Impact of climate change on the blue water footprint of agriculture on a regional scale

In the case study of Tangshan city, Hebei Province, China, this paper analyzes the temporal change of the blue agricultural water footprint (WF) during 1991–2016 and discusses the applicability of different climate change models during 2017–2050. Results show effective rainfall, wind speed and maximum temperature are leading factors influencing the blue agricultural WF. Relative error analysis indicates that the HadGEM2-ES model is the most applicable for climate change projections in the period of 2017–2050. Agricultural blue WF is about 1.8 billion m in RCP2.6, RCP4.5 and RCP8.5 emission scenarios, which is almost equal to the average value during 1991–2016. This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/). doi: 10.2166/ws.2018.046 om https://iwaponline.com/ws/article-pdf/19/1/52/507052/ws019010052.pdf er 2019 Huiping Huang Yuping Han (corresponding author) Dongdong Jia North China University of Water Resources and Electric Power, Zhengzhou 450045, China E-mail: han0118@163.com Huiping Huang Collaborative Innovation Center of Water Resources Efficient Utilization and Support Engineering, Zhengzhou 450046, China Yuping Han Henan Key Laboratory of Water Environment Simulation and Treatment, Zhengzhou 450046, China


INTRODUCTION
Due to the uneven distribution of water at spatial and temporal scales, more than two billion people live in waterstressed areas (Oki & Kanae ). Simulations have shown that 59% of the world's population will face water shortage by 2050 (Rockström et al. ). Agricultural water use currently accounts for ∼70% of freshwater withdrawals both globally and in China. The water footprint (WF) of a product was first defined in 2003 as the volume of freshwater used for production at the place where the product is actually produced (Hoekstra & Hung ) and it consists of green WF, blue WF, and grey WF. In this paper, grey WF is not considered in the WF calculation for crops. Green WF is defined as rainwater consumption that is stored in soil and evaporated during production during crop growth. Blue WF refers to surface and groundwater that is consumed by irrigation and evaporated, and the agricultural blue WF is considered as the theoretical irrigation water requirement.
Climate change is a global issue that may have dramatic impacts on ecosystems, social economics, and agriculture.
Changes in the availability of water, particularly for agriculture, due to climate change have been observed and reported globally (Burn & Hesch ). Temperature increase and precipitation decrease in the Zayandeh-Rud River Basin is expected to cause increasing irrigation water demand from 2015 to 2055 (Gohari et al. ).
Based on crop data and changes in the seasonal timing of water demand, irrigation water requirements in the Guadalquivir River Basin of Spain are projected to increase 15%-20% by the 2050s (Rodríguez-Díaz et al. ). The satellite weather application platform (SWAP) model, global change models (GCM), and regional climate models (RCM) have been used to predict local impacts of climate change on irrigation water demand in Turkey (Yano et al. ). The expected water requirements of global irrigation, without measures to alleviate climate change impacts, may increase ∼20% by 2080 (Fischer et al. ). The impacts of climate change on agricultural water use, especially on irrigation water demand, have been reported in numerous studies in China (Zhang & Cai ). Existing studies have shown that the influence of climate change on regional irrigation water demand varies greatly among different areas (Zang et al. ; Wang et al. ). Research on the effects of climate change on the agricultural blue WF is of great significance for guiding agricultural management to cope with climate change.
At present the climate change model is the main approach to predicting future climate and its associated human response. However, initial and boundary conditions, scenarios, observations, model parameters, and structure of climate simulations all cause some degree of uncertainty in simulation results, and uncertainty is more prominent at the regional scale (IPCC ; Knutti & Sedlácěk ).
Therefore, evaluating the adaptability of climate change models to different regions is essential for climate change model use.
The main objectives of this paper were to: (1) identify key meteorological variables influencing the blue WF of major crops and calculate the relative error of these variables with different climate change models during 1991-2016; and (2) determine an applicable model for a specific region and use it to calculate the WF in 2017-2050. The results will help guide regional water use management.

Site description and data
Located in the east of Hebei, a province in north China, Tangshan covers an area of ∼13,472 km 2 (Figure 1). The city is characterized by a typical mainland monsoon climate with annual precipitation of 644.2 mm. The per capita water resources in Tangshan are 329 m 3 , which are less than 15% of that in China and water resources in the region are in short supply. Agricultural water consumption accounts for about 60% of the total regional water use and crop irrigation water takes up more than 90% in this region.
Daily meteorological data from 1991 to 2016 were acquired from the China Meteorological Data Service Center (http://data.cma.cn/). Parameters included average air temperature, average relative humidity, average wind speed, sunshine hours, and precipitation. Time series of crop yield were taken from annals of statistics for Tangshan.
Data from the coupled model inter-comparison project phase 5 (CMIP5) released by the Earth System Grid Federation were downloaded from the internet with a grid distance of 25 km, and data used in calculations were interpolated to 0.5 degrees with a bilinear interpolation method.

Agricultural blue WF
The green and blue WF of primary crops can be calculated with the following equations (Hoekstra & Chapagain ): where WF green is a crop's green WF (m 3 /kg), WF blue is the blue WF (m 3 /kg), CWU green and CWU blue are green and blue water consumption (m 3 /ha), 10 is a constant to convert water depth (mm) into water volume (m 3 /ha), and Y is crop yield (m 3 /ha).
For a certain type of crop in an irrigation district, consumption of green and blue water for the crop's water requirements during the entire growth period is (Novo where c stands for crop type, q represents area, ET g [c,q,m] and ET b [c,q,m] are the green and blue water of the mth month, CWR is the crop water requirement, P eff is effective precipitation, and CWU b [c,q,t] is the blue water requirement of the tth month, which directly reflects the theoretical need for irrigation. During computation, the CWR is assumed to be equal to crop evapotranspiration (ET c , mm/day), which can be calculated by multiplying the potential reference evapotranspiration (ET 0 , mm/day) and the crop coefficient (K c ).
In this study, ET 0 was calculated using the Penman-Monteith equation recommended by the FAO in 1998. The parameter K c is obtained based on current values in Tangshan City and Hebei Province combined with parameters from FAO-56.

Meteorological yield
Crop yield in any given year is strongly influenced by at least three kinds of factors: meteorological variability, agricultural economic practice, and random noise. Briefly, crop yield can be decomposed as where y represents crop yield, y t denotes trend yield (mainly determined by social productivity), y w is meteorological yield (mainly determined by climatic conditions), and Δy stands for the yield component produced by random elements that can be ignored in the actual calculation.

RESULTS AND DISCUSSION
Applicability of climate change model

Agricultural WF in Tangshan
In Tangshan, the main cultivated crops include soybeans, sorghum, oil crops, cotton, vegetables, tubers, muskmelon, watermelon, winter wheat, tobacco, summer maize, and rice. Among the 12 kinds of crops, the WF of summer maize, winter wheat, oil crops, cotton, vegetables, and rice accounted for more than 94% of the entire agricultural WF of the city. Therefore, to simplify other analyses in this paper, these six crops were selected to represent the major crops of Tangshan.

Agricultural blue WF
The agricultural blue WF is closely related to CWR and effective precipitation, which reflect the combined effect of growth period, precipitation, topography, and meteorological conditions. In Tangshan city, the WF of summer maize is the highest, followed by winter wheat; however, the blue WF of winter wheat exceeds that of summer maize because the growth period of summer maize is consistent with the rainy season (June-September), which makes it possible for some of its water requirement to be satisfied by green water. 1.17 billion m 3 and 2.82 billion m 3 , with an average value of 1.79 billion m 3 . The volume of the blue WF for grain crops including winter wheat, summer maize, and rice was 80% of that for the whole region. Therefore, reducing the irrigation water demand for these three types of crops is critical for water conservation.
Simulation accuracy of meteorological factors with different climate models where RE is the relative error of climatic parameters, x sim is simulated multi-year average data, x obs is mean annual measured data, TE is the total error of climatic parameters, and p i is the proportion of total blue WF for winter wheat, summer maize, rice, cotton, oil crops, and vegetables, respectively. The results are shown in Table 1 ((a)-(e)).

Implications of water resources management
As discussed in this paper, reducing winter wheat and summer maize yield can have a large impact on regional blue WF. In contrast, significant increases in vegetables cause only small changes to the regional blue WF. This is because the blue WF per unit mass of vegetables is only 14.72 m 3 /ton, whereas that of winter wheat and summer maize is 846.4 and 407.8 m 3 /ton, respectively. Therefore, adjustments to planting structure can be implemented to mitigate water scarcity. Tangshan city should plant more vegetables and limit wheat planting. In addition, an interregional virtual water compensation scheme can provide a practical solution for water shortages.

CONCLUSIONS
In this paper, we presented a comprehensive analysis of the effects of climate change on regional agricultural blue WF based on meteorological and economic data, using the city of Tangshan as an example. Winter wheat, summer maize, rice, cotton, oil crops, and vegetables account for 94% of the total agricultural WF, and blue WF contributes 53.08% of the total agricultural WF. This demonstrates that these six crop varieties are major factors in the regional agricultural water resources system. Effective precipitation, wind