Quantitative analyzes of virtual water net exports under the impacts of natural changes and human activities in the last 20 years in Shandong Province, China

Virtual water trade in a region is affected by both nature and humans. To study the contributions of human activities to virtual water trade quantitatively, an innovative method of quantitative comparison and analysis is put forward. At ﬁ rst, the climates are adjusted into a uni ﬁ ed standard. Then the impacts of increment and reduction of foreign water are studied. Additionally, the impacts of water management policy are studied according to the comparable water quotas. Results show that with the development of economy, an N-shaped trend and inverted U-shaped trend exist with regard to the net exports of agricultural and industrial virtual water, respectively. The net imports of virtual water have bene ﬁ cial effects to the water environments of water de ﬁ cient areas, while the net exports have negative effects. In 1997, the net exports of agricultural and industrial virtual water reduced by 20.13% and 49.67% respectively due to the cut-off of Yellow River channel compared with that under the average Yellow River water diversion. In 2017, they increased by 1.32% and 41.99% respectively because of the South-to-North Water Transfer project and reduced by 10.01% and 20.39% respectively under the effects of the most stringent water management policy.

Researches on virtual water trade can promote the perfection and development of water policies (Sun et al. 2021). For water policies making, factors such as physical water, virtual water, science, climates should be considered simultaneously, and adjustments of space and time are also needed (Katyaini et al. 2020;Xu et al. 2020).
Shandong Province is one of the water scarce provinces in Northern China. The annual average local water resources only account for 1.1% of the total water resources of the country. Per capita water resources are less than 1/6 of the national average, only 1/24 of the world's average, and it belongs to the severely water deficient areas identified by the United Nations. What's more, the spatial and temporal distribution of water resources is uneven in Shandong Province. The annual rainfall was various from 1997 to 2017, with maximum rainfall of 936 mm in 2003 and minimum rainfall of 420 mm in 2002. Average rainfall in 20 years was 680 mm, and the standard deviation was 114 mm.
Problems of water shortage in Shandong Province can be solved by increasing water supply, controlling water usage, and improving water use efficiency (Deng et al. 2018;Fu et al. 2021). Besides the local water resources, as foreign water, 7 billion m 3 of Yellow River water is allocated to Shandong Province by the central government, which is an important source of water supply in Shandong Province. However, affected by the cut-off of the Yellow River in 1997-1999, the average diversion amount of Yellow River water in Shandong Province was only 1.99 billion m 3 in the three years. The diversion amount of Yellow River water did not return to the normal level until the year of 2000. The diversion amount was 5.36 billion m 3 in 2000 and since then there has been no cut-off of Yellow River channel because of unified management of whole basin by central government. Additionally, with the implementation of South-to-North Water Transfer (SNWT) project, the SNWT water has been transferred to Shandong Province since 2014, with a total of 2.19 billion m 3 until 2019. Furthermore, in order to control the use of water, the Shandong Provincial Government has issued a series of documents to implement the most stringent water management policy since 2015, including three red lines of total water consumption, water use efficiency and limited pollutant carrying capacity of water function area, which has achieved good results. Of course, the problem of water shortage can also be solved through virtual water trade (Deng et al. 2015).
Nevertheless, the above researches suffer from two shortcomings. Firstly, virtual water trade is not only related to local natural conditions, but also affected by human activities. The above researches fail to distinguish the two factors, so that they can't quantitatively analyze the impacts of nature and human activities on virtual water trade respectively. Secondly, the time series of virtual water trade research in Shandong Province is short, and the longest time series so far is 15 years . Therefore, this study aims to analyze the dynamic changes of virtual water trades in Shandong Province over the past 20 years and propose a new comparative analysis method. Impacts of the increment and reduction of foreign water on virtual water net exports are analyzed within a year. Impacts of the most stringent water management policy on virtual water net exports are analyzed at annual level. This method can quantitatively analyze the impacts of different foreign water supply sources and water management policy on virtual water trade, so as to provide bases for managers to formulate comprehensive water management policies of physical water and virtual water, and promote the sustainable utilization and development of regional water resources.

Study area
Shandong province is located in eastern coast of China, where the latitude is 34°-39°N and longitude is 114°-123°E, with temperate monsoon climate. The Yellow River water is an important water source in Shandong Province. In order to alleviate the water shortage in Qingdao, Yellow River water has been diverted from Binzhou to Qingdao. In addition, the East Route of South-to-North Water Transfer project passes through Shandong Province. After been regulated and stored by Dongping Lake in Taian, the Yangtze River water is supplied to north area of Yellow River and eastern Shandong Province respectively. In eastern Shandong Province, Yangtze River water is transferred by part the route of Yellow River water diversion project. Schematic diagram of water transfer projects in Shandong Province is shown in Figure 1.

Calculation of virtual water net export
Virtual water net export can be calculated as follows in Equation (1).
where DW i j and W i jC are the virtual water net export and total water consumption of sector j in the year of i, respectively, m 3 . Y i jT , Y i jE and Y i jI are the total output value, export value and import value of sector j in the year of i, respectively, CNY. W i jC =Y i jT is the water consumption per unit output value, m 3 /CNY.

Quantitative comparison and analysis of virtual water net export changes
In this paper, a new method is proposed to analyze quantitatively the impacts of human activities on the virtual water net exports. The impacts of foreign water are studied into two scenarios: the Yellow River water reduction and the SNWT water increment. Firstly, the common comparative basis is that the rainfalls of the five statistical years are assumed to be the same, with the local water supply and consumption being consistent correspondingly. Consequently, the differences of virtual water trade caused by different foreign water supplies in different hydrological years can be eliminated. Secondly, it is considered that water saving and using levels are changeless, that is, the water consumption per unit output value is changeless, then the relationships between output value and water consumption are established. Finally, if the adjusted water supply amount is larger than the actual water consumption in the statistical year, the export value will increase, causing the increment of the virtual water exports. On the contrary, virtual water imports will increase. Accordingly, the quantitative impacts of foreign water on virtual water net exports are discussed. For the water saving measures, it is the impacts of the most stringent water management policy on virtual water net exports since 2015 according to the comparable water quotas that are paid attention to.

Adjustment principle of balance between water demand and supply for virtual water trade
If the adjusted water supply of sector j increases compared with the actual water supply in the year of i in a region, the outputs will also increase, so the exports of products will increase under the premise of the unchanged local demand. Conversely, if the adjusted water supply of sector j reduces compared with the actual water supply, the outputs will reduce too, so the imports of products will increase. From the perspective of output value, export value and import value increase respectively. Thus, the virtual water net exports can be calculated as follows in Equations (2) and (3).
where W iw jC , W i jC and DW iw j are adjusted water supply, actual total water consumption and adjusted virtual water net export of sector j in the year of i respectively, m 3 . a i j is the change rate of water supply of sector j in the year of i.

Quantification under the same annual rainfall
The annual rainfalls at 20, 50, 75, and 95% frequencies were 796, 670, 579, 463 mm, respectively in Shandong Province from 1956 to 2000. The rainfalls were 553, 420, 773, 651, 636 mm in 1997 (dry year), 2002 (extraordinary dry year), 2007 (slight wet year), 2012 (normal year), and 2017 (slight dry year), respectively. Although the hydrological characteristics of these 5 years were mostly dry years, 2007 was a slight wet year and 2012 was a normal year, which made these five years have a certain representativeness of wet, normal and dry years. Therefore, rainfall is assumed to be 606 mm, being the annual average of five statistical years. The corresponding total water supply is calculated as Equation (4).
where i ¼ 1,2,3,4,5 represent the year of 1997, 2002, 2007, 2012 and 2017, respectively. W i 0 S0 is the adjusted total water supply under the same annual rainfall in the year of i, m 3 . W i S is the actual water supply in the year of i, m 3 . Water consumption of sector j is expressed as Equation (5).
where W i 0 jC is adjusted water consumption of sector j under the same annual rainfall in the year of i, m 3 . Then virtual water net exports can be calculated according to Equations (2) and (3).

Impacts of the reduction of Yellow River water
The adjusted total water supply considering the Yellow River water can be calculated as Equation (6).
where W i 00 S1 is the adjusted total water supply considering the Yellow River water in the year of i, m 3 . W i Ye is the quantity of Yellow River water in the year of i, m 3 .
In order to study the impact of the cut-off of Yellow River channel on virtual water exports in 1997, a virtual quantity of Yellow River water is defined. The adjusted total water supply considering the virtual Yellow River water can be calculated as Equation (7).
where i ¼ 1,2,3,4 represent the year of 2002,2007,2012,2017. W i 00 S2 is adjusted total water supply considering the virtual Yellow River water in the year of i, m 3 .
The adjusted water consumption of sector j can be calculated as follows in Equations (8) and (9).
where W i 00 jC1 and W i 00 jC2 are adjusted water consumptions of sector j when consider the actual and virtual Yellow River in the year of i, respectively, m 3 .
Similarly, virtual water net exports can be calculated according to Equations (2) and (3).

Impacts of the increment of SNWT water
When considering the SNWT project, the SNWT water is assumed to be consumed by industry entirely because most of the SNWT water is distributed to industry. If the industrial water supply is sufficient, the previously occupied agricultural water by industry will be returned to agriculture. Subsequently, the agricultural water supply will increase. The impacts of SNWT water on agricultural virtual water net exports are discussed in this case. What's more, agriculture refers to generalized agriculture in this paper, including four industries of planting, forestry, animal husbandry, and fishery. Consequently, the adjusted water consumptions of agriculture and industry can be calculated as follows in Equation (10) under the effects of the Yellow River water and SNWT water.
where W i 000 jC is the water consumption of sector j under the effects of the Yellow River water and SNWT water in the year of i, m 3 . W i SNWT is the quantity of SNWT water in the year of i, m 3 . Similarly, virtual water net exports can be calculated according to Equations (2) and (3).

Impacts of the water management policy
The impacts of the most stringent water management policy on virtual water net exports are studied by comparable water quotas. In the calculation of comparable water quota, the effects of SNWT water should be removed. The calculation of comparable water quota is shown in Equation (11).
where u i j is comparable water quota of sector j in the year of i, m 3 / CNY. Y i 0 jT is the total comparable output values of sector j and the base year is 2007, CNY.

Data
The input and output values of agriculture and industry are from Input-Output Tables of Shandong Province in 1997, 2002, 2007, 2012 and 2017. The rainfalls, groundwater funnel areas, quantities of Yellow River water and SNWT water are from Water Resources Bulletins of Shandong Province in 1997-2017. The gross domestic product (GDP) and discharge amounts of industrial wastewater are from the Shandong Statistical Yearbooks in 1998-2018.

Virtual water net exports of agriculture and industry
The actual agricultural and industrial virtual water net exports are calculated by Equation (1). Virtual water net exports and per capita GDP changes are shown in Figures 2 and 3.
From Figure 2, with the development of economy, an N-shaped trend exists with regard to the net exports of agricultural virtual water. In the first stage, net exports increased slowly in the period from 1997 to 2002. In the second stage, net exports decreased. The net exports of agriculture did not become imports until the GDP per capita was 14540 CNY in 2004. Then the net imports were increasing. The net imports reached the maximum of 7,524.54 million m 3 when the GDP became 48763 CNY in 2013. In the third stage, the net imports decreased, but it was still net imports in 2017 and the amount was 1,220.61 million m 3 .
From Figure 3, with the development of economy, an inverted U-shaped trend exists with regard to the net exports of industrial virtual water. With the increase of GDP per capita, industrial virtual water remains net exports. In the first stage, the net exports increased. The net exports reached the maximum when the GDP became 44464 CNY in 2012 and the amount was 128.38 million m 3 . In the third stage, the net exports were decreased. The amount was 53.98 million m 3 when the GDP was 63162 CNY in 2017.
Compared virtual water net exports of agriculture with industry, it is find that agriculture is a big water user.

Relationships between virtual water net exports and water environments
The current total water supply amount is 24.7 billion m 3 , of which groundwater supply accounts for 53%, surface water supply accounts for 25% and foreign water accounts for 22% in Shandong Province. Besides, the development and utilization rate of local surface water is 32%, and the exploitation rate of shallow groundwater is 64%. Overexploitation of water resources has negative impacts on groundwater firstly. Additionally, the uneven distributions of rainfalls at inter-annual and inter-decadal levels and the constructions of reservoirs make the ecological water amount of the river seriously insufficient. Although the ecological water quantity of the river is required to be 10% of the average annual runoff quantity, it is difficult to achieve for most large and medium-sized reservoirs. In non-flood season, river channels are mainly recharged by reclaimed water. Hence the quality of the treated industrial wastewater has important influence on the quality of the ecological water of the river.
Therefore, in order to study the impacts of virtual water net exports on water environments, groundwater funnel areas and treated industrial wastewater amounts are taken as examples.

Uncorrected Proof
As is shown in Figure 4, with the economic growth, the groundwater funnel area reduced firstly and then increased in 1997-2017. In 2002-2017, it had the same trend with agricultural virtual water net exports roughly. Broadly speaking, in 2002-2012, for every 1000 CNY of GDP per capita was increased, the groundwater funnel area reduced by 106.44 km 2 ; in 2012-2017, the groundwater funnel area increased by 92.43 km 2 . Similarly, for every 1000 CNY of GDP per capita was increased, the agricultural virtual water net export reduced by 275.70 million m 3 in 2002-2012 and increased by 237.30 million m 3 in 2012-2017. Thus, in 2002-2012, for every 1 million m 3 of agricultural virtual water net export reduced, the groundwater funnel area reduced by 0.38 km 2 . In 2012-2017, for every 1 million m 3 of agricultural virtual water net export increased, the groundwater funnel area increased by 0.39 km 2 .
As is shown in Figure 5, with the economic growth, the treated industrial wastewater increased firstly and then reduced in 1997-2017, being consistent with the changes of industrial virtual water net exports. For every 1000 CNY of GDP per capita was increased, in 1997-2012, the discharge amount of treated industrial wastewater increased by 24.49 million m 3 ; in 2012-2017, it reduced by 19.12 million m 3 correspondingly. Similarly, for every 1000 CNY of GDP per capita was increased, the industrial virtual water net exports increased by 2.60 million m 3 in 1997-2012 and reduced by 2.80 million m 3 in 2012-2017. Thus, in 1997-2012, for every 1 m 3 of industrial virtual water export increased, the discharge amount of wastewater increased by 9.41 m 3 . In 2012-2017, for every 1 m 3 of industrial virtual water export reduced, the discharge amount of wastewater reduced by 6.82 m 3 .
From the above correlation analysis, it can be concluded that the change trends of groundwater funnel area and treated industrial wastewater are related to the virtual water net exports of agriculture and industry respectively to a certain extent. As the agricultural virtual water net exports increase, groundwater funnel areas increase, which will lead to insufficient groundwater supply and land subsidence, worsening the ecological environment. Similarly, as the industrial virtual water net exports increase, the discharges of industrial wastewater increase, which will put pressure on wastewater treatments in factories. At the same time, the treated wastewater discharged into the river will also change the original ecological environment and have a certain impact on the water environment of the river. Fortunately, with the development of economy, the net imports of agricultural virtual water are maintained and the net exports of industrial virtual water are reduced, which are beneficial to the ecological environment of Shandong Province. It can be said that the increment of net imports or the reduction of net exports of industrial virtual water not only reduce the intakes of physical water, but also reduce the  Uncorrected Proof discharges of industrial wastewater and water pollutions. There is no doubt that it is one of the effective measures for industrial water saving, and can achieve many benefits at one stroke. What's more, the increment of net imports or reduction of net exports of agricultural virtual water can reduce the groundwater overdraft areas and increase the ecological water consumptions of the river, bringing eco-environmental benefits.

Impacts of foreign water
The virtual water net exports of agriculture and industry under different foreign water supplies are calculated by Equations (2)-(10). The results are presented in Figure 6.
As is depicted in Figure 6, when the Yellow River water and the SNWT water are considered, there are increments on the virtual water net exports of agriculture and industry compared with that under the adjusted same local water supply. When considering the Yellow River water, the agricultural virtual water net exports of five statistical years increased by 4.31%, 22.65%, 16.71%, 1.29% and 6.32%, respectively compared with that under the adjusted same local water supply. Correspondingly, the industrial virtual water net exports increased by 52.02%, 150.19%, 67.08%, 82.07% and 85.43%, respectively. The results show that the Yellow River water impacts industrial virtual water net exports greater than the agricultural virtual water net exports. Subsequently, the impacts of Yellow River water reduction and SNWT water increment on virtual water net exports are studied taking 1997 and 2017 as examples respectively. The impacts of SNWT water increment on virtual water net exports are studied taking 2017 as an example. The results are shown in Tables 1 and 2. From Table 1, the net export of agricultural virtual water was 265.69 million m 3 under the adjusted same local water supply (W 0 S0 ) in 1997. After considering the Yellow River water (W 00 S1 ), the net export of agricultural virtual water increased by 11.46 million m 3 and 4.31% compared with that under W 0 S0 . After considering the virtual Yellow River water (W 00 S2 ), the net export of agricultural virtual water increased by 81.32 million m 3 and 30.61% compared with that under W 0 S0 . The net export of agricultural virtual water reduced by 69.86 million m 3 and 20.13% in W 00 S1 compared with that under W 00 S2 . For industrial virtual water, the net export was 23.57 million m 3 under the adjusted same local water supply (W 0 S0 ) in 1997. After considering the Yellow River water (W 00 S1 ) and virtual Yellow River water (W 00 S2 ), the net exports of industrial virtual water increased by 12.26 million m 3 and 47.62 million m 3 respectively compared with that in W 0 S0 , increasing 52.02% and Uncorrected Proof 202.04% respectively. The net exports of industrial virtual water reduced by 35.36 million m 3 and 49.67% under W 00 S1 compared with that under W 00 S2 . Therefore, it can be seen that the virtual water net exports of agriculture and industry reduced by 20.13% and 49.67% respectively due to the cut-off of Yellow River channel compared with that without a cut-off.
From Table 2, the net export of agricultural virtual water was À1,195.22 million m 3 under the adjusted same local water supply (W 0 S0 ) in 2017. After considering the Yellow River water (W 00 S1 ), the net export of agricultural virtual water increased by 75.56 million m 3 and 6.32% compared with that under W 0 S0 . When considering the quantity of water returned by industry, the net export of agricultural virtual water increased by 90.38 million m 3 and 7.56% under the combined effects of foreign water (W 00 S1 þ W SNWT ) compared with that under W 0 S0 . The net exports of agricultural virtual water increased by 14.82 million m 3 and 1.32% under W 00 S1 þ W SNWT compared with that under W 00 S1 . For industrial virtual water, the net export was 75.63 million m 3 under the same local water supply (W 0 S0 ) in 1997. The net export increased by 64.61 million m 3 and 123.49 million m 3 respectively under the W 00 S1 and W 00 S1 þ W SNWT compared with that under W 0 S0 , increasing 85.43% and 163.28% respectively. The net exports of industrial virtual water increased by 58.88 million m 3 and 41.99% in W 00 S1 þ W SNWT compared with that under W 00 S1 . Therefore, it can be seen that if the SNWT water is consumed by industry, the net export of industrial virtual water will increase by 41.99% compared with that without SNWT water. If the industry returns the corresponding amount of water to agriculture, the net export of agricultural virtual water will increase by 1.32% compared with that without SNWT water.

Impacts of water management policy
Comparable water quotas of agriculture and industry are calculated by Equation (11). The results are shown in Table 3.
It can be seen from Table 3 that the comparable water quotas of agriculture and industry are reducing year by year. For agricultural water quota, it reduced by 24.26% in 2012 compared with that in 2007, and reduced by 31.84% in 2017 compared with that in 2012. For industrial water quota, it reduced by 35.29% and 48.49% respectively. The effects of SNWT water on the comparable water quotas have been removed. Moreover, it is assumed that the water saving level in the two five years is the same. Therefore, if the most stringent water management policy had not been implemented in 2015, the water quotas of agriculture and industry in 2017 should be 20,415.68 m 3 /million CNY and 176.54 m 3 /million CNY, respectively. It can be Uncorrected Proof concluded that because of the implementation of the policy, water quotas of agriculture and industry reduced by 10.01% and 20.39%, respectively compared with that without the policy in 2017. That is, the reductions of virtual water net exports were consistent with that of the water quotas in 2017. In summary, the increments of agricultural and industrial virtual water net exports from 1997 to 2002 are mainly due to the increments of Yellow River water. The increment of agricultural virtual water net exports from 2012 to 2017 is mainly due to the SNWT water, while the reduction of industrial virtual water net export was mainly due to the water management policy.

CONCLUSIONS
In this paper, the virtual water net exports of agriculture and industry are studied under the effects of both nature and human activities over the past 20 years in water-short Shandong Province. Several main conclusions of this research are as follows.
From 1997 to 2017, with the development of economy, an N-shaped trend and inverted U-shaped trend exist with regard to the net exports of agricultural and industrial virtual water, respectively. Agricultural virtual water changed from net export to net import in 2004, and has been net import ever since. Industrial virtual water has constantly been the net export.
Virtual water net imports are beneficial to water environments of water deficient areas, while net export is harmful. The correlation analysis showed that in 2002-2012, for every 1 million m 3 of agricultural virtual water net export was reduced, the groundwater funnel area reduced by 0.38 km 2 . In 2012-2017, for every 1 million m 3 of agricultural virtual water net export was increased, the groundwater funnel area increased by 0.39 km 2 . In 1997-2012, for every 1 m 3 of industrial virtual water export was increased, the wastewater increased by 9.41 m 3 . In 2012-2017, for every 1 m 3 of industrial virtual water export was reduced, the wastewater reduces by 6.82 m 3 . The increment of net imports or the reduction of net exports of industrial virtual water not only reduce the physical water intakes, but also reduce the discharge of industrial wastewater and water pollutions. It is one of the effective measures of industrial water saving, and can achieve many benefits at one stroke. What's more, the increments of net imports or reduction of net exports of agricultural virtual water can reduce the groundwater overdraft area and increase the ecological water consumption of the river, bringing eco-environmental benefits.
From 1997 to 2002, the increments of agricultural and industrial virtual water net exports were affected by the Yellow River water. In 1997, the net exports of agricultural and industrial virtual water reduced by 20.13% and 49.67% respectively because of the cut-off of Yellow River channel. In 2017, if the SNWT water is consumed by industry, the net export of industrial virtual water will increase by 41.99% compared with that without SNWT water. If the industry returns the corresponding amount of water to agriculture, the net export of agricultural virtual water will increase by 1.32% compared with that without SNWT water. Because of the implementation of the policy, the reductions of agricultural and industrial virtual water net exports reduced by 10.01% and 20.39% respectively compared with that without the policy in 2017. Therefore, from 2012 to 2017, the increment of agricultural virtual water net exports was affected by SNWT water, while the reduction of industrial virtual water net exports was affected by water management policy.
In this paper, a new comparative analysis method is proposed. The climates are adjusted into a unified standard, and the impacts of human activities on the virtual water net exports are studied quantitatively under the premise of considering the economic developments and water-saving levels. However, the statistics of Input-Output Tables are only performed on the year with last digits of 2 and 7. Data are relatively less because of the long time interval of five years, but certain trends can be reflected. This paper provides the direction for the quantitative analysis of virtual water trade under the effects of human activitiesforeign water and policy. The reasons of the changes of virtual water trade are explained. It can be concluded that the application of the virtual water tools have led to a higher precision in the quantification and in the coherence and quality of water management decisions than by using classical hydrological evaluation and quantification procedures. From a new perspective, it provides a scientific basis for the decision-making of integrated water management of physical water and virtual water in Shandong Province, as well as other water shortage areas.