EGU Poster Outline - Division of Geological and Planetary

advertisement
Identifying Aerosol Parameters for Trace Gas Retrievals from Near Infrared
Satellite Measurements
Vijay Natraj1, Hartmut Bösch2 and Yuk L. Yung1
1
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA, USA
2
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA
Introduction
One of the major sources of uncertainty in the retrieval of trace gas abundances from spaceborne measurements is the type, amount and vertical distribution of aerosols in the atmosphere.
As this schematic shows, aerosol scattering can introduce errors in the retrieved column amount
of the trace gas by adding uncertainty to the photon path length. Retrieval of tropospheric gases
requires high precision. For example resolving the sources and sinks of CO2 requires column CO2
to be retrieved at 1 ppm precision. Knowledge of aerosol properties is thus crucial.
<path length schematic>
Global Climatology
Kahn and coworkers used results from a collection of global transport models to identify
climatologically probable groupings of component aerosols. They boiled the transport model
results down to a small number of aerosol component groupings, which they called mixing
groups, which encompass the climatologically probable combinations of component aerosols for
all locations and months. They found that five mixing groups were needed to span the
climatology, with each group being further subdivided based on the relative proportion of the
component aerosols. A total of 13 types were enough to adequately describe the observed global
aerosol climatology.
<Kahn climatology>
All the mixing groups contain sulfate particles. Those groups that contain sea-salt particles are
called maritime; “continental” refers to those groups that do not have sea salt among the four
most abundant component particles but do have accumulation mode dust. The other aerosol
components contributing to each group determine whether the classification is “dusty,”
“carbonaceous,” or “black carbon.” The color scheme is the following: most common maritime
classes appear in shades of blue; the most common continental classes are brown. For those
that remain, classes rich in black carbon are gray, those having high carbonaceous aerosol
fraction are green, and the ones abundant in coarse dust are yellow.
Optical Properties
We computed the aerosol optical properties using Mie and T-matrix codes depending on their
shape. In correspondence with standard practice, we used a lognormal size distribution. We also
took full account of polarization because for near infrared retrievals it could be a significant
component of the error budget, especially for polarization-sensitive instruments such as those on
GOME, SCIAMACHY and OCO.
A sample scattering matrix looks like this. We have plotted the phase function and the (1,2)
element, which accounts for linear polarization It is clear that the 13 types do not all have different
radiative effects and that polarization causes a wider spread.
<phase plots for components and the 13 types>
Forward Model Details
Here are some details of the forward model calculations. We considered a scene over Park Falls,
Wisconsin, in July. The solar zenith angle for this case is about 31 degrees, which corresponds to
a 150 degree scattering angle for nadir viewing. We assumed an exponential extinction for the
aerosol, with a scale height of about 1 km and a total optical depth of 0.1. The Orbiting Carbon
Observatory (OCO) spectral regions (0.76 μm O2 A band and CO2 bands at 1.61 μm and 2.06
μm) were used. The multiple scattering code RADIANT was used, with a single scattering
approximation for polarization. We used a Lorentzian instrument lineshape function, with
resolving powers of 17000 for the O2 A band and 20000 for the CO2 bands.
Weighting Functions
We computed weighting functions for each of the spectral regions to determine the change in
observed radiance when the aerosol optical depth was changed. The color coding is such that
each mixing group is represented by one color, with different linestyles used for the subdivisions.
Just by looking at the weighting functions it is clear that the 13 different aerosol groups do not all
have different effects as far as the retrieval is concerned.
<weighting function plots>
Retrieval Groups
We attempted to split these groups into retrieval groups based on the variation of the single
scattering albedo and extinction cross section with wavelength and found surprisingly good
results. We obtained 5 different retrieval groups, which represent aerosol types exhibiting
different physics. For example, groups 5 and 4 differ because group 4 contains accumulation
mode dust, which is absorbing while group 5 contains seasalt which has a single scattering
albedo of 1, even though they are very similar otherwise.
<extinction and ssa plots>
Sensitivity Tests
We did some sensitivity tests to find out the error due to choosing a wrong aerosol type within a
retrieval group. The results show that these errors are smaller than the measurement noise and
the smoothing error.
<sensitivity analysis results>
Conclusions
We used the weighting function behavior of basic aerosol mixing groups to arrive at aerosol
retrieval groups. Sensitivity tests indicated that choosing a wrong aerosol type within a retrieval
group introduced minimal errors. It should be mentioned that we need to test more scattering
geometries to ensure that a weighting function analysis would give us the same retrieval groups.
Also, more realistic aerosol vertical profiles need to be considered.
References
1. Kahn et al.
2. de Rooij et al.
3. Mishchenko et al.
4. OCO
5. Radiant
Download