This deteriorates the air quality of the densely populated and industrialized areas such as Beijing, Shanghai and the Pearl River Delta, and increases the background pollution levels world-wide (1). The EU AMFIC project, led by KNMI, kicked off in September 2007 and aims at monitoring and forecasting the air quality in China by using satellite observations and model simulations, together with ground observations from collaborating Chinese institutes. The combination of these instruments and tools offers a unique possibility to investigate trends in air pollution and the effectiveness of air quality policy.
The following sections describe the use of satellite instruments to monitor air quality, the implementation of an operational air quality model for China, and the use of both satellite observations and model simulations to estimate the effect of the air quality measures during the 2008 Beijing Olympic Games.
Ground measurements of air quality in China are often sparse and inaccessible, which make satellite observations the obvious tool to monitor country-wide pollution levels on a daily basis and to observe concentration trends on longer time scales. The Ozone Monitoring Instrument (OMI) is a nadir looking solar backscatter spectrometer, which measures in the ultraviolet and visible wavelength range to infer trace gases, such as ozone and nitrogen dioxide (NO2). Since it was launched in 2004 onboard NASA’s EOS-Aura satellite it observes the Earth’s atmosphere at a spatial resolution of about 20 km. In October 2006 a comparable instrument, GOME-2, was launched onboard the MetOp-A satellite. GOME-2 measures at a wider spectral range, but at a lower spatial resolution of about 40 km. Both instruments have near-daily global coverage.
Trace gas columns are retrieved from the depth of typical absorption structures in the atmospheric reflectance spectrum. Radiative transfer calculations are used to relate the measured absorption to a vertical column concentration, taking into account the viewing angle and atmospheric conditions, such as clouds, surface reflectance and the vertical profile of the trace gas. For nitrogen dioxide, the main interest for air quality monitoring is its amount in the troposphere. To separate the tropospheric column from the total column a data assimilation scheme is applied that provides the stratospheric column. Subtraction from the total column provides the tropospheric column.
The satellite instruments have a fixed overpass time: GOME-2 measures in the morning (9:30 local time) and OMI measures in the afternoon (at 13:30). This allows us to probe the diurnal cycle of nitrogen dioxide (Figure 1). Besides the diurnal cycle also weekly and seasonal cycles are monitored. This variability in the amount of nitrogen dioxide is due to chemical conversions depending on sunlight (photo-dissociation) and to time-dependent emissions of nitrogen dioxide and related species.
Nitrogen dioxide has significant natural sources (e.g. soil emissions, wildfires and lightning), but in populated areas NO2 sources are predominantly anthropogenic (e.g. fuel combustion and human-induced biomass burning). By monitoring the concentration of nitrogen dioxide over several years, information on the trend of the emission sources can be determined. For the period 1996-2006, a time series has been constructed by combining the observations of GOME (launched in 1995) with SCIAMACHY (launched in 2002; almost same overpass time) of the daily global nitrogen dioxide concentrations (2). The monthly NO2 columns for these ten years have been fitted with a linear function superposed on an annual seasonal cycle on a grid with a spatial resolution of 1x1 degree. Western-Europe and the East coast of the US show slightly negative trends due to emission regulations (Figure 2), but positive trends are found for Asian cities with a strong economical growth like Teheran (Iran), Novosibirsk (Russia) and especially for the booming cities in the East of China.
For a better interpretation of satellite measured air pollution and for the conversion of measured column data to emission rates, an appropriate model is required which is able to simulate pollutant concentrations. CHIMERE is a regional chemistry transport model which is successfully used in Europe for air quality forecasts on urban to continental scales. For the AMFIC project (Air quality Monitoring and Forecasting in China) we implemented CHIMERE on a 0.25x0.25 degree resolution over East China, enclosing all important populated and industrialized areas. The model simulates the evolution of 44 gaseous species and aerosols in 8 atmospheric layers in the troposphere up to 500 hPa. The meteorological data needed for calculating the transport, deposition and the chemistry of the species is taken from the European Centre for Medium-Range Weather Forecasts (ECMWF).
The sparse available ground measurements from China makes model validation complicated, but not impossible (3). Figure 3 shows comparisons of CHIMERE simulations with daily averaged measured surface concentrations of NO2 published by the Beijing Environmental Protection Bureau (BJEPB), and PM10 (particulate matter of 10 microns or less in diameter) surface concentrations measured by the British Broadcasting Company (BBC) at 13:00 local time. Both comparisons illustrate the capability of CHIMERE of capturing the day-to-day variability of pollutant concentrations.
By using meteorological forecast data, the model is able to generate daily air quality forecasts up to five days ahead. The result for East China and for the greater urban areas of Beijing, Shanghai, Hong Kong, Shenyang, Qingdao and Seoul are published on an English-Chinese bilingual website (see Figure 4). The daily model output is also used as boundary condition for a higher resolution model of Beijing and Shenyang by the Flemish Institute for Technical Research (VITO) and the street level model of Beijing by Cambridge Environmental Research Consultants (CERC). On request of the Dutch National Olympic Committee, during the 2008 Beijing Olympic Games a dedicated website was made available by KNMI to inform the athletes on the forecasted air quality and meteorology at all Olympic venues.
Heavy air pollution in Beijing, mainly originating from dense traffic, construction activities, industry, and coal-fired power plants, is a major concern for local authorities. To prevent high levels of air pollution during the Beijing Olympic Games (8-24 August 2008) and the Paralympics (6-17 September 2008), important measures inside and outside the city have been taken, including the temporarily shut down of polluting industry, the suspension of construction activities, and traffic restrictions. Traffic within the ring roads was restricted to cars with even number plates on even days and with odd numbers on odd days, 300.000 high-emission vehicles were banned from the city’s roads, and the use of governmental and commercial vehicles was restricted (see Figure 5).
To study the effect of the air quality measures on tropospheric nitrogen dioxide concentrations, we compare the NO2 observations over the Beijing area by OMI and GOME-2 before and during the Olympic Games. We take advantage of the high spatial resolution and daily global coverage of OMI, and the stronger anthropogenic nitrogen dioxide signal (due to its earlier overpass in the day) of GOME-24). As can be seen in Figure 6, the nitrogen dioxide concentration over Beijing during the Olympic Games is significantly less than before. This, however, can partly be explained by favourable meteorological conditions during this period: predominant northerly winds bring in clean air masses from the sparsely populated mountain areas, and more precipitation on more rainy days washes out the air pollution over the city. By comparing the satellite observations with CHIMERE simulations based on pre-Olympic nitrogen dioxide emission estimates, we compensate for these atypical meteorological conditions (5). Differences between observation and simulation can only be explained by changes in anthropogenic emissions.
The model results are interpolated to the time and the footprint of the satellite observation. For the pre-Olympic reference period (2 May to 30 June 2008) we see good agreement between simulations and observations for the Beijing area. When the air quality measures are enforced, the observations drop with respect to the simulations, which are based on an unchanged emission scenario. During the Beijing Olympic Games, the GOME-2 and OMI columns show a reduction of 59%–69% with respect to pre-Olympic values. Figure 6 shows the geographic extent of the concentration reductions as observed by GOME-2. In the pre-Olympic period both satellite and model show high concentrations in the populated and industrialized areas. During the Olympic period, the satellite observes decreased nitrogen dioxide concentrations for Beijing, whereas the other cities continue to show high concentrations. Highest concentration reductions are found in and around Beijing and the industrial areas in the south and south-east (60%–70%). The surrounding cities of Tianjin and Shijiazhuang show smaller reductions of about 30% and 20%, respectively. In the two months after the Olympic Games the nitrogen dioxide concentrations are still reduced with 40%, mainly due to the prolonged air quality measures and the reduced economic activity. One year afterwards, however, nitrogen dioxide levels have returned to their high pre-Olympic values.
Satellite observation of tropospheric columns of nitrogen dioxide are extremely useful to analyze air pollution levels world-wide. The long-term data record from 1995 onwards shows strongly increasing pollution levels in China and slowly decreasing levels in Western Europe and Eastern USA. Observed trends can be attributed to economic growth and emission reduction measures.