Climate change and shifts in cropping systems together exacerbate China's water scarcity

This study selected eight major crops in China, including wheat (Triticum aestivum L.) (winter wheat and spring wheat), maize (Zea mays L.) (spring maize and summer maize), paddy rice (Oryza sativa L.) (early-season, mid-season, and late-season rice), soybean (Glycine max L.), millet (Panicum miliaceum L.), potato (Solanum tuberosum L.), cotton (Gossypium hirsutum L.), and rapeseed (Brassica napus L.). Crop phenology data came from the crop growth and development data set (http://data.cma.cn/data/cdcindex/cid/4ee1c7fce9cd6a5f.html) of the China Meteorological Data Sharing Service System. The crop water demand coefficients (K_c) of different crops were provided by the FAO (www.fao.org/3/X0490E/x0490e0b.htm), and they were calibrated based on the local climatic conditions following the method provided by FAO.

The planted areas of different crops were derived from the county-level crop statistical data for 2850 counties in 1990, 1995, 2000, 2005, 2010, and 2015, provided by the Institute of Agricultural Information, Chinese Academy of Agricultural Sciences (CAAS). Due to the lack of statistical data, the study area did not include Hong Kong, Macao, and Taiwan, and when a county unit lacked a continuous dataset, the county was designated as a no data area.

The IWDs of different crops were calculated using long-term climatic observation data and crop phenological parameters. We collected 30 year daily climate data (1987–2016) from 839 national weather stations in the China Meteorological Data Sharing Service System (http://data.cma.cn), and the dataset included daily average temperature, maximum temperature, minimum temperature, precipitation, average relative humidity, average wind speed, and sunshine hours.

IWD was calculated for each of the eight crops from values of reference crop evapotranspiration, crop water demand coefficients, and effective precipitation during the crop growing seasons.

2.2.1. Reference crop evapotranspiration

The FAO Penman–Monteith equation (www.fao.org/3/X0490E/x0490e08.htm) was used to calculate the daily evapotranspiration of reference crops (Allen et al 1998):

$$\begin{equation}E{T_0} = \frac{{0.408{{\Delta }}\left( {{R_n} - G} \right) + \gamma \frac{{900}}{{{T_{mean}} + 273}}{u_2}\left( {{e_s} - {e_a}} \right)}}{{{{\Delta }}\, + \,\gamma \left( {1 + 0.34{u_2}} \right)}}{\text{ }}\end{equation} \tag{ 1 }$$

where R_n is the net radiation at the crop surface (MJ · m⁻² · d⁻¹); G is the soil heat flux density (MJ · m⁻² · d⁻¹); T_mean is the daily average temperature (°C); u₂ is the wind speed at 2 meters height (m/s); es is the saturation vapour pressure (kPa); e_a is the actual vapour pressure (kPa); Δ is the slope vapour pressure-temperature curve (kPa ·°C⁻¹), and γ is the psychrometric constant (kPa ·°C⁻¹).

2.2.2. Crop water requirement

Daily water requirement for each crop was calculated from reference crop evapotranspiration (ET₀) multiplied by the crop water demand coefficients (K_c) for each day. We followed the methods provided by FAO to calibrate K_c for different growth stages of each crop. The calibrated K_c was then used to construct the K_c curve in order to determine daily K_c. Finally, the ET_c for crop i on day j of growth stage k was calculated as (Allen et al 1998):

$$\begin{equation}E{T_{c,i,k,j}} = {K_{c,i,k,j}} \times E{T_{0,i,k,j}}\end{equation} \tag{ 2 }$$

where ET_c,i,k,j, K_c,i,k,j, ET_0,i,k,j are, respectively, the crop water requirement, crop water demand coefficients, and reference evapotranspiration for crop i on day j of growth stage k. The water requirement in the k^th stage was given as:

$$\begin{equation}E{T_{c,i,k}} = \mathop \sum \limits_{j = 1}^n E{T_{c,i,k,j}}\end{equation} \tag{ 3 }$$

where ET_c,i,k is the crop water requirement during growth stage k; j (1, 2,...,n) indicates the j^th day of stage k; ET_c,i,k,j is the water requirement on day j of growth stage k for crop i.

2.2.3. Effective precipitation

Effective precipitation is defined as the fraction of the total precipitation that is available to the crop and does not run off (Petra and Stefan 2002). In this study, we calculated the effective precipitation during the growth stages of each crop at different sites. The growth stages of different crops were divided according to regional agricultural phenology. By matching crop growth stages with the weather stations in this region, we could obtain a data set of weather stations with corresponding crop growth stages. Equation (4) gives a commonly used method for measuring effective precipitation (Petra and Stefan 2002):

$$\begin{equation}{P_{eff,i,k,j}} = \left\{ \begin{array}{*{20}{c}} {\frac{{{P_{i,k,j}}\left( {4.17 - 0.2{P_{i,k,j}}} \right)}}{{4.17}},{P_{i,k,j}} < 8.3mm/d{\text{ }}} \\ {4.17 + 0.1{P_{i,k,j}},P \ge 8.3mm/d{\text{ }}} \end{array}\right.\end{equation} \tag{ 4 }$$

where P_eff,i,k,j, α_i,k,j, P_i,k,j are, respectively, the effective precipitation, utilization coefficient, and precipitation for crop i on day j of growth stage k.

The effective precipitation for growth stage k was given as:

$$\begin{equation}{P_{eff,i,k}} = \mathop \sum \limits_{j = 1}^n {P_{eff,i,k,j}}\end{equation} \tag{ 5 }$$

where P_eff,i,k is the effective precipitation during growth stage k, j (1, 2,...,n) is the j^th day in this stage, and P_eff,i,k,j is the effective precipitation on day j of growth stage k.

2.2.4. Irrigation water demand

We calculated IWD for each crop (I_i) as (Moseki et al 2019):

$$\begin{equation}I_{i,k}^{{\text{unit}}} = \left\{ \begin{array}{*{20}{c}} {E{T_{c,ik}} - {P_{eff,i,k}},{\text{ }}E{T_{c,ik}} \ge {P_{eff,i,k}}} \\ {0,{\text{ }}E{T_{c,ik}} < {P_{eff,i,k}}} \end{array}\right.\end{equation} \tag{ 6 }$$

$$\begin{equation}I_i^{{\text{unit}}} = \mathop \sum \limits_k I_{i,k}^{{\text{unit}}}\end{equation} \tag{ 7 }$$

where $I_{i,k}^{{\text{unit}}}$ and $I_i^{{\text{unit}}}$ are the IWDs for growth stage k and for the entire growing season of crop I per unit area, respectively.

2.2.5. County-level IWD

By using the 'inverse distance weighted interpolation method' and 'zonal statistics as table' tools in ArcGIS software (www.esri.com/en-us/arcgis/about-arcgis/overview), we obtained the IWD per unit area for each crop in each county. To avoid the unrepresentative impact of extreme weather in a single year on the inter-annual climate change effect, we used five-year average values of climate data rather than single-year values to quantify climate impacts and to calculate the IWD per unit area for each crop. For example, the climate effect or IWD per unit area in 1990 was the average value over the 1988–1992 period, the average value of 1993–1997 was used for 1995, and the same was done for the other years of 2000, 2005, 2010, and 2015. County-level IWD (I) was calculated from the average IWD per unit area and total planted area of each crop in each county of China:

$$\begin{equation}I = \mathop \sum \limits_i I_i^{{\text{unit}}}{P_i}\end{equation} \tag{ 8 }$$

where I is the regional total IWD and P_i is planted area of crop i.

In order to quantify the contributions of various factors to IWD and water scarcity in China, we used LMDI to quantitatively distinguish the different effects of climate change, changes in planted area, and changes in crop mix on IWD. This method originated from Divisia index decomposition, a form of index decomposition analysis (IDA) (Ang 2004). In the 1980s, IDA was used to distinguish the impacts of product structure and energy intensity on electricity consumption of industrial sectors. In recent years, new versions of IDA and the LMDI method have been widely used in the fields of energy and environment (Timilsina and Shrestha 2009, Xu et al 2014, Ang 2015), as well as in agricultural water resource analysis (Zhao and Chen 2014, Zhao et al 2017, Zhang et al 2018, Zou et al 2018). Moreover, due to its simplicity and ability to completely decompose the factors influencing IWD, LMDI can be combined with geographic information systems to identify the contributions of multiple factors at different space-time scales (Fatima et al 2019, Yao et al 2019, Wen and Li 2020).

The LMDI approach was used to analyze the driving factors of changes in IWD across China. The decomposition model of IWD was developed based on published research studies (Xie and Su 2017, Zhang et al 2018, Zou et al 2018):

$$\begin{equation}I = \mathop \sum \limits_i {I_i} = \mathop \sum \limits_i P\frac{{{P_i}}}{P}\frac{{{I_i}}}{{{P_i}}} = \mathop \sum \limits_i P{S_i}{C_i}\end{equation} \tag{ 9 }$$

where I is the regional total IWD; I_i is the IWD of crop i; P is the planted area, which refers to the planting scale or crop spatial extent; S_i= (P_i/P) is the proportion of the total planted area that is planted to crop i, and represents the crop mix. Crop mix is defined and measured as the proportion of each of the eight crops to the total planted area, which reflects the difference in regional IWD caused by different crop types and combinations; C_i = (I_i/P_i) is the IWD per unit area for crop i, which is determined by the water requirement and effective precipitation at different crop growth stages, thus mainly representing the climate change effect (including precipitation, temperature, and other climatic factors).

The LMDI approach includes two kinds of models (i.e. additive decomposition and multiplicative decomposition). We used additive decomposition because of its simplicity and consistency of results (Ang 2015):

$$\begin{equation}{I^T} - {I^0} = {{\Delta }}{I_{tot}} = {{\Delta }}{I_{area}} + {{\Delta }}{I_{mix}} + {{\Delta }}{I_{climate}}\end{equation} \tag{ 10 }$$

where I^T and I⁰ are the IWD for periods T and 0; ΔI_tot is the change in IWD; ΔI_area, ΔI_mix, and ΔI_climate represent the effects of planted area, crop mix, and climate change, respectively. Finally, the effect of each driving factor on the change in IWD was calculated as (Ang 2015):

$$\begin{equation}{{\Delta }}{I_{area}} = \mathop \sum \limits_i {w_i}In\left( {\frac{{{P^T}}}{{{P^0}}}} \right)\end{equation} \tag{ 11 }$$

$$\begin{equation}{{\Delta }}{I_{mix}} = \mathop \sum \limits_i {w_i}In\left( {\frac{{{S_i}^T}}{{{S_i}^0}}} \right)\end{equation} \tag{ 12 }$$

$$\begin{equation}{{\Delta }}{I_{climate}} = \mathop \sum \limits_i {w_i}In\left( {\frac{{{C_i}^T}}{{{C_i}^0}}} \right)\end{equation} \tag{ 13 }$$

$$\begin{equation}{w_i} = \frac{{{I_i}^T - {I_i}^0}}{{In{I_i}^T - In{I_i}^0}}\end{equation} \tag{ 14 }$$

where the superscripts 0 and T denote year 0 and year T.

As shown in figure 1(a), cotton had the largest IWD per unit area, followed by rice, wheat, and maize. Total crop planted area in China increased during the past 25 years, especially from 2005 to 2015 (figure 1(b)). The total planted area increased from 1.02 × 10⁸ ha in 2005 to 1.17 × 10⁸ ha in 2015, with a mean growth of 1.49 × 10⁶ ha per year. There was a notable change in the composition of the planted area. Maize comprised a greater share of total planted area compared with other crops as time went on, increasing from 19.79% in 1990 to 34% in 2015. Moreover, the planted area of maize peaked at 3.98 × 10⁷ ha in 2015, followed by rice (2.9 × 10⁷ ha). Rice also comprised a large share of the total planted area, but its planted area has changed ony slightly in the past 25 years. In contrast, the planted area of wheat declined from 2.96 × 10⁷ ha in 1990 to 2.22 × 10⁷ ha in 2005. Planted area of both potato and rapeseed increased from 1990 to 2015, although the planted areas of both were much small than seen for other crops.

Zoom In Zoom Out Reset image size

From 1990 to 2015, China's total IWD increased by 60.4 × 10⁹ m³, with different trends in different periods (figure 1(c)). Based on the variation pattern for national IWD, we found that the IWD has increased considerably between 2000–2005, 2005–2010 and 2010–2015. These three periods exhibited IWD increases of 18.4 × 10⁹ m³, 28.3 × 10⁹ m³, and 9.7 × 10⁹ m³, respectively. Maize, rice, and wheat accounted for more than 75% of total IWD. Rice used the most irrigation water (averaging 95.7 × 10⁹ m³), followed by wheat (averaging 69.7 × 10⁹ m³). The proportion of IWD attributed to maize increased greatly from 17.3% in 1990 to 32.1% in 2015.

The additive decomposition results for different time periods are shown in figure 2. From 1990 to 2015, changes in planted area, crop mix, and climate change increased IWD by 45.4 × 10⁹ m³, 0.8 × 10⁹ m³, and 10.8 × 10⁹ m³, respectively. The change in planted area had the most important effect, followed by the effect of climate change. However, the effects of different driving factors in different periods were varied. Climate change had negative effects in the periods of 2005–2010 and 2010–2015 (figure 2), with a decrease in IWD of 4.9 × 10⁹ m³, and 4.3 × 10⁹ m³ respectively. In contrast, planted area increased IWD in both periods of 2000–2005, 2005–2010, and 2010–2015, especially in period 2005–2010 (+34.8 × 10⁹ m³). At the national scale, crop mix was a relatively weak driving factor, and exerted different effects in different time periods.

Zoom In Zoom Out Reset image size

The total effect (the combination of changes in planted area and changes in IWD per unit area) of each crop on changes in IWD across China varied greatly (table 1). The effect of maize on the increase in IWD from 1990–2015 was much larger than observed for any other crop, and was followed by wheat and potato. Rice showed large, opposite effects for 1990–1995 (−13.59 × 10⁹ m³) and 2000–2005 (12.77 × 10⁹ m³). During 2000–2005, most crops contributed to an increase in IWD except for wheat and millet. During 2005–2010 and 2010–2015, the most influential crop was maize, which increased IWD in China by 21.20 × 10⁹ m³ and 22.32 × 10⁹ m³, respectively.

As shown in the figure 3, maize cultivation dominated the increased IWD in Northeast China (NEC) and central Northwest China (NWC). Other crops (except for wheat, maize, and rice) contributed greatly to the increased IWD in western NWC and the Tibet Region (TR). In most regions of North China (NC), wheat and other crops contributed to the decreased IWD. Most regions of South China (SC) were dominated by the cultivation of rice, which decreased IWD. The effects of different crop types compensated for each other across different regions, and it was one of the reasons that the shift in crop mix was the least influential of the factors nationally.

Zoom In Zoom Out Reset image size

The spatial distribution of changes in IWD showed a significant difference between northern and southern China from 1990–2015 (figure 4). The IWD increased in the north and decreased in the south. Northern China largely suffered from an intense increase in IWD, while reduction in IWD occurred in large portions of the three southern regions (the Yangtze River Region (YRR), South China (SC), and the eastern part of Southwest China (SWC).

Zoom In Zoom Out Reset image size

Generally, these spatial patterns reflected the dominant effect of changes in planted area, and the secondary effect of climate change and crop mix on the historical changes in regional IWD. Particularly in northern China, planted area effect contributed the most to the increase in IWD in most parts of NEC and NWC (figure 4(b)). Also in the western part of YRR, the negative effect of crop mix was notably offset by the positive effect of planted area. While in the northern part of SWC, it reflected a relatively significant impact of climate change on changes in regional IWD (figure 4(d)). Overall, planted area had the largest effect in most regions of China, and climate change and crop mix played a complementary role in IWD changes.

More detailed information regarding the effect of the three driving factors on interannual changes in IWD is shown in figure 5. In NEC and NWC, the changes in regional IWD were mostly affected by planted area during the periods of 2005–2015 and 1990–2015. The climate change effect in NC resulted in overall increases in regional IWD for all periods except 2000–2005. In contrast, the climate change effect decreased IWD in YRR for almost all periods except during 2000–2005. The IWD was essentially not changed by any of the three driving factors in SC and TR as compared with other regions.

Zoom In Zoom Out Reset image size

This study demonstrated that the planted area expansion of some water-intensive crops, particularly the rapid increase in planted area of maize in northern China, was the main contributor to increased IWD and water scarcity across China. From 1990 to 2015, the planted area of eight main crops increased by 14.7%, which not only contributed to gains in grain production, but also imposed heavy pressures on the environment (Lu and Tian 2013, Huang et al 2017). With continued climatic warming, the planted areas of some warm-season crops may further expand. Prior studies reported that the planting boundaries for paddy rice have tended to move northward (Ye et al 2014), and the northern limits of maize production in NEC have also shifted to the north (Liu et al 2013, Yang et al 2015). Rapid cropping expansion and intensification can increase regional IWD. For example, maize water demand in the West Liaohe Plain of NEC increased by 0.32 billion m³ from 2000 to 2010 due to the increase in maize acreage (Yang et al 2016). In order to mitigate water scarcity, policy makers should properly control planted area and make it match availability of regional water resources.

Our quantitative analysis of the changes in IWD and its driving factors demonstrated that planted area was a major factor that caused unsustainable use of water resources from 1990 to 2015. In the past, there has been a spatial mismatch between food production and water resources in China, i.e. water resources in northern China are scarce, but in the south they are abundant. However, this study found that over the past recent decades, changes in total planted area and crop mix have made the north-south mismatch more serious, thereby undermining China's long-term food and water security. A related study also showed that over the past decades most of the increase in irrigation water consumption in China was concentrated in NC and NWC, and was largely driven by maize expansion (Zuo et al 2018). Similarly, a region-specific study in China showed that the expansion of the total cultivated areas also increased agriculture water consumption (Wang et al 2020). It is strongly suggested that China must mitigate its water shortage problems by properly controlling the scale of some water-intensive crops. These strategies will be essential for promoting adaptation to climate change while ensuring China's food and water security.

Although climate change was secondary to total planted area in affecting IWD, the effects of climate change were regionally important. Additionally, climate change is likely to become more important as potential area for crop expansion will diminish due to finite land area. Besides, large-scale agricultural expansion could also influence the near-surface climate (He et al 2020). In recent decades, the impact of climate change on crop water demand has been investigated from many aspects. As the crop water demand is mainly determined by potential evapotranspiration and the physiological characteristics of crops, the observed climate warming and the various trends of solar radiation, relative humidity, and precipitation have posed great impacts on crop water demand and regional agricultural water resources (Dai et al 2015, Um et al 2020). In addition, there are great spatial and temporal differences in the trends of crop water demand (Xing et al 2016, Wang et al 2017). Therefore, considering the spatiotemporal complexity and uncertainty of the future climate, it is essential for decision makers to develop strategies to mitigate climate-induced water shortages while maintaining food production.

Due to the difficulty of obtaining the actual amounts of irrigation water applied to each crop across China over the study period, this study assumed that the crop water demand was fully met by supplementary irrigation, which may slightly overestimate or underestimate the actual IWD (Ahmadi et al 2019, Elhani et al 2019, Fullana-Pericas et al 2019). The rapid economic growth and urbanization associated with an increasing population will demand greater crop production and IWD (Dalin et al 2015, Xu et al 2015). Improving water use efficiency may help to reduce regional agricultural water consumption (Zhao and Chen 2014). Because the driving factors of changes in IWD were determined based on historical data in this study, further studies may consider possible climate change scenarios in the future, which could provide a more comprehensive reference for agricultural water management.

Due to data availability and research design limitations, some important factors were not accounted for in the current study. For example, interactions between climate, planting area, and crop type; separating the climate change factors into temperature and precipitation; and considering the contribution to ET_c of soil water storage from a previous growth stage when calculating IWD. We expect to address these problems in future research. In particular, accounting for soil water storage may be accomplished by using some process-based crop models such as Apsim, DSSAT, etc.