A fast forward model for IASI and TANSO-FTS CH4 retrievals 

Atmospheric methane (CH₄) is measured continuously from space, providing valuable information at global scales for atmospheric monitoring. CH₄ measurements from space can be based on observations in the shortwave infrared (SWIR), leading to a more uniform sensitivity to the atmospheric column, as well as thermal infrared observations (TIR) which provide useful information on the CH₄ content in the upper troposphere and lower stratosphere. Among the various instruments measuring in the TIR, the Infrared Atmospheric Sounding Interferometer (IASI) series of instruments onboard the METOP satellites have been observing the Earth’s atmosphere for more than 15 years. Also, the Thermal And Near infrared Sensor for carbon Observation Fourier-Transform Spectrometer (TANSO-FTS) on-board the GOSAT 1 and 2 satellites monitor the atmosphere in the SWIR and TIR since 2009. Both instruments provide valuable information for the retrieval atmospheric CH₄. CH4 retrievals in the TIR can be demanding in terms of computational time. The large number of species in the CH4 region, and the high radiometric accuracy of current and upcoming instruments (e.g. IASI, IASI-NG) demand highly accurate radiative transfer modelling (RTM), to be carried out on a fine spectral and vertical grid. These constraints usually lead to long processing time when using full-physics RTMs (e.g. ASIMUT-ALVL). Other fast RTMs exists (e.g. RTTOV), but they often cannot be easily modified to include a specific species or spectroscopy, and do not support all instruments. To allow for faster processing of the already large datasets available, we developed a model based on the Principal Component-based Radiative Transfer Model (PCRTM) approach (Liu et al., 2006) to perform CH4 inversion in the TIR with IASI and TANSO-FTS. Instead of predicting channel radiance directly, the Principal Component-based Radiative Transfer Model (PCRTM) predicts the Principal Component (PC) scores of these quantities parameterized by a relatively small number of monochromatic RT simulations, leading to significant savings in computational time. The model returns the channel radiances and the jacobians for all species of interest. In this work, we will detail the approach that was taken for the development of the fast RTM and will compare the results with a full-physics RTM. The impact of some internal parameters on the current model will also be discussed, as well as possible improvements in the future.   Reference: Xu Liu, William L. Smith, Daniel K. Zhou, and Allen Larar, "Principal component-based radiative transfer model for hyperspectral sensors: theoretical concept," Appl. Opt. 45, 201-209 (2006)