Abstract
Recent years have shown significant decrease in concentrations levels of particulate matter (PM10) and nitrogen dioxide (NO2) in Belgium. For ozone (O3), no such trend is found. Recent years, however, did not feature many periods with unfavourable meteorological dispersion conditions, casting some ambiguity on the underlying reasons for the decrease. This study tries to separate the impact of weather effects from emission reductions in the long-term trend. We build a statistical model explaining the daily averaged concentrations based on 32 meteorological parameters, the day of the week, the month of the year and the year itself, for the period 2004–2014. The 32 meteorological parameters are those considered to train the neural network prediction model OVL. Many of these meteo variables have only a small predictability and are intercorrelated with each other. Therefore, only those meteo parameters are used that have a significant impact on concentration levels. This procedure is applied for the complete time series and for each air quality monitoring location separately. In order to avoid overfitting, the same analysis is done, restricted to the data of even-numbered years, and the regression is then applied to the odd-numbered years. It is shown that the statistical parameters remain reasonably constant, which proves that the amount of overfitting is not significant. The results show, on average over all measurement locations, a range of yearly meteorological effects of 1.9 µg/m3 for NO2, 3.1 µg/m3 for PM10 and 2.7 µg/m3 for O3. Meteorology combined with the residuals of the statistical fit show a range of 1.2 µg/m3 for NO2, 2.9 µg/m3 for PM10 and 4.4 µg/m3 for O3. Finally, the long-term trend shows a range of 5.3 µg/m3 for NO2, 11.1 µg/m3 for PM10 and 2.3 µg/m3 for O3, with clearly decreasing trends for NO2 and PM10, and an oscillating trend for O3. Differences between rural, urban background, urban and industrial stations exist but are rather small. We can conclude that the major trend in air pollution (Belgium) is a long-term trend, linked to emission changes, and it can be expected that the concentration decreases of the last years will not suddenly disappear in the near future given unchanged policy. Furthermore, it can be concluded that emission reductions at the local, regional, European and worldwide scale are the dominant factors explaining the improvement of air quality.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Reference
Hooyberghs J, Mensink C, Dumont G, Fierens F, Brasseur O (2005) A neural network forecast for daily average PM concentrations in Belgium. Atmos Environ 39:3279–3289. doi:10.1016/j.atmosenv.2005.01.050
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Questions and Answers
Questions and Answers
Questioner: Paul Makar
Question: I was wondering whether clouds, cloud rainfall were included in the set of meteo variables, since they would influence O3 via photolysis rates?
Answer: The meteovariables (day of the week and month of the year are added later) that were taken into account are shown in the table below:
P1 | 2-m temperature |
---|---|
P2 | 30-m temperature |
P3 | 10-m wind velocity |
P4 | 10-m wind direction |
P5 | 30-m wind velocity |
P6 | 30-m wind direction |
P7 | Boundary layer height (corresponding to critical bulk Richardson = 0.5) |
P8 | Height of the layer where the transport length < 100 m |
P9 | Transport length at the first level (about 25-m height) |
P10 | Mean transport length in the layer 0–50 m (from the surface) |
P11 | Mean transport length in the layer 0–100 m |
P12 | Total cloud cover (range0 to 1) |
P13 | Low cloud cover (range0 to 1) |
P14 | Medium cloud cover (range0 to 1) |
P15 | High cloud cover (range0 to 1) |
P16 | Horizontal transport in the boundary layer |
P17 | Height of the layer where the transport length < 200 m |
P18 | Height of the layer where the transport length < 300 m |
P19 | Height of the layer where the transport length < 500 m |
P20 | Height of the layer where the transport length < 1000 m |
P21 | “S parameter” of Bultynck-Malet classification (multiplied by a factor 1000) |
P22 | “λ parameter” of Bultynck-Malet classificationLambda = LOG10(106 * ABS(S)) |
P23 | Relative humidity (%) at 300-m height |
P24 | Relative humidity in the layer 0–50 m |
P25 | Relative humidity at 100-m height (not layer 0–100 m!) |
P26 | Wind velocity at 50-m height |
P27 | Wind velocity at 100-m height |
P28 | Wind velocity at 200-m height |
P29 | Wind velocity at 300-m height |
P30 | Wind velocity at 500-m height |
P31 | Wind velocity at 750-m height |
P32 | Wind velocity at 1000-m height |
Cloud cover is thus taken directly into account, precipitation only with related variables (such as relative humidity).
Questioner: Tony Dore
Question: For analysis of trends, consistency in monitoring techniques is important. Your air concentrations showed more annual variability at the start of the time series. Has there been any change in quality control during the monitoring period?
Answer: There have been changes between monitoring techniques, and thus probably to associated quality control. However, at least for NO2 and PM10, these changes cannot account for the large trends that have been found in the results.
Rights and permissions
Copyright information
© 2018 Springer International Publishing AG
About this paper
Cite this paper
Lefebvre, W., Maiheu, B., Hooyberghs, H., Fierens, F. (2018). Is the Recent Decrease in Belgian Air Pollution Concentration Levels Due to Meteorology or to Emission Reductions?. In: Mensink, C., Kallos, G. (eds) Air Pollution Modeling and its Application XXV. ITM 2016. Springer Proceedings in Complexity. Springer, Cham. https://doi.org/10.1007/978-3-319-57645-9_38
Download citation
DOI: https://doi.org/10.1007/978-3-319-57645-9_38
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-57644-2
Online ISBN: 978-3-319-57645-9
eBook Packages: Earth and Environmental ScienceEarth and Environmental Science (R0)