This study develops a Weighting Featured Monochromatic Radiance (WFMR) method to efficiently and accurately compute the vector radiance, including multiple scattering at the near-infrared band. The monochromatic radiances within a spectral band that are of a high correlation are grouped to represent the average band radiance. The simulated radiance from WFMR is validated against a large variety of OCO-2 (the Orbiting Carbon Observation 2) profiles. It is shown that the dominant errors introduced by the WFMR method are below 0.1%, 0.025%, and 0.12% for O2 A band, weak and strong CO2 absorption bands, respectively, and the computational speed is, however, about 80 times faster than that of the reference line-by-line calculation. The technique performs very well and captures the realistic characteristics of the atmospheric radiation distribution. Besides, for the WFMR method, its numerical accuracy can be defined, and its linearization to form the Jacobians matrix is very straightforward. These advantages are particularly well suited for remote sensing applications.