4.3. PM2.5 and O3 Collaborative Pollution Response to Each Precursor
To quantitatively estimate the impact of each precursor on the PM2.5 and O3 collaborative pollution of each region, sensitivity analyses based on the PM2.5 and O3 estimation MTL models are applied to determine the contribution of each factor. Taking the concentration of each precursor in 2010 as a baseline, a scenario of changes in this precursor concentration was modeled by replacing the concentration of this precursor in each day of 2011–2020 with values from the corresponding date in 2010, while all other influencing factors of PM2.5 and O3 collaborative pollution remained unchanged. Subsequently, the impact of this specific precursor on PM2.5 and O3 was calculated as the PM2.5 and O3 difference between the simulated original PM2.5 and O3, and the PM2.5 and O3 simulated with this special precursor time series were replaced by its concentration from the corresponding date in 2010.
Figure 4 depicted the temporal patterns of the annual response of PM
2.5 and O
3 pollution to each precursor during 2011–2020 estimated by the MTL models in the NCP, the PRD, and the YRD. The results showed that among these seven PM
2.5 precursor factors, four influencing factors with the largest PM
2.5 pollution response in these three regions were SO
2, HCHO, NO
2, and SO
X, and a 10-year average value of PM
2.5 response for these four factors were larger than 0.30 μg m
−3. The variations of these four precursors had the greatest impact on PM
2.5 pollution in these three regions, which means that emission reduction in these four precursors is the most effective measure to mitigate PM
2.5 pollution in the NCP, the PRD, and the YRD. However, the two most crucial precursors of PM
2.5 are different across these three regions. The two most important influencing factors on PM
2.5 pollution in the NCP (YRD) are SO
2 and HCHO, with 10-year averaged PM
2.5 response values of 2.64 μg m
−3 (2.01 μg m
−3), and 1.71 μg m
−3 (0.95 μg m
−3), respectively. However, the two most important impact factors on PM
2.5 pollution in the PRD are SO
2 and NO
2, with 10-year averaged PM
2.5 response values of 1.24 μg m
−3 and 0.88 μg m
−3, respectively.
In terms of O3 pollution, it indicated that among these O3 precursors, three influencing factors with the largest O3 pollution response in these three regions were HCHO (the by-product of many VOC oxidation processes, and a critical proxy of VOC), NO2, and SOX, and the 10-year average O3 response values for these three factors were larger than 0.6 μg m−3. The variations of these three precursors had the greatest impact on O3 pollution in these three regions, which means that the emission reduction in VOC, NO2, and SOX is the most effective measure to mitigate O3 pollution in the NCP, the PRD, and the YRD. Moreover, the two most crucial precursors of O3 are different across these three regions. The two most important influencing factors on O3 pollution in the NCP (YRD) are HCHO and NO2, with 10-year averaged O3 response values of 1.00 μg m−3 (2.89 μg m−3) and 0.70 μg m−3 (0.55 μg m−3), respectively. However, the two most important impact factors on O3 pollution in the PRD are HCHO and SOX, with 10-year averaged O3 response values of 1.79 μg m−3 and 0.47 μg m−3, respectively.
In terms of the temporal patterns of PM2.5 and O3 pollution responses to precursor variations, the responses of PM2.5 and O3 pollution to different precursor factors in different regions have different temporal variation features. For PM2.5 pollution, the temporal variation patterns of the PM2.5 pollution response to SO2 in the NCP, the YRD, and the PRD were similar. And, the response values of PM2.5 pollution to SO2 in these three regions increased from 2011 to 2015, peaked in 2015, decreased from 2015 to 2017, reached a minimum value in 2017 (2018 for the PRD), then elevated again from 2017 (2018 for the PRD) to 2020, and reached a second peak in 2020. However, the temporal variation patterns of the response of PM2.5 to HCHO in these three regions were different. The response values of PM2.5 to HCHO in the NCP increased from 2011 to 2015, peaked in 2015, and decreased from 2015 to 2020, while the response values of PM2.5 pollution to HCHO in the PRD and the YRD showed an overall decreasing trend from 2011 to 2020. The temporal variation patterns of the response values of PM2.5 pollution to NO2 in these three regions were similar, and the response values of PM2.5 to NO2 increased from 2011 to 2015, peaked in 2015, and then decreased from 2015 to 2020. In addition, the temporal variation patterns of the response values of PM2.5 pollution to SOX in these three regions were different, and the response values of PM2.5 to SOX in the NCP increased from 2011 to 2014, peaked in 2014, and then decreased from 2014 to 2020, while the response values of PM2.5 to SOX in the PRD decreased from 2011 to 2016, reached a minimum value in 2016, and then increased from 2016 to 2020. Different from the response of PM2.5 to SOX in the NCP and the PRD, the response values of PM2.5 to SOX in the YRD showed an overall decreasing trend from 2011 to 2020. For O3 pollution, the temporal variation patterns of the O3 response to HCHO in the NCP, the YRD, and the PRD were similar. The response values of O3 to HCHO in these three regions increased from 2011 to 2014, peaked in 2014, and decreased from 2014 to 2020. The temporal variation patterns of the response values of O3 pollution to NO2 in these three regions were similar, and the response values of O3 to NO2 increased from 2011 to 2015 (2014 for the PRD), peaked in 2015 (2014 for the PRD), and then decreased from 2015 (2014 for the PRD) to 2020. However, the temporal variation patterns of the response of O3 to SOX in these three regions were different. The response values of O3 to SOX in the NCP and the YRD showed an overall increasing trend from 2011 to 2020, while the response values of O3 pollution to SOX in the PRD decreased from 2011 to 2014, reached a minimum value in 2014, and then increased from 2014 to 2020. In addition, the temporal variation patterns of the response values of O3 pollution to SO2 in these three regions were different, and the response values of O3 to SOX in the NCP and the YRD increased from 2011 to 2014, peaked in 2014, and then decreased from 2014 to 2020, while the response values of O3 to SO2 in the PRD increased from 2011 to 2015, reached a minimum value in 2015, decreased from 2015 to 2017, reached a minimum value in 2017, and then increased from 2017 to 2020. Different from the response of O3 to HCHO, SOX, NO2, and SO2 in these three regions, the response values of O3 to VOC showed an overall increasing trend from 2011 to 2020.
Figure 5 depict the spatial patterns of the annual response of PM
2.5 pollution to each precursor during 2011–2020 estimated by the MTL models in the NCP. The results indicated that among these precursors, SO
2, HCHO, SO
X, and NO
2 were the four most important precursors for PM
2.5 pollution in the NCP, which was consistent with the results presented in
Figure 4a. Among these precursors, the amplitude of PM
2.5 response to SO
2 was the largest, with the values of the PM
2.5 response to SO
2 higher than +12 μg m
−3 in parts of Shandong Province in 2015 and the PM
2.5 response to SO
2 lower than −18 μg m
−3 in parts of the Jing-Jin-Ji (Beijing, Tianjin, and Hebei provinces) region in 2020. This may be related to the isotropic feedback of PM
2.5 pollution on SO
2 [
58]. Compared with 2010, the SO
2 concentration in (shown in
Figure S1) 2011 in the NCP was relatively higher, so the response value of PM
2.5 to SO
2 in 2011 was positive. From 2012 to 2013, SO
2 concentrations in the NCP were lower than those in 2010, especially in the Jing-Jin-Ji region (Beijing, Tianjin, and Hebei provinces) and the Lu-Yu region (Shandong and Henan provinces), so the response values of PM
2.5 pollution to SO
2 were negative from 2012 to 2013. During 2014–2017, SO
2 concentrations in the NCP increased significantly and were larger than those in 2010, especially in Hebei, Shandong, Henan, and Shanxi provinces, so the response values of PM
2.5 to SO
2 in 2014–2017 were positive, and the response value was largest in 2015. In contrast, SO
2 concentrations from 2014 to 2017 in Beijing and Tianjin were lower than those in 2010, so the response values of PM
2.5 to SO
2 in Beijing and Tianjin were negative, and the response value was largest in 2017. From 2018 to 2020, SO
2 concentrations in the NCP were smaller than those in 2010, especially in the Jing-Jin-Ji region and the Lu-Yu region, so the response value of PM
2.5 to SO
2 in 2018–2020 was negative, and the amplitude of the response value was largest in 2020.
In addition to SO
2, the amplitude of the response of PM
2.5 pollution to HCHO (the by-product of many VOC oxidation processes, and a critical proxy of VOC) was the second largest among all precursors, with the PM
2.5 response values to HCHO higher than +6 μg m
−3 in parts of the Ji-Lu-Yu (Hebei, Henan, and Shandong provinces) region in 2011. This may be related to the positive feedback of PM
2.5 pollution on HCHO (also known as VOC). Compared with 2010, the HCHO (i.e., VOC) concentration in 2011–2020 in the main region of Ji-Lu-Yu was relatively higher, so the response value of PM
2.5 to VOC in 2011–2020 was positive. However, the response value of PM
2.5 to VOC in the main regions of the NCP increased during 2011–2015, reached the peak value in 2015, and then decreased during 2015–2020. Meanwhile, HCHO concentrations (shown in
Figure S2) and the anthropogenic emissions of VOC (shown in
Figure S3) showed an overall increasing trend from 2010 to 2020. This may be related to the nonlinear relationship between HCHO (VOC emissions) and PM
2.5 pollution in the NCP [
57,
59,
60,
61]. Before 2015, VOC emissions were not saturated for PM
2.5 pollution, and the positive response of PM
2.5 pollution increased gradually with the increase in VOC emissions. However, VOC emissions began to saturate for PM
2.5 pollution after 2015, and the contribution of rising VOC emissions to the increase in PM
2.5 pollution levels is gradually declining, where the positive response of PM
2.5 pollution gradually decreased with the increase in VOC emissions between 2015 and 2020.
Besides SO
2 and HCHO, the amplitude of the response of PM
2.5 pollution to SO
X is the third largest among all precursors, with the PM
2.5 response values to SO
X higher than +4 μg m
−3 in parts of the Ji-Lu-Yu region in 2014. This may be due to the positive feedback of PM
2.5 pollution on SO
X. Compared with 2010, the SO
X concentration (shown in
Figure S4) in 2011–2020 in the main region of Ji-Lu-Yu was relatively higher, so the response value of PM
2.5 to SO
X in 2011–2020 was positive. However, the response value of PM
2.5 to SO
X in the main regions of the NCP increased during 2011–2014, reached the peak value in 2014, and then decreased during 2014–2020. Meanwhile, SO
X concentrations (shown in
Figure S6) increased from 2010 to 2014, and then decreased during 2014–2020.
Moreover, the amplitude of the response of PM
2.5 pollution to NO
2 is the fourth largest among all precursors, with PM
2.5 response values to NO
2 higher than +3 μg m
−3 in parts of the Ji-Lu-Yu region in 2015. This may be due to the positive feedback of PM
2.5 pollution on NO
2 [
33,
62,
63]. Compared with 2010, the NO
2 concentration (shown in
Figure S5) in 2011–2012 in the central and southern NCP was lower, so the response value of PM
2.5 to NO
2 in 2011–2012 was negative. However, the NO
2 concentration in 2013–2017 in the main part of the NCP was relatively higher, so the response value of PM
2.5 to NO
2 in 2013–2017 was positive. From 2018 to 2020, the NO
2 concentration in the main part of the NCP was relatively lower, so the response value of PM
2.5 to NO
2 in 2018–2020 was negative. Compared with these four precursors, the response of PM
2.5 pollution to CAMS-VOC, aerosol component, and NO
X were relatively smaller.
Figure 6 depicts the spatial patterns of the annual response of O
3 pollution to each precursor during 2011–2020 estimated by the MTL models in the NCP. The results indicated that among these precursors, HCHO, NO
2, SO
X, and VOC were the most important four precursors for O
3 pollution in the NCP, which was consistent with the results shown in
Figure 4b. Among these precursors, the amplitude of the O
3 response to HCHO was the largest, with O
3 response values to HCHO higher than +5 μg m
−3 in northwestern parts of the NCP. This may be due to the different response patterns of O
3 to HCHO in different regions. In the Ji-Lu-Yu region, the response of O
3 pollution to HCHO was negative, and the level of O
3 pollution decreased with the increase in HCHO concentration. Compared with 2010, the HCHO concentration (shown in
Figure S2) in 2011–2020 in the Ji-Lu-Yu region was relatively higher, so the response value of O
3 to HCHO was negative during 2011–2020. In contrast, the responses of O
3 pollution to HCHO in the northwestern part of the NCP (Shanxi province and northwest part of Hebei province) and southeastern part of the NCP (Anhui province, the Jiangsu province, and the southeast part of Shandong province) were negative from 2011 to 2016, and then turned positive from 2017 to 2020. The level of O
3 pollution in the northwestern and northwest parts of the NCP increased with the decrease in HCHO concentration between 2011 and 2016, and then increased with the increase in HCHO concentration between 2017 and 2020. Compared with 2010, the HCHO concentration in 2011–2016 in the northwestern and northwest parts of the NCP was relatively lower, and the response value of O
3 to HCHO in this region was positive during 2011–2016; however, the HCHO concentration in 2017–2020 in this region was relatively higher, and the response value of O
3 to HCHO in this region was positive during 2017–2020.
For NO
2, the response value of O
3 pollution to NO
2 in the main region of the NCP decreased from 2011 to 2016, reached a minimum value in 2016, and then increased from 2016 to 2020. This may be related to the negative feedback of O
3 pollution on NO
2 [
64,
65]. The NO
2 concentration in 2011–2012 in the main region of the NCP was relatively lower than that in 2010, and the response value of O
3 to NO
2 in 2011–2012 was positive. Compared with 2010, NO
2 concentration (shown in
Figure S5) in the NCP increased from 2013 to 2016 and was relatively higher, and the response value of O
3 pollution to NO
2 was negative and decreased from 2013 to 2016. Compared to 2010, NO
2 concentrations in the NCP were relatively higher between 2017 and 2019 and relatively lower in 2020, and NO
2 concentrations decreased from 2017 to 2020. Meanwhile, the response value of O
3 pollution to NO
2 decreased from 2017 to 2020, and the response value was negative between 2017 and 2019 and was positive in 2020.
During 2011–2020, the response value of O
3 pollution to SO
X in the NCP increased between 2011 and 2014, reached the peak value in 2014, and then decreased from 2014 to 2020. This may be related to the isotropic feedback of O
3 pollution on SO
X. The SO
X emission in the northern Hebei province, Beijing, and Tianjin in 2011–2015 were smaller than that in 2010 (shown in
Figure S4), so the response value of O
3 to SO
X was negative in this region from 2011 to 2015, while the SO
X in the border area of the Hebei, Henan, and Shandong provinces in 2011–2015 were larger than that in 2010, so the response value of O
3 to SO
X was positive in this region during 2011–2015. Meanwhile, SO
X emission in the NCP during 2016–2020 was lower than that in 2010; therefore, the response value of O
3 to SO
X in 2016–2020 is negative in the NCP.
For VOC, the response of O
3 to VOC in the NCP increased from 0.1 μg m
−3 in 2011 to 2 μg m
−3 in 2020. This may be related to the isotropic feedback of O
3 pollution on VOC in the NCP [
66,
67,
68]. VOC emissions in the NCP increased from 2011 to 2020 and were relatively higher than that in 2010 (shown in
Figure S3), and the response value of O
3 pollution to VOC was positive and increased from 2011 to 2020. Compared with these four precursors, the response of O
3 pollution to SO
2, NO
X, and aerosol components were relatively smaller.
Figure 7 depicts the spatial patterns of the annual response of PM
2.5 pollution to each precursor during 2011–2020 estimated by the MTL models in the YRD. Similar with the NCP, the results indicated that among these precursors, SO
2, HCHO, NO
2, and SO
X were the four most important precursors for PM
2.5 pollution in the YRD, which was consistent with the results presented in
Figure 4c. Among all precursors, the amplitude of the PM
2.5 response to SO
2 was the largest, with the values of the PM
2.5 response to SO
2 higher than +9 μg m
−3 in the south part of Jiangsu province in 2015 and the PM
2.5 response to SO
2 lower than −12 μg m
−3 in the north parts of the YRD in 2020. This may be due to the isotropic feedback on PM
2.5 pollution to SO
2. Compared with 2010, the SO
2 concentration (shown in
Figure S6) in 2011 in the YRD was relatively high, so the response value of PM
2.5 to SO
2 in 2011 was positive. From 2012 to 2013, the SO
2 concentration in the YRD was lower than that in 2010, especially in the north part of the YRD, so the response values of PM
2.5 pollution to SO
2 were negative from 2012 to 2013. However, the SO
2 concentration in the YRD increased significantly in 2014–2017, and was larger than that in 2010, so the response values of PM
2.5 to SO
2 in 2014–2017 were positive. Meanwhile, the amplitude of the response value increased between 2014 and 2015, reached the peak value in 2015, and then decreased in 2016–2017. From 2018 to 2020, the SO
2 concentration in the YRD decreased significantly and was lower than that in 2010, especially in the north part of the YRD, so the response value of PM
2.5 to SO
2 in 2018–2020 was negative, and the response value decreased in 2018–2020.
In addition to SO
2, the amplitude of the response of PM
2.5 pollution to HCHO is the second largest among all precursors, with the PM
2.5 response values to HCHO higher than +2.5 μg m
−3 in the north parts of the YRD in 2011. This may be due to the different response patterns of PM
2.5 pollution to HCHO in different regions. The response of PM
2.5 to HCHO in the southern part of the YRD (including the southern part of the Anhui province, the northern part of Zhejiang province, and Shanghai) were likely due to negative feedback between 2011 and 2020. HCHO concentration (shown in
Figure S7) in the southern part of the YRD increased in 2011–2020, and was higher than that in 2010, but the responses of PM
2.5 to HCHO were negative during 2011–2020 in this region. In the scenarios simulated by the MTL model, simulated PM
2.5 concentration decreased in 2011–2020 when it was compared with that in 2010. However, the response of PM
2.5 to HCHO in the northern part of the YRD (including the northern part of the Anhui province and the Jiangsu province) were likely due to negative feedback between 2011 and 2015 and turned to be positive feedback between 2016 and 2020. The HCHO concentration in the northern part of the YRD decreased in 2011–2015, and was lower than that in 2010, and the responses of PM
2.5 to HCHO were positive during 2011–2015 in this region, while HCHO concentration in this region increased in 2016–2020, and was higher than that in 2010, and the responses of PM
2.5 to HCHO were positive during 2016–2020 in this region.
Besides SO
2 and HCHO, the amplitude of the response of PM
2.5 pollution to NO
2 is the third largest among all precursors, with PM
2.5 response values to NO
2 higher than +5 μg m
−3 in the central and southern YRD in 2015. This may be due to the positive feedback of PM
2.5 pollution on NO
2 [
33,
62,
63]. Compared with 2010, the NO
2 concentration (shown in
Figure S8) in 2011–2012 in the northeastern part of the YRD was lower, so the response value of PM
2.5 to NO
2 in 2011–2012 was negative in this region, while the NO
2 concentration in 2011–2012 in the central part of the YRD was higher, so the response value of PM
2.5 to NO
2 in 2011–2012 was positive in this region. Moreover, the NO
2 concentration in 2013 in the YRD was relatively higher than that in 2010, so the response value of PM
2.5 to NO
2 in 2013 was positive. From 2014 to 2020, the NO
2 concentration in the northern YRD was relatively lower than that in 2010, so the response value of PM
2.5 to NO
2 in 2014–2020 was negative in this region, while NO
2 concentration in the central and southern YRD was relatively higher than that in 2010, so the response value of PM
2.5 to NO
2 in the central and southern YRD was positive from 2014 to 2020.
Moreover, the amplitude of the response of PM
2.5 pollution to SO
X is the fourth largest among all precursors, with PM
2.5 response values to SO
X higher than +3 μg m
−3 in the Anhui province in 2013. This may be due to the positive feedback of PM
2.5 pollution to SO
X. SO
X concentrations in 2011, 2013–2015, and 2019 in the YRD were relatively higher than that in 2010 (shown in
Figure S9), so the response values of PM
2.5 to SO
X in 2011, 2013–2015, and 2019 were positive. However, SO
X concentrations in 2012 in the northern YRD were relatively lower than that in 2010, so the response values of PM
2.5 to SO
X in 2012 were negative in northern YRD, while SO
X concentrations in 2012 in southern YRD were relatively higher than that in 2010, so the response values of PM
2.5 to SO
X in 2012 were positive in the southern YRD. In addition, SO
X concentrations in 2016–2018 and 2020 in the YRD were relatively lower than that in 2010, so the response values of PM
2.5 to SO
X in 2016–2018 and 2020 were negative in the YRD. Compared with these four precursors, the response of PM
2.5 pollution to CAMS-VOC, aerosol components, and NO
X were relatively smaller.
Figure 8 depicts the spatial patterns of the annual response of O
3 pollution to each precursor during 2011–2020 estimated by the MTL models in the YRD. It indicated that among these precursors, HCHO, NO
2, SO
X, and VOC were the most important four precursors for O
3 pollution in the YRD, which is consistent with the results presented in
Figure 4d. Among these precursors, the amplitude of the O
3 response to HCHO was the largest in 2015, with the O
3 response to HCHO higher than +9 μg m
−3 in eastern parts of the Jiangsu province. This may be due to the different response patterns of O
3 to HCHO in different regions. The response of O
3 to HCHO in the southern part of the YRD (including the southern part of the Anhui province, the northern part of the Zhejiang province, and Shanghai) were likely due to negative feedback between 2011 and 2020. HCHO concentration (shown in
Figure S7) in the southern part of the YRD increased in 2011–2020, and was higher than that in 2010, but the responses of O
3 to HCHO were negative during 2011–2020 in this region. However, the response of O
3 to HCHO in the northern part of the YRD (including the northern part of the Anhui province and the Jiangsu province) were likely due to negative feedback between 2011 and 2015, and turned to be positive feedback between 2016 and 2020. HCHO concentration in the northern part of the YRD decreased in 2011–2015, and was lower than that in 2010, and the responses of O
3 to HCHO were positive during 2011–2015 in this region, while HCHO concentration in this region increased in 2016–2020 and was higher than that in 2010, and the responses of O
3 to HCHO were positive during 2016–2020 in this region.
For NO
2, the response value of O
3 pollution to NO
2 in the main region of the YRD decreased from 2011 to 2016, reached a minimum value in 2016, and then increased from 2016 to 2020. This may be related to the negative feedback of O
3 pollution on NO
2 [
64,
65]. The NO
2 concentration (shown in
Figure S8) in 2011–2012 in northeastern YRD was relatively lower than that in 2010, and the response value of O
3 to NO
2 in 2011–2012 was positive, while NO
2 concentration in 2011–2012 in the central part of the YRD was higher, so the response value of O
3 to NO
2 in 2011–2012 was negative in this region. Moreover, the NO
2 concentration in 2013 in the main part of the YRD was relatively higher than that in 2010, so the response value of O
3 to NO
2 in 2013 was negative. From 2014 to 2020, the NO
2 concentration in northern YRD was relatively lower than that in 2010, and the response value of O
3 to NO
2 in 2014–2020 was positive in this region, while the NO
2 concentration in central and southern YRD were relatively higher than that in 2010, and the response value of O
3 to NO
2 in central and southern YRD were negative from 2014 to 2020.
Moreover, the amplitude of the response of O
3 pollution to SO
X is the third largest among all precursors, with O
3 response values to SO
X lower than –3 μg m
−3 in the Anhui province in 2020. This may be due to the different response patterns of O
3 pollution to SO
X in different regions. The responses of O
3 to SO
X in the YRD were likely due to negative feedback in 2011. From 2012 to 2013, the responses of O
3 to SO
X in the Jiangsu province were likely due to positive feedback, while the responses of O
3 to SO
X in the Anhui and Zhejiang provinces were likely due to negative feedback. In addition, from 2014 to 2020, the responses of O
3 to SO
X in the YRD were likely due to positive feedback. In 2011, SO
X concentrations in the southwestern region of the YRD were relatively higher than that in 2010 (shown in
Figure S9), so the response values of O
3 to SO
X in 2011 were negative, while SO
X concentrations in the northeastern region of the YRD were relatively lower than that in 2010, and the response values of O
3 to SO
X in this region were positive. However, from 2012 to 2013, SO
X concentrations in the northeastern region of the YRD were relatively lower than that in 2010, so the response values of O
3 to SO
X in this region were negative, while SO
X concentrations in the other parts of the YRD were relatively higher than that in 2010, and the response values of O
3 to SO
X in these regions were negative. In addition, during 2014–2020, SO
X concentrations in the main part of the YRD were relatively lower than that in 2010, so the response values of O
3 to SO
X were negative in this region; however, SO
X concentrations in western YRD in 2014, and in Shanghai and the Jiangsu province in 2017, were higher than those in 2010, and, therefore, the values of the O
3 response to SO
X were positive in these regions.
For VOC, the response of O
3 to VOC in the YRD increased from 0.1 μg m
−3 in 2011 to 1.5 μg m
−3 in 2020. This may be related to the isotropic feedback of O
3 pollution on VOC in the YRD [
66,
67,
68]. VOC emissions in the YRD increased from 2011 to 2020 and were relatively higher than that in 2010 (shown in
Figure S10), and the response value of O
3 pollution to VOC was positive and increased from 2011 to 2020. Compared with these four precursors, the response of O
3 pollution to SO
2, NO
X, and aerosol components were relatively smaller.
Figure 9 depicts the spatial patterns of the annual response of PM
2.5 pollution to each precursor during 2011–2020 estimated by the MTL models in the PRD. Similar with the NCP and the YRD, the results indicated that among these precursors, SO
2, NO
2, HCHO, and SO
X were the four most important precursors for PM
2.5 pollution in the PRD, which are consistent with the results present in
Figure 4e. Among these precursors, the amplitude of the PM
2.5 response to SO
2 was the largest, with the values of the PM
2.5 response to SO
2 higher than +6 μg m
−3 in the western region of the PRD in 2015 and the PM
2.5 response to SO
2 lower than −12 μg m
−3 in the central part of the PRD in 2020. This may be due to the isotropic feedback of PM
2.5 pollution on SO
2 [
58]. Compared with 2010, SO
2 concentration (shown in
Figure S11) in 2011, and 2013–2014, in the PRD was relatively higher, so the response value of PM
2.5 to SO
2 in 2011, and 2013–2014, were positive. In 2012, SO
2 concentration in the PRD was lower than that in 2010, and the values of the PM
2.5 response to SO
2 were negative. In the northwestern part of the PRD, the SO
2 concentration was larger than that in 2010, but it decreased significantly from 2015 to 2020, and, therefore, values of the PM
2.5 response to SO
2 were all positive in 2015–2020, but the spatial extent covered by positive values gradually decreased. In contrast, the SO
2 concentration in the southeastern part of the PRD was lower than that in 2010, and it decreased significantly from 2015 to 2020, and, therefore, values of the PM
2.5 response to SO
2 were all negative in this period, but the spatial extent covered by negative values gradually increased.
In addition to SO
2, the amplitude of the PM
2.5 response to NO
2 was the second largest among all precursors, with values of the PM
2.5 response to NO
2 higher than +5 μg m
−3 in western PRD in 2014. This may be due to the positive feedback of PM
2.5 pollution on NO
2 [
33,
62,
63]. In 2011–2013, NO
2 concentration (shown in
Figure S12) in the PRD was lower than that in 2010, so values of PM
2.5 response to NO
2 were negative in this region. From 2014 to 2020, the NO
2 concentration in the main part of the PRD was relatively higher than that in 2010, so value of the PM
2.5 response to NO
2 was positive. Moreover, NO
2 concentrations in the southeastern part of the PRD between 2015 and 2019 and in the southern part of the PRD in 2020 were relatively lower than that in 2010, so the response value of PM
2.5 to NO
2 was negative.
Besides SO
2 and NO
2, the amplitude of the response of PM
2.5 pollution to HCHO is the second largest among all precursors, with the PM
2.5 response values to HCHO higher than +2 μg m
−3 in north parts of the PRD in 2011. This may be due to the different response patterns of PM
2.5 pollution to HCHO in different regions. The response of PM
2.5 to HCHO was likely due to negative feedback in northwestern PRD in 2011 and the entire PRD in 2013, while the response of PM
2.5 to HCHO was likely due to positive feedback in central and southeastern PRD in 2011, and in the entire PRD in 2012 and 2014–2020. The HCHO concentration (shown in
Figure S13) was lower in the northwestern YRD in 2011, and the entire PRD in 2013, than that in 2010, and then values of the PM
2.5 response to HCHO were positive. Moreover, the HCHO concentration was higher in central and southeastern PRD in 2011, and in the entire PRD in 2012 and 2014–2020, than that in 2010, and then values of the PM
2.5 response to HCHO were positive. HCHO concentration increased between 2014 and 2020; however, the positive values of the PM
2.5 response to HCHO decreased in this period. This may be related to the nonlinear relationship between HCHO (VOC emissions) and PM
2.5 pollution in the PRD [
57,
59,
60,
61]. After 2014, HCHO (VOC emissions) began to saturate for PM
2.5 pollution, and the contribution of rising HCHO (VOC emissions) to the increase in PM
2.5 pollution levels gradually declined, and the positive response of PM
2.5 pollution gradually decreased with the increase in VOC emissions between 2015 and 2020.
Moreover, the amplitude of the response of PM
2.5 pollution to SO
X is the fourth largest among all precursors, with the PM
2.5 response values to SO
X higher than +1 μg m
−3 in the southwestern province in 2011. This may be due to the positive feedback of PM
2.5 pollution on SO
X. SO
X concentrations decreased between 2011 and 2020 and were higher in 2011–2014 than that in 2010 (shown in
Figure S14), so the response values of PM
2.5 to SO
X in 2011–2014 were positive, while SO
X concentrations were lower in 2015–2020 than that in 2010, and, therefore, values of the PM
2.5 response to SO
X in 2015–2020 were negative. Compared with these four precursors, the response of PM
2.5 pollution to CAMS-VOC, aerosol components, and NO
X were relatively smaller.
Figure 10 depicts the spatial patterns of the annual response of O
3 pollution to each precursor during 2011–2020 estimated by the MTL models in the PRD. It indicated that among these precursors, HCHO, NO
2, SO
X, SO
2, and VOC were the most important five precursors for O
3 pollution in the PRD, which are consistent with the results present in
Figure 4f. Among these precursors, the amplitude of the O
3 response to HCHO was the largest in 2015, with the values of the O
3 response to HCHO higher than +4 μg m
−3 in northern parts of the PRD. This may be due to the different response patterns of O
3 to HCHO in different regions. The response of O
3 to HCHO in central PRD was likely due to negative feedback in 2011–2012 and in 2014–2020, while the response of O
3 to HCHO in the other regions of the PRD, except the central part, was likely due to positive feedback in 2012 and in 2014–2020. Except for the central region of the PRD, the response of O
3 to HCHO was likely due to negative feedback in 2011, and in contrast, the response was likely due to positive feedback in 2013. The HCHO concentration was higher in central PRD in 2011–2012, and 2014–2020, than that in 2010, but the responses of O
3 to HCHO were negative, while the HCHO concentration was higher in the other regions of the PRD, except the central part, in 2012 and 2014–2020 than that in 2010, and then the responses of O
3 to HCHO were positive. The HCHO concentration was lower in 2011 in the other regions of the PRD, except the central part, than that in 2010, but the responses of O
3 to HCHO were positive. However, the HCHO concentration (shown in
Figure S13) was lower in central PRD in 2013 than that in 2010, and, therefore, the responses of O
3 to HCHO were negative, while the HCHO concentration was lower in the other regions of the PRD, except the central part, in 2013 than that in 2010, but the responses of O
3 to HCHO were positive.
Moreover, the amplitude of the response of O
3 pollution to SO
X is the second largest among all precursors, with O
3 response values to SO
X lower than –3 μg m
−3 in southern PRD in 2011. This may be due to the positive feedback of O
3 pollution on SO
X. SO
X concentrations decreased between 2011 and 2020 and were higher in 2011–2014 than that in 2010 (shown in
Figure S14), and, therefore, the response values of O
3 to SO
X in 2011–2014 were positive, while SO
X concentrations were lower in 2015–2020 than that in 2010, and, therefore, values of the O
3 response to SO
X in 2015–2020 were negative.
For NO
2, the response value of O
3 pollution to NO
2 in the main region of the PRD decreased from 2011 to 2016, reached a minimum value in 2016, and then increased from 2016 to 2020. This may be related to the different response patterns of O
3 pollution to NO
2 in the PRD [
64,
65]. The response of O
3 to NO
2 was likely due to negative feedback in 2011–2013, while the response of O
3 to NO
2 was likely due to positive feedback in central PRD from 2014 to 2020. However, the response of O
3 to NO
2 was likely due to positive feedback in northeastern PRD in 2014–2015 and 2017–2020, while the response of O
3 to NO
2 was likely due to negative feedback in this region in 2016. In addition, the response of O
3 to NO
2 was likely due to negative feedback in northwestern PRD in 2014–2020, while the response of O
3 to NO
2 was likely due to positive feedback in southwestern PRD from 2015 to 2020. However, the response of O
3 to NO
2 was likely due to negative feedback in southwestern PRD in 2014. The NO
2 concentration in 2011–2013 was relatively lower than that in 2010, and the response value of O
3 to NO
2 was positive (shown in
Figure S12), while the NO
2 concentration in central PRD was higher during 2014–2020 than that in 2010, so the response value of O
3 to NO
2 was positive. Moreover, the NO
2 concentration in northeastern PRD was relatively higher than that in 2010, so the response value of O
3 to NO
2 was positive, while the NO
2 concentration was relatively higher in this region in 2016 than that in 2010, but the response value of O
3 to NO
2 was negative in 2016. The NO
2 concentration was relatively higher in northwestern PRD in 2014–2020 than that in 2010, but the response value of O
3 to NO
2 in 2014–2020 was negative, while the NO
2 concentration was relatively lower in southwestern PRD during 2015–2020 than that in 2010, and the response value of O
3 to NO
2 was negative. However, the NO
2 concentration was relatively higher in southwestern PRD than that in 2010, and the response value of O
3 to NO
2 was negative in this region from 2015 to 2020.
For SO
2, the response value of O
3 pollution to SO
2 in the main region of the PRD increased from 2011 to 2015, reached the peak value in 2015, and then decreased from 2016 to 2020. This may be due to the different response patterns of O
3 pollution to SO
2. The response of O
3 to SO
2 was likely due to negative feedback in 2012, while the response of O
3 to SO
2 was likely due to positive feedback in 2011 and in 2013–2020. Compared with 2010, SO
2 concentration in 2011 and 2013–2014 in the PRD were relatively higher (shown in
Figure S11), so the response value of O
3 to SO
2 in 2011 and 2013–2014 were positive. In 2012, the SO
2 concentration in the PRD was lower than that in 2010, and the values of the O
3 response to SO
2 were negative. In the northwestern part of the PRD, the SO
2 concentration was larger than that in 2010, but it decreased significantly from 2015 to 2020, and, therefore, values of the O
3 response to SO
2 were all positive in 2015–2020, but the spatial extent covered by positive values gradually decreased. In contrast, the SO
2 concentration in the southeastern part of the PRD was lower than that in 2010, and it decreased significantly from 2015 to 2020, and, therefore, values of the O
3 response to SO
2 were all negative in this period, but the spatial extent covered by negative values gradually increased.
For VOC, the response of O
3 to VOC in the PRD increased from 0.1 μg m
−3 in 2011 to 2.0 μg m
−3 in 2020. This may be related to the isotropic feedback of O
3 pollution on VOC in the PRD [
66,
67,
68]. VOC emissions in the PRD increased from 2011 to 2020 and were relatively higher than that in 2010 (shown in
Figure S15), and the response value of O
3 pollution to VOC was positive and increased from 2011 to 2020. Compared with these four precursors, the response of O
3 pollution to SO
2, NO
X, and aerosol components were relatively smaller.