In studies related to the scattering characteristics of objectives of physical context, the Bidirectional Reflection Distribution Function (BRDF) is the preferred option for describing the anisotropy of objectives. The geometric attenuation factor in BRDF model has a great impact on the accuracy of BRDF model, which is related to the target surface undulation, roughness and refractive index of the material. The current research and optimization of geometric attenuation factor is mostly from the perspective of surface geometric structure, such as target surface masking and shadowing. The wavelength is not included in influencing factors. However, the relative value of wavelength and surface undulation height directly affects the determination of target surface roughness, thus interfering the geometric attenuation factor, which seriously affects the accuracy of BRDF model. It is necessary to take the wavelength as a part of geometric attenuation factor, study their functional relationship and condense their law of influence.This paper establishes an empirical functional model of the geometric attenuation factor by optimizing the existing geometric attenuation factor from experimental data. Based on the six-parameter spectral BRDF model, the measurement data of target plants is used to inverse the six-parameter values at different incident wavelengths, and the influence law of wavelength on the six-parameter is condensed. The numerical relationships among error values, wavelength, and geometric attenuation factor are analyzed to calculate the expression for the geometric attenuation factor as a function of wavelength. The feasibility of the optimization idea in this paper is verified by comparing the Root Mean Square Errors (RMSE) before and after adding the coefficients related to wavelength.The results show that adding a coefficient, which is exponentially correlated to wavelength, to geometric attenuation factor can significantly improve the accuracy of the six-parameter spectral BRDF model at large scattering angles and reduce the modeling error due to the influence of wavelength. The optimized geometric attenuation factor better fits the measured value versus wavelength curve, and the RMSE of target plants is reduced by about 40% compared to pre-optimization. The wavelength has a weak effect on the a, c parameters, while the effect on the b, k_b, k_d, k_r parameters is more pronounced. Wavelength mainly affects the range of numerical changes and the band location of trend turning point. The specific pattern is affected by the material and surface roughness of target plants after preliminary analysis.