1. Introduction
The non-point source (NPS) pollution has become the main source of pollution in most of China’s rivers and lakes, causing the deterioration of water quality [
1]. In fact, total nitrogen and phosphorus produced by agricultural NPS pollution have become a constraint to China’s sustainable development [
2]. It is well known that land use and climate change are two essential factors affecting water quality through NPS progress. In fact, climate change affects the hydrological cycle of the watershed by changing the physical and chemical processes, migration and transformation capacity of pollutants as well as the ability of water bodies to dilute pollutants [
3,
4], resulting in deteriorating surface water quality and bringing new challenges for the environmental management of watershed [
5]. Moreover, land use as a key factor affecting the properties of the underlying surface which determines the basic parameters of the watershed runoff generation and soil erosion processes [
6], has a significant impact on the production and output of pollutants in the soil [
7]. Therefore, it is of great significance to study the impacts of land use and climate change on non-point source pollution processes at the watershed scale.
In recent years, various mathematical methods and hydrological climate models have been used to quantify the impact of land use and climate change on watersheds. In future climate scenarios, precipitation and temperature will change, affecting non-point source processes. Zhang et al. [
8] estimated the impacts of climate change on streamflow and non-point source pollutant loads by combining the circulation model (HadCM3) with the soil and water assessment tool (SWAT) hydrological model. Narsimlu et al. [
9] evaluated the future impacts of climate change on water resources using the SWAT model combined with the sequential uncertainty fitting (SUFI-2) algorithm. Li et al. [
10] simulated the changes in NPS pollutant loads for a period of 81 years (2019–2099) by applying the SWAT with six climate model. Numerous studies have also focused on assessing the impact of the land use pattern on NPS pollution. Wang et al. [
11] combined the genetic algorithm (GA) with the land use model to achieve an optimized land use pattern to control NPS pollution. The impacts of static and dynamic land use input conditions on the performance of the NPS model and the search for appropriate dynamic land use input to improve model accuracy were also evaluated [
12]. The combined impact of climate change and land use were also studied. Moreover, Bai et al. [
13] investigated the combined effects of land use and climate change on ecosystem services by using models and environmental setting scenarios where two indicators were developed to evaluate the effects of land use and climate change on these ecosystem services. In addition, Bai et al. [
14] used cellular automata (CA) and hydrological models to study the response mechanism of NPS pollution loads to land use change under different precipitation scenarios.
Previous research has indicated that the impacts of global climate change and land use on NPS load are significant; thus, effective management to alleviate the negative effects is indispensable. Furthermore, studies have shown that appropriate mitigation measures can greatly reduce the output of non-point source pollutants [
15,
16]. For example, Jiang et al. [
17] evaluated the reduction of high-level nonpoint source (NPS) pollution discharges in the highland agricultural catchment by applying technical measures using SWAT. In addition, Kaini et al. [
18] coupled the genetic algorithm (GA) with SWAT to find an optimal combination of structural measures to meet treatment goals at a watershed scale. Furthermore, Jeon et al. [
19] proposed an evaluation methodology to quantify future changes in BMPs on total phosphorus (TP) loads in the river system as a function of climate change.
Over the last 50 years, the measured discharges of the Haihe River Watershed have shown a significant decrease resulting from the impact of human activities and climate change. In fact, the inflow of the Miyun Reservoir, which is the largest reservoir in the Haihe River Watershed, has been continuously reduced in the past 30 years. Indeed, climate change contributed 25% and 45% to the reduction of runoff from the main rivers entering the Miyun Reservoir—the Chaohe River and Baihe River respectively, which seriously affected the water supply safety and sustainable development of Beijing [
20]. At present, the eutrophication degree of Miyun Reservoir is mesotrophic and the tendency to eutrophication is evident [
21] and agricultural non-point source pollution has become the main factor affecting the water quality of Miyun Reservoir. Therefore, it is of great significance to study the impact of climate change and human activities on the hydrology and ecology of the watershed [
22].
On the other hand, previous studies mentioned above have certain limitations, such as the use of future climate data from old emission scenarios data source. Moreover, future land use scenarios did not consider spatial allocation, resulting in uncertainties in the assessment of impacts on hydrology and water quality. Therefore, in this paper we evaluated the non-point source control strategy and the pollutant reduction effect in the context of land use change under future climate scenarios with SWAT model. The main objectives of this study were: (1) to estimate the future climate change by quantile mapping methods (2) to analyze the land use change pattern of the past, present and future by CLUE-S model; (3) to estimate the stream flow and nutrient loading under climate change; (4) to investigate the impact of the characteristics of land use change patterns on streamflow and nutrient loading in future climate scenarios. This research presents a certain theoretical and practical significance for the future non-point source pollution control and management and urban drinking water safety. It also provides a management model for future non-point source pollution under climate change conditions in other river watersheds.
3. Results and Discussion
3.1. Model Calibration and Validation
3.1.1. SWAT Model Calibration and Validation
The simulation results of runoff and total nitrogen at the two stations are shown in
Figure 2. The water quality evaluation indicators (R
2 and NSE value) during the verification period were lower than the regular water quality evaluation indicators. The results of calibration for the SWAT parameters in the Baihe and Chaohe watersheds are shown in
Table 3. The calibrated SWAT model can accurately describe the process of hydrological cycle and the process of migration and transformation of pollutants in the watershed. It can also be used to analyze the impact of climate change and land use change scenarios on the runoff and aquatic environment of the watershed.
3.1.2. CLUE-S Model Calibration and Validation
The CLUE-S model was initialized after setting all the model parameter files. After many iterations, a land use simulation map in 2008 was obtained. The Kappa index was calculated by analyzing the CLUE simulation results using land use status map from 2008. The Kappa index was 0.67, indicating that the accuracy of the model simulation results can reach 67%. Depending on the level of consistency of the Kappa index, the results of the CLUE-S simulation of the Miyun Reservoir watershed in 2008 and the results of the historical land use status map were highly consistent. Thus, we can assume that the simulation results of the CLUE-S model have high credibility and that the calibrated model can be used for the analysis of future land use change scenarios.
3.2. Future Climate Change Analysis
According to the performance evaluation results and data availability of 44 CMIP5 GCMs in the Haihe River Basin in precipitation and average temperature [
33], two climate models (ACCESS1.3 and HadGEM-ES) and two typical emission scenarios (RCP4.5 and RCP8.5) were used to predict future trends in precipitation and temperature in the Miyun Reservoir watershed during two future evaluation periods (2020–2042 and 2060–2082), which were compared with the base period (1988–2010).
3.2.1. Variation in Future Temperature
The average maximum temperature and the 95th value of the Miyun Reservoir watershed are shown in
Figure 3. Compared with the average maximum temperature of the MRW (13.5 °C) during the base period, the future maximum temperature under the two emission scenarios of RCP4.5 and RCP8.5 showed an increasing trend. The variations in the maximum temperature under different emission scenarios were obviously different. Indeed, the temperature under RCP8.5 scenario from 2060 to 2082 showed an annual increase of 4.5 °C, which was the highest. This variation was essentially the same as that of the greenhouse gas emission scenario in which, higher greenhouse gas emissions led to a greater temperature increase. From the perspective of the future maximum temperature generated by different GCMs, the maximum temperature variation generated by the HadGEM-ES model was the largest under the same emission scenario and evaluation period. The 95th value of the highest temperature in the base period was 30.3 °C and the 95th value of the highest temperature under different climate combinations showed an increasing trend. Among them, the temperature increase was the largest under RCP8.5 scenario from 2060 to 2082 with a value of 5.2 °C. Like the average maximum temperature, the HadGEM-ES model generated the largest change in maximum temperature under the same emission scenario and evaluation period. In order to better describe changes in future maximum temperature and to reduce prediction uncertainty, the ensemble average method was used to reflect future changes in maximum temperature in the watershed. During 2020–2042, the average maximum temperature will increase by 1 °C and 1.4 °C, whereas during 2060–2082, it will increase by 2.5 °C and 4.1 °C under RCP4.5 and RCP8.5 scenarios, respectively.
Figure 4 shows the predicted values of the monthly and seasonal averages of the highest temperature in the MRW under different combination scenarios. As can be seen in the next two evaluation periods, the two climate models ACCESS1.3 and HadGEM-ES will show a warming trend at both monthly and seasonal time scales, which is consistent with the change trend of the annual maximum temperature. In different evaluation periods in the future, the monthly and seasonal average rates of the highest temperature increase in the period from 2060 to 2082 will be significantly higher than those in the period from 2020 to 2042. It is noteworthy that the rise in the maximum temperature in spring and winter was greater than that in summer and autumn.
The average minimum temperature and the 5th value of the MRW are shown in
Figure 5. Compared with the annual average minimum temperature (0.8 °C) during the base period, the future minimum temperature in the MRW under RCP4.5 and RCP8.5 showed an increasing trend and the variations in the minimum temperature under different emission scenarios were obviously distinct. Among them, during the period from 2060 to 2082, the temperature increase was greatest under RCP8.5 scenario with a value of 5.3 °C, which was basically consistent with the trend of temperature variation under the greenhouse gas emission scenario. From the perspective of the future minimum temperature generated by different GCMs, the HadGEM-ES model generated the biggest change in the minimum temperature under the same emission scenario and evaluation period. The 5th value of the lowest temperature in the base period was −17.6 °C and the 5th value of the lowest temperature in the MRW under different climate combinations showed an increasing trend. Among them, the temperature increase was the highest under RCP8.5 scenario from 2060 to 2082 with a value of 6.1 °C. Under RCP4.5 scenario from 2020 to 2042, the minimum temperature of the ACCESS1.3 climate model had the largest variation of 5th, while under RCP8.5 scenario, the HadGEM-ES climate model had the greatest change. During the period from 2060 to 2082, the minimum temperature generated by the ACCESS1.3 model under RCP4.5 and RCP8.5 scenarios had the largest change in the 5th. The ensemble average method was also used to reflect the future minimum temperature changes in the watershed. During 2020–2042, the average minimum temperature increase will be 1.4 °C and 1.7 °C, whereas during 2060–2082, it will be 3.2 °C and 5.9 °C, under RCP4.5 and RCP8.5 scenarios, respectively.
Figure 6 shows the predicted values of the monthly and seasonal averages of the minimum temperature in the MRW under different combination scenarios. It was depicted that in the next two evaluation periods, ACCESS1.3 and HadGEM-ES will show a warming trend at monthly and seasonal time scales, which is consistent with the change trend of the annual minimum temperature. In different evaluation periods in the future, the monthly and seasonal average rates of the minimum temperature in the period from 2060 to 2082 will increase significantly more than those in the period from 2060 to 2082. Note that the minimum temperature increase in spring and winter was higher than that in summer and autumn.
3.2.2. Variation in Future Precipitation
The annual precipitation in the MRW is shown in
Figure 7. During 2020–2042, the annual average precipitation will increase by 17.1 mm and 16.9 mm, whereas during 2060–2082, it will increase by 33.7 mm and 50.6 mm, under RCP4.5 and RCP8.5 scenarios, respectively.
Figure 8 shows the predicted values of the monthly and seasonal averages of precipitation in the MRW under different combination scenarios. Compared with the maximum and minimum temperatures, the future changes in precipitation will be much more complicated. At monthly time scale, the five monthly precipitations of 1, 2, 8, 9 and 10 all showed an increasing trend under different climate combinations. Autumn and winter precipitation also showed an increasing trend.
According to the empirical downscaling method, the future climate of the MRW will show a trend of warming and humidification, which is consistent with the results obtained by Bao et al. [
34] and the future climate change of the Haihe River Watershed [
35]. All GCMs and RCPs showed that the annual and monthly maximum and minimum temperatures in the Miyun Reservoir watershed will gradually increase. Unlike temperature, future changes in precipitation at the monthly scale will be bidirectional and will depend on the GCMs selected and the evaluation time period. The two-way change in precipitation may be attributed to the simulation ability of GCM, because the precipitation simulation of different GCMs may be inconsistent with the magnitude and direction of the change in a specific area.
3.3. Climate Change Impact on Streamflow and Sediments
The spatial distribution of runoff changes in the watershed under different climate scenarios is shown in
Figure 9. At different time periods, watershed runoff varied greatly under greenhouse gas emission scenarios and global climate models. From 2020 to 2042, the reduction of watershed flow mainly occurred in the upper and middle reaches of the Chaohe and Baihe Rivers with a range of variation of 0–10 mm under ACCESS1.3-RCP8.5 and HadGEM-ES-RCP4.5 scenarios. Under the two other models, the runoff of the watershed basically showed an increasing trend. The sub-watershed runoff increased by 20 mm mainly concentrated in the lower reaches of the Chaohe and Baihe Rivers, compared with the base period watershed runoff basically showed an increasing trend at the sub-watershed level (from 2060 to 2082). Sub-watersheds with increasing runoff were mainly found in No. 27, 32 and 34 sub-watersheds of the Baihe River Watershed and No. 19, 20 and 22–29 sub-watersheds of Chaohe River.
The spatial distribution of sediment changes in the watershed under different climate scenarios is shown in
Figure 10. The sediment yield of the river watershed varied significantly under the different emission scenarios and global climate models over the different evaluation periods. From 2020 to 2042, the sediment yield in the watershed under ACCESS1.3-RCP4.5 and HadGEM-ES-RCP8.5 scenarios, basically showed an increasing trend concentrated in the 23rd sub-watershed of the Baihe River and in the 15th and 23rd of the Chaohe River. The 29th sub-watershed under ACCESS1.3-RCP8.5 and HadGEM-ES-RCP4.5 scenarios showed greater spatial differences in sediment yield in the watershed. Sediment yield in 13 sub-watersheds within the ACCESS1.3-RCP8.5 watershed was reduced, ranging from −0.484 to −0.001 kg/ha, with an average value of −0.06 kg/ha. Under HadGEM-ES-RCP4.5 scenario, there were 33 sub-watersheds in the watershed with reduced sediment yield output changes, ranging from −0.597 to −0.007 kg/ha, with an average value of −0.1 kg/ha; during the period from 2060 to 2082, the sediment yield in the watershed showed an increasing trend at the sub-watershed level. Compared with the base period, the sub-watersheds with an increase of more than 1 ton/ha of sediment yield in the sub-watershed were mainly distributed in the middle and lower reaches of the Chaohe River and the west of the Baihe River. The sub-watersheds with increased sediment yield in the future were mainly the No. 23 sub-watersheds of the Baihe River Watershed and the 5, 15, 16, 20, 23–25, 28 and 29 sub-watersheds of Chaohe River.
Figure 11 depicts the spatial distribution changes in the total nitrogen loading in the watershed under different climate scenarios. The total nitrogen load of the watershed varied greatly with the different emission scenarios and global climate models. From 2020 to 2042, the changes in total nitrogen load in the sub-watershed within the watershed, under ACCESS1.3-RCP4.5 and HadGEM-ES-RCP8.5 scenarios, showed an overall increasing trend. The watershed total nitrogen load increased mainly in Chaohe 17, Sub-watersheds 19, 20, 23, 24 and 28. Under ACCESS1.3-RCP8.5 and HadGEM-ES-RCP4.5 scenarios, the spatial difference in the total nitrogen load in the watershed was relatively large. Among them, under HadGEM-ES- RCP4.5 scenario, the total nitrogen load of 29 sub-watersheds in the watershed decreased, ranging from −0.707 to −0.008 kg/ha, with an average value of −0.12 kg/ha; during the period from 2060 to 2082, the total nitrogen load in the watershed showed an increasing trend at the sub-watershed level. Compared with the base period, the sub-watersheds with a variation in total nitrogen load of more than 1 kg/ha in the sub-watershed were distributed mainly in the upper and lower reaches of the Chaohe River and in the middle part of the Baihe River. Specifically, these sub-watersheds were the 18 sub-watersheds of the Baihe River Watershed and the 1, 5–8, 15–17, 19, 20, 23–25 and 28 sub-watersheds of Chaohe River.
Figure 12 depicts the spatial distribution of changes in total phosphorus load in the watershed under different climate scenarios. The total phosphorus load in the watershed varied considerably under different emission scenarios and global climate models. From 2020 to 2042, the variation in total phosphorus load in watershed sub-catchments under ACCESS1.3-RCP4.5, ACCESS1.3-RCP8.5 and HadGEM-ES-RCP8.5 scenarios showed an increasing trend and sub-catchments with a total variation in phosphorus output greater than 0.021 kg/ha were the No. 23 sub-watersheds of Baihe River and No. 15–17, 23 and 29 sub-watersheds of Chaohe River. For HadGEM-ES-RCP4.5 scenario, there were 22 sub-watersheds in the watershed showing a decreasing trend, ranging from −0.014 to −0.001 kg/ha, with an average value of −0.004 kg/ha; during the period from 2060 to 2082, the total phosphorus load in the watershed was maintained on an increasing trend at the sub-watershed level. Compared with the base period, the sub-watersheds with a total variation in phosphorus load greater than 0.021 kg/ha were distributed mainly in the middle and lower reaches of the Chaohe River and in the western part of the Baihe River. Specifically, these sub-watersheds were No. 6, 10, 14 and 23 of the Baihe River Watershed and No. 2, 5, 6, 12, 15–17, 19–21, 23–25, 28 and 29 of Chaohe River. Sub-watershed.
Table 4 describes watershed evaporation, runoff, total nitrogen and total phosphorus loading over different future evaluation periods by collective average. The results in the future evaluation, were all greater than the values of the corresponding variables in the reference period and the range of changes in the variables from 2060 to 2082 was greater than that in the 2020 to 2042 period.
Figure 13 depicts the estimated monthly average discharge, sediment, total nitrogen and total phosphorus loadings under different future climate change scenarios. It can be seen that there was no significant change in these parameters from January to May and December. Concerning the future change in the precipitation in the river watershed, it will be mainly concentrated in the flood season. Under HadGEM-ES-RCP8.5 climate scenario, runoff, sediment and total phosphorus decreased in July during 2020–2042 and 2060–2082 but showed an increasing trend in other months.
In the future, the spatial distribution of runoff, sediment, total nitrogen and total phosphorus loading in watershed will vary significantly with the evaluation period, emission scenarios and global climate models. During the period from 2060 to 2082, all sub-watersheds in the watershed basically showing an increasing trend; during the period from 2020 to 2042, under HadGEM-RCP4.5 scenario, half of the sub-watersheds in the watershed had runoff, sediment, total nitrogen and total phosphorus showing a decreasing trend. Sub-watersheds with a large increase in pollutants in the future were located mainly in the lower reaches of the Chaohe River Watershed. The results of the ensemble average of the GCM model showed that the evapotranspiration, runoff, total nitrogen and total phosphorus load in the future evaluation period were all greater than the baseline period and the changes in variables from 2060 to 2082 were greater than those in the period from 2020 to 2042.
3.4. Future Land Use Scenarios
Previous studies have used scenario analysis to assess the coupling impact of the future land use and future climate change on the hydrological process and non-point source output of the watershed. In the CLUE-S model, the establishment of scenario plans was mainly accomplished by defining different land use requirements and then combined with the spatial allocation module based on the grid of different land use types in the model to predict land use changes under different scenarios.
Based on the analysis of the land use structure, spatial distribution and change characteristics of the Miyun Reservoir watershed and regional planning, three scenarios were developed. Land use demand under different scenarios is used to input into the CLUE-S model to analyze the future spatial distribution change characteristics. The regional planning and regional policy documents involved in this research include “Chengde City Master Plan (2016–2030)”, “Zhangjiakou City Master Plan (2016–2030)”, “Beijing City Master Plan (2016–2030)”, “Beijing Major Function Zone Planning”, “Miyun New Town Planning (2005–2020)”, “Chaobai River Green Ecological Development Zone Comprehensive Planning (2010–2020)”, “Beijing Miyun Reservoir Huairou Reservoir He Jingmi Water Diversion Canal Protection and Management Regulations”.
Three scenarios are proposed as follows shown in
Figure 14:
Historical trend scenario: The future demand for land use follows the linear change trend of land use from 2000 to 2008. The overall performance of future land use changes is as follows: forest and urban land will increase and the area of grassland, water bodies, unused land and arable land will decrease. This scenario describes a scenario where there is no future intervention in land use change policies.
Ecological protection without consideration of spatial allocation scenario: Many water and soil conservation projects are currently being implemented in the Miyun River Watershed, such as: Taihang Mountain Greening Project, Beijing-Tianjin Sand Source Control Project. In the meantime, since the Miyun Reservoir is the source of drinking water in Beijing, environmental protection of the watershed is particularly important. Therefore, in the future, the annual growth rate of forest land, grassland and urban land will be 1.5, 0.5 and 1.5 times the historical trend scenario, respectively.
Ecological protection with consideration of spatial allocation scenario: Based on the results of the historical SWAT simulation from 1988 to 2010 and taking into account the spatial output characteristics of non-point source pollution in the watershed under future climate change and the Miyun Reservoir Watershed Protection Zone Division, specific regional preference variables were added to increase the probability of conversion of cultivated land to forest land in the secondary protection areas of the Miyun Reservoir watershed, watersheds above 25 °C and downstream sub-catchments of the Miyun Reservoir watershed. In this study, the regional weighting factor for forest land was set at 0.6 and the weighting factor for other land use types was set to 0. The rate of change of the different land use types was consistent with the ecological protection scenarios that did not consider the spatial allocation.
3.5. Climate Change and Land Use Change Impacts on Streamflow and NPS Loading
After importing the future land use map into the SWAT model database and using the SWAT model to simulate the non-point source pollution load in the Miyun Reservoir watershed, the annual average pollution load in the three land use scenarios was analyzed, noting that the simulation period was from 1988 to 2010. Comparative analysis was used to define the load reduction of sediment, total nitrogen and total phosphorus and the values are shown in
Table 5,
Table 6 and
Table 7, GCM1 and GCM2 mentioned in the tables stand for ACCESS1.3 and HadGEM-ES, respectively.
The simulation results showed that the ecological protection scenario considering the spatial configuration has a better reduction effect on the output of sediment and nutrients. In the historical reference period, the reductions in sediment, total nitrogen and total phosphorus were 12.88 × 104−ton, 280.918 ton and 8.116 ton, respectively. Under future climate change scenarios, the amount of sediment reduction will increase and the reduction of total nitrogen and total phosphorus will decrease. The average reduction rates of sediment, total nitrogen and total phosphorus were 11.4%, 6.3% and 7.4%, respectively. In the SWAT model, the surface runoff is calculated using the SCS (soil conservation service) method and its runoff will decrease as the value of the runoff curve number decreases. Since the CN value of forest land is smaller than that of cultivated land, the runoff will be reduced after returning farmland to forest. Corresponding reduction, the improvement of soil water retention capacity in the watershed can effectively reduce soil erosion, thereby reducing the output load of sediment; at the same time, as the large area of arable land is reduced, the overall fertilizer use in the watershed will be greatly reduced. The output of nitrogen and phosphorus nutrients will also be greatly reduced. At the same time, in the ecological protection scenario considering the spatial configuration, the sub-watersheds where the pollution load will increase in the future climate change scenario are considered, which will help to further reduce the sediment and nutrients.
4. Conclusions
Based on the empirical downscaling method, this study used two typical emission scenarios (RCP4.5 and RCP8.5) and two GCMS (ACCESS1.3 and HadGCM-ES) to generate future climate scenario data for the Miyun Reservoir basin. In general, the future climate will show a trend of warming and humidification.
It was found that future changes in the spatial distribution of watershed runoff, sediment, total nitrogen and total phosphorus load in the watershed will differ greatly depending on the evaluation period, emission scenarios and global climate models. During the period from 2060 to 2082, all sub-basins in the basin basically showed an increasing trend; during the period from 2020 to 2042 and under HadGEM-RCP4.5 scenario, half of the sub-basins in the basin had water production, sediment production, total nitrogen and total phosphorus showing a decreasing trend. In addition, the amount of load change showed a decreasing trend. It was also found that sub-basins with a strong increase in pollutants in the future, are mainly located in the lower reaches of the Chaohe River Basin. The results of the GCM ensemble average showed that the evapotranspiration, water production, total nitrogen and total phosphorus output load in the future evaluation period were all greater than the baseline period and the changes in variables during the period from 2060 to 2082 were greater than the changes in variables from 2020 to 2042.
After assessing the impact of climate change on runoff and NPS loading by the SWAT model, three scenarios were generated by the CLUE-S model to evaluate the reduction of sediment and nutrients in the watershed. After analyzing the relationship between the spatial distribution of land use types and driving factors in historical periods, the integrated impact of climate change and land use was evaluated. The results showed that land use change measures have a good reducing effect on the output of sediment and nutrients. Under ecological protection with consideration of spatial configuration scenario, the average reduction rates of sediment, total nitrogen and total phosphorus, were 11.4%, 6.3% and 7.4%, respectively. These results were explained by the fact that the ecological protection scenario takes fully into account the MRW protection zone policy, the policy of returning farmland to forest and the spatial variation of pollutants in the watershed under future climate scenarios. Therefore, the addition of region-specific preference variables under land use change setting provides better pollutant control effects under future climate scenarios.