22)
t
ln VWIi  ln VWIi 0 Csi0 t
ln VWIi  ln VWIi 0 Pi0
8 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670
Fig. 9. The boxplot of each driving forces of total water footprints of each city in the
Beijing-Tianjin-Hebei region (Agricultural cities are underlined).
2.4. Data
The meteorological data on daily precipitation, temperature,
initial moisture content, humidity, sunshine, and wind speed for
AquaCrop modelling were derived from the China Meteorological
Data Sharing Service System (2018). Data on soil texture corre-
sponded to each station in Beijing-Tianjin-Hebei region according
to latitude and longitude was taken from the ISRIC Soil and Terrain
database for China (Dijkshoorn et al., 2008). Statistics on GDP,
population, crop production, and acreage for each city were ob-
tained from CNKI (2018). The international trade and crop con-
sumption data was derived from FAOSTAT (FAO, 2018). The map of
traffic networks was obtained from the Resources and
Environmental Sciences Data Center (2018). The WF per unit of
crop of other countries was from and Mekonnen and Hoekstra
(2011). The provincial average yield levels of each crop in China
was obtained from the National Bureau of Statistics of China (NBSC,
2018).
Fig. 10. The boxplot of each driving forces of virtual water import of each city in the Beijing-Tianjin-Hebei region.
3. Results
3.1. WFs in crop production in the BTH region
As shown in Fig. 3, the WF per unit of crop decreased (by 24%
and 35% for cereals and cash crops, respectively) with an increase in
the average crop yield for the entire BTH region (by 24% and 32% for
cereals and cash crops, respectively). The share of the blue WF per
unit of crop was 29% for cereals and 16% for cash crops. The blue WF
per unit of crop showed a relatively rapid decrease (approximately
4% for both cereals and cash crops) when compared with the green
WF per unit of crop.
Table 2 shows the WF per unit of crop by 2013 and the relative
changes in the WF per unit of crop over the study period per
considered crop per city in the BTH region. Most cities had a
declining WF; Chengde and Qinhuangdao showed a relatively more
rapid rate of decline. Comparing the WF in different cities, high WFs
were concentrated in the central and southern parts of the BTH
region, where there was higher potential ET than other places in the
region (Fig. 2a). Among the crops, soybean exhibited the highest
WF (2203 m3/t) by 2013 while vegetables exhibited the lowest WF
(65 m3/t) by 2013.
Over the period of 2000e2013, the average annual total WFs in
crop production in the BTH region was 43.3 billion m3/y, of which
21% was blue WF. Although crop production in the BTH region
increased by 47% from 2000 to 2013, the total WFs of crop pro-
duction was almost stable and increased only slightly due to de-
creases of the WF per unit of crop (Table 2). Maize, wheat, and
vegetables accounted for over 70% of the total WFs in the BTH re-
gion. In terms of crop patterns in the total WFs, maize accounted for
34% of the total WFs in 2000, followed by wheat (31%) and vege-
tables (11%). In 2013, maize and vegetables increased by 7.6% and
2.3%, respectively, but wheat decreased by 4.2%, which was related
to the planting scale of each crop (the crop harvested area of maize
and vegetables increased by 20.6% and 19.7% but the area of wheat
decreased by 14.5%).
As can be seen in Fig. 4, high total WFs of crop production were
concentrated in the southern part of BTH region, containing Han-
dan, Baoding, and Cangzhou cities. In 2013, Baoding city had a
 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670 9

higher total WF than other areas, accounting for 14.4% of the total
WFs in the BTH region. The share of total blue WFs ranged from 13
to 25% (Langfang and Baoding, respectively) in 2013 and was
relatively small in areas with large precipitation. During the period
of 2000e2013, the proportion of total blue WFs in Beijing
decreased significantly from 42% in 2000 to 23.4% in 2013. This is
because the harvested area of all crops in Beijing decreased, with
wheat exhibiting the greatest decline of 70%. Moreover, the total
blue WFs of wheat accounted for a relatively large proportion (40%).
In the case of total WFs, most cities showed an increasing trend
except for Beijing, Tianjin, Tangshan, Xingtai, and Langfang. The
decline in total WFs was greatest in Beijing (53%), whereas that of
the other four cities was below 10%.
3.2. Crop-related inter-city VW flows in the BTH region
The BTH region as a whole was a net VW import area during
2000e2013, and the amount of net VW imports tripled from 3.6
billion m3 in 2000 to 11.7 billion m3 in 2013. The weight of blue VW
imports decreased from 43% to 8% during the same period. Fig. 5
shows the VW flows structure of crops for the entire BTH region.
The VW imports of the BTH region were predominantly rice, potato,
and soybean. VW exports were mainly composed of maize and
wheat. Soybean has been totally supported by places outside the
BTH region.
Fig. 6 shows the net VW imports for the cities in the BTH region
in 2000 and 2013. Tangshan, Xingtai, and Hengshui in Hebei
Province were the only three VW exporting cities in 2000. By 2013,
the number of VW export cities had increased significantly and VW
was predominantly exported from the southern regions. As the
largest VW export city, Tangshan City exhibited minimal changes in
VW exports, mainly due to the export of wheat, vegetables, and
specifically maize carrying VW. In 2000, Tangshan City assumed
95% of the maize VW exports. Hengshui changed from a major VW
exporter in 2000 to an importer in 2002. Because there is no potato
and rice planting in Hengshui, VW imports increased with demand.
At the same time, the VW exports of wheat, millet, and groundnuts
decreased with a decrease of crop planting area. On the contrary,
over the period 2000e2013, Baoding, Cangzhou, Shijiazhuang, and
Handan were transformed from VW importers to exporters due to
the increase in vegetables planted areas (by more than 20% and by a
maximum increase in Cangzhou of 48%). In addition, the VW ex-
ports of wheat, groundnuts, and cotton in Cangzhou also increased
because of the increase in crop yield.
Fig. 7 shows the direction of VW flows within the BTH region.
Chengde to Shijiazhuang had the largest VW flow, mainly due to
the VW flows caused by the transport of maize. Tangshan, the
largest VW exporter in 2000e2013, increased its export of maize to
Handan and Xingtai and its VW export related to vegetables to
Beijing. The VW exports of Qinhuangdao decreased significantly
due to the reduction of maize VW exports. The direction of blue VW
flows was approximately the same as that of the total VW flows
(Fig. 8). In 2000, the largest blue VW flow occurred in Chengde to
Shijiazhuang, which is 18.3% of the total VW flow, but the value was
reduced by half in 2013, to approximately 12.4% of the total VW
flow. In 2013, the blue VW exports from each city accounted for
10.8e52.4% of the total VW exports (from Handan to Hengshui and
Chengde to Hengshui, respectively).
3.3. Driving forces of total WFs in crop production and relative
inter-city VW flows
3.3.1. Driving forces of total WFs in crop production
Fig. 9 shows the annual driving forces of the total WFs of crop
production for each city in the BTH region. According to the
decomposition results, the per capita GDP (a proxy for economic
development) is the main factor enhancing the WFs, which led to
an approximate increase in total WFs of 165.5% over the period
2000e2013. Population had a positive effect, but only caused a
13.3% increase. The unit GDP productivity was the strongest in-
hibitor, resulting in a 134.5% decrease in the total WFs. The struc-
ture of production and the unit water productivity together
reduced the total WFs of most cities during the study period, by
14.8% and 28.5%, respectively. Apart from the effect of production
structure, the effects of other drivers were mainly due to maize,
wheat, and vegetables. Regarding the structure of production, it is
mainly due to the reduction of the wheat harvested area, resulting
in a decrease in the total WFs of crop production.
At the city scale, the population effect had a greater impact on
cities with more rapid population growth in Beijing and Tianjin;
however, due to the influence of urbanization level, the total WFs
growth caused by population growth is also weakened (Xu et al.,
2015). The impact of population growth on the total WFs is weak
in other cities where the population change is less intense. In
addition to the population effect, the other four driving factors had
a large impact on agricultural cities, such as Cangzhou, Handan, and
Baoding, especially the per capita GDP and unit GDP productivity
effect. Agricultural production is the foundation of economic
development for agricultural cities. Therefore, a higher per capita
GDP leads to stronger agricultural development, a sharp increase in
crop yields, and an increase in the total WF of crop production.
Similarly, agricultural cities are more sensitive to changes in unit
GDP productivity than non-agricultural cities. For agricultural cit-
ies, the production structure has gradually changed to be

Table 3
Comparison of water footprint per unit of crop production (m3/t) (total green plus blue unit water footprint) between the current results and previous studies in each province
for different period.

Sources Beijing Tianjin Hebei

Current Mekonnen and Zhuo et al. Current Mekonnen and Zhuo et al. Current Mekonnen and Zhuo et al.
Study Hoekstra (2011) (2016a) Study Hoekstra (2011) (2016a) Study Hoekstra (2011) (2016a)

Year 2003 1996e2005 2008 2003 1996e2005 2008 2003 1996e2005 2008

Wheat 1026 1346 889 892 1340 896 758 1295 1274
Maize 826 929 731 999 932 925 807 925 1008
Rice 853 1065 0 732 1071 0 553 1039 728
Millet 2287 1739 2246 0 1776 1376 1326 1698 1064
Potato 0 312 0 0 312 0 124 238 294
Soybean 2966 3795 0 4517 3848 0 2119 3723 2672
Groundnuts 1699 2270 2899 1575 2311 1068 1244 2161 912
Cotton 2046 1875 0 1756 1888 1299 1800 1857 1344
Vegetables 107 203 134 81 204 135 63 202 134
10 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670
Table 4
Comparison of net virtual water import (billion m3/y) between the current results and previous studies estimating the virtual water flows with input-output method in each
province.
Sources Study subjects Beijing
2002 2007
Zhang et al. (2011) Agriculture sector 0.86 / /
Wang et al. (2013) Agriculture sector 0.71 1.88
Liu et al. (2019b) Including other sectors / 0.8
Zhao et al. (2017a) Including other sectors / /
Current Study Specific crops 3.48 4.15
dominated by vegetables and maize, which consume less water,
while the proportion of wheat production has decreased signifi-
cantly. It is because that the wheat harvested area decreased 17%
from 2000 to 2013, as the government encourage farmers to sow
crops that consume less water such as maize and vegetables.
3.3.2. Driving forces of crop-related VW import for each city
Fig. 10 shows the annual effect of each driving factor on the VW
imports of each city. Clearly, the per capita GDP still played the
largest role in increasing the VW imports, resulting in an increase of
approximately 46%. The population effect, as the second largest
positive factor, resulted in an approximate increase of 7.1%. The unit
GDP consumption ability and the crop consumption structure were
inhibiting factors, causing a decrease in VW imports of 38.3% and
6.2%, respectively. The intension of VW flows effect had a variable
influence in different cities; combined, it decreased VW imports by
9.7%.
The driving factors played a significant role in Beijing and
Tianjin because the VW imports in these regions showed a signif-
icant increase whereas other cities exhibited a decline. Meanwhile,
the population and per capita GDP of Beijing and Tianjin grew faster
than that of other cities, leading to enhanced promotion of VW
imports. Cities with high economic development levels such as
Beijing and Tianjin mainly rely on VW imports to meet food de-
mands. Rapid per capita GDP and population growth lead to a
subsequent rapid increase in the food demand (Beijing and Tianjin
exhibited the largest growth rate of food consumption during the
study period; i.e., a 2-fold and 1.9-fold increase, respectively);
therefore, VW imports were greatly promoted. The intension of VW
flows effect played a positive role in Beijing and Tianjin and the
effect is reversed in other cities. It is because that the intension of
Beijing and Tian are considerably higher than other cities as they
are main VW importers in the BTH region. The other cities only
have VW import for maize and potato. The crop consumption
structure effect played an inhibitory role, mainly due to the
reduction in the proportion of rice consumption (caused by the
reduction of seed and waste consumption). Given that rice was the
main importing crop of the BTH region, this reduction in rice
consumption greatly contributed to the reduction of VW imports.
The unit GDP of consumption ability had a large inhibitory effect on
VW imports as the GDP was growing faster than consumption so
that the unit GDP of consumption appeared a downward trend
from 2000 to 2013.
4. Discussion
The current study evaluated the regional differences in the
driving forces behind both WFs in crop production and relative
inter-city VW flows in the BTH region from 2000 to 2013. Through
the study case, the spatial differences in regional driving forces in
both crop-related WFs and inter-regional VW flows are shown at a
sub-national scale, for the first time. Here, we compare the quan-
titative results on crop WFs of this study to published research for
Tianjin Hebei
2010 2012 2002 2007 2010 2012 2002 2007 2010 2012
/ / / / / / / / /
/ / / / / / / / / /
0.1 0.2 / 0.8 0.8 0.8 / 0.2 0.8 0.4
11.72 / / / 9.52 / / / 7.72 /
5.86 8.18 2.41 3.39 4.01 4.68 1.66 5.04 5.6 8.11
different periods (this study for 2013, Mekonnen and Hoekstra
(2011) for 1996e2005 and Zhuo et al. (2016a) for 2008). As
shown in Table 3, the calculation results are different due to dif-
ferences in the calculation methods and input parameters, but all
differences are within an acceptable range.
In the city level, there is a lack of both trade data and input-
output tables. This is the first time to quantify the crop-related
inter-city virtual water network within the Beijing-Tianjin-Hebei
region. Therefore, the shown inter-city virtual water flows cannot
be calibrated quantitatively, neither for numerical uncertainty
analysis. However, qualitative assessment of the reliability of the
results is possible. We compare the current results to previous
studies which calculated the inter-provincial virtual water flows
based on input-output tables for the same region (see Table 4). The
current results are reliable in terms of the magnitude and direction
of virtual water flows in each province. Undoubtedly, the actual
crop trade network could also be influenced by many other factors,
such as the price, quality, policy guidance and personal preferences.
Therefore, it should be in caution when implementing current re-
sults for certain local issues. Local calibration through surveys are
highly recommended.
Regarding the driving force analysis, we conclude that the main
promoting driving forces are essentially the same as those of pre-
vious studies (Tian et al., 2016; Zhao and Chen, 2014). Besides, our
study found that the intension of a same driving force differs for
different areas. Furthermore, we also confirm that the evolution of
VW flow patterns is determined by economy, not water resource
endowments or water resource production efficiency. However,
water management in agricultural production still focuses on water
resources, which largely ignores the interaction between economic
benefits and water consumption, water productivity and VW flows
between regions, and its impact on water shortages.
According to our analysis, we make the following suggestions
for improving water resources in the BTH region. 1) Since 2000, the
total blue WF has decreased significantly by 15.4% due to the pro-
motion of water-saving technologies. The water-saving irrigation
area of Hebei province increased by 47% from 2000 to 2013 (CNKI,
2018). Therefore, agricultural water-saving technology should
continue to be developed, which cannot be achieved without
strong support by the government. Specifically, the agricultural
water-saving technology includes not only the engineering water-
saving technology (e.g., drip- and micro-irrigation), but also
water-saving tillage like organic mulching, biological water-saving
technology by planting drought-resistant varieties, chemical re-
agent water saving and managerial water-saving technology like
soil moisture monitoring (Zhang et al., 2019). 2) The export of large
amounts of VW in Hebei Province has created tremendous pressure
on local water resources. Cities in this province should work
together to develop characteristic-based agriculture and appro-
priately reduce the cultivation of water-intensive crops such as
soybean and cotton in water scarce region, finding a balance be-
tween food security, economic development, and environmental
protection. Meanwhile, the compensating duty can be developed
 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670
and paid to VW exporters in terms of water consumed for exports.
3) Economic level greatly affects the VW flow pattern, thereby
affecting local water resource conditions. All cities, especially major
agricultural production cities, should vigorously develop other in-
dustries, ensure reasonable use of the VW strategy to meet food
demands, and maximize the efficiency of water resource utiliza-
tion. For instance, the major agricultural production cities can
import the water-intensive crops to reduce local water resources
pressure in circumstances that the economy flourishes.
The current study is not limited to the study case itself. Many
nations, basins or sub-national regions have visible spatial het-
erogeneities in terms of water resource endowments, economic
development levels, as well as roles in water productivities and VW
networks at various levels (e.g. Tuninetti et al., 2017b; Dang et al.,
2015; Katyaini and Barua, 2017). Given the varied local processes
towards United Nations Sustainable Development Goals (UN, 2015)
on water, environment, food and economic securities, the current
study case sheds lights on that the regional differences should be
considered in the analysis of driving forces in WFs and VW flows for
multiple regional scales in different places in the world. The
method used in this study is not only applicable to our study area,
but also can be applied to other regional scales and industries.
However, there are still some limitations to this study. First, we only
considered the amount of water consumed during crop growth and
did not consider its environmental impact; i.e., we did not calculate
the grey WF. Second, due to a lack of data, four assumptions were
made when quantifying the VW flows between cities. One is that
the provinces transport crops through provincial capitals. The
second is that crop export occurs when the production volume is
greater than the demand. The third is crops are transported to re-
gions where trade costs are minimal. The last one is the estimation
of crop consumption. The uncertainties of virtual water network
are com from the assumptions made in the model. This may not be
in line with the actual situation; however, compared with previous
studies, the general VW transport directions and magnitudes are
approximately equivalent. Therefore, these results are reliable.
Finally, in terms of driving force evaluation, the selected driving
factors are still socio-economic and the research objective is
focused only on agriculture. It has been reported that the WF in
crop production were also sensitive to the climatic characteristics
(Tuninetti et al., 2015; Zhuo et al., 2014). Apparently, further ex-
plorations on possible interactions between the natural and socio-
economic effects on WFs is necessary as well.
5. Conclusion
This study quantified the WFs in crop production and relative
inter-city VW flows in the BTH region from 2000 to 2013 and
identified their driving factors. Our main conclusions are as follows:
1) The WF per unit mass of crop production in the BTH region from
2000 to 2013 exhibited a downward trend with increasing crop
yields. The blue WF accounted for less than the green WF and its
proportion decreased each year. Although there were no sig-
nificant changes in total WF of crop production during the study
period, the crop structure changed greatly.
2) The BTH was always a VW import region but the roles of each
cities were different. Tangshan, Cangzhou, Handan, Shi-
jiazhuang, and Xingtai accounted for all VW exports in 2013;
therefore, these cities require a different approach regarding
water resource management.
3) Per capita GDP had the largest positive effect on both WFs and
VW flows. Economic productivity and consumption capacity
were the largest inhibitors of WFs and VW flows. The degree of
influences of each driving forces was directly related to the
11
regional conditions. The regional differences of driving forces
cannot be ignored.
A bigger picture showing the spatial heterogeneities in driving
forces of WFs and VW flows in places other than the shown study
case at different management levels are highly recommended for
future studies. Meanwhile, the scope of the study can also be
further extended to all water using sectors in the future.
CRediT authorship contribution statement
Meng Li: Software, Methodology, Validation, Formal analysis,
Writing - original draft. Zenghui Xu: Software, Methodology,
Validation, Formal analysis, Writing - original draft. Shan Jiang:
Writing - review & editing. La Zhuo: Conceptualization, Writing -
review & editing, Supervision. Xuerui Gao: Writing - review &
editing. Yong Zhao: Conceptualization, Writing - review & editing,
Supervision. Yilin Liu: Visualization, Data curation. Wei Wang:
Visualization, Data curation. Jiming Jin: Writing - review & editing.
Pute Wu: Conceptualization, Writing - review & editing,
Supervision.
Declaration of competing interest
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
Acknowledgments
This work was financially supported by the National Key R&D
Program of China [2016YFC0401407, 2018YFF0215702], the Open
Research Fund of State Key Laboratory of Simulation and Regulation
of Water Cycle in River Basin at China Institute of Water Resources
and Hydropower Research [IWHR-SKL-201702], the National Nat-
ural Science Foundation of China Grants [51809215] and the 111
Project [No. B12007]. The work was partially developed within the
framework of the Panta Rhei Research Initiative of the international
Association of Hydrological Sciences (IAHS).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jclepro.2020.123670.
References
Ang, B.W., Zhang, F., Choi, K., 1998. Factorizing changes in energy and environ-
mental indicators through decomposition. Energy 23 (6), 489e495. https://
doi.org/10.1016/S0360-5442(98)00016-4.
Ang, B.W., 2004. Decomposition analysis for policymaking in energy. Energy Pol. 32,
1131e1139. https://doi.org/10.1016/s0301-4215(03)00076-4.
Cai, B., Zhang, W., Hubacek, K., Feng, K., Li, Z., Liu, Y., Liu, Y., 2019. Drivers of virtual
water flows on regional water scarcity in China. J. Clean. Prod. 207, 1112e1122.
https://doi.org/10.1016/j.jclepro.2018.10.077.
Chukalla, A.D., Krol, M.S., Hoekstra, A.Y., 2015. Green and blue water footprint
reduction in irrigated agriculture: effect of irrigation techniques, irrigation
strategies and mulching. Hydrol. Earth Syst. Sci. 19 (12), 4877e4891. https://
doi.org/10.5194/hess-19-4877-2015.
CMA, 2018. China Meteorological Data Sharing Service System. China Meteorolog-
ical Administration, Beijing, China. http://data.cma.cn.
CNKI, 2018. China Yearbooks Full-Text Database. China National Knowledge Infra-
structure, Beijing, China. http://data.cnki.net/Yearbook/Navi?type=type&c-
ode=A" xlink:role="http://www.elsevier.com/xml/linking-roles/research-data.
Chouchane, H., Krol, M.S., Hoekstra, A.Y., 2018. Virtual water trade patterns in
relation to environmental and socioeconomic factors: a case study for Tunisia.
Sci. Total Environ. 613e614, 287e297. https://doi.org/10.1016/
j.scitotenv.2017.09.032.
Dalin, C., Hanasaki, N., Qiu, H., Mauzerall, D.L., Rodriguez-Iturbe, I., 2014. Water
resources transfers through Chinese interprovincial and foreign food trade.
12 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670
Proc. Natl. Acad. Sci. U.S.A. 111 (27), 9774e9779. https://doi.org/10.1073/
pnas.1404749111.
Dang, Q., Lin, X., Konar, M., 2015. Agricultural virtual water flows within the United
States. Water Resour. Res. 51, 973e986. https://doi.org/10.1002/2014wr015919.
Dietz, T., Rosa, E.A., 1994. Rethinking the environmental impacts of population,
affluence, and technology. Hum. Ecol. Rev. 1, 277e300. https://doi.org/10.2307/
24706840.
Dijkshoorn, J.A., van Engelen, V.W.P., Huting, J.R.M., 2008. Soil and Landform
Properties for LADA Partner Countries (Argentina, China, Cuba, Senegal and the
Gambia, South Africa and Tunisia). ISRIC Report 2008/06. ISRIC World Soil In-
formation, Wageningen, the Netherlands. http://www.isric.org/isric/Webdocs/
Docs/ISRIC_Report_2008_06.pdf.
D’Odorico, P., Rulli, M.C., 2013. The fourth food revolution. Nat. Geosci. 6 (6),
417e418. https://doi.org/10.1038/ngeo1842.
Ehrlich, P.R., Holdren, J.P., 1971. Impact of population growth. Science 171,
1212e1217. https://doi.org/10.1126/science.171.3977.1212.
FAO, 2018. FAOSTAT On-Line Database. Food and Agriculture Organization of the
United Nations. http://faostat.fao.org.
Feng, K., Chapagain, A.K., Suh, S., Pfister, S., Hubacek, K., 2011. Comparison of
bottom-up and top-down approaches to calculating the water footprints of
nations. Econ. Syst. Res. 23 (4), 371e385. https://doi.org/10.1080/
09535314.2011.638276.
Gao, J., Zhuo, L., Liu, Y.L., Xie, P.X., Wang, W., Li, M., Gao, X.R., Wu, P.T., 2020. Effi-
ciency and sustainability of inter-provincial crop-related virtual water transfers
in China. Adv. Water Resour. 138, 103560. https://doi.org/10.1016/
j.advwatres.2020.103560.
Hoekstra, A.Y., 2003. Virtual Water Trade: Proceedings of the International Expert
Meeting on Virtual Water Trade. IHE Delft, The Netherlands. http://www.
ayhoekstra.nl/pubs/Report11.pdf.
Hoekstra, A.Y., 2013. The Water Footprint of Modern Consumer Society. Routledge,
London, UK.
Hoekstra, A.Y., Chapagain, A.K., Aldaya, M.M., Mekonnen, M.M., 2011. Water Foot-
print Assessment Manual: Setting the Global Standard. Earthscan, London, UK.
Hoekstra, A.Y., Mekonnen, M.M., 2012. The water footprint of humanity. Proc. Natl.
Acad. Sci. U.S.A. 109 (9), 3232e3237. https://doi.org/10.1073/pnas.1109936109.
Hoekstra, A.Y., Mekonnen, M.M., 2016. Imported water risk: the case of the UK.
Environ. Res. Lett. 11 (5), 055002 https://doi.org/10.1088/1748-9326/11/5/
055002.
Hoekstra, A.Y., Wiedmann, T.O., 2014. Humanity’s unsustainable environmental
footprint. Science 344 (6188), 1114e1117. https://doi.org/10.1126/
science.1248365.
Hoekstra, R., van den Bergh, J.C.J.M., 2003. Comparing structural and index
decomposition analysis. Energy Econ. 25 (1), 39e64. https://doi.org/10.1016/
S0140-9883(02)00059-2.
Hsiao, T.C., Heng, L., Steduto, P., Rojas-Lara, B., Raes, D., Fereres, E., 2009. Aqua-
Cropdthe FAO crop model to simulate yield response to water. III. Parameter-
ization and testing for maize. Agron. J. 101 (3), 448e459. https://doi.org/
10.2134/agronj2008.0218s.
Jagermeyr, J., Pastor, A., Biemans, H., Gerten, D., 2017. Reconciling irrigated food
production with environmental flows for Sustainable Development Goals
implementation. Nat. Commun. 8, 15900. https://doi.org/10.1038/
ncomms15900.
Katyaini, S., Barua, A., 2017. Assessment of interstate virtual water flows embedded
in agriculture to mitigate water scarcity in India (1996-2014). Water Resour.
Res. 53, 7382e7400. https://doi.org/10.1002/2016wr020247.
Leontief, W., Ford, D., Brody, A., Carter, A.P., 1972. Air pollution and the economic
structure: empirical results of input-output computations. In: Input-output
Techniques. North-Holland Publishing Company, North-Holland, pp. 9e30,
1972.
Liu, H., Du, K., Li, J., 2019a. An improved approach to estimate direct rebound effect
by incorporating energy efficiency: a revisit of China’s industrial energy de-
mand. Energy Econ. 80, 720e730. https://doi.org/10.1016/j.eneco.2019.02.012.
Liu, W., Antonelli, M., Kummu, M., Zhao, X., Wu, P.T., Liu, J., Zhuo, L., Yang, H., 2018.
Savings and losses of global water resources in food-related virtual water trade.
WIREs Water 6, e1320. https://doi.org/10.1002/wat2.1320.
Liu, X., Du, H.B., Zhang, Z.K., Crittenden, J.C., Lahr, M.L., Moreno-Cruz, J., Guan, D.,
Mi, Z.F., Zuo, J., 2019b. Can virtual water trade save water resources? Water Res.
163, 114848. https://doi.org/10.1016/j.watres.2019.07.015.
Ma, M., Cai, W., Cai, W., 2018. Carbon abatement in China’s commercial building
sector: a bottom-up measurement model based on Kaya-LMDI methods. Energy
165, 350e368. https://doi.org/10.1016/j.energy.2018.09.070.
Ma, J., Hoekstra, A.Y., Wang, H., Chapagain, A.K., Wang, D., 2006. Virtual versus real
water transfers within China. Philos. Trans. Roy. Soc. B-Biol. Sci. 361 (1469),
835e842. https://doi.org/10.1098/rstb.2005.1644.
Mekonnen, M.M., Hoekstra, A.Y., 2011. The green, blue and grey water footprint of
crops and derived crop products. Hydrol. Earth Syst. Sci. 15 (5), 1577e1600.
https://doi.org/10.5194/hess-15-1577-201.
Mousavi, B., Lopez, N.S.A., Biona, J.B.M., Chiu, A.S.F., Blesl, M., 2017. Driving forces of
Iran’s CO2 emissions from energy consumption: an LMDI decomposition
approach. Appl. Energy 206, 804e814. https://doi.org/10.1016/
j.apenergy.2017.08.199.
NBSC, 2018. National Data, China. National Bureau of Statistics of China. http://data.
stats.gov.cn.
Qian, Y., Tian, X., Geng, Y., Zhong, S., Cui, X., Zhang, X., Moss, D.A., Bleischwitz, R.,
2019. Driving factors of agricultural virtual water trade between China and the
belt and road countries. Environ. Sci. Technol. 53, 5877e5886. http://dx.10.
1021/acs.est.9b00093.
Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2009. AquaCropdthe FAO crop model to
simulate yield response to water. II. Main algorithms and software description.
Agron. J. 101 (3), 438e477. https://doi.org/10.2134/agronj2008.0140s.
Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2017. Reference Manual, Chapter2,
AquaCrop Model, Version 6.0. Food and Agriculture Organization of the United
Nations, Rome, Italy.
RESDC, 2018. Resources and Environmental Sciences Data Center. Chinese Academy
of Sciences. http://www.resdc.cn/Default.aspx.
Steduto, P., Hsiao, T.C., Fereres, E., 2007. On the conservative behavior of biomass
water productivity. Irrigat. Sci. 25, 189e207. https://doi.org/10.1007/s00271-
007-0064-1.
Steduto, P., Hsiao, T.C., Raes, D., Fereres, E., 2009. AquaCropdthe FAO crop model to
simulate yield response to water. I. Concepts and underlying principles. Agron.
J. 101 (3), 426e437. https://doi.org/10.2134/agronj2008.0139s.
Sun, S., 2019. Water footprints in Beijing, Tianjin and Hebei: a perspective from
comparisons between urban and rural consumptions in different regions. Sci.
Total Environ. 647, 507e515. https://doi.org/10.1016/j.scitotenv.2018.07.343.
Su, W., Chen, S., Bale zentis, T., Chen, J., 2020. Economy-water nexus in agricultural
sector: decomposing dynamics in water footprint by the Lmdi. Technol. Econ.
Dev. Econ. 26, 240e257. https://doi.org/10.3846/tede.2020.11908.
Tamea, S., Carr, J.A., Laio, F., Ridolfi, L., 2014. Drivers of the virtual water trade. Water
Resour. Res. 50 (1), 17e28. https://doi.org/10.1002/2013WR014707.
Tan, R., Lin, B., 2018. What factors lead to the decline of energy intensity in China’s
energy intensive industries? Energy Econ. 71, 213e221. https://doi.org/10.1016/
j.eneco.2018.02.019.
Tian, Y., Ruth, M., Zhu, D., 2016. Using the IPAT identity and decoupling analysis to
estimate water footprint variations for five major food crops in China from 1978
to 2010. Environ. Dev. Sustain. 19 (6), 2355e2375. https://doi.org/10.1007/
s10668-016-9860-1.
Tuninetti, M., Tamea, S., D’Odorico, P., Laio, F., Ridolfi, L., 2015. Global sensitivity of
high-resolution estimates of crop water footprint. Water Resour. Res. 51,
8257e8272. https://doi.org/10.1002/2015WR017148.
Tuninetti, M., Tamea, S., Laio, F., Ridolfi, L., 2017a. A Fast Track approach to deal with
the temporal dimension of 369 crop water footprint. Environ. Res. Lett. 12,
074010 https://doi.org/10.1088/1748-9326/aa6b09.
Tuninetti, M., Tamea, S., Laio, F., Ridolfi, L., 2017b. To trade or not to trade: link
prediction in the virtual water network. Adv. Water Resour. 110, 528e537.
https://doi.org/10.1016/j.advwatres.2016.08.013.
United Nations (UN), 2015. Transforming our world: the 2030 agenda for sustain-
able development. A/RES/70/1. https://sustainabledevelopment.un.org/
post2015/transformingourworld.
Wang, M., Feng, C., 2018. Decomposing the change in energy consumption in
China’s nonferrous metal industry: an empirical analysis based on the LMDI
method. Renew. Sustain. Energy Rev. 82, 2652e2663. https://doi.org/10.1016/
j.rser.2017.09.103.
Wang, Q., Jiang, R., 2019. Is China’s economic growth decoupled from carbon
emissions? J. Clean. Prod. 225, 1194e1208. https://doi.org/10.1016/
j.jclepro.2019.03.301.
Wang, R., Hertwich, E., Zimmerman, J.B., 2016. (Virtual) water flows uphill toward
money. Environ. Sci. Technol. 50 (22), 12320e12330. https://doi.org/10.1021/
acs.est.6b03343.
Wang, Z., Huang, K., Yang, S.S., Yu, Y.J., 2013. An input-output approach to evaluate
the water footprint and virtual water trade of Beijing, China. J. Clean. Prod. 42,
172e179. https://doi.org/10.1016/j.jclepro.2012.11.007.
Wang, Z., Zhang, L., Ding, X., Mi, Z., 2019. Virtual water flow pattern of grain trade
and its benefits in China. J. Clean. Prod. 223, 445e455. https://doi.org/10.1016/
j.jclepro.2019.03.151.
Xiong, Y.L., Tian, X., Liu, S.W., Tang, Z.P., 2020. New patterns in China’s water
footprint: analysis of spatial and structural transitions from a regional
perspective. J. Clean. Prod. 245, 118942. https://doi.org/10.1016/
j.jclep.ro.2019.118942.
Xu, Y., Huang, K., Yu, Y., Wang, X., 2015. Changes in water footprint of crop pro-
duction in Beijing from 1978 to 2012: a logarithmic mean Divisia index
decomposition analysis. J. Clean. Prod. 87, 180e187. http://dx.10.1016/j.jclepro.
2014.08.103.
Zhang, B., Fu, Z.T., Wang, J.Q., Zhang, L.X., 2019. Farmers’ adoption of water-saving
irrigation technology alleviates water scarcity in metropolis suburbs: a case
study of Beijing, China. Agric. Water Manag. 212, 349e357. https://doi.org/
10.1016/j.agwat.2018.09.021.
Zhang, Z., Shi, M., Yang, H., 2012. Understanding Beijing’s water challenge: a
decomposition analysis of changes in Beijing’s water footprint between 1997
and 2007. Environ. Sci. Technol. 46 (22), 12373e12380. https://doi.org/10.1021/
es302576u.
Zhang, Z.Y., Yang, H., Shi, M.J., 2011. Analyses of water footprint of Beijing in an
interregional inputeoutput framework. Ecol. Econ. 70, 2494e2502. https://
doi.org/10.1016/j.ecolecon.2011.08.011.
Zhao, C., Chen, B., 2014. Driving force analysis of the agricultural water footprint in
China based on the LMDI method. Environ. Sci. Technol. 48 (21), 12723e12731.
https://doi.org/10.1021/es503513z.
Zhao, C., Chen, B., Hayat, T., Alsaedi, A., Ahmad, B., 2014. Driving force analysis of
water footprint change based on extended STIRPAT model: evidence from the
Chinese agricultural sector. Ecol. Indicat. 47, 43e49. https://doi.org/10.1016/
j.ecolind.2014.04.048.
 M. Li et al. / Journal of Cleaner Production 279 (2021) 123670
Zhao, D., Tang, Y., Liu, J., Tillotson, M.R., 2017a. Water footprint of Jing-Jin-Ji urban
agglomeration in China. J. Clean. Prod. 167, 919e928. https://doi.org/10.1016/
j.jclepro.2017.07.012.
Zhao, D., Hubacek, K., Feng, K., Sun, L., Liu, J., 2019. Explaining virtual water trade: a
spatial-temporal analysis of the comparative advantage of land, labor and water
in China. Water Res. 153, 304e314. https://doi.org/10.1016/j.watres.2019.01.025.
Zhao, X., Liu, J., Liu, Q., Tillotson, M.R., Guan, D., Hubacek, K., 2015. Physical and
virtual water transfers for regional water stress alleviation in China. Proc. Natl.
Acad. Sci. U.S.A. 112 (4), 1031e1035. https://doi.org/10.1073/pnas.1404130112.
Zhao, X., Li, Y.P., Yang, H., Liu, W.F., Tillotson, M.R., Guan, D., Yi, Y., Wang, H., 2018.
Measuring scarce water saving from interregional virtual water flows in China.
Environ. Res. Lett. 13, 054012 https://doi.org/10.1088/1748-9326/aaba49.
Zhao, X., Tillotson, M.R., Liu, Y.W., Guo, W., Yang, A.H., Li, Y.F., 2017b. Index
decomposition analysis of urban crop water footprint. Ecol. Model. 348, 25e32.
https://doi.org/10.1016/j.ecolmodel.2017.01.006.
Zhao, X., Zhang, X., Shao, S., 2016. Decoupling CO2 emissions and industrial growth
in China over 1993e2013: the role of investment. Energy Econ. 60, 275e292.
13
https://doi.org/10.1016/j.eneco.2016.10.008.
Zhuo, L., Liu, Y.L., Yang, H., Hoekstra, A.Y., Liu, W.F., Cao, X.C., Wang, M.R., Wu, P.T.,
2019. Water for maize for pigs for pork: an analysis of inter-provincial trade in
China. Water Res. 166, 115074. https://doi.org/10.1016/j.watres.2019.115074.
Zhuo, L., Mekonnen, M.M., Hoekstra, A.Y., 2014. Sensitivity and uncertainty in crop
water footprint accounting: a case study for the Yellow River Basin. Hydrol.
Earth Syst. Sci. 18 (6), 2219e2234. https://doi.org/10.5194/hess-18-2219-2014.
Zhuo, L., Mekonnen, M.M., Hoekstra, A.Y., 2016a. The effect of inter-annual vari-
ability of consumption, production, trade and climate on crop-related green and
blue water footprints and inter-regional virtual water trade: a study for China
(1978-2008). Water Res. 94, 73e85. https://doi.org/10.1016/
j.watres.2016.02.037.
Zhuo, L., Mekonnen, M.M., Hoekstra, A.Y., Wada, Y., 2016b. Inter- and intra-annual
variation of water footprint of crops and blue water scarcity in the Yellow
River basin (1961e2009). Adv. Water Resour. 87, 29e41. https://doi.org/10.1016/
j.advwatres.2015.11.002.