Tomographic retrieval of atmospheric parameters from infrared limb emission observations

Abstract

Typical inversion of limb-sounding measurements assumes local horizontal homogeneity of the atmosphere. This simplification corresponds to spectral radiance errors that can exceed the noise level of a typical infrared instrument by a factor of 10 and causes errors in retrieved state parameters. To avoid these errors and to take the horizontal structure of the atmosphere into account, a two-dimensional (2D) tomographic sequential estimation approach is described. Application to temperature retrievals from simulated measurements yields typical retrieval errors of the order of 1 K, and a one-dimensional retrieval with the same synthetic measurements shows differences to the true values up to 10 K in regions with strong horizontal inhomogeneities. The horizontal resolution of the 2D retrieval is even better (up to 40 km) than the horizontal tangent point spacing.