China’s water–energy nexus: greenhouse-gas emissions from groundwater use for agriculture

China is the world’s largest emitter of greenhouse gases (GHGs) and the agricultural sector in China is responsible for 17–20% of annual emissions and 62% of total freshwater use. Groundwater abstraction in China has increased rapidly from 10 km3 yr−1 in the 1950s to more than 100 km3 yr−1 in the 2000s, such that roughly 70% of the irrigated area in northern China is now groundwater-fed. Pumping of water for irrigation is one of the most energy consuming on-farm processes; however, to date this source of GHG emissions in China and elsewhere has been relatively neglected. We derive the first detailed estimate of GHG emissions from groundwater pumping for irrigation in China, using extensive village survey data from 11 provinces, broadly representative of the situation during the mid-2000s. The 11 provinces cover roughly half of China’s irrigated cropland and we upscale to the national level using government statistics for the remaining 20 provinces. Our results show emissions of 33.1 MtCO2e, just over half a per cent of the national total. Groundwater abstraction represents an important source of GHG emissions that has been rapidly increasing and which at present is largely unregulated. Water scarcity in China is already driving policies to improve water conservation. These results suggest that significant potential exists to promote the co-benefits of water and energy saving in order to meet national planning targets.


Introduction
China faces its own 'perfect storm' as rapid economic transition drives increasing per capita demand for water, food and energy with far-reaching environmental consequences. Food production in China depends fundamentally on water because of the extensive use of irrigation. China is the world's second largest irrigator with agricultural water use accounting for 62% of total freshwater use (Ministry of Water Resources 2009). Irrigated agriculture has been essential in sustaining growth in national crop production, but multiple drivers are generating immense pressure on water resources and water supply for irrigation, especially in northern China. This has greatly promoted groundwater abstraction which has increased from roughly 10 km 3 yr −1 during the 1950s to more than 100 km 3 yr −1 in the 2000s , Shah 2009). Today, roughly 70% of the irrigated area in northern China is groundwater-fed  and irrigation produces approximately 70% of total grain production (Amarasinghe et al 2005). Studies on direct energy use in farm operations show that pumping water for irrigation is one of the highest on-farm energy consuming processes (e.g. Lal 2004, Mushtaq et al 2009. Whilst China is the world's highest emitter of greenhouse gases (GHGs), per capital emissions are low relative to OECD country averages. The agricultural sector in China is responsible for roughly 17-20% of annual GHG emissions (Wang et al 2010) of which an unknown proportion originates from groundwater pumping for irrigation. In spite of the rapid growth in groundwater use there has been no detailed attempt to quantify the related energy use and GHG emissions of groundwater abstraction. Indeed, the broader 'energy for water' dimensions of the water-energy nexus are under-recognised and poorly quantified (Rothausen and Conway 2011). China's 12th Five Year Plan (2011-5) aims to increase irrigation water use efficiency by 3% by 2015 and emphasizes the importance of improving groundwater resource management to control over-exploitation. However, these targets are to be achieved whilst increasing total grain production by roughly 13% to 450 million tonnes and decreasing national energy consumption per unit of GDP by 16%. Response measures will need to be both water and energy efficient to ensure consistent policy outcomes.
Research on climate change in China suggests that the interactive effects of climate change and other socio-economic drivers could lead to substantial changes in total production by the 2040s (Xiong et al 2009). Water availability plays a particularly significant role in limiting potential crop production, due to the combined effects of higher crop water requirements and increasing demand for non-agricultural use of water. Increasing frequency of extremes in heat and water stress have also been shown to induce higher crop failure rates (Challinor et al 2010). Successful adaptation policies will be essential in order for China to produce enough to keep pace with per capita growth in demand and the effects of other drivers such as land use change. However, production oriented policy goals may often fail to consider wider issues of sustainability, such as the intensity of fossil fuel and water use required to meet specific targets.
In this study we use extensive survey data to develop a detailed method to estimate the GHG emissions from groundwater pumping for irrigation in China. Section 2 describes the data and methods and section 3 presents the results and a sensitivity analysis of provincial and national emissions. Section 4 compares our results with those from other countries and considers other energy uses for agricultural water in China. We end by discussing the wider context of agricultural water management and the challenges and opportunities for reducing both energy and water use.

Methodology
Our method comprises several steps (figure 1). First, we estimate energy use and associated GHG emissions from groundwater abstraction for irrigation in 11 surveyed provinces. Then, we upscale the survey results and calculate the emissions from the remaining 20 provinces of China to obtain a national estimate of GHG emissions. The methods and data sources for each step are described in detail as follows.

The 11 surveyed provinces
We use extensive survey data collected in rural China by the Centre for Chinese Agricultural Policy (CCAP), of the Chinese Academy of Sciences. CCAP conducted several large field surveys during the period between 2004 and 2006, covering more than 500 randomly selected villages in 11 provinces, representing the main groundwater using provinces. Sample screening for villages with no tubewells and quality control checking for survey errors of groundwater levels and pump lifts reduced the number of villages to 366 for our analysis. Figure 2 shows the location of the 366 villages and the 11 provinces. The surveys covered population, cultivated area, groundwater and surface water irrigated area, the number of deep and shallow tubewells and their corresponding pump lift, groundwater level, power sources of pumps (electricity or diesel) and socio-economic characteristics of the villages. Full details of the surveys and sampling techniques can be found in Wang et al (2006Wang et al ( , 2007Wang et al ( , 2009).

Energy use rate.
The main components of energy use associated with irrigated agriculture are primarily related to processes required to apply water to crops in the field by lifting, conveying, and in some cases, pressurizing it (Rothausen and Conway 2011). Lift relates to the vertical distance over which water is raised prior to application and is the most crucial factor but energy use varies both with the source of water (groundwater or surface water) and the application method (i.e. whether it is gravity fed or pressurized irrigation system). The efficiencies of the power generation and supply, the pump and pipeline system also influence energy use. Here we focus on energy for abstraction of water and for this purpose we apply a basic theoretical physical relationship, which prescribes the energy required to lift 1 m 3 of water (with a density 1000 kg m −3 ) up 1 m at 100% efficiency is 0.0027 kWh (see equation (1), Rothausen and Conway 2011) Energy (kWh) = 9.8 m s −2 × Lift (m) × Mass (kg) 3.6 × 10 6 × Efficiency (%) .
Given a lack of detailed knowledge of the efficiency of the individual pumps, we select efficiencies according to available information on pumping efficiency, Chinese conditions and studies of energy use for irrigation pumping in India where the pumping efficiencies are around 30% (Nelson et al 2009, Shah 2009). Overall efficiencies in diesel powered pumping systems result from the compounded efficiencies of the diesel engine (typically 15-40%), transmission (60-100%), pump (40-80%) and pipes (30-95%), giving an overall efficiency of 0.5-27% (Fraenkel 1986), for which we have selected the approximate mid-point of 15%. With higher efficiencies of electrical motors of 75-85% (Fraenkel 1986, Kay 2008 and the direct coupling of motor and pump within down-hole submersible pumps, higher overall efficiencies of 30-60% are achieved within electrical pumping systems, for which we have used 40%. The overall efficiency of using electricity to power pumps is reduced further by inefficiencies in the transmission and distribution (T&D) of electricity. There is some disagreement about the T&D losses in China but according to three studies of the rural electricity network it is between 10 and 19% (Li et al 2007, Sun 2006, Xia 2003   (2) 2.1.3. Agricultural water use. Detailed and/or comprehensive information on actual water use for irrigation (including how much is from groundwater) for all provinces in China are unavailable. The most appropriate data are the provincial level statistics of total agricultural water use reported in the China National Statistical Yearbook (National Statistical Bureau of China 2006). Although these data include all agricultural uses of water, the field survey data showed there is very limited use of water for non-irrigated agricultural production such as livestock (less than 5%). These data for the year 2005 are the best available and accurate enough for the purpose of our study. To determine the proportion of total agricultural water derived from groundwater we multiply it by the percentage of groundwater irrigated land relative to the total irrigated land across all of the surveyed villages, in each province. This allows a calculation of GHG emissions from groundwater pumping for irrigation in each of the 11 surveyed provinces.
To determine GHG emissions in the 20 non-surveyed provinces we extrapolate pump lift (the most essential factor for energy use, see equation (1)) from observations of groundwater levels for different aquifer types for year 2005, from unpublished data and published data in the China Groundwater Level Yearbook from GEO-Environmental Monitoring Institute (China GEO-Environmental Monitoring Institute 2006). Pump lift is greater than groundwater level because pumping causes a localized lowering of the aquifer groundwater level. To define a relationship between groundwater levels and pump lifts, a linear regression of these variables is used based on the data from the 366 surveyed villages. The correlation between groundwater level (x) and average pump lift (y) is shown in equation (3) with the pump lift being an average of 20 m deeper than groundwater level (supplementary figure S1 available at stacks.iop.org/ERL/7/014035/mmedia). This relationship was used with the groundwater level observed in the 20 non-surveyed provinces to generate an estimate of pump lift.

National GHG emissions
This step involves simply adding together the emissions from the surveyed and non-surveyed provinces to obtain a national estimate of total GHG emissions from groundwater use for irrigation.

Results
The Energy use rates and GHG emission rates are shown in figures 3(d) and (e), respectively, with details of the 11 surveyed provinces presented in table 2. The GHG emission rates reflect the mix of pump lift and power source so that the provinces show considerable differences in their rates, with a range of 0.15-0.60 kgCO 2 e m −3 . Shaanxi has the highest energy use rate of 0.64 kWh m −3 which is directly related to the greater pump lift. In Henan, however, roughly two-thirds of pumps are diesel powered causing its GHG emission rate to drop from fourth equal lowest energy use rate (0.30 kWh m −3 ) by province to second lowest GHG emission rate (0.18 kgCO 2 e m −3 ) highlighting the importance of the power source (see sensitivity analysis below). Overall, the 11 surveyed provinces emit a total of 24.13 MtCO 2 e with Hebei the largest contributor by far, accounting for 27.4% of the total, due to its mix of high proportion of deep tubewells (83 m pump lift) and high agricultural groundwater use (see figures 3(a)-(c) and 2 for location). In contrast, Shandong, with higher agricultural groundwater use but no deep tubewells, produces almost half the emissions of Hebei (table 2). Table 3 presents the results for the 20 non-surveyed provinces. Despite a 100 km 3 larger total agricultural water use, the total GHG emissions are 8.97 MtCO 2 e, much less than the 11 surveyed provinces. This is because the groundwater use in the 20 non-surveyed provinces is nearly three times lower than in the 11 surveyed provinces. The average emission rate is almost 30% lower than the surveyed provinces due to the shallower groundwater levels in the non-surveyed provinces (primarily located in southern China, see figure 3(b)). The total national GHG emissions from groundwater use for agriculture in China are 33.1 MtCO 2 e, around 3% of agricultural emissions and 0.58% of national total emissions (5670 MtCO 2 e, National Statistical Bureau of China 2006). Figure 3(f) shows the concentration of highest GHG emissions in provinces located in north and north-east China.

Sensitivity analysis
We explore the sensitivity of our results to varying assumptions about efficiency and pump lift. Figure 4 shows the average GHG emission rate for the 366 surveyed villages with a range of values. Our estimate uses 85% T&D efficiency, 40% electric pump efficiency and 15% diesel pump efficiency,   , it is likely that in many areas pump lifts will increase. An average 5 m deeper pump lift would increase total GHG emissions by almost 10%. Yet figure 4 shows that changes in the efficiency of electric pumps have the largest effect on GHG emission rates. The GHG emissions would more than double with a 20% decrease in electric pump efficiency, whereas a 20% increase in electric pump efficiency will reduce the emission rate by a third. Changes in diesel pump efficiency have the least effect. Accurate data on pump efficiency and its influencing factors will be important for estimating GHG emissions and appropriate pump selection and regular maintenance crucial for limiting emissions.

Comparison with other studies
National estimates of energy use and GHG emissions from water delivery for irrigation are, to our knowledge, only available for India and the US. India is the world's largest irrigator and has experienced even greater development of groundwater abstraction than China, with the area of groundwater irrigation quadrupling since 1950 (Shah et al 2004). Emissions from groundwater pumping for irrigated rice were estimated to be 58.7 MtCO 2 e in 2000 (Nelson et al 2009

Other energy requirements for agricultural water use in China
Irrigation in China relies heavily on surface water from rivers and reservoirs which sometimes also require energy use for pumping and conveyance. However, information on this is sparse and to our knowledge no quantification of aggregate energy use exists. Currently, power equipment for pumping surface water resources in China has a capacity of 22 million kW (Irrigation and Drainage Center 2007), and the related energy use is likely to be considerable. Imbalances in local and inter-province water resources distribution and demand have exacerbated problems of water scarcity and enhanced the energy intensity of water supply. For example, China has many large-scale inter-basin water transfers. Out of the seven established, two of them include a total of 24 pumping stations to irrigate 2.89×10 6 ha (Shao et al 2003). A further seven projects are planned, including the south to north water transfer scheme. Four of the proposed projects are to include a total of 37 pumping stations to irrigate 5.80 × 10 6 ha (Shao et al 2003). The construction and operation of this water supply infrastructure will require energy as will water quality treatment where this is necessary. While the abstraction/supply of water is proportional to the lift and distance over which it is pumped/transported, the energy required to apply water to fields is dependent on the irrigation method/technology. Across the largest proportion of irrigated land in China water is applied as flood/gravity irrigation, yet water conservation policies are promoting the introduction of pressurized irrigation systems (e.g. sprinkler, drip) which require energy to operate. These technologies generally provide a larger degree of control and depending upon other factors can improve water use efficiency considerably, so that any additional energy use may be offset by water savings. However, only 3.5% of the irrigated areas in our surveyed villages were equipped with pressurized irrigation systems and there is little evidence for widespread or effective adoption of water-saving technologies in China (Lohmar et al 2003). Many argue that a lack of economic incentives to save water is a barrier to the uptake of water-saving practices (e.g. Blanke et al 2007), while others emphasize that problems with water rights (e.g. institutional and policy dimensions of ensuring stability of supply and sufficient access) are also critical (Webber et al 2008). Nevertheless, potential exists to realize co-benefits in water and energy savings through improved irrigation technology and this opportunity may be under-recognised in the design and cost benefit analysis of water-saving practices and policy in China and elsewhere (Rothausen and Conway 2011).

A future outlook
Agriculture in China is by far the largest water user with irrigated agriculture generating around 70% of total grain production, yet the national crop deficit was 11% in 2000 (Amarasinghe et al 2005). With a growing population and rising living standards, the demand for food, particularly animal products, is increasing. Shifts in food consumption patterns and population growth may lead to additional unconstrained requirements of 407-515 km 3 water for food production by 2030, compared with 2003(Liu and Savenije 2008. Such large increases would likely induce further water resource deterioration and higher energy intensive supply. Intensification and expansion of irrigated agriculture is often seen as necessary to meet future demand and to achieve growth in the agricultural sector and such measures will require more water and more energy for abstraction, transportation and application to crops. Given China's continued rapid economic transition, the demand for water in the urban and industrial sectors is also likely to increase (Kahrl andRoland-Holst 2008, Xiong et al 2009). Re-allocation of already scarce water resources from agriculture to other sectors with higher output/water use ratios may become increasingly necessary (Kahrl and Roland-Holst 2008). During the past 60 years, with expansion of industrialization and urbanization, the share of water allocated to agriculture has declined from 97% in 1949 to 62% in 2009 (Ministry of Water Resources 2009). During this process, increasing (and sometimes unsustainable) exploitation of groundwater resources has played an important role in sustaining irrigated agriculture. Between 1978 and 2003, the number of groundwater tubewells across China more than doubled  and with increasing privatization of tubewells it is likely that development will continue to exacerbate declining trends in groundwater levels. Groundwater utilization in northern China is over 70% of known resources in the region. In contrast, 70% of all China's groundwater resources are located in southern China, where less than 30% of the resources are currently in use . Drought in parts of southern China (e.g. Qui 2010) may promote groundwater use to secure yields, and multi-climate model averages project drier conditions in southern China out to the 2050s (Li et al 2011). These pressures point towards further groundwater development across China with implications for energy use.
Strategic reform of water resources allocation has been initiated by the Chinese Government with attempts to regulate agricultural water use through a water use rights system . The system is designed to gradually reduce surface water quotas for farmers to transfer agricultural water to other sectors. However, studies show that farmers compensate the 'loss' of surface water by using more groundwater to maintain production practices and yields (Zhang 2007, Pittock 2011). Regulation of groundwater use is well known to be difficult (e.g. Shah et al 2003). These conditions have promoted a shift in policy from supply-side measures towards more water 'efficient' agriculture and irrigation technology innovations. China is to an increasing degree investing in water-saving technologies for farming. Programmes under China's Agenda 21: White Paper on Population, Environment and Development and 'Dryland Farming' are major initiatives that provide support for selected research facilities which target increasing water utilization efficiency by 20-30% (PRC 1994). Various approaches to improve regulation and compliance are also being tested as part of water-saving programmes (World Bank 2009). Since 1991, the water use intensity of Chinese agriculture has decreased, as agricultural investments in channel lining, field levelling, pressurized irrigation systems etc have reduced inefficiencies and water losses (Wu et al 2010).
Yet, many of these technologies require energy and, pressurized irrigation systems especially, are often coupled with groundwater use. It will be a challenge to harmonize water conservation with the 12th Five Year Plan targets of reducing energy and carbon intensity by 16% and 17%, respectively (Hannon et al 2011). Water-saving technologies need to be assessed in relation to energy trade-offs and potential co-benefits. Our sensitivity analysis highlights the importance of optimization of the efficiency of electric pumps as a promising way of decreasing energy use (section 3.1). Research from India also shows widespread low efficiencies of agricultural pumps, with significant potential for economic (and hence energy) savings through improved standards (Sant and Dixit 1996). The use of renewable energy sources for pumping and application of water should also be explored, although projections of renewable energy dissemination in India show that water pumping technologies may not reach their maximum potential even after 25 years (Purohit and Kandpal 2005).

Conclusion
This first detailed estimate of GHG emissions from groundwater pumping for irrigation in China is 33.1 MtCO 2 e for the mid-2000s, 0.58% of total national emissions (in 2006). Groundwater pumping for irrigation alone accounts for roughly 3% of the total emissions from agriculture in China. To date this source of GHG emissions has been relatively ignored; our results highlight their significance in relation to national GHG emissions and underline the importance of energy use for groundwater irrigation. The method used extensive primary survey data originally designed for understanding agricultural water use to upscale from village to province to national level. In exploring this aspect of China's water-energy nexus it is clear that government monitoring systems and data sets do not capture energy use very well. This is also the case for surface water use such as inter-basin transfers. Groundwater abstraction in China has grown rapidly since the 1950s, and without intervention, continued growth could lead to substantial increases in energy use and GHG emissions. Sensitivity analysis of factors affecting energy use in abstraction highlights the importance of pump efficiency, suggesting potential for interventions to target pump set standards. Water scarcity in China is already driving policies to improve water conservation and the results obtained here suggest that quantification and promotion of the co-benefits of water and energy savings could be considerable.