Monitoring of the global distribution of water vapor, clouds and aerosols is essential for the understanding of the Earth's energy budget. All three quantities are interconnected in several ways: for example clouds can form when relative humidity exceeds saturation levels, the properties and amount of aerosols determine cloud properties (acting as cloud condensation nuclei), and precipitation is an effective removal process for aerosols. Because water vapor, clouds and aerosols are also involved in several climate feedback mechanisms they are important for the understanding and prediction of climate change. Water vapor, clouds and aerosols are highly variable making global observations from satellite highly desirable. Although instruments like GOME, SCIAMACHY and OMI (GOME-type instruemnst) were not primarily designed for the observation of these quantities, they can provide very valuable information on the global distribution of water vapor, clouds and aerosols.
Cloud and aerosol observations
GOME-type instruments measure the backscattered solar radiance over continuous wavelength ranges with moderate spectral resolution. In the recorded ‘earthshine spectra' the fingerprints of many atmospheric trace gases (including water vapor, see below) can be found. However, the wealth of spectral information is achieved at the expense of a coarse spatial resolution (typically of the order of tens to hundreds of km). Thus GOME-type satellite instruments are usually not the preferred instruments for cloud and aerosol retrievals. Cloud and aerosol retrievals are mainly based on radiance measurements from sensors with moderate spatial resolution (and are often based on only one or a limited set of wavelengths). It is interesting to note here that even the spatial resolution of the so called polarization monitoring devises (PMDs) of GOME-type instruments is still much coarser than that of imaging instruments, e.g. MERIS and MODIS. Despite their limited spatial resolution, cloud and aerosol retrievals from GOME-type instruments are also possible. They have important advantages: The measured absorptions of atmospheric oxygen, O2, and the oxygen collisional complex, [O2]2, allow the determination of the cloud height (typically the height of the cloud center), because clouds shield part of the atmosphere below (cf. Kuce et al., 1994; Koelemeijer et al., 2001, Loyola et al., 2007; Kokhanovsky et al., 2007, 2011; Wagner et al., 2008; Lelli et al., 2014). In a similar way, measurements of the Ring effect (i.e. the filling-in of the Fraunhofer lines due to rotational Raman scattering) can be used to derive cloud altitude information (cf. Joiner and Bhartia, 1995¸ Vasilkov et al., 2008). It was shown, that from the combination of such measurements with simultaneous satellite measurements in the thermal IR, information on multi-layer clouds can be derived (Joiner et al., 2010). Such cloud properties are important for the correction of tropospheric trace gas observations from satellite instruments. They also provide useful input for the quantification of the radiative effects of clouds. It is interesting to note that cloud retrievals from GOME-type instruments are usually based on relative quantities (e.g. the depth of an O2 absorption band). Thus the derived cloud and water vapor products are rather insensitive to instrumental degradation and are therefore especially well suited for trend studies (e.g. Wagner et al., 2006; Mieruch et al., 2008, 2014; Lelli et al., 2014).
For aerosol retrievals, the large pixel sizes cause particular problems, because aerosol retrievals are heavily influenced by clouds (even for very small cloud fractions). In addition, the surface contribution to the measured top-of-atmosphere reflectance is often uncertain and often larger than the contributions of aerosols. One way around these issues is the use of the Absorbing Aerosol Index (AAI), which is determined from reflectance ratios at selected UV wavelengths (usually around 340 and 380 nm). This method was originally developed for TOMS observations (e.g. Hsu et al., 1996; De Graaf et al., 2005). AAI observations are possible also for cloudy scenes. In fact, if aerosols are located above clouds, the sensitivity towards absorbing aerosols is even increased. AAI measurements are very useful for the detection of biomass burning aerosols and volcanic ash. Recently, it was shown that also non-absorbing aerosols (like sulphate particles or secondary organic aerosols) can be detected by this method (Penning de Vries et al., 2009). However, from these UV aerosol indices (for both absorbing or non absorbing aerosols) quantitative information cannot easily be deduced, since UV aerosol indices (UVAI) depend not only on the aerosol single scattering albedo, but also on the aerosol optical depth and the layer height. In contrast, the most recent quantitative aerosol retrievals using OMI observations are based on more detailed radiative transfer simulations, which allow the simultaneous retrieval of both aerosol optical depth (AOD) and the single scattering albedo (SSA) under cloud free conditions (Torres et al., 2013), and the retrieval of AOD of aerosols above clouds (Torres et al., 2012, Ahn et al., 2014; Jethva et al.,2014). Another recent development is the combination of satellite observations of aerosol properties and trace gas amounts in order to determine aerosol properties and sources on a global scale from satellite observations (Veefkind et al., 2011; Torres et al., 2013; Penning de Vries et al., 2015).
A few studies also derived stratospheric aerosol profiles from SCIAMACHY limb observations (e.g. Ernst et al., 2012, Savigny et al., 2015). Such observations are very important because of their daily global coverage, which is not possible for occultation measurements. Recent studies investigated the effect of non-homogeneous aerosol distributions after strong volcanic eruptions on limb aerosol retrievals and indicated the need to correct for these effects on the retrieved aerosol extinction and layer height (Penning de Vries et al., 2014).
Water vapor observations
Atmospheric absorptions by water vapor occur in many spectral ranges (from the blue to the near IR) which are covered by GOME-type instruments. In general, the strength of the water vapor absorption increases towards larger wavelengths. Thus, depending on the viewing geometry and the atmospheric water vapor content, different spectral ranges for water vapor retrievals can be selected (e.g. Noël et al., 1999; Casadio et al., 2000; Lang et al., 2003; Wagner et al., 2003; Wagner et al., 2013; Wang et al., 2014). One advantage of water vapor retrievals from GOME-type instruments is that they are sensitive to the surface-near layers, where the water vapor concentrations are highest. In addition, they can be applied over both land and oceans. This allows in particular the study of ENSO-related changes on a global scale (Wagner et al., 2005; Loyola et al., 2006). A fundamental disadvantage of these retrievals is the relatively strong influence of clouds, which can cause large uncertainties for individual observations.
Compared to other tropospheric trace gases, the retrieval of the atmospheric water vapor column is challenging in several aspects: In many spectral ranges, the water vapor absorptions are quite strong. Thus saturation effects have to be corrected, and also the radiative transfer is rather complex (in principle line by line calculations are required). Thus for many retrieval algorithms the determination of appropriate AMFs is not based on radiative transfer simulations but on simultaneous measurements of the oxygen absorption (e.g. Wagner et al., 2005; Noël et al., 2004). This approach leads to very ‘stable' water vapor products, which are almost independent of external information and thus well suited for trend studies. Future water vapor retrieval algorithms will include full radiative transfer simulations, which will improve the accuracy for individual observations.
Kelly Chance (SAO), Steffen Dörner (MPIC), Michael Grzegorsky (EUMETSAT), Margherita Grossi (DLR), Joanna Joiner (NASA/Goddard), Alexander Kokhanovsky (EUMETSAT), Luca Lelli (IUP Bremen), Sebastian Mieruch (KIT), Stefan Noël (IUP Bremen), Marloes Penning de Vries (MPIC), Holger Sihler (MPIC), Gijsbert Tilstra (KNMI), Omar Torres (NASA/Goddard), Huiqun Wang (SAO)