Comparison of weighting function calculations by linearized radiative transfer models
Abstract
Linearized radiative transfer models provide efficient algorithms to calculate the weighting functions and the layer air mass factors (AMFs), which are used for solution of inverse problem and for determination of gas distribution in the atmosphere. Scattering limb geometry, which is used in a few space observation systems (OSIRIS, GOMOS, SCIAMACHY), is most sensitive for specialties of algorithms for calculation of AMFs, because it has the sharpest AMFs. A comparison of the layer AMF computations by MCC++ spherical model and CDI pseudospherical model was performed. Calculations were carried out at 325, 345 and 600 nm for a geometry coincided to SOLSE/LORE experiment on measurement of scattering limb brightness. The AMFs were computed for atmospheric layers of 1 km thick. The comparison allows identification of three main atmospheric zones characterized by specific manifestations of the peculiarities of the computational algoriths. (1) The first zone is located two layers over the tangent height. In this zone, the difference between two models did not exceed 1%. The AMFs are high here (as a rule, they range from 40 to 8). The AMFs are predominantly determined by single scattering. (2) The second zone is located below the tangent height. In this zone, the layer AMFs are determined completely by multiple scattering. The AMFs range typically between 1.8 and 2.2 and increase up to 3.8 for the troposphere and the visible spectral region. The absolute difference between two models is still small; however, the relative difference reaches 30% for heights of 50-60 km. Apparently, these differences are caused by discrepancies occurring in the multiple scattering computations performed with the spherical and pseudospherical models, which is known issue for these heights. Notice that the retrieval errors should be scarcely affected by these discrepancies because the AMFs in the second zone are small. (3) The third zone represents two layers located just over the tangent height h0. Because of weak absorbtion in layers above, the AMFs in the third zone sharply varies with height approximately like . This peculiarity causes a strong dependence of the computed AMFs on the algorithm and its actual height resolution. It is necessary to give attention to the agreement between the resolutions of the computational algorithm and measuring instrument. For this zone, the typical AMFs range between 60 and 80. The difference between the AMFs of this zone is no more than 1.5%.
- Publication:
-
EGS - AGU - EUG Joint Assembly
- Pub Date:
- April 2003
- Bibcode:
- 2003EAEJA....12154P