Search

This paper proposes a deep learning model that integrates a convolutional neural network and a gated recurrent unit with groups of neighboring stations to accurately predict PM2.5 concentrations at 25 stations in Seoul, South Korea. The deep learning model uses observations obtained from one Korea Meteorological Administration (KMA) station, 25 National Institute of Environmental Research (NIER) stations, and 28 automatic weather stations (AWSs) throughout Seoul. To train the deep learning model, we use all available meteorological and air quality data observed between 2015 and 2017. With the trained model, we predict PM2.5 concentrations at all 25 NIER stations in Seoul for 2018. This study also proposes a geographical polygon group model that determines the optimal number of neighboring NIER stations required to increase the accuracy of PM2.5 concentration predictions at the target station. Comparing the model measures for each of the 25 monitoring sites in 2018, we find that the geographical polygon group model achieves an index of agreement of 0.82–0.89 and a Pearson correlation coefficient of 0.70–0.83. Compared to the method using only meteorological and air quality data from one target station (average IOA = 0.77) to predict PM2.5 concentrations at the 25 stations in Seoul, the proposed method using geographical correlation-based neighboring NIER stations as polygonal groups (average IOA = 0.85) improves the PM2.5 prediction accuracy by an average of about 10%. This approach, based on deep learning, can be updated to predict air pollution or air quality indices up to several days in advance. [ABSTRACT FROM AUTHOR]

This study uses a deep learning approach to forecast ozone concentrations over Seoul, South Korea for 2017. We use a deep convolutional neural network (CNN). We apply this method to predict the hourly ozone concentration on each day for the entire year using several predictors from the previous day, including the wind fields, temperature, relative humidity, pressure, and precipitation, along with in situ ozone and NO2 concentrations. We refer to a history of all observed parameters between 2014 and 2016 for training the predictive models. Model-measurement comparisons for the 25 monitoring sites for the year 2017 report average indices of agreement (IOA) of 0.84–0.89 and a Pearson correlation coefficient of 0.74–0.81, indicating reasonable performance for the CNN forecasting model. Although the CNN model successfully captures daily trends as well as yearly high and low variations of the ozone concentrations, it notably underpredicts high ozone peaks during the summer. The forecasting results are generally more accurate for the stations located in the southern regions of the Han River, the result of more stable topographical and meteorological conditions. Furthermore, through two separate daytime and nighttime forecasts, we find that the monthly IOA of the CNN model is 0.05–0.30 higher during the daytime, resulting from the unavailability of some of the input parameters during the nighttime. While the CNN model can predict the next 24 h of ozone concentrations within less than a minute, we identify several limitations of deep learning models for real-time air quality forecasting for further improvement. [ABSTRACT FROM AUTHOR]

Six generalized machine learning (ML) ensemble models were developed to predict the real-time hourly ozone concentration of the following day. These models were used to forecast hourly ozone concentrations of the following day for all of 2017 in the city of Seoul, South Korea. To prepare the training dataset, it was referred to observed meteorology and air pollution parameters of the 2014–2016 period. The ensemble models fuse two regression models: a low-ozone peak model and a high-ozone model. For both, extremely randomized trees and deep neural networks were used. A regularization approach was also adopted that adjusts the model toward capturing higher ozone peaks by resampling the training dataset based on the peaks. Adopting the proposed ML ensemble forecasting method over single-model ML techniques as a part of mainstream practice for air quality forecasting will be beneficial for several reasons. For one, the proposed method, which captures daily maximum ozone concentrations during the high-ozone season (April–September), reduces the ozone peak prediction error by 5 to 30 ppb. In addition, compared to station-specific (independent) ML models with more frequent low-ozone values, models are trained with a uniformly distributed dataset, so they are more generalizable in nature. As a result, unlike station-specific models, they retain their accuracy (yearly IOA = 0.84–0.89) in all stations with an IOA increment. Proposed models also make predictions several times faster, requiring only one-time training for predicting an entire station network. Based on a categorical analysis of the training dataset, an algorithm was proposed for selecting the most suitable model for each month. The "best" model further improves the accuracy of both the ML ensemble and individual models by up to 2.4%. This study shows that the ML ensemble modeling approach is a fast, reliable, and robust technique that can benefit environmental decision-makers in urban regions. [ABSTRACT FROM AUTHOR]