Studies for the Odin sub-millimetre radiometer: I. Radiative transfer and instrument simulation

Abstract

The Odin satellite mission will include radiometric measurements of the Earth's atmosphere in a limb-sounding mode, using frequencies between 480 and 580 GHz, with the overall aim of retrieving vertical distributions of atmospheric constituents. The current paper, being one of a three-part series, addresses primarily the modelling of atmospheric radiative transfer and the effect of instrumental properties: the forward model. Such a model is required for the retrieval process and this presentation puts emphasis on refraction, sensor characteristics, systematic model errors, and some implementation aspects. Refraction must be considered below about 15 km and an efficient algorithm to include this effect is presented. Sensor parts treated are the antenna, the side-band filter, and the spectrometer. The forward model is also essential for determining the needed weighting functions. A semi-analytical expression for species-abundance weighting functions is derived. To form a common basis for the article series, a comprehensive formalism is reviewed and general issues, such as the separation between fixed and variable uncertainties, discussed. As a complement to the theoretical characterization, limited to linear situations, the possibility of using repeated simulations is also described. PACS Nos.: 42.68A, 07.07D, 07.57K

Qpack, a general tool for instrument simulation and retrieval work

“…In [8] it is pointed out that the sensor part of the forward model can be implemented by a multiplication between a response matrix, H, and i. That is,…”

“…(4), was suggested by [8] but became first implemented in conjunction with ARTS-1-0-x. The idea is based on the fact that the response of the different sensor parts corresponds typically to some linear operation.…”

“…Linear retrievals follow Eq. (8). Nonlinear cases require an iterative search for the solution where the Marquardt-Levenberg approach is used.…”

“…Despite that a lower number of iterations could be achieved, the total calculation time was little affected and could even be increased. The reason for this is that the extra calculation burden of going to a new iteration is to determine K i , and as this part was rapid in the used forward model [8], it could be more time consuming to test more values for g (requiring the calculation of F ðx i ; bÞ) than just going to the next iteration. As the calculation of Jacobians in ARTS uses analytical expressions, as in [8], the same behaviour was expected in Qpack and only a simple scheme for the updating of g was implemented.…”

ARTS, the Atmospheric Radiative Transfer Simulator – version 2.2, the planetary toolbox edition

“…Right from the start the code was open source (GNU's Not Unix (GNU) public license); the current version is freely available at www.radiativetransfer.org (last access: 16 April 2018). At the start, the model focused on simulating clear-sky limb observations of Earth's atmosphere in the millimeter and submillimeter spectral range because that was the main interest of the authors (Eriksson et al, 2002;Buehler et al, 2005b). Pretty soon, the interests widened, and ARTS adopted new capabilities such as simulating down-looking meteorological microwave sensors and active radio link measurements (Eriksson et al, 2003).…”