A feasibility study for the detection of the diurnal variation of tropospheric NO₂ over Tokyo from a geostationary orbit

Nitrogen oxides (NO_x = NO + NO₂) are well-known anthropogenic air pollutants emitted in the process of fossil fuel combustion from power plants and vehicles mainly in the form of NO. NO_x itself is harmful to human health and has a key role in the formation of tropospheric ozone, which is also toxic to the human body and the dominant component of photochemical smog; the photolysis of NO₂ first produces O(³P) and NO, and a three-body reaction including O(³P) quickly forms ozone. The NO formed in the photolysis of NO₂ would oxidize ozone again without other reactions. In polluted air, however, NO is successively oxidized by radical compounds such as organic peroxy radical (RO₂) and the hydroperoxyl radical (HO₂) before consuming ozone, and ozone formation can proceed without consuming ozone (Hobbs, 2000).

Since the mid 1990s, space-borne measurements based on spectroscopic techniques have been developed to complement the ground-based monitoring network. The Global Ozone Monitoring Experiment (GOME) (Burrows et al., 1999) and its successor Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY) (Bovensmann et al., 1999) are the pioneers for the global tropospheric trace gas measurements from space using backscattered solar radiation. The measurements of GOME and SCIAMACHY revealed global long-term trends in the tropospheric NO₂ including China, where a large increase rate of the tropospheric NO₂ was found (Richter et al., 2005), constraints for the global inventories of NO_x emissions (Leue et al., 2001, Martin et al., 2003), and global distributions of NO_x sources by the identification of the seasonal cycles of the sources (van der et al., 2008). Following GOME and SCIAMACHY, measurements with improved spatial resolutions and more frequent global surface coverage are performed by the Ozone Monitoring Instrument (OMI) (Levelt et al., 2006) and GOME-2 (Callies et al., 2000). The observations from OMI and GOME-2 facilitate studying the spatiotemporal distribution of the tropospheric NO₂ over urban and industrialized regions (Wang et al., 2007, Kim et al., 2009, Mijling et al., 2009, Sitnov, 2009, Witte et al., 2009, Zhang et al., 2009).

The satellites carrying those sensors have sun-synchronous low earth orbits (LEO), and the observations of a given location would be conducted at a fixed local time (except at high latitudes where several orbits from the same day overlap) when only one sensor was utilized. Therefore, the measurements based on only one sensor cannot obtain the information on the diurnal variation of a species. To overcome this limitation of the observation with one sensor, a satellite constellation with several instruments was utilized. Boersma et al. (2008) showed that the difference of NO₂ observed by SCIAMACHY (observing at LT10–11) and OMI (at LT13–14) is up to 40% in summer, depending on the sources of NO₂. The authors attributed the difference between SCIAMACHY and OMI’s results primarily to daytime photochemical loss of NO₂, weakened by a broad daytime maximum of anthropogenic NO_x emissions. Boersma et al. (2009) showed further detailed results on the seasonal difference of the diurnal variations of the tropospheric NO₂ over urban areas; SCIAMACHY (observing at LT10–11) > OMI (at LT13–14) in summer but vice versa in winter. The authors discussed the reason of the seasonal difference using a chemical transport model (CTM) and concluded that the seasonally varying photochemistry mainly causes the difference. Such a satellite constellation with instruments in morning and noon orbits has contributed to the understanding on the most important diurnal trends in chemistry and emissions. It could further be complemented if measurements from additional instruments in the afternoon were available.

In contrast to LEO, a geostationary earth orbit (GEO) enables hourly measurements throughout the day. Hourly observations with GEO are particularly useful for the study of the diurnal variation of tropospheric species such as NO₂ because;

• •
  data from one instrument are more consistent than data from a constellation,
• •
  hourly data can track emissions e.g. during rush hours which is not possible with a constellation, and
• •
  hourly data at high spatial resolution allow tracking of transport.

Hourly NO₂ data from a geostationary sensor ingested into data assimilation systems could also provide the necessary input to take air quality forecasts further and improve them up to the point that they can reliably forecast violations of air quality standard hours in advance. A GEO measurement has another advantage against the existing satellite constellation; GEO can solve a temporal sampling problem of the existing LEO measurements. We define the temporal sampling problem of LEO as the shortage of the temporal frequency in observations. The LEO observations of a given location are conducted once per day at most. As clouds often prevent measurements down to the surface, the real sampling frequency at a given location tends to be reduced to less than once per day. The introduction of GEO can create a much larger number of cloud free observations, solving the temporal sampling problem of the existing LEO measurements. Note that the introduction of GEO would result in another problem: a spatial sampling problem. Since GEO instruments can only cover part of the Earth, they introduce a spatial sampling problem in that the rest of the Earth is not covered at all. Therefore, while solving the temporal sampling problem, the use of a GEO-instrument generates a spatial sampling problem. Indeed, the existing LEO measurements had a priority on solving the spatial sampling problem and obtained a global coverage of the tropospheric NO₂. To solve the spatial sampling problem in a GEO-based measurement system, several simultaneous measurements, e.g., in Asia, Europe and America will be needed. As an alternative, GEO and LEO observations can be combined to create a complete global picture.

Since the beginning of the last decade, the possibility of GEO-based measurements of tropospheric tracers has been discussed as a contribution to the local- and regional scale air quality monitoring and forecasts (Kelder et al., 2005, Edwards, 2006). In these discussions, several GEO missions such as GEO TROPSAT (Little et al., 1997) and GeoTROPE (Bovensmann et al., 2002, Burrows et al., 2004) were proposed in the past. Presently, the Geostationary Coastal and Air Pollution Events (GEO-CAPE) mission is being studied in the United States, and the Sentinel-4 mission is proceeding for launch in Europe. In Asia, the Korean Geostationary Environment Monitoring Spectrometer (GEMS) is planned to be operating onboard of a GEO satellite. In Japan, the Geostationary Mission for Meteorology and Air Pollution (GMAP-Asia) was proposed by the Japan Society of Atmospheric Chemistry (Akimoto et al., 2008, Akimoto et al., 2009). In those missions, the development of the sensor is one of the key issues, as the requirements for the sensor specification in GEO (36,000 km altitude) become much more severe than for LEO (<1,000 km) because of the much weaker intensity of light coming from the Earth’s atmosphere to GEO.

In the present study, we conducted a feasibility analysis for the detection of diurnal variations of the tropospheric NO₂ from a GEO satellite. As a target region, we focused on Tokyo, which is the largest polluted urban area in Japan. We estimated the precision and its dependence on local time and season by numerical simulations to determine the smallest acceptable signal-to-noise ratio (SNR) for the detector if diurnal variations of NO₂ are to be studied. First, we assume an ideal sensor with constant SNR for the simulations. The results obtained here are general, because no specific sensor specification is assumed in the simulations. Based on these general results, we assume a realistic sensor specification similar to that currently discussed in the GMAP-Asia project and investigate to what extent this hypothetical sensor can detect the diurnal variation of the tropospheric NO₂.