Abstract: The special geography of the plateau area leads to a series of problems in boiler operation. In this study, the absorption coefficients and total emissivity of flue gas were determined under different air pressures during air combustion using the Line-By-Line (LBL) method based on the HITEMP2010 database (High-temperature molecular spectroscopic database). The effects of pressure, temperature, and molar fraction (H₂O and CO₂) on the radiative properties of flue gas were analyzed. An improved Weighted-Sum-of-Gray-Gases (WSGG) correlation, which relates the absorption coefficients to temperature and total pressure, was proposed. The results show that the reduced total pressure diminishes the total emissivity of flue gas. The maximum differences in total emissivity along the path lengths for the four working conditions with a pressure drop from 0.101 325 to 0.061 655 are 0.093 4, 0.084 5, 0.091 1, and 0.084 3, respectively. For a larger molar fraction, the effect of pressure on the total emissivity is greater for shorter path lengths but not for longer ones. Similarly, the higher temperature would reduce the total emissivity of flue gas. The maximum differences in total emissivity along the path lengths for the four working conditions with a temperature increase from 1 000 K to 2 500 K are 0.273 6, 0.270 5, 0.251 5, and 0.250 5, respectively. For a larger molar fraction, temperature has a greater effect on the total emissivity for shorter path lengths but not for longer ones. Furthermore, increasing the molar fraction enhances the total emissivity of flue gas. The maximum differences in total emissivity along the path lengths for the four working conditions with a molar fraction increase from 1 to 2 are 0.088 1, 0.100 4, 0.088 9, and 0.100 6, respectively. For a higher temperature or lower pressure, the effect of molar fraction on the total emissivity is smaller for shorter path lengths but greater for longer ones. The maximum relative error of the improved WSGG model for the total emissivity of flue gas under different working conditions is 3.67%. It is a significant reduction in the error compared to that of the existing WSGG model. Therefore, the improved WSGG model is more accurate for air combustion atmosphere and sub-atmospheric pressure.