Impacts of temperature and precipitation on the spatiotemporal distribution of water resources in Chinese mega cities: the case of Beijing

Water shortages in China have hindered development of mega cities, especially Beijing. Assessing the impact of temperature and precipitation on water resources is important. This study analyzed spatiotemporal variations and impacts of temperature and precipitation on water resources in Beijing from 1956 to 2013, using statistical and spatial analysis. The results showed the following. (1) Temperature and precipitation affect water resources variously from region to region; their correlation in mountains is lower than in other areas. Precipitation redistribution caused by terrain reduces water resources. (2) The inter-annual variabilities of precipitation, temperature and water resources are different among five water resource divisions. Because of ‘rain-slope’, Beisanhe’s precipitation is larger than others; Yongdinghe’s precipitation is less than others due to ‘rainshadow’; suffering from urban heat island effect, Beisihe and Daqinghe-plain’s temperature is higher than others; Beisanhe and Beisihe’s water resources are greater than others due to area differences. (3) Water resources are positively correlated with precipitation and negatively with temperature. (4) In recent years, precipitation and water resources decreased and temperature rose. Population growth, land use/land cover change, urbanization and pollution affected precipitation, temperature and water resources. Imported water cannot completely solve water shortages. With increasing water demand, precipitation and temperature will significantly influence water resources in Beijing. 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/wcc.2017.038 s://iwaponline.com/jwcc/article-pdf/8/4/593/239551/jwc0080593.pdf Pengpeng Jia Dafang Zhuang Yong Wang (corresponding author) State Key Laboratory of Resources and Environmental Information System, Institute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China E-mail: wangy@igsnrr.ac.cn Pengpeng Jia University of Chinese Academy of Sciences, Beijing 100049, China


INTRODUCTION
Many mega cities in China have suffered from severe water shortages, hindering the development of regional economies and sustainable development (Tian et al. ).
In recent decades, many scholars have studied the impact of climate change on water resources.As climate change is becoming more prominent, associated with changing precipitation (Christensen et al. ), analysis on the impact of precipitation on water resources has become a main direction of research (Cui & Li ).Using long time series data to analyze trends and cycles of precipitation and water resources is the basis for research on the impact of precipitation on water resources (Arnell et al. ).Water resource yield and distribution are affected as a direct result of fluctuation in intensity, frequency and distribution of precipitation (Ligaray et al. ).But the effect of climate change on water resources is a complex process (Rochdane et al. ).It is insufficient to analyze the impact of precipitation on water resources so as to have an in-depth understanding of the impacts of climate change on water resources.Moreover, the length and the accuracy of historical data limit research on changes in precipitation and water resources (Labat et

al. ).
There is an increasing consensus that changing climatic In this study, the cumulative departure method, linear regression analysis, correlation analysis and spatial analysis were used to analyze spatial and temporal variations.We evaluate the impact of changes in temperature and precipitation on water resources, using 1956 to 2013 as a reference period.The results of the study can provide data support and method references to ease water resource pressure in Beijing and provide management plans for future development of urban water resources.

Method description
In this study, the cumulative departure curve was used to analyze inter-annual variations in precipitation; the binary standardized linear regression method was used to establish the quantitative assessment model for the impact of precipitation and temperature on water resources; the population carrying capacity of water resources was used to evaluate the interaction between population scale and water resources; the inter-annual variation and spatial variability of precipitation, temperature and water resources in WRD_3rd were studied using geographic information system (GIS) spatial analysis methods as shown in Table 2.

RESULTS AND ANALYSES
Inter-annual variability analysis seasons alternate (Figure 3).1956 to 1960 was a relatively wet period in Beijing; mean annual precipitation was significantly greater than the multi-year mean precipitation.After 1960, the amount of precipitation dropped sharply, followed by continuous drought.The severe drought in the 1990s was

Methods description Calculation formula Explanation
Cumulative departure curve is a method to judge the trend of change from curve directly For time series X, the accumulated anomaly at time t is X t .By calculating n times' cumulative departure values Standardized regression coefficients are used in multiple regression to compare the importance of variables The carrying capacity of water resources depends on the total water resources, water supply capacity coefficient and per capita water consumption, three indicator variables     This may be the main reason for variation in water resource yield among the five WRD_3rd.

Impact and correlation analysis
Inter-annual variability analysis

Regional variability analysis
In this study, annual precipitation, temperature and water resource variations in different regions of Beijing were analyzed and their correlation coefficients were calculated, using data from WRD_3rd.The correlation coefficient of precipitation, temperature and water resource is greater than that of precipitation and water resource (Table 3). Although

Quantitative evaluation model
Precipitation and temperature have a great influence on water resources.Therefore, to maximize the fitting of precipitation and temperature impact on water resources, this study established: a unitary standard linear regression Equation ( 7) to study the relationship between precipitation (P) and water resource yield (W ); and a binary standard linear regression Equation ( 8) to study the correlation between precipitation (P), temperature (T ) and water resource yield (W ).
Analysis of the correlation coefficients (0.89, 0.91) from these two equations found that the binary linear fitting Equation ( 8) is better than the unitary Equation ( 7).
Therefore, in this study, the water resource yield was used as a dependent variable and precipitation and temperature were used as independent variables to establish a quantitative evaluation model of water resources.The model is used to analyze the impact of precipitation and temperature on water resources.Although both changes in temperature and precipitation will affect water resource yield, the impact of precipitation is much greater than temperature.

Model validation
In order to verify the accuracy of the quantitative evaluation model for the impact of precipitation and temperature on water resources, mean annual precipitation and temperature data from 1956 to 2013 was used to fit water resources (Figure 15).The fitted value of water resources and its trend are similar to the actual conditions, except for 1985 and 2006.The relative error over 57 years is less than 6%, except for 1984 (Table 4).Therefore, it is clear that this model has better accuracy.

DISCUSSION Correlation differentiation under different spatial scales
In this study, urban functional regions of Beijing were used as a minimum research unit (Table 5) and results showed that under the filter effect of human activities, such as administrative boundaries, land cover types, and urbanization, water resource is highly correlated with precipitation and temperature.However, there is no significant difference among these regions.To sum up, the influence of human activities masks differences across regions.Therefore, it is illogical to use the data of urban functional regions or administrative divisions to study correlations between precipitation, temperature, and water resources in different regions and different spatial scales in Beijing.In addition, results will be accurate if natural water resources divisions or basins are used to calculate these correlations.

Influence of water quality on water resources
In the exploitation of water resources, water quality is one of the important indexes.According to the application of water resources, water quality is divided into five categories: I, II, and III are suitable for drinking, while IV and V are suitable for industry, agriculture and ecology (ZHB ; ZHB ).
By analyzing the proportion of I, II, III water yield in the evaluated water resources, we can find the change in the amount of available drinking water.The analysis of the proportion of I, II, III water yield in the evaluated water resources of Beijing (both surface water and groundwater) from 1999 to 2014 (Figure 16) showed that I, II, III groundwater yield is less than 50% of the evaluated groundwater and the proportion has slight variation.
Combined with the change of total groundwater resources in Beijing (Figure 5), it can be found that the groundwater resources reduced sharply in total amount and the groundwater resources suitable for drinking amounts to less than half of the total amount.As can be seen from Figure 16,   between them has grown in the last 15 years (Figure 17).6), which accounted for nearly 30% of the total water resources that year in Beijing.The project rapidly increased the water resources population carrying capacity in Beijing, but with increasing water consumption it will stop or even decline again.Therefore, the project can only temporarily solve the problem; it is difficult to completely solve the water resource problem in Beijing.
Since the 1980s, global climate change predictions and their impact on water resources have become an unavoidable issue.The IPCC (Intergovernmental Panel on Climate Change) released its fifth assessment report, Climate Change 2013: The Physical Science Basis, in Paris.According to the 2013 IPCC report, from 1880 to 2012, the average land-ocean surface temperature trended linearly upward, increasing by 0.85 W C. The linear temperature rate increase in the most recent 50 years is about 0.13 W C per 10 years (IPCC ).The climate change trend in China is consistent with global climate change.Since 1913, the average surface temperature in China has risen by 0.91 W C and the temperature rise has been particularly evident in the last 60 years, with an average annual increase of about 0.23 W C, which is almost twice as fast as the global average (Guo et al. ).In addition, with abundant water resources, but lower per capita availability, China is included in the most water-stressed countries in the world.
trends, especially temperature and precipitation, can change the hydrological cycle, which influences water resources (Jordan et al. ).Selecting appropriate hydrological models based on specific conditions of the region to simulate water resource yield, and quantitatively assess the impact of climate change on water resources in combination with measured data, is the main method to study climate change impacts on water resources (Emam et al. ; Mahmood et al. ).The main reason for declining water resources is increasing temperature and decreasing precipitation (Kundzewicz et al. ).However, the impact of temperature and precipitation on water resources may vary from region to region (Emam et al. ; Mahmood et al. ).Hence, the analysis of the variable impact of climate change on water resources between regions is the focus of current research.The hydrological cycle of a region is mainly influenced by regional terrain, climatic conditions, land use/cover and population scale (Kundzewicz et al. ).The speed at which temperature, precipitation and land use/cover have changed has accelerated as a direct result of population growth and human activities, thereby affecting the interception, infiltration, and evaporation process of the hydrological cycle in the spatial and temporal domain (Madhusoodhanan et al. ; McGrane ).Research has evaluated the impact of climate change on water resources, but has not been comprehensive enough (Rochdane et al. ; Ligaray et al. ).Numerous studies might not effectively reflect spatial and temporal distribution of water resources and cannot reflect spatial correlation; nevertheless, spatial attribute is the most important feature of geographical elements such as water resource divisions and basins.Geographic information system technology is used to collect, store, manage, analyze, display and describe the geographic distribution data on the surface of the earth, using computer system support (Longley et al. ).It is an important method to analyze water resource differences in distribution and spatial correlation, which has the unique advantage in describing temporal and spatial variations in water resources and analysis of basin spatial differences.
Analysis of 57 years of annual mean precipitation data from 1956 to 2013 in Beijing (Figure 2) shows that the maximum was 1,404.6 mm in 1959 and the minimum was 261.4 mm in 1965.Since the 1960s, precipitation has been declining, and since the 1990s the rate of decline has accelerated significantly.Linear regression show that the downward trend of precipitation in Beijing was significant, and the decreasing range was about 37.4 mm per 10 years.There were obvious cyclical changes in Beijing precipitation as the wet and dry development of a dry spell that began in 1960.After the 1960s, the average precipitation had a negative anomaly, which may be the main reason for the continuous Beijing water shortage (Song et al. ; Zhang & Xia ).According to the annual mean temperature data of Beijing from 1956 to 2013 (Figure 4), the highest mean annual temperature was 14 W C in 2007, the lowest was 10.5 W C in 1956, and since the 1960s, temperature has been on the rise.Especially since the 1980s, the rate of increase has accelerated.Linear regression shows that the temperature in Beijing increased significantly, with an uptrend of about 0.4 W C per 10 years, which is higher than the recent 50 year global trend of 0.13 W C per 10 years.This shows that Beijing has a sensitive response to global warming.In addition to the large-scale climate background such as global warming (Gao & Fu ), underlying surface changes caused by urbanization, artificial waste heat, excessive greenhouse gas emissions and other human activities are the main reasons for the temperature increase in Beijing (Xu ; Zhang et al. ; Lin & Yu ; Mehta et al. ).Especially after reform and opening, Beijing has entered a period of rapid development; cultivated land is occupied by urban construction land; and large-scale land cover changes are the main reasons for the rapid temperature increase in Beijing since the 1980s.Water resource yield Water resources yield is the sum of surface streamflow and precipitation infiltration recharge.Based on water resource yield data in Beijing from 1956 to 2013 (Figures 2 and 4),

Figure 1 |
Figure 1 | Overview of the study area.
Figure 2 | Annual mean precipitation, water resource yield and linear fitting (LF) from 1956 to 2013.

Figure 5 |
Figure 5 | Inter-annual variation of groundwater depth in plain area from 1960 to 2013.

Figure 7 |
Figure 7 | Annual mean temperature in mountainous area (AMTIM), annual mean temperature in plain area (AMTIP) and linear fitting (LF) from 1989 to 2013.

Figure 8 |
Figure 8 | Annual mean water resource yield in mountainous area (AMWIM), annual mean water resource yield in plain area (AMWIP) and linear fitting (LF) from 1989 to 2013.

From
Figure 10 | Temperature spatial distribution of five WRD_3rd from 1989 to 2013.

Figure 11 |
Figure 11 | Water resource yield spatial distribution of five WRD_3rd for the period 1989-2013.

Figure 13 |
Figure 13 | Annual mean temperature, per capita water consumption and quadratic polynomial fitting (QPF) from 1979 to 2013.
precipitation is the only source of water recharge, temperature affects the evaporation rate of water resources and other consumption processes.Therefore, it is better to analyze the binary impact of precipitation and temperature on water resource yield.Influenced by land cover types and the hydrological cycle of WRD_3rd, the correlation coefficients between precipitation, temperature and water resources are different in different areas.The correlation coefficients of BSIH and BSH are 0.93 and 0.96 (p < 0.01) while the YDH correlation coefficient is only 0.54 (p ¼ 0.2).Overlay analysis between urban functional regions and spatial correlation of precipitation, temperature and water resources in WRD_3rd (Figure 14) shows that precipitation, temperature and water resources in the regions of new urban development and ecological conservation development are moderately correlated, while in the regions of inner city and urban expansion, they are highly correlated.In addition, in the mountainous regions of western Beijing, precipitation, temperature and water resources are moderately correlated.Spatial distribution is closely related to factors such as altitude, slope, and land cover.Despite fluctuations in precipitation and temperature in mountainous regions, redistribution of precipitation caused by terrain reduced water resources where streamflow gathered inconveniently (Xu ).Forest land cover in the mountains influenced infiltration, evaporation and water resources.Therefore, it led to spatiotemporal changes in water quantity and quality (Hua et al. ; Zhu ; Markovic & Koch ; Cai et al. ).Terrain and forestland in mountains changed hydrological processes and water balance, affected the amount of surface water and groundwater, and decreased the correlation between precipitation, temperature and water resources.

Figure 14 |
Figure14| Overlay analysis between urban functional regions and spatial correlation of precipitation, temperature and water resources in WRD_3rd.

Figure 15 |
Figure 15 | Actual and fitted values curves of water resource from 1956 to 2013.
the proportion of I, II, III surface water yield (river, lake and reservoir) in the evaluated water resources decreased year by year and the rate of decline accelerated after 2005.Combined with the change of total surface water resources in Beijing (Figures2 and 4), it can be found that the proportion of surface water resources suitable for drinking is decreasing while the total amount of surface water resources decreases continuously.Therefore, with the decrease of total water resources, the deterioration of water quality is also an important factor that leads to the shortage of water resources in Beijing.Influence of population scale variations on water resources The resident population of Beijing has increased dramatically from 1979 to 2013 (Figure 17); by the end of 2013, the resident population reached 21.148 million.Under the influence of economic development and urbanization, population growth in Beijing has directly led to an increasing demand for water consumption.In addition, with technological and industrial development, water demand is also increasing (Tong ).However, water resources in Beijing are decreasing and water resource shortages are becoming more serious, leading to the rapid decline in the population carrying capacity of water resources.Since the 1990s, the population carrying capacity of water resources in Beijing has been lower than the resident population and the gap

Figure 16 |
Figure 16 | Annual proportion of I, II, III water resources yield, quadratic polynomial fitting (QPF) in evaluated water of Beijing from 1999 to 2014.
These changes indicate that compared with current population growth, water resources in Beijing have been unable to satisfy the demand from urban development and water shortages will become more severe.Influence of outside water transfer projects on water resources The South-to-North Water Transfer Project is a strategic project in China that allocated water resources of the Yangtze River basin to water shortage areas in north and northwest China to meet the needs of mega cities.From 2008 to 2014, Beijing has been in a dry spell and its water resources cannot meet the city's development needs.As a result, the Jing-Shi Water Transfer Project, part of the South-to-North Water Transfer Project, transferred nearly 1.5 billion m 3 of water from Hebei and Shanxi provinces (Table 6).However, the cyclical variation of water resources and climate in these places is similar to Beijing, which limited the project's impact.Therefore, over-exploitation of underground water resources has become the only way to alleviate water shortage pressures.The South-to-North Water Transfer Project was put into use on December 7, 2014; the project provides 1 billion m 3 of water to Beijing every year, equivalent to 1/3 of the multi-year mean water resources.This effectively alleviated the water shortage pressure and Beijing gradually reduced the amount of groundwater exploitation.At present, the groundwater decline has slowed and it is expected that the groundwater depth will gradually rise until 2025 (Xinhua Net ).In 2015, the South-to-North Water Transfer project transferred 880 million m 3 of water resources to Beijing (Table

Figure 17 |
Figure 17 | Population carrying capacity of water resources (PCCW) and resident population (RP) from 1979 to 2013.

Table 1 |
Times (year)and frequency (%) of drought of Beijing in the period 1271-2000

Table 2 |
Main research methods of this research

Table 3 |
Correlation coefficients of precipitation, temperature and water resources in WRD_3rd

Table 4 |
Comparing the actual and fitted values of water resource

Table 5 |
Correlation coefficients of precipitation, temperature and water resource in different urban functional regions

Table 6 |
Annual variability of outside water resources from 2008 to 2015