Methane is known to be the second most significant greenhouse gas, and expanding observations of its atmospheric content is important and relevant. The IKFS-2 Fourier transform spectrometer is installed onboard the Russian Meteor-M No. 2 spacecraft. The primary purpose of this instrument is to perform meteorological sounding of the atmosphere. However, the spectral range (660-2000 cm-1) of these measurements contains methane absorption bands. This may allow the use of the outgoing radiation spectra, measured aboard the Meteor-M No. 2 spacecraft, to estimate atmospheric methane content.
We developed an algorithm for determining methane content from these spectra, based on a neural network approach and using TROPOMI data for training. Various approaches to training artificial neural networks (ANNs) and their configurations were considered.
These approaches yield similar results. To verify the accuracy of the obtained total methane content (TMC) values, we processed spectra obtained from the first satellite of this series for the period 2015-2022 and compared the results with ground-based measurements from the NDACC and TCCON networks. The average differences between satellite and ground-based measurements of the columnar methane mixing ratio for different measuring stations range from 5 to 35 ppbv (0.3 - 2.5%), and the standard deviations of these differences range from single digits to 27 ppbv (from tenths to 1.5%). The temporal variability of TMC at various scales is assessed, and examples of its spatial distribution and variability over Russian territories are provided.