This paper presents an improved Empirical Orthogonal Functions （EOF） model for estimating methane profiles from spaceborne hyperspectral infrared observations based on profile dataset from European Centre for Medium-Range Weather Forecasts （ECMWF） and radiative transfer model RTTOV9.3. The model was applied to Atmospheric Infrared Sounder （AIRS） observations

validated by ground-based Fourier Transform Spectrometer （FTS） observations and AIRS v5.0 CH4 products. Compared with FTS measurements

the Root Mean Square （RMS） relative error of CH4 mixing ratio of EOF retrieval was smaller than that of AIRS v5.0 CH4 product for data lower than 300 hPa

and the relative error of CH4 column amount of EOF retrieval was also smaller. Compared with AIRS v5.0 CH4 product

the coefficient of determination for CH4 profiles retrieved from EOF model was 0.9715

and the RMS relative errors were smaller than 2.5%. The validation results show that the EOF model could provide a good initial value for physical retrieval and is a promising operational approach due to high stability and efficiency.