The GEISA spectroscopic database has been significantly improved (Jacquinet-Husson et al., 2005) as a prerequisite for the optimal channel selection for CO2 retrievals from the AIRS instrument (Crevoisier et al., 2003).
A detailed error analysis of the differential absorption method has been performed (Dufour and Bréon, 2003), which forms the principle of SCIAMACHY instrument as well as the forthcoming OCO instrument (Crisp et al., 2004). This study showed that the knowledge of the spectroscopic parameters and the impact of scattering by aerosols and undetected clouds constitute the two most limiting factors affecting the accuracy of the retrievals. Indeed, first retrievals from the SCIAMACHY instrument also show encouraging CO2 column concentration patterns, which, however, indicate that indeed aerosols may induce a serious bias of up to 3ppm. In order to be useful for inversions, the retrievals must be corrected for this bias (Houweling et al., 2005). An application of the differential absorption method using ground based solar spectra from the Kitt Peak astronomical observatory shows the potential of the method: column retrievals using the 1.6µm wave band indicate an error on the order of 1.5 ppm (RMS) compared to a forward model simulation (Dufour et al., 2004).
The first retrievals of CO2 from space using the observations from the TOVS instrument on the NOAA polar satellites exhibit credible upper troposphere CO2 concentration patterns (Chedin et al., 2003). This pioneering work demonstrates that remote sensing of CO2 from space is actually possible. These early space borne CO2 retrievals clearly reveal several scientifically very interesting features, such as the day-night patterns detected in the TOVS retrievals that may be linked to diurnal CO2 emissions by biomass burning in the Tropics (Chedin et al., 2005).
Detailed comparison of CO2 column retrievals from TOVS and AIRS instrument with in situ measurements and model simulations
Retrievals from the existing systems measuring the atmospheric infrared radiation, TOVS and AIRS, reveal credible upper troposphere CO2 concentration patterns as compared by in situ measurements and forward model simulations. The zonal averaged annual cycle and atmospheric growth rate pattern in the Tropical region from TOVS retrievals exhibits an accuracy of about 2ppm (RMS) compared to forward model simulations and in situ aircraft measurements (Peylin et al., 2005 submitted). An even better agreement is found by comparing the AIRS retrievals with the forward model simulations and in situ aircraft measurements (Tiwari et al., 2005, submitted).
The operational 4D-Var data assimilation system of the ECMWF has been adapted and applied to estimate the upper troposphere CO2 concentration from the infrared radiances determined by the AIRS instrument (Engelen et al., 2004, Engelen and McNally, 2005). As an interesting side effect the concurrent CO2 retrieval together with other meteorological variables also improves the weather model forecast.
Satellite based CO2 retrievals, if accurate enough may significantly complement and enhance the present in situ observational network. The required accuracy level, however is very high: depending on the sensor and orbit characteristics a single column retrieval must be determined better than 1% in order to allow for a significant improvement in the determination of surface sources and sinks (Houweling et al., 2004).