Abstract:Abstract. As air pollution is regarded as the single largest environmental health risk in Europe it is important that communication to the public is up to date and accurate and provides means to avoid exposure to high air pollution levels. Long- and short-term exposure to outdoor air pollution is associated with increased risks of mortality and morbidity. Up-to-date information on present and coming days' air quality helps people avoid exposure during episodes with high levels of air pollution. Air quality forecasts can be based on deterministic dispersion modelling, but to be accurate this requires detailed information on future emissions, meteorological conditions and process-oriented dispersion modelling. In this paper, we apply different machine learning (ML) algorithms – random forest (RF), extreme gradient boosting (XGB), and long short-term memory (LSTM) – to improve 1, 2, and 3 d deterministic forecasts of PM10, NOx, and O3 at different sites in Greater Stockholm, Sweden. It is shown that the deterministic forecasts can be significantly improved using the ML models but that the degree of improvement of the deterministic forecasts depends more on pollutant and site than on what ML algorithm is applied. Also, four feature importance methods, namely the mean decrease in impurity (MDI) method, permutation method, gradient-based method, and Shapley additive explanations (SHAP) method, are utilized to identify significant features that are common and robust across all models and methods for a pollutant. Deterministic forecasts of PM10 are improved by the ML models through the input of lagged measurements and Julian day partly reflecting seasonal variations not properly parameterized in the deterministic forecasts. A systematic discrepancy by the deterministic forecasts in the diurnal cycle of NOx is removed by the ML models considering lagged measurements and calendar data like hour and weekday, reflecting the influence of local traffic emissions. For O3 at the urban background site, the local photochemistry is not properly accounted for by the relatively coarse Copernicus Atmosphere Monitoring Service ensemble model (CAMS) used here for forecasting O3 but is compensated for using the ML models by taking lagged measurements into account. Through multiple repetitions of the training process, the resulting ML models achieved improvements for all sites and pollutants. For NOx at street canyon sites, mean squared error (MSE) decreased by up to 60 %, and seven metrics, such as R2 and mean absolute percentage error (MAPE), exhibited consistent results. The prediction of PM10 is improved significantly at the urban background site, whereas the ML models at street sites have difficulty capturing more information. The prediction accuracy of O3 also modestly increased, with differences between metrics. Further work is needed to reduce deviations between model results and measurements for short periods with relatively high concentrations (peaks) at the street canyon sites. Such peaks can be due to a combination of non-typical emissions and unfavourable meteorological conditions, which are rather difficult to forecast. Furthermore, we show that general models trained using data from selected street sites can improve the deterministic forecasts of NOx at the station not involved in model training. For PM10 this was only possible using more complex LSTM models. An important aspect to consider when choosing ML algorithms is the computational requirements for training the models in the deployment of the system. Tree-based models (RF and XGB) require fewer computational resources and yield comparable performance in comparison to LSTM. Therefore, tree-based models are now implemented operationally in the forecasts of air pollution and health risks in Stockholm. Nevertheless, there is big potential to develop generic models using advanced ML to take into account not only local temporal variation but also spatial variation at different stations.

Indoor sulfur dioxide prediction through air quality modeling and assessment of sulfur dioxide and nitrogen dioxide levels in industrial and non-industrial areas

Seasonal Trends of Indoor Fine Particulate Matter and Its Determinants in Urban Residences in Nanjing, China

Indoor PM2.5 forecasting and the association with outdoor air pollution: a modelling study based on sensor data in Australia

Construction and evaluation of hourly average indoor PM2.5 concentration prediction models based on multiple types of places

Modelling of Carbon Monoxide and Suspended Particulate Matter Concentrations in a Rural Area Using Artificial Neural Networks

Enhancing indoor PM2.5 predictions based on land use and indoor environmental factors by applying machine learning and spatial modeling approaches

Simultaneous Determination of Nitrogen Dioxide and Sulfur Dioxide in the Air by Passive Individual Samplers

Temporal and Spatial Variation of PM2.5 in Indoor Air Monitored by Low-Cost Sensors.

A Sectional Aerosol Model for Submicron Particles in Indoor Air.

Refining Children's exposure assessment to NO2, SO2, and O3: Incorporating indoor-to-outdoor concentration ratios and individual daily routine

Improving 3-day deterministic air pollution forecasts using machine learning algorithms

Characterizing the Indoor-Outdoor Relationship of Fine Particulate Matter in Non-Heating Season for Urban Residences in Beijing

Indoor, Outdoor and Regional Profiles of PM2.5 Sulfate, Nitrate and Carbon

Estimating the Spatiotemporal Variation of NO2 Concentration Using an Adaptive Neuro-Fuzzy Inference System

Estimating long‐term time‐resolved indoor PM 2.5 of outdoor and indoor origin using easily obtainable inputs

Detecting and quantifying PM2.5 and NO2 contributions from train and road traffic in the vicinity of a major railway terminal in Dublin, Ireland

Feasibility Study on the Use of NO2 and PM2.5 Sensors for Exposure Assessment and Indoor Source Apportionment at Fixed Locations

Emission Rates Due to Indoor Activities: Indoor Aerosol Model Development, Evaluation, and Applications

A Random Forest Approach to Estimate Daily Particulate Matter, Nitrogen Dioxide, and Ozone at Fine Spatial Resolution in Sweden

Evaluating air quality and criteria pollutants prediction disparities by data mining along a stretch of urban-rural agglomeration includes coal-mine belts and thermal power plants

Prediction of PM2.5 Concentration Using Spatiotemporal Data with Machine Learning Models