Trend and Trend Reversal Analysis of Ambient NO2 Concentration in Europe
A recent study based on a compiled dataset of satellite NO2 retrievals covering ~21 years revealed changes in the NO2 trends, both from positive to negative and vice versa, over some European and Asian regions, contradicting the expected monotonic decreasing trend. In this study, we analyzed the trends in the surface level NO2 over the European Environment Agency (EEA) member countries. Some studies used station data up to 2013 while some only after 2013. In this study, we compiled a monthly-averaged hourly surface NO2 dataset that covers the whole period of observation, including before and after 2013. As a result, data availability in some cities extended to about 40 years. To compare the station surface NO2 with the satellite NO2 column retrievals, we converted the surface NO2 into tropospheric NO2 column by matching the surface NO2 with that of the NO2 profile assimilated in the European Centre for Medium-Range Weather Forecasts (ECMWF). 18 European cities were later selected where the converted NO2 column agreed well with the satellite NO2 column. We also converted the satellite NO2 column to surface NO2 using the ECMWF profile, but our selection remained unchanged. The long-term trend, changes in diurnal and seasonal cycles, and trend reversals of the selected cities were studied. The selected cities all showed a decreasing trend, and most of the cities already comply with the annual NO2 limit by EEA. All the cities showed similar diurnal and seasonal cycles, but the morning peaks of NO2 over some cities have shifted 1–2 hours towards noon, which may imply a shorter NO2 lifetime over those cities. Trend reversals were detected in most of the cities. However, the time of the reversals detected in ground station data could be quite different from those in the satellite time series, the latter mostly in the early 2000s. The trend reversal points seem to be a part of decadal variability, whose cause needs further investigation. Given the limitation in the trend reversal detection algorithm over finite time series, the reversal points detected in the previous study using short NO2 records may subject to large uncertainties.