Air quality forecasting using artificial neural networks with real time dynamic error correction in highly polluted regions.

Air pollution is an important issue, especially in megacities across the world. There are emission sources within and also in the regions around these cities, which cause fluctuations in air quality based on prevailing meteorological conditions. Short term air quality forecasting is used not to just possibly mitigate forthcoming high air pollution episodes, but also to plan for reduced exposures of residents. In this study, a model using Artificial Neural Networks (ANN) has been developed to forecast pollutant concentration of PM10, PM2.5, NO2, and O3 for the current day and subsequent 4 days in a highly polluted region (32 different locations in Delhi). The model has been trained using meteorological parameters and hourly pollution concentration data for the year 2018 and then used for generating air quality forecasts in real-time. It has also been equipped with Real Time Correction (RTC), to improve the quality of the forecasts by dynamically adjusting the forecasts based on the model performance during the past few days. The model without RTC performs decently, but with RTC the errors are further reduced in forecasted values. The utility of the model has been demonstrated in real-time and model validations were performed for the whole year of 2018 and also independently for 2019. The model shows very good performance for all the pollutants on several evaluation metrics. Coefficient of correlations for various pollutants varies from 0.79-0.88 to 0.49-0.68 between the Day0 to Day4 forecasts. Lowest deterioration of performance was observed for ozone over the four days of forecasts. Use of RTC further improves the model performance for all pollutants.

Health Impact Assessment of a Predicted Air Quality Change by Moving Traffic from an Urban Ring Road into a Tunnel. The Case of Antwerp, Belgium.

BACKGROUND: The Antwerp ring road has a traffic density of 300,000 vehicles per day and borders the city center. The 'Ringland project' aims to change the current 'open air ring road' into a 'filtered tunneled ring road', putting the entire urban ring road into a tunnel and thus filtering air pollution. We conducted a health impact assessment (HIA) to quantify the possible benefit of a 'filtered tunneled ring road', as compared to the 'open air ring road' scenario, on air quality and its long-term health effects. MATERIALS AND METHODS: We modeled the change in annual ambient PM2.5 and NO2 concentrations by covering 15 kilometers of the Antwerp ring road in high resolution grids using the RIO-IFDM street canyon model. The exposure-response coefficients used were derived from a literature review: all-cause mortality, life expectancy, cardiopulmonary diseases and childhood Forced Vital Capacity development (FVC). RESULTS: Our model predicts changes between -1.5 and +2 µg/m³ in PM2.5 within a 1,500 meter radius around the ring road, for the 'filtered tunneled ring road' scenario as compared to an 'open air ring road'. These estimated annual changes were plotted against the population exposed to these differences. The calculated change of PM2.5 is associated with an expected annual decrease of 21 deaths (95% CI 7 to 41). This corresponds with 11.5 deaths avoided per 100,000 inhabitants (95% CI 3.9-23) in the first 500 meters around the ring road every year. Of 356 schools in a 1,500 meter perimeter around the ring road changes between -10 NO2 and + 0.17 µg/m³ were found, corresponding to FVC improvement of between 3 and 64ml among school-age children. The predicted decline in lung cancer mortality and incidence of acute myocardial infarction were both only 0.1 per 100,000 inhabitants or less. CONCLUSION: The expected change in PM2,5 and NO2 by covering the entire urban ring road in Antwerp is associated with considerable health gains for the approximate 352,000 inhabitants living in a 1,500 meter perimeter around the current open air ring road.

Impact of trees on pollutant dispersion in street canyons: A numerical study of the annual average effects in Antwerp, Belgium.

Effects of vegetation on pollutant dispersion receive increased attention in attempts to reduce air pollutant concentration levels in the urban environment. In this study, we examine the influence of vegetation on the concentrations of traffic pollutants in urban street canyons using numerical simulations with the CFD code OpenFOAM. This CFD approach is validated against literature wind tunnel data of traffic pollutant dispersion in street canyons. The impact of trees is simulated for a variety of vegetation types and the full range of approaching wind directions at 15° interval. All these results are combined using meteo statistics, including effects of seasonal leaf loss, to determine the annual average effect of trees in street canyons. This analysis is performed for two pollutants, elemental carbon (EC) and PM10, using background concentrations and emission strengths for the city of Antwerp, Belgium. The results show that due to the presence of trees the annual average pollutant concentrations increase with about 8% (range of 1% to 13%) for EC and with about 1.4% (range of 0.2 to 2.6%) for PM10. The study indicates that this annual effect is considerably smaller than earlier estimates which are generally based on a specific set of governing conditions (1 wind direction, full leafed trees and peak hour traffic emissions).

Improving local air quality in cities: to tree or not to tree?

Vegetation is often quoted as an effective measure to mitigate urban air quality problems. In this work we demonstrate by the use of computer models that the air quality effect of urban vegetation is more complex than implied by such general assumptions. By modelling a variety of real-life examples we show that roadside urban vegetation rather leads to increased pollutant concentrations than it improves the air quality, at least locally. This can be explained by the fact that trees and other types of vegetation reduce the ventilation that is responsible for diluting the traffic emitted pollutants. This aerodynamic effect is shown to be much stronger than the pollutant removal capacity of vegetation. Although the modelling results may be subject to a certain level of uncertainty, our results strongly indicate that the use of urban vegetation for alleviating a local air pollution hotspot is not expected to be a viable solution.

Dispersion modelling of traffic induced ultrafine particles in a street canyon in Antwerp, Belgium and comparison with observations.

The aim of this study is to investigate the dispersion of ultrafine particles and its spatial distribution in a street canyon and its neighbourhood with the 3D CFD model ENVI-met®. The performance of the model at street scale is evaluated and the importance of the boundary conditions like wind field and traffic emissions on the UFP concentration is demonstrated. To support and validate the modelled results, a short-term measurement campaign was conducted in a street canyon in Antwerp, Belgium. The UFP concentration was measured simultaneously with P-TRACK (TSI Model 8525) at four different locations in the canyon. The modelled UFP concentrations compare well with the measured data (correlation coefficient R from 0.44 to 0.93) within the standard deviation of the measurements. Despite the moderate traffic flow in the street canyon, UFP concentrations in the canyon are in general double of the background concentrations, indicating the high local contribution for this particle number concentration. Some of the observed concentration profiles are not resembled by the model simulations. For these specific anomalies, further analysis is performed and plausible explanations are put forward. The role of wind direction and traffic emissions is investigated. The performance evaluation of ENVI-met® shows that in general the model qualitatively and quantitatively describes the dispersion of UFP in the street canyon study.

Size resolved ultrafine particles emission model--a continues size distribution approach.

A new parameterization for size resolved ultrafine particles (UFP) traffic emissions is proposed based on the results of PARTICULATES project (Samaras et al., 2005). It includes the emission factors from the Emission Inventory Guidebook (2006) (total number of particles, #/km/veh), the shape of the corresponding particle size distribution given in PARTICULATES and data for the traffic activity. The output of the model UFPEM (UltraFine Particle Emission Model) is a sum of continuous distributions of ultrafine particles emissions per vehicle type (passenger cars and heavy duty vehicles), fuel (petrol and diesel) and average speed representative for urban, rural and highway driving. The results from the parameterization are compared with measured total number of ultrafine particles and size distributions in a tunnel in Antwerp (Belgium). The measured UFP concentration over the entire campaign shows a close relation to the traffic activity. The modelled concentration is found to be lower than the measured in the campaign. The average emission factor from the measurement is 4.29E+14 #/km/veh whereas the calculated is around 30% lower. A comparison of emission factors with literature is done as well and in overall a good agreement is found. For the size distributions it is found that the measured distributions consist of three modes--Nucleation, Aitken and accumulation and most of the ultrafine particles belong to the Nucleation and the Aitken modes. The modelled Aitken mode (peak around 0.04-0.05 µm) is found in a good agreement both as amplitude of the peak and the number of particles whereas the modelled Nucleation mode is shifted to smaller diameters and the peak is much lower that the observed. Time scale analysis shows that at 300 m in the tunnel coagulation and deposition are slow and therefore neglected. The UFPEM emission model can be used as a source term in dispersion models.