A Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP Analysis
Abstract Introduction PM2.5 pollution is a significant environmental and health concern in Thailand, with levels intensifying during the dry season. However, the lack of long-term PM2.5 data limits understanding of historical trends and meteorological influences. Objective This study aims to reconst...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-03-01
|
| Series: | Aerosol and Air Quality Research |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44408-025-00007-z |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850206355745406976 |
|---|---|
| author | Nishit Aman Sirima Panyametheekul Ittipol Pawarmart Sumridh Sudhibrabha Kasemsan Manomaiphiboon |
| author_facet | Nishit Aman Sirima Panyametheekul Ittipol Pawarmart Sumridh Sudhibrabha Kasemsan Manomaiphiboon |
| author_sort | Nishit Aman |
| collection | DOAJ |
| description | Abstract Introduction PM2.5 pollution is a significant environmental and health concern in Thailand, with levels intensifying during the dry season. However, the lack of long-term PM2.5 data limits understanding of historical trends and meteorological influences. Objective This study aims to reconstruct historical PM2.5 data from 1981 to 2022 and analyze the influence of various contributing factors across six key provinces in Thailand: Chiang Mai (CM), Lampang (LP), Khon Kaen (KK), Bangkok (BK), Chonburi (CB), and Songkhla (SK). Methods A Light Gradient Boosting Machine (LightGBM) model was developed using meteorological and aerosol-related variables from the Thai Meteorological Department and MERRA-2. The model was trained on PM2.5 data spanning 2012–2022, depending on availability for each province. Model performance was evaluated across diurnal, monthly, and annual scales and then used for historical reconstruction of PM2.5 data. SHAP analysis was used to determine the important predictor variables affecting PM2.5 prediction. Results The LightGBM model accurately predicted PM2.5 across all provinces, showing better performance for daily prediction than for hourly prediction. Model accuracy was higher during clean hours than during haze hours. Good agreement between observed and predicted PM2.5 was found on different time scales (diurnal, monthly, and annually). CM shows a non-significant PM2.5 trend, limiting insights into meteorological effects, while LP exhibits significant decreases in PM2.5 and PM2.5_emis, indicating positive weather impacts on air quality. In contrast, regions like KK, BK, and CB display worsening meteorological influences, with non-significant or increasing PM2.5 trends despite declines in PM2.5_emis. In SK, removing meteorological effects reveals a decreasing PM2.5 trend, underscoring the critical role of meteorology. SHAP analysis identified visibility, gridded PM2.5, and specific humidity at 2 m as common and important predictor variables over all the provinces, along with additional variables that were not consistent over different provinces. Conclusion The LightGBM model effectively reconstructs historical PM2.5 levels and provides insight into meteorological influences on air quality. Based on the findings of the study, some policy implications have also been provided. Graphical abstract |
| format | Article |
| id | doaj-art-b9b25d4a8fe14b17a0cf3bc21f4cf7b5 |
| institution | OA Journals |
| issn | 1680-8584 2071-1409 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Springer |
| record_format | Article |
| series | Aerosol and Air Quality Research |
| spelling | doaj-art-b9b25d4a8fe14b17a0cf3bc21f4cf7b52025-08-20T02:10:50ZengSpringerAerosol and Air Quality Research1680-85842071-14092025-03-01251-411710.1007/s44408-025-00007-zA Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP AnalysisNishit Aman0Sirima Panyametheekul1Ittipol Pawarmart2Sumridh Sudhibrabha3Kasemsan Manomaiphiboon4Department of Environmental and Sustainable Engineering, Faculty of Engineering, Chulalongkorn UniversityDepartment of Environmental and Sustainable Engineering, Faculty of Engineering, Chulalongkorn UniversityPollution Control Department, Ministry of Natural Resources and EnvironmentThai Meteorological Department, Ministry of Digital Economy and SocietyThe Joint Graduate School of Energy and Environment, King Mongkut’s University of Technology ThonburiAbstract Introduction PM2.5 pollution is a significant environmental and health concern in Thailand, with levels intensifying during the dry season. However, the lack of long-term PM2.5 data limits understanding of historical trends and meteorological influences. Objective This study aims to reconstruct historical PM2.5 data from 1981 to 2022 and analyze the influence of various contributing factors across six key provinces in Thailand: Chiang Mai (CM), Lampang (LP), Khon Kaen (KK), Bangkok (BK), Chonburi (CB), and Songkhla (SK). Methods A Light Gradient Boosting Machine (LightGBM) model was developed using meteorological and aerosol-related variables from the Thai Meteorological Department and MERRA-2. The model was trained on PM2.5 data spanning 2012–2022, depending on availability for each province. Model performance was evaluated across diurnal, monthly, and annual scales and then used for historical reconstruction of PM2.5 data. SHAP analysis was used to determine the important predictor variables affecting PM2.5 prediction. Results The LightGBM model accurately predicted PM2.5 across all provinces, showing better performance for daily prediction than for hourly prediction. Model accuracy was higher during clean hours than during haze hours. Good agreement between observed and predicted PM2.5 was found on different time scales (diurnal, monthly, and annually). CM shows a non-significant PM2.5 trend, limiting insights into meteorological effects, while LP exhibits significant decreases in PM2.5 and PM2.5_emis, indicating positive weather impacts on air quality. In contrast, regions like KK, BK, and CB display worsening meteorological influences, with non-significant or increasing PM2.5 trends despite declines in PM2.5_emis. In SK, removing meteorological effects reveals a decreasing PM2.5 trend, underscoring the critical role of meteorology. SHAP analysis identified visibility, gridded PM2.5, and specific humidity at 2 m as common and important predictor variables over all the provinces, along with additional variables that were not consistent over different provinces. Conclusion The LightGBM model effectively reconstructs historical PM2.5 levels and provides insight into meteorological influences on air quality. Based on the findings of the study, some policy implications have also been provided. Graphical abstracthttps://doi.org/10.1007/s44408-025-00007-zPM2.5MERRA-2Machine learningMeteorological normalizationExplainable machine learningShapley value |
| spellingShingle | Nishit Aman Sirima Panyametheekul Ittipol Pawarmart Sumridh Sudhibrabha Kasemsan Manomaiphiboon A Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP Analysis Aerosol and Air Quality Research PM2.5 MERRA-2 Machine learning Meteorological normalization Explainable machine learning Shapley value |
| title | A Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP Analysis |
| title_full | A Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP Analysis |
| title_fullStr | A Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP Analysis |
| title_full_unstemmed | A Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP Analysis |
| title_short | A Visibility-Based Historical PM2.5 Estimation for Four Decades (1981–2022) Using Machine Learning in Thailand: Trends, Meteorological Normalization, and Influencing Factors Using SHAP Analysis |
| title_sort | visibility based historical pm2 5 estimation for four decades 1981 2022 using machine learning in thailand trends meteorological normalization and influencing factors using shap analysis |
| topic | PM2.5 MERRA-2 Machine learning Meteorological normalization Explainable machine learning Shapley value |
| url | https://doi.org/10.1007/s44408-025-00007-z |
| work_keys_str_mv | AT nishitaman avisibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT sirimapanyametheekul avisibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT ittipolpawarmart avisibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT sumridhsudhibrabha avisibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT kasemsanmanomaiphiboon avisibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT nishitaman visibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT sirimapanyametheekul visibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT ittipolpawarmart visibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT sumridhsudhibrabha visibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis AT kasemsanmanomaiphiboon visibilitybasedhistoricalpm25estimationforfourdecades19812022usingmachinelearninginthailandtrendsmeteorologicalnormalizationandinfluencingfactorsusingshapanalysis |