key: cord-0783969-hj0fnxis authors: Li, Huiming; Dai, Qian'ying; Yang, Meng; Li, Fengying; Liu, Xuemei; Zhou, Mengfan; Qian, Xin title: Heavy metals in submicronic particulate matter (PM(1)) from a Chinese metropolitan city predicted by machine learning models date: 2020-07-08 journal: Chemosphere DOI: 10.1016/j.chemosphere.2020.127571 sha: ca9c2431920933ecb95380ef7a9aa6f0ae1c4e2f doc_id: 783969 cord_uid: hj0fnxis The aim of this study was to establish a method for predicting heavy metal concentrations in PM(1) (aerosol particles with an aerodynamic diameter ≤ 1.0 μm) based on back propagation artificial neural network (BP-ANN) and support vector machine (SVM) methods. The annual average PM(1) concentration was 26.31 μg/m(3) (range: 7.00–73.40 μg/m(3)). The concentrations of most metals were higher in winter and lower in autumn and summer. Mn and Ni had the highest noncarcinogenic risk, and Cr the highest carcinogenic risk. The hazard index was below safe limit, and the integrated carcinogenic risk was less than precautionary value. There were no obvious differences in the simulation performances of BP-ANN and SVM models. However, in both models many elements had better simulation effects when input variables were atmospheric pollutants (SO(2), NO(2), CO, O(3) and PM(2.5)) rather than PM(1) and meteorological factors (temperature, relative humidity, atmospheric pressure and wind speed). Models performed better for Pb, Tl and Zn, as evidenced by training R and test R values consistently >0.85, whereas their performances for Ti and V were relatively poor. Predicted results by the fully trained models showed atmospheric heavy metal pollution was heavier in December and January and lighter in August and July of 2019. For the period covering the COVID-19 outbreak in China, from January to March 2020, most of the predicted element concentrations were lower than in 2018 and 2019, and the concentrations of nearly all metals were lowest during the nationwide implementation of countermeasures taken against the pandemic. Urban air pollution is one of the most serious environmental issues confronting In this study, the machine learning models BP-ANN and SVM were used to 153 simulate the concentrations of particle-bound heavy metals. Atmospheric pollutant 8 concentrations (SO 2 , NO 2 , CO, O 3 and PM 2.5 ), meteorological factors (temperature, 155 relative humidity, atmospheric pressure and wind speed) and the PM 1 concentration 156 served as input variables. A total of four models were developed according to the 157 input variables and methods (Table 1) . All data were randomly partitioned into two 158 sets, with 75% for the training set and 25% for the test set. The optimal BP-ANN and 159 SVM models were chosen based on higher correlation coefficients and lower errors in 160 the training and test stages. between the predicted concentration and the expected output does not meet the 169 accuracy requirement, the error is propagated from the output layer such that the 170 weight and threshold are adjusted until the accuracy requirement is met. In this study, 171 MATLAB R2013a was used to establish the BP-ANN models. The SVM models were established using MATLAB R2013a and libsvm-3.21. The volume-and mass-based concentrations of heavy metals during the four 235 seasons are presented in Table 2 and Table S2 was highest in spring (Table 2) whereas the Cr concentration was higher in autumn 240 than in spring, and the Ni concentration was slightly higher in summer than in autumn. Ni, Pb and Tl were lowest in summer (Table S2) . As shown in Table S3 , for heavy metals in PM 1 , the concentrations recorded in Pb were all less than the precautionary value set for children and adults (10 −4 ). The The volume-related concentrations of heavy metals in PM 1 were simulated using 316 BP-ANN and SVM models ( and test R values were <0.8. According to the EF (Fig. S1 ), Pb and Zn were greatly (Table S12) . By contrast, the seasonal decreases in the concentrations of As 396 and Mn followed the pattern spring > winter > autumn > summer, and those of Co and 397 Ti the pattern spring > autumn > summer > winter. The COVID-19 pandemic struck China at the beginning of 2020 (Flahault, 2020). and then increased to some extent in March (Fig. 4(b) ). This result is consistent with In this study, the heavy metals in PM 1 from Nanjing, China were investigated. During 2018, the annual average PM 1 concentration was 26.31 μg/m 3 and most 418 element concentrations followed a seasonal pattern: winter>spring>autumn>summer. As, Cd, Pb and Zn were greatly enriched, whereas Co, Mn and V originated mainly formation mechanism of heavy haze-fog pollution Prediction of viscosity of 595 imidazolium-based ionic liquids using MLR and SVM algorithms An improved retrieval method of atmospheric 598 parameter profiles based on the BP neural network No noncarcinogenic or carcinogenic risk resulted from inhalation of metals in PM 1 .Many metals had better simulation effects when atmospheric pollutants as inputs.BP-ANN and SVM models both performed better for Pb, Tl and Zn than for Ti and V.Predicted metal contents were lower during COVID-19 outbreak than in 2018 or 2019. ☐ √ √ √ √The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: