Climate Change Impacts on Agricultural and Industrial Water Demands in the Beijing–Tianjin–Hebei Region Using Statistical Downscaling Model (SDSM)

Abstract

As a politically and culturally important city cluster, the Beijing–Tianjin–Hebei (BTH) region is the most prominent area in China where the imbalance between the supply and demand of water resources restricts the sustainable and healthy development of the regional social economy. In the context of global warming, research into water demand prediction that takes climate change into consideration would be more in line with the strategic goal of the low-carbon sustainable development of future cities. At the same time, the prediction of agricultural water demands against a background of climate change is urgently needed, while industrial water consumption is weakly correlated with climate change, an investigation of the statistical relationship between the two is needed. Thus, in this paper, future climate data from the BTH region under the scenarios RCP2.6, RCP4.5 and RCP8.5 were generated using a statistical downscaling model, and then coupled with agricultural and industrial water demand prediction models to simulate and analyze the impact of climate change on the agricultural and industrial water demands, respectively. The results show that during the forecast period (2020–2035), the reference crop evapotranspiration (ET₀) growth rates in the Beijing, Tianjin and Hebei areas under the RCP2.6 scenario are 1.438 mm·a⁻¹, 1.393 mm·a⁻¹ and 2.059 mm·a⁻¹, respectively. Under the RCP4.5 scenario, they are 2.252 mm·a⁻¹, 2.310 mm·a⁻¹ and 2.827 mm·a⁻¹, respectively. Under the RCP8.5 scenario, they are 3.123 mm·a⁻¹, 2.310 mm·a⁻¹ and 2.141 mm·a⁻¹, respectively. Furthermore, under each climate scenario, the increase in evapotranspiration in the Hebei area is the largest, followed by that in the Tianjin area, and that in the Beijing area is the smallest. For water consumption per CNY 10,000 of industrial added value during the forecast period, under the three different climate scenarios, a downward trend is seen in the Beijing area, with rates of 0.158, 0.153 and 0.110, respectively, but in the Tianjin area, there is an upward trend, with an upward tendency in rates of 0.170, 0.087 and 0.071, and an upward trend in the Hebei area, with an upward tendency in rates of 0.254, 0.071 and 0.036, respectively. This study will help the BTH region to rationally allocate agricultural and industrial water against the background of future climate change, and strengthen the coordination and cooperation between the different regions to promote the healthy and sustainable development of the cities.

1. Introduction

Climate change refers to changes in the average state of the climate system caused by natural factors and human activities. Climate change can have a significant impact on hydrological processes [1], and can also affect the demand for water resources. With successive increases in global warming, there are increasing regional changes in soil moisture, precipitation, and mean temperature, and changes in extreme weather are expected to be greater in intensity and frequency [2]. The uneven spatial and temporal distribution of water resources is a prominent feature of China’s water resources distribution. Under the dual impact of climate change and human activities, many city clusters in China are facing increasingly severe challenges to water security [3].

China is vigorously promoting the development strategy of the city cluster, and the development of city clusters is closely related to changes in water demand. As the economic core region in northern China, the Beijing–Tianjin–Hebei (BTH) city cluster has a poor natural endowment of water resources, but its population and economy are especially concentrated. Furthermore, it is affected by rapid industrialization and urbanization, and has become one of the regions most affected by the global water cycle disturbance [4]. On the whole, the contradiction between regional sustainable development and the serious shortage of water resources carrying capacity in the BTH region has existed for a long time and has been deepening.

Thus, this paper takes the BTH city cluster as the research area and tries to solve the contradiction.

Of all industries, agriculture is the most sensitive to the effects of climate change. Agricultural water includes water for crops and water for forestry, animal husbandry and freshwater aquaculture. The crop water requirement is the total amount of water consumed by transpiration by the plants and soil evaporation between the plants during the whole growth period and is an important factor that must be considered by irrigation decision-making and water resource planning [5]. Compared with that of crops, water consumption by forestry, animal husbandry and aquaculture industries is less affected by climate and is difficult to quantify. Therefore, in terms of agricultural water demand, this paper mainly considers the change in water demand related to crops.

Reference crop evapotranspiration (ET₀) is the only factor that appears in both the water balance and the surface energy balance equations [6], and is an important part of the crop water requirement calculation performed in climate and hydrological research [7]. ET₀ reflects the atmospheric evaporation capacity and the vegetation water demand. Only meteorological factors can affect ET₀ without considering the influence of crop characteristics and soil conditions. Researchers have carried out in-depth studies of ET₀ from general areas to regions with special climatic and geographical conditions, such as around the Mediterranean Sea [8,9]. In such studies, information describing crop water consumption is typically obtained by recording regional evapotranspiration and surface energy flux [10]. In the context of climate change, the study and analysis of temporal and spatial variations and complex periodic variations in evapotranspiration can provide a relevant basis for irrigation water management [11,12]. With progress in research and the continuous development of technology, regional evapotranspiration can now be calculated more accurately by integrating innovative technologies such as remote sensing, or by improving the precision of the crop coefficient, so as to ensure regional water security and crop yield through good water resource management [13,14]. Currently, the majority of studies on evapotranspiration primarily focus on investigating its sensitivity to various meteorological factors [15,16,17] and its variations across different geographical conditions and climates [18,19]. Specifically, research on evapotranspiration in the BTH area mainly centers around understanding its temporal and spatial changes as well as historical patterns [20]. This paper aims to predict future trends in evapotranspiration within the BTH area under diverse climatic conditions, while also providing practical recommendations for regional water resource management cooperation based on forecast results and the developmental status of the respective regions, thereby facilitating enhanced regional strategic coordination.

Industrial water management has an important impact on urban water consumption. Due to the continuous advancement of industrialization and urbanization, efficient urban water demand management is becoming more important, and the impact of industrial operations on the effective use and sustainability of water resources is attracting increasing attention [21]. At present, for industrial water demand, the main prediction methods are the quota method, regression analysis, grey theory prediction, the neural network method, and system dynamics [22,23,24]. Compared with socio-economic factors, meteorological factors have a relatively weak impact on industrial water demand. In view of this, most studies of industrial water demand focus on the impact of socio-economic factors [25,26] and ignore the impact of climate factors. Current relevant studies have included considerations such as industrial added value, industrial water repetition rate, an industrial technological progress index, industrial water price and policy-based water-saving rates, and various combinations of these have been incorporated into forecast models of industrial water demand [27,28]. In addition, a challenge arises in the process of establishing an industrial water demand prediction model in China due to excessively large forecast values. Therefore, this paper aims to focus on the impact of climate change on industrial water demand by incorporating meteorological factors into the existing model. This approach not only enables us to predict the changing trends in industrial water demand but also provides a method for quantifying the relationship between climate change and industrial water demand.

In a departure from the literature, our main contribution is that we apply the method of coupling an SDSM and a water demand model to China’s important region, BTH. First of all, in order to improve the spatial resolution of future climate information output by atmospheric circulation models, the statistical downscaling model (SDSM) was established for the BTH region, and a 95% confidence level was selected in the calibration process to reduce the uncertainty of simulation results. The calibration results show that the model can explain the daily temperature change in the study area well. Then, three climate scenarios RCP2.6, RCP4.5 and RCP8.5 were simulated based on the SDSM constructed by the CanESM2 atmospheric circulation model, and the daily maximum temperature, minimum daily temperature and daily precipitation data of meteorological stations in the BTH region from 2020 to 2050 were simulated. The data of Beijing, Tianjin and Hebei were obtained by Kriging interpolation of weather station data by ArcGIS. Secondly, the industrial water demand model established in this paper is different from other studies that mainly analyzed the impact of social and economic factors such as industrial water price, water use repetition rate and policy saving rate on industrial water demand. On that basis, this paper also fitted climate factors into the industrial water demand model through correlation analysis and multiple linear regression. Finally, meteorological data were fed into industrial and agricultural water demand models to simulate and predict the change in water demand against the background of climate change. The results can be compared across three climate scenarios to obtain the impact of future climate change on agricultural and industrial water demands. Additionally, inter-regional comparisons of forecast results can be conducted. This study focuses on the unique city cluster of BTH as the research region. After predicting the changing trend in water demand under various climate scenarios, this paper emphasizes the promotion of strategic cooperation in urban water resources and aims to provide relevant foundations and feasible recommendations for future water resource management collaboration by integrating the changing trends and developmental status of water demands across different regions.

On the whole, this study mainly used the SDSM to generate precipitation and temperature climate scenarios related to representative concentration path scenarios (RCP 2.6, RCP 4.5 and RCP 8.5). The simulation results were coupled with agricultural and industrial water demand models to compare the forecast results of water demand in different regions and under different climate scenarios, so as to provide a relevant basis for future agricultural and industrial water resources management in the BTH region, and at the same time make relevant policy recommendations to cope with climate change.

2. Study Area and Data Sources

2.1. Study Area

The BTH region is located between 113°04′~119°53′ east longitude and 36°01′~42°37′ north latitude, bordered by Bohai Bay to the east, the Taihang Mountain region to the west, the Yellow River area to the south, and the Mongolia Plateau to the north. The study area covers an area of 217,000 km², accounting for 2.3% of the national land area. Within the BTH region, resource distribution, climate characteristics, and water resource systems are all similar and the region is a resource-based water-shortage area in China.

Total water resources in 2021 in the BTH region were 6.13 billion m³ (Beijing), 3.98 billion m³ (Tianjin), and 37.66 billion m³ (Hebei), accounting for about 1.6% of the national total water resources (2964 billion m³). Hebei is an important grain-producing area in China and agricultural water consumption in this province has reached 53.4% of the total regional water consumption. Tianjin has a relatively equal distribution of all kinds of water consumption, while Beijing has a relatively high proportion of domestic water consumption. The BTH region contributes about 10% of the national GDP, 6% of the food production and contains 8% of the population, and currently limited water resources restrict the sustainable development of the region [29].

2.2. Data Sources

The climate data used in this paper mainly include historical meteorological observational data and CMIP5 climate model data. The daily measured meteorological data of 20 meteorological stations in the BTH region were obtained from the Water Resources Bulletin, Water Statistical Yearbook and provincial and municipal statistical yearbooks of the BTH region, including daily minimum temperature, maximum temperature and precipitation for the period 1973 to 2005.

CMIP5 model data was obtained from the first set of NCEP reanalysis data on the website of https://psl.noaa.gov accessed on 12 March 2023, including 26 atmospheric circulation forecast factors. The atmospheric circulation model was HadCM3, the climate model was CanESM2. Based on China’s current carbon peaking and carbon neutrality policies in response to global warming, and considering the frequency of extreme weather in the future, this study selected representative concentration paths RCP2.6, RCP4.5 and RCP8.5 as proposed by the IPCC for future hypothetical predictions.

For the industrial water demand model, the data involved in the annual industrial water repetition rate and industrial water price in the BTH region from 2000 to 2019 were derived from the China Environmental Statistical Yearbook.

3. Methodology

3.1. Statistical Downscaling Model

The SDSM used in this study is a combination of multiple linear regressions and random weather generators proposed by Wilby and Dawson [30]. It is practical, simple and easy to operate [31]. The SDSM utilizes outputs from a global climate model (GCM). This is a mathematical model that describes the physical processes of the Earth’s atmosphere, ocean and land based on the Navier–Stokes equations of a rotating sphere. It is widely used in climate change research and weather forecasting [32]. However, current GCMs only provide future climate scenarios at large scales and cannot be directly applied at regional scales [33]. To overcome this limitation, the SDSM can use multiple regression technology to extract station-scale climate information from GCM output results. Using the coupling principle of random weather generators and multiple regressions, the SDSM can establish relationships between GCM variables and local scale variables. The SDSM is then used in two stages (Figure 1): (1) establishment of the statistical relationship between predictors from regional or station meteorological data and predictors from a GCM; (2) generation of future local-scale daily time series data using the GCM output data. The SDSM is widely used in climate change studies around the world [34,35].

Figure 1. Flowchart illustrating the operation of the SDSM. The predictands (i.e., daily observed maximum temperature, minimum temperature and rainfall) are first extracted for 20 meteorological stations in the BTH region. Time series of 26 predictor variables are then extracted from NCEP for the same time period (1973 to 2000). Correlation analysis is then performed between the predictands and the 26 predictor variables to obtain the statistical relationship between the predictors and the predictands. The results selection of the predictors for each station are presented in

Table 1. The predictors corresponding to the daily maximum temperature (T_max), daily minimum temperature (T_min) and daily precipitation (P) at 20 meteorological stations. Definitions of the predictor variables are given in the notes below the table.

Meteorological Station	Tmax	Tmin	P
Beijing, Miyun, Baoding	plth p1zh p5_z p500 p5zh temp	p1zh p5_z p500 shum temp	p1_f p1zh p500 p8_f prcp s850
Tianjin, Binhai, Tanggu	p1_u p1zh p5zh p500p8zh shum temp	p1zh p5_z p500 shum temp	p5_f p8_f p8_v prcp s500 s850
Chengde, Fengning, Weichang	p1_v p1zh p5_z p5zh p8zh temp	p1zh p5_z prcp s500 shum temp	p1_f p1zh p500 p8_f prcp s500
Tangshan, Zunhua, Leting	p1_u p1zh p5_z p500 p5zh temp	p1zh p5_z p500 s500 shum temp	p1_f p1zh p8_v prcp s500 s850
Qinglong	p1_u p5zh p8th p8zh temp	p1zh p5_z prcp s500 shum temp	p1_f p1zh p8_v prcp s500 s850
Shijiazhuang, Nangong, Xingtai, Cangzhou	p1_u p1th p500 p5zh p8zh shum temp	p1th p1zh p5_z p500 shum temp	p1zh p5_v p8_v prcp s500 s850
Huailai, Weixian, Zhangjiakou	p1zh p5_z p500 p5zh p8_u p8zh temp	p1zh p5_z prcp s500 shum temp	p1_f p1zh p500 prcp s500 s850

Note: predictors are: p1_f—sea level wind speed; p1_u—sea level meridional temperature; p1_v—sea level zonal temperature; p1th—sea level wind direction; p1zh—sea level divergence; p5_f—500 hPa wind speed; p5_v—500 hPa zonal temperature; p5_z—500 hPa vorticity; p500—500 hPa geopotential height; p5zh—divergence of 500 hPa; p8_f—wind speed of 850 hPa; p8_u—meridional temperature of 850 hPa; p8_v—zonal temperature of 850 hPa; p8th—850 hPa wind direction; p8zh—850 hPa divergence; prcp—precipitation; s500—500 hPa relative humidity; s850—850 hPa relative humidity; shum—near surface humidity; temp—near surface temperature.

.

Figure 1. Flowchart illustrating the operation of the SDSM. The predictands (i.e., daily observed maximum temperature, minimum temperature and rainfall) are first extracted for 20 meteorological stations in the BTH region. Time series of 26 predictor variables are then extracted from NCEP for the same time period (1973 to 2000). Correlation analysis is then performed between the predictands and the 26 predictor variables to obtain the statistical relationship between the predictors and the predictands. The results selection of the predictors for each station are presented in Table 1.

After the statistical relationships between predictors and predictands had been established, the SDSM was calibrated. Historical parameters were inputted to the model and the model was run. The simulation results were then obtained and compared to the actual observational data to verify the accuracy of the model. After model calibration, daily data for the period 2020 to 2050 were obtained from HadCM3 and used as input to the SDSM in order to predict daily maximum and minimum temperature time series at each of the 20 meteorological stations under the three future climate scenarios of RCP2.6, RCP4.5 and RCP8.5 (

Table 2. Details of the three future climate scenarios (RCPs) used in this study. The paths of greenhouse gas (GHG) emissions differ in the three climate change scenarios.

Climate Scenario	Assumed Development Path
RCP2.6	The lowest GHG emissions scenario. Emissions initially rise and then fall until reaching stability.
RCP4.5	Intermediate GHG emissions scenario with moderate emissions. This scenario is consistent with the requirements of future Chinese economic development planning.
RCP8.5	The highest GHG emissions scenario. Emissions continue to rise.

). ArcGIS 10.6 was then used to perform Kriging spatial interpolation using the meteorological station predictions.

The three climate scenarios selected in this paper correspond to different degrees of climate intervention policies and greenhouse gas emission levels. Through the comparative analysis of the forecast results under different climate scenarios, the influence of different degrees of climate policy intervention on them can be indirectly obtained. At the same time, the uncertainty of future forecast results would be evaluated better.

3.2. Agricultural Water Demand Model

In order to quantitatively evaluate the impact of climate change on crop water requirements, reference crop evapotranspiration was selected as the evaluation index. Crop water requirement can be estimated using the product of ET₀ and a crop coefficient [36]. However, the crop coefficient varies in space and time, due to inherent variability in emergence date, land use pattern, antecedent precipitation, emissivity, vegetation amount, and atmospheric boundary conditions [37]. Hence, to avoid further error, we use ET₀ alone to reflect the degree of impact of climate change on overall crop water requirements.

In this study, the Hargreaves formula was used to calculate ET₀ at each of the 20 meteorological stations. The method estimates the reference crop evapotranspiration (ET₀) value using temperature and solar radiation [38]. Compared with the traditional Penman–Monteith formula, this method does not require climate data that is unavailable from the observational meteorological stations [39,40]. Temperature and solar radiation are easy to measure quantitatively, and are used to estimate ET₀ as follows:

$$E{T}_0=\frac{0.0023R_a}{\lambda }\times (T_{mean}+T_{off})\times \left(T_{max}-T_{min}\right){ }^{0.5}$$

(1)

where ET₀ is the reference crop evapotranspiration (mm·d⁻¹); λ is the latent heat of water vapor (2.45 MJ·kg⁻¹); T_mean is the daily average air temperature (°C); T_off is a temperature constant (recommended value is 17.8 °C [41]); T_max is the daily maximum air temperature (°C); T_min is the daily minimum air temperature (°C); and R_a is the radiation intensity of the upper atmosphere (MJ·m⁻²·d⁻¹) and is given by:

$$R_a=\frac{118.08}{\pi }{d}_r\left[{\omega }_s\sin (\phi )\sin (\delta )+\cos (\phi )\cos (\delta )\sin ({\omega }_s)\right]$$

(2)

$${\omega }_s=\arccos \left[-\tan (\phi )\tan (\delta )\right]$$

(3)

$$d_r=1+0.033\cos \left(\frac{2\pi }{D}J\right)$$

(4)

$$\delta =0.409\sin \left(\frac{2\pi }{D}J-1.39\right)$$

(5)

where d_r is the reciprocal of the relative distance between the sun and the Earth; ω_s is the solar hour angle (rad); φ is the geographical latitude (rad); δ is the solar magnetic decrement angle (rad); J is the ordinal number of a day in a year, ranging from 1 (1 January) to 365 (31 December) in common years and from 1 to 366 in leap years; and D represents the number of days in the current year, 366 in leap years and 365 in common years.

After calculating ET₀ time series using the observed and predicted meteorological data for each of the observational stations located in the BTH region, average ET₀ values for the Beijing, Tianjin and Hebei sub-areas were estimated using spatial interpolation. The data was then divided into the baseline period (2005–2019) and the forecast period (2020–2035) with a period of 15 years. Comparing the baseline period as a whole with the forecast period, the impact of future climate change on the overall evapotranspiration change trend in the forecast period can be more intuitive. And it would be easier to evaluate the ET₀ trends for each sub-area under the three climate scenarios (i.e., RCP2.6, RCP4.5 and RCP8.5)

3.3. Industrial Water Demand Model

The quota method [42] was used to calculate the industrial water demand (W_industry). This method considers the influence of climate change on the water consumption per CNY 10,000 of industrial added value. W_industry (m³) is estimated as follows:

$$W_{industry}=u_{industry}\times GD{P}_{industry}$$

(6)

where GDP_industry is the industrial added value (CNY 10,000); and u_industry represents the water consumption per CNY 10,000 of industrial added value (m³·10,000 yuan⁻¹).

The industrial water demand prediction model of Yingjie et al. [43] was used to estimate values of u_industry. This model utilizes an industrial water price, an industrial water reuse rate and the air temperature as follows:

$$\mathit{ln}{u}_{indusry}=b_0+b_1\mathit{ln}{x}_1+b_2\mathit{ln}{x}_2+b_3\mathit{ln}{x}_3$$

(7)

where b₀, b₁, b₂, b₃ denote regression coefficients; x₁ is the utilization rate of industrial water; x₂ is the industrial water price (yuan·m⁻³); and x₃ is the absolute change in temperature.

The statistical software package SPSS 26.0 was used to carry out correlation analysis and multiple linear regressions on the data from 2000 to 2019 in order to establish the regression coefficients b₀, b₁, b₂, b_3. Predictions of x₁, x₂ and x₃ for the period 2020 to 2035 under the three climate scenarios (RCP2.6, RCP4.5 and RCP8.5) were obtained by SDSM and various planning data in relevant documents. They were inputted to the fitted formula to obtain the water consumption per CNY 10,000 of industrial added value and evaluate the simulation effect of the established model.

4. Results and Discussion

4.1. SDSM in the BTH Region

The SDSM was established using data for the period 1973 to 1990, and the model performance was then evaluated using data from the period 1991 to 2005. Two evaluation parameters were calculated for the validation period, the standard error (SE) and the explanatory variance (R-squared), and these are summarized in

Table 3. Results of the SDSM performance evaluation at each of the 20 meteorological stations for the period 1991 to 2005. Values presented are the average percentage of explained variance (R²), and the standard error (SE) of daily predicted values of maximum temperature (T_max), daily minimum temperature (T_min) and daily precipitation (P).

Station	Tmax		Tmin		P
	R2	SE (°C)	R2	SE (°C)	R2	SE (°C)
Beijing	0.559	2.640	0.702	1.670	0.445	0.260
Miyun	0.610	2.430	0.699	1.620	0.410	0.290
Tianjin	0.622	2.400	0.709	1.570	0.365	0.260
Binhai	0.592	2.600	0.517	2.280	0.332	0.290
Tanggu	0.579	2.500	0.623	1.910	0.362	0.240
Baoding	0.593	2.600	0.707	1.580	0.483	0.320
Cangzhou	0.627	2.480	0.732	1.600	0.533	0.170
Chengde	0.695	2.300	0.73	1.780	0.536	0.200
Fengning	0.66	2.600	0.683	2.140	0.480	0.220
Huailai	0.69	2.400	0.713	1.790	0.562	0.150
Leting	0.64	2.270	0.711	1.800	0.622	0.210
Qinglong	0.655	2.370	0.736	1.860	0.627	0.220
Nangong	0.668	2.430	0.699	1.610	0.429	0.150
Shijiazhuang	0.625	2.610	0.688	1.670	0.411	0.230
Tangshan	0.653	2.260	0.741	1.680	0.409	0.220
Weichang	0.702	2.440	0.724	1.860	0.224	0.310
Weixian	0.456	3.410	0.651	2.430	0.415	0.140
Xingtai	0.661	2.530	0.691	1.650	0.329	0.180
Zhangjiakou	0.705	2.420	0.759	1.680	0.614	0.140
Zunhua	0.623	2.470	0.702	1.960	0.335	0.320

.

It can be seen from Table 3 that the explained variance of the model for daily maximum temperature and daily minimum temperature simulations varied between 0.559 and 0.705, with most values being greater than 0.6. This suggests that the SDSM explained the daily temperature variation well. Due to the complex factors affecting precipitation, the explained variance of daily precipitation was relatively lower, but to a certain extent the SDSM can be assumed to simulate precipitation to an acceptable degree. In general, the SDSM demonstrated reasonable performance and hence can be used to generate predictions of daily maximum air temperature, minimum air temperature and precipitation under RCP2.6, RCP4.5 and RCP8.5 climate scenarios.

4.2. Daily Maximum and Minimum Temperatures under Future Climate Scenarios

The simulated changes in daily maximum and minimum air temperatures in the BTH region under the three climate scenarios (RCP2.6, RCP4.5 and RCP8.5) for the period 2020 to 2050, are presented in Figure 2.

There is a difference of 0.14 to 1.55 °C between the forecast period (2020–2050) and the base period (2005–2019). The difference was greatest in the RCP2.6 scenario. The data for the forecast period are shown in the

Table 4. The average daily maximum and minimum temperatures over the period 2020 to 2050 in the BTH under the three scenarios.

Climate Scenario	Average Daily Maximum Temperature (°C)			Average Daily Minimum Temperature (°C)
	Beijing	Tianjin	Hebei	Beijing	Tianjin	Hebei
RCP2.6	18.92	19.01	18.78	6.94	9.35	6.81
RCP4.5	18.29	18.79	17.95	6.59	9.00	6.43
RCP8.5	19.10	19.31	19.05	7.33	9.69	7.20

below.

During the simulation period, the daily maximum and daily minimum temperatures in the BTH region fluctuate and show an overall upward trend. Under the RCP8.5 climate scenario, the daily maximum temperature and daily minimum temperature in each region increase the most, while under the RCP2.6 and RCP4.5 climate scenarios, the daily maximum temperature and daily minimum temperature increase less. This shows that governments need to implement policy interventions to effectively control carbon emissions, reduce the impact of greenhouse effect as much as possible, and control the temperature increase in future climate change.

4.3. Evapotranspiration under Future Climate Scenarios

The predicted evapotranspiration results of the three sub-areas of the BTH region under the three climate scenarios are presented in Figure 3.

It can be seen from the data listed in the

Table 5. The mean ET₀ of the BTH region in the base period (2005–2019) and forecast period (2020–2035), and the overall growth rate under each scenario in the forecast period.

	Average ET0 (mm, 2005–2019)	Average ET0 (mm, 2020–2035)			Growth Rate (mm·a−1, 2020–2035)
		RCP2.6	RCP4.5	RCP8.5	RCP2.6	RCP4.5	RCP8.5
Beijing	1203.747	1237.447	1242.986	1240.096	1.438	2.252	3.123
Tianjin	1151.658	1189.936	1206.202	1213.139	1.393	2.310	2.310
Hebei	1095.822	1224.169	1224.169	1240.621	2.059	2.827	2.141

that, under the three climate scenarios, the order of ET₀ growth in the three regions of Beijing, Tianjin and Hebei is always RCP8.5 > RCP4.5 > RCP2.6 from 2005 to 2035.

A comparison of ET₀ variation trends in the BTH region under the three climate scenarios shows that, although ET₀ in the region is on the rise under all three climate scenarios, the increase in ET₀ in each sub-area is the largest under the RCP8.5 climate scenario, followed by the RCP4.5 climate scenario, and under the RCP2.6 climate scenario, the increase in ET₀ in each region is the smallest. The main reason for this situation is that in the RCP8.5 climate scenario, the future temperature increase is relatively large, and more water will be released by soil evaporation and crop transpiration. As a result, the increase in evapotranspiration will be large, and the agricultural water demand will also increase. At the same time, the differences between climate scenarios also indicate that for future climate change and even some extreme climates, the negative impact on agricultural water use will be effectively reduced under the condition of government policy intervention.

The regional comparison of ET₀ in the three sub-areas shows that the overall increase in ET₀ in Hebei is the largest, followed by that in Tianjin, with the smallest being in the Beijing area. The main reason for this result is that agricultural water accounts for the largest proportion of overall water consumption in the Hebei area, while agricultural water accounts for the smallest proportion in the Beijing area. Therefore, under the same climate scenario, the fluctuation in water demand caused by climate influence in the Hebei area is greater, while the fluctuation in the Beijing area is relatively small. The results can provide a reference to guide overall crop irrigation and water management in the BTH region in the future. According to the specific fluctuations in evapotranspiration in different regions under different climate scenarios in the future, combined with relevant factors such as regional water use structure and agricultural irrigation technology level, we can propose corresponding practical measures to deal with the negative impact of climate change on water demands in the future. In addition, the strategic cooperative relationships between city clusters can be fully utilized to coordinate the supply and demand of water resources between regions and formulate corresponding cooperation policies, so as to facilitate the rational distribution of water resources in order to guarantee the safety of agricultural water utilization.

Based on the above comparative analysis results of evapotranspiration changes in different climate scenarios and different regions, we propose the following suggestions. From the obvious gap between the increase in evapotranspiration under the three climate scenarios, it can be seen that under the trend of global warming in the future, if no active measures are taken to control carbon emissions and global temperature is allowed to rise, the increase in evapotranspiration will lead to an increase in agricultural water demand in many regions like the BTH region, especially in areas like Hebei where agricultural water consumption is dominant. However, effective human intervention in carbon emissions through the formulation of relevant policies, such as the two-carbon policy currently formulated in China, will achieve good results in the future when dealing with the negative effects of climate change on water resources. In view of the management of agricultural water use in the three regions against the background of future climate change, as the proportion of agricultural water consumption in Hebei and Tianjin is large, emphasis should be placed on improving irrigation technology to control agricultural water demand.

4.4. Industrial Water Consumption under Future Climate Scenarios

The water consumption per CNY 10,000 of industrial added value in the BTH region under the three different climate scenarios of RCP2.6, RCP4.5 and RCP8.5 is presented in Figure 4.

As we can see (

Table 6. The fluctuation range and growth rate of the water consumption per CNY 10,000 of industrial added value in the BTH region under three climate scenarios during the forecast period (2020–2035).

	Fluctuation Range (m3)	Growth Rate
		RCP2.6	RCP4.5	RCP8.5
Beijing	3.27–6.63	−0.158	−0.153	−0.110
Tianjin	13.82–28.28	0.170	0.087	0.071
Hebei	12.45–17.89	0.254	0.071	0.036

), under the three climate scenarios of RCP2.6, RCP4.5 and RCP8.5, the overall change in water consumption per CNY 10,000 of industrial added value showed a downward trend in Beijing, but an upward trend in Tianjin and Hebei.

A comparison of the water consumption per CNY 10,000 of industrial added value in the BTH region under the three different climate scenarios shows that there are no significant differences in the change trend of industrial water demand in each sub-area. The reason for this is that the influence of temperature on industrial water demand is relatively weak. The change in industrial water demand is mainly controlled by regional water price, water repetition rate and water saving rate. By comparing the water consumption per CNY 10,000 of industrial added value between the three sub-areas, it can be concluded that the order of the water consumption per CNY 10,000 of industrial added value is Hebei > Tianjin > Beijing under all three climate scenarios. In addition, the future water consumption per CNY 10,000 of industrial added value in the Beijing area will decrease, while that in the Tianjin and Hebei areas will increase. The main reason for this is that there is a big difference between the water price and the industrial water repetition rate in the three sub-areas.

The above results show that the model of industrial water demand used in this study, which includes temperature as a factor, predicts a very limited impact from climate change on industrial water demand in the BTH region, but it also provides a reference calculation method for the change in industrial water demand in the BTH region under the extreme climate conditions that may occur in the future; that is, by using multiple linear regression and correlation analysis, climate factors such as temperature and rainfall are included in the calculation model of industrial water demand. In addition, the differences in the trends in industrial water demand among the three regions are mainly caused by social and economic factors such as the repetition rate of industrial water use and the price of industrial water. The industrial development status of the three regions is very different, which indicates that we should focus on combining the characteristics of industrial development and strategic cooperation in each region to provide a relevant basis for the overall industrial transfer in the BTH region in the future.

According to the industrial development of the different regions in recent years, the industrial water consumption in the Beijing area has been reduced due to the relocation of enterprises. The expansion of infrastructure investment in Tianjin has led to an increase in industrial water consumption. Hebei Province, as the sub-area with the highest water demand in the BTH region, has a much larger industrial scale than Beijing and Tianjin. Although its industrial water consumption has increased due to some relocating of manufacturing industry, improvements in both industrial development and water use efficiency have gradually reduced this consumption. According to the results of this study, industrial water usage in the three regions will continue to change with the same trend in each climate change context up to 2035. Therefore, for the industrial water allocation in the BTH region against the background of future climate change, targeted measures should be implemented according to the characteristics of industrial development in each region. For example, Tianjin should address the input–output imbalance of a number of resources, including water resources, while Hebei should focus on improving water efficiency. At the same time, with the continuous development of the coordinated development policy of the BTH region, the overall process of promoting industrial transfer in Beijing, Tianjin and Hebei should be “upgraded first and then transferred”, so as to achieve mutual benefit and a win-win situation.

4.5. Uncertainty Analysis

In this paper, a statistical downscaling method was used to simulate the variation trend of ET₀ and the industrial water consumption in the BTH region under three typical concentration paths (RCP2.6, RCP4.5 and RCP8.5) using output data from the CanESM2 model in the CMIP5 model. It should be noted that, in this paper, only the CanESM2 model data was used to analyze the ET₀ changes in the BTH region, which may have introduced some uncertainties. In subsequent studies, multiple model data, different downscaling methods and calculation models containing more climate impact factors should be used to evaluate the magnitude of the uncertainties and to increase the accuracy of the simulation results. In addition, some parameters involved in the ET₀ model and in the industrial water consumption model (e.g., the latent heat value of water vapor and the constant value of temperature included in the Hargreaves formula; the industrial water repetition rate and industrial water price) have large associated uncertainties. Further improvement of the accuracy of these parameters would also make the simulation results of water demand more accurate.

5. Conclusions

In this paper, a climate downscaling model was used to simulate future regional climate change under the RCP2.6, RCP4.5 and RCP8.5 scenarios. The output from this model was used to drive models of agricultural and industrial water demand so as to simulate and predict future water demand in the BTH region. The following conclusions were reached:

(1)

During the forecast period (2020–2035), the ET₀ growth rates in the Beijing, Tianjin and Hebei areas under the RCP2.6 scenario are 1.438 mm·a⁻¹, 1.393 mm·a⁻¹ and 2.059 mm·a⁻¹, respectively. Under the RCP4.5 scenario, they are 2.252 mm·a⁻¹, 2.310 mm·a⁻¹ and 2.827 mm·a⁻¹, respectively. Under the RCP8.5 scenario, they are 3.123 mm·a⁻¹, 2.310 mm·a⁻¹ and 2.141 mm·a⁻¹, respectively. Under each climate scenario, the increase in evapotranspiration in the Hebei area is the largest, followed by that in the Tianjin area, and that in the Beijing area is the smallest. The order of increase in evapotranspiration in all three areas under the different climate scenarios is RCP8.5 > RCP4.5 > RCP2.6.

(2)

During the forecast period (2020–2035), under the three different climate scenarios, the water consumption per CNY 10,000 of industrial added value in the Beijing area shows a downward trend, with rates of 0.158, 0.153 and 0.110, respectively. The water consumption per CNY 10,000 of industrial added value in the Tianjin area shows an upward trend, with upward tendency rates of 0.170, 0.087 and 0.071 under the three climate scenarios, respectively. The water consumption per CNY 10,000 of industrial added value in the Hebei area also shows an upward trend, with upward tendency rates of 0.254, 0.071 and 0.036 under the three climate scenarios, respectively. The results of this study provide a statistical method of assessing the impact of climate change on industrial water demand in the BTH region.

(3)

In the context of future climate change, with the increase in temperature, the agricultural demand for water will increase significantly, and the industrial demand for water will be less affected, but will also rise. Therefore, in the future allocation process of agricultural and industrial water resources in the Beijing–Tianjin–Hebei region, on the one hand, targeted measures should be proposed to deal with the negative effects of climate change based on the development status and water use characteristics of each region, and on the other hand, the water use structure of each region should be fully considered to allow the strategic cooperative relationship between the city clusters to develop. Through the rational distribution of water resources in the different regions, the constraints of water supply and demand on social and economic development can be solved, so as to promote the healthy and sustainable development of the study region.

However, there are still some areas that need to be improved. First of all, compared with the traditional Penman–Monteith formula, the temperature-based Hargreaves formula adopted in this paper is relatively simple to calculate, and the meteorological parameters considered to change with climate change only include the minimum, maximum and average temperature. For the important parameter of solar radiation involved in the formula, it was calculated through parameters such as geographical latitude, which were easy to measure, and only its specific changes over time were taken into account. Therefore, the influencing factors of this method are relatively simple. As a future improvement, more climate-influencing factors could be added based on the Penman–Monteith method according to research needs. And it will make the prediction result more realistic. Secondly, there were relatively few variables in the industrial water demand model established in this paper. Future innovations and improvements to industrial models may use methods such as neural networks and system dynamics to include as many relevant variables as possible to improve the prediction results. Finally, the Coupled Model Intercomparison Project Phase 6 (CMIP6) provides a scenario framework for global socio-economic development and climate change by combining Shared Socioeconomic Pathways (SSP) and Representative Concentration Pathways (RCP). Unlike the RCP scenario in CMIP5, CMIP6 takes into account future socio-economic development, and the GCM data are more accurate and complex than the previous CMIP5 scenario. In future studies, CMIP6 data can be combined to make a more accurate prediction of water demand under different development scenarios in the future, and relevant studies will have more value to reflect future changes and hot spots.