Using machine learning to quantify sources of light-absorbing water-soluble humic-like substances (HULISws) in Northeast China

The Sources and Atmospheric Processes of Strong Light‐Absorbing Components in Water Soluble Brown Carbon: Insights From a Multi‐Proxy Study of PM2.5 in 10 Chinese Cities

“…In contrast, the HULIS-C directly emitted from BB (∼1.6 m 2 /gC) and coal combustion (∼2.0 m 2 /gC) generally have higher MAE 365 values (Tang et al, 2020), these HULIS may enhance the MAE 365 of HULIS-C in cold seasons. The observed MAE 365 values of HULIS-C fall within the range of values previously reported in urban sites in China (Beijing: 1.80 ± 0.40 m 2 /gC, Xi'an: 1.80 ± 0.30 m 2 /gC, Nanjing: 1.71 ± 0.30 m 2 / gC, Changchun: 1.07 ± 2.70 m 2 /gC) (Bao et al, 2022;Hong et al, 2022;Huang et al, 2020), and they are higher than those in remote sites in China (Tibet: 0.23-0.73 m 2 /gC) and Central European (Hungary: MAE at 355 nm: 0.32 ± 0.04 m 2 /gC) (Utry et al, 2013;Wu et al, 2018). Particularly in cold seasons, we found that the MAE 365 of HULIS-C in the area without central hearing (1.66 ± 0.54 m 2 /gC) was comparable to those in areas affected by BB (Central and Southern Europe: Thessaloniki: 0.71-1.40 m 2 /gC, Mainz: 0.37-1.04 m 2 /gC, Ostrava: 0.83-1.61 m 2 /gC, Kladno: 1.01-1.8 m 2 /gC 7, Amazon basin: 0.62-0.76 m 2 /gC, p = 0.06, Student's t-test) (Hoffer et al, 2006;Voliotis et al, 2017).…”

Nitrogen‐Containing Functional Groups Dominate the Molecular Absorption of Water‐Soluble Humic‐Like Substances in Air From Nanjing, China Revealed by the Machine Learning Combined FT‐ICR‐MS Technique

“…After that, the Abs 365 of HULIS WS (target variable) were allocated to each molecular marker (input variables) using the attribution technique combined with the FSL Abs (Davies et al., 2021; Hong, Cao, Fan, Lin, Bao, et al., 2022; Wei et al., 2015), and the main light‐absorbing molecules were selected by the first‐order difference of molecular absorption among each molecule (Text S7 in Supporting Information ).…”

“…FSL is an algorithm of the ML that is aimed to learn the underlying pattern from a few samples (Parnami & Lee, 2022), and has been widely used in previous object detection (Kisantal et al., 2019), cheminformatics (Chen et al., 2023), and environmental studies (Huang et al., 2023). Here, we used the synthetic minority over‐sampling technique (SMOTE) (Chawla et al., 2002) combined with the random forest (RF) algorithm provided by Ranger package (Fan et al., 2023; Hong, Cao, Fan, Lin, Bao, et al., 2022; Wright & Ziegler, 2017) to successfully build an FSL model without the risk of overfitting (has been proved in a 55 × 35799 ESI‐FT‐ICR‐MS data set, see details in Text S8 of the Supporting Information ) (Belgiu & Drăguţ, 2016; Cortes‐Ciriano & Bender, 2015; Jablonka et al., 2020), and then proved by the validation data set and model outputs (Text S9 in Supporting Information ) (Arulkumaran et al., 2017).…”

“…Data: The observation data of ESI-FT-ICR-MS and the collected UV-Vis spectra data from the chemical database in this work are available in Tables S8 and S9 of the Supporting Information S1 which can be found via the link https://doi.org/10.17605/OSF.IO/BKX9H (Hong et al, 2023). Data: The observation data of ESI-FT-ICR-MS and Abs 365 value of dissolved organic matters (DOM) used in this work was collected from Jiang, Li, et al (2020).…”

“…Based on this, some over-sampling techniques like the synthetic minority over-sampling technique (SMOTE) (Chawla et al, 2002) and other ML-based data augment methods (Mumuni & Mumuni, 2022) were provided to ensure good performance. This work uses the ML approach combined with the SMOTE technique (described in Text S10 of the Supporting Information S1) to allocate the Abs values of HULIS WS mixtures to each molecule level by the molecular marker using the ESI-FT-ICR-MS technique (Hong, Cao, Fan, Lin, Bao, et al, 2022;Jiang, Li, Tang, Cui, et al, 2022). Further, the molecular absorption was allocated to the functional groups by combining the molecular absorption data collected from the chemical database-Reaxys (Goodman, 2009).…”