In order to observe two kinds of greenhouse gases, CO2 and CH4, making the biggest contribution to global warming, a ground-based Fourier transform near-infrared spectral remote sensing system was developed to record the perpendicular incidence sun spectra from February 2012 to April 2013 in Hefei continuously. The measured total transmittances in the atmosphere were obtained from perpendicular incidence sun spectra. Methods of line-by-line and low-order polynomial approximation were used to model the total atmospheric transmittances in forward model. The measured transmittance spectra were fitted iteratively using the modeled transmittance spectra in the regions of CO2 6 150～6 270 and CH4 5 970～6 170 cm-1 in order to obtain their column concentrations. The column-average dry-air mole fractions of CO2 and CH4 were obtained with the internal standard function of O2 column concentrations. CO2 and CH4 daily average values of column-average dry-air mole fractions changed with a larger fluctuation and obvious seasonal periodicity. Their monthly average values were consistent as a whole, although there were different characteristics. Compared with the results reported by Japanese greenhouse-gas satellite in the area of Waliguan, there was a time lag corresponding to peak and trough of CO2 content and the change from peak to trough costed a long time. CH4 content showed variation tendency of unique peak and trough, higher in summer and lower in winter, compared with average values of nationwide CH4 column concentrations based on SCIAMACHY data. The variation characteristics were related to complex factors such as the balance of source and sink, meteorological and climate conditions, and required long-term observation and further study.