Rotational Raman scattering in the Earth's atmosphere ... - SRON

16 Chapter 1
The answers to these questions are given in Chapters 2 and 3, respectively. In Chapter 2 we present
a model approach that is based on the doubling-adding method. The power of this approach is that
the multiple scattered radiation, both elastically and inelastically scattered, is obtained in a straightforward
way. One starts with the analytical single scattering solution and uses the extended doublingadding
equations to obtain the multiple scattered radiation. With this scalar model the approximation
of including only one order of Raman scattering is investigated.
Chapter 3 describes a different approach where the effect of inelastic Raman scattering is described
by a perturbation to the radiative transfer problem which includes multiple Rayleigh scattering.
Here, the perturbation series describes increasing orders of Raman scattering. For numerical
implementation the perturbation series is truncated at first order, which means that one order of Raman
scattering is included in addition to multiple elastic scattering. This eases the implementation
and results in a very efficient vector radiative transfer model including inelastic Raman scattering.
With this model the impact of the scalar approximation on the reflectivity spectrum including the
Ring effect can be investigated.
Accurate simulation of the Ring effect relies heavily on the use of an accurate input solar spectrum.
The Ring effect that is induced by the solar Fraunhofer lines is directly linked to the exact
amplitude and spectral shape of the Fraunhofer lines in the solar spectrum. In the literature it is common
to rely on a reference solar spectrum. However, large deviations exist between these reference
solar spectra [Gueymard, 2004]. In addition, these spectra show significant differences with the solar
spectra measured by instruments such as GOME, SCIAMACHY, OMI and GOME-2 (e.g. Fletcher
and Lodge [1996]). This raises the question:
3. Since the input solar spectrum is so crucial for the simulation of the Ring effect, what is the best
modeling approach?
In Chapter 4 we describe a newly developed approach in which the maximum amount of information
that is contained in the daily measured solar spectrum is employed to simulate the spectral finestructure
in the reflectivity measurements instead of relying on a reference input solar spectrum.
The Ring effect in the measurement is sensitive to presence of clouds (e.g Joiner and Bhartia
[1995], de Beek et al. [2001]) and aerosols (e.g Stam et al. [2002], Langford et al. [2007]) in the
observed scene. The cloud information that can be retrieved from the Ring effect, such as cloud top
height, can be used to improve trace gas retrievals for cloudy scenes. This is important since most of
the GOME, SCIAMACHY, OMI and GOME-2 observations are influenced by clouds due to the large
field-of-view of the considered spectrometers (see footprint size in Table 1.1). Two other approaches
are commonly used in the literature for the same purpose, but these are based on using absorption
features of oxygen and the oxygen dimer in the reflectivity spectrum, i.e. the O 2 A band near 760 nm
and absorption by O 2 -O 2 at 477 nm. In the last chapter, Chapter 5, we consider the question:
4. How does the capability to retrieve cloud parameters from the Ring structures compare with
other commonly used approaches that use absorption bands of oxygen and the collisional complex
of oxygen?
The different approaches for the retrieval of cloud parameters are evaluated, and the most appropriate
method is proposed for present and future GOME-type instruments.