4 Application of the Program McArtimInversion

In order to demonstrate the capabilities of the non linear inversion procedure in combination with a MC RTM this section discusses the retrieval of an aerosol profile from a simulated measurement which consists of logarithmic sun normalised radiances log (I∕I0) at 300nm and 470nm recorded in different heights. For both wavelength aerosol parameters ϖ0 = 1 and g = 0.67 has been chosen and the ground albedo was set to 0.05. The geometry description file is listed in figure 2.


 
1Height=0.12 
2Longitude=0.0 
3Latitude=0.0 
4Azimuth=123.0 
5Elevation=0.0 
6Omega=0.01 
7SZA=35 
8SRAA=-134

Figure 2: File geometry.conf.


The height definition in line 1 is ignored, since it is overwritten in the simulation script (file content listed in Fig. 3).


 
1simulation=simulation.conf 
2simulation.wavelength_nm=477 
3raytracer.no_quadrature_trajectories=100000 
4intensity.b_logarithmic_output=yes 
5intensity.fn_outfile=logI.txt 
6intensity.enable=yes 
7combine 
8        geometry.Height=0.05:0.4:8.1 
9end

Figure 3: File script.conf. The detector altitude is varied in steps of 0.4km between 0.05km and 8.1km.


The resulting logarithmic intensities are plotted in the left panel of figure 4


PICPIC
PICPIC

Figure 4: Top-left: Simulated logarithmic sun normalised radiances. Top-right: effective number of scatter events for ground, Rayleigh and aerosol scattering, where the aerosol phase function was approximated by the Henyey-Greenstein model with g = 0.67. Bottom panels: profile evolution during the non linear inversion at 300nm (left) and 477nm (right).


In principle, it should be possible to invert the aerosol profile from the 21 measurements, so the a priori information was not considered and it has beed tried to retrieve the profile from the measurement only. The cost function was

  2      ⃗        T  -1 ⃗
χ (⃗x ) = [F (⃗x) - ⃗y] Sɛ [F (⃗x) - ⃗y ].
(34)

Figure 4 shows the profile evolution in different steps of the iteration. The difference of the RT at 300nm and 477nm is the ratio of the Rayleigh extinction coefficient ɛR to the aerosol extinction coefficient ɛA. This can be seen cleary from the top-right panel of Fig. 4: at 300nm and above 4km ɛA is smaller than ɛR, where at 477nm ɛA is always greater than ɛR. In consequence, the 477nm measurement is generally more sensitive towards the presence of an aerosol.