A simple scheme to process solar diffuse radiation was added in a bidirectional reflectance distribution （BRDF） model of vegetation canopy. The canopy was regarded as turbid media in BRDF model. With ground remote sensing data of soybean canopy

in visible spectral band

the modified BRDF model was inverted. The inversion experiments were organized to include those considering initial values of model parameters

contribution of solar diffuse radiation

and the influences of solar zenith and azimuth during remote sensing data collection. These experiments can be benefit to the description of the physical processes in BRDF model

understanding of the model inversion and the selection of proper remote sensing data for the inversion. （1） For the BRDF model used in inversion experiments

the appropriate estimation of LAI initial value is helpful to obtain good model inversion. When LAI value scope is one times or a bit more larger than the actual data and coincides with LAI scope of random initial values

the results of inversion are good. （2） With the consideration of solar diffuse radiation in BRDF model

the inversion can give more physically reasonable explanations. （3） The ground remote sensing data

collected under small zenith （

<

45°） and less solar azimuth deviation （

<

45°） to 180°

support stable inversion of the BRDF model. These conditions hint the period for data collection from 10:00 a.m. to 2:00 p.m. （4） Within 31°—61° of solar zenith and 136°\

2

58° of solar azimuth

the multi angle observations of canopy reflectance makes the influence on the results of LAI by BRDF model inversion insignificant. （5） The influence of solar diffuse radiation on the results of LAI by BRDF model inversion was not significant when diffuse part was not large in total solar radiation. If only focusing on LAI in BRDF model inversion

it is possible to use the ground remote sensing data without atmospheric corrections. Though these results were obtained when processing ground remote sensing data

it is still potential to be applied in the selection and processing of satellite remote sensing data. The above conclusions are useful to the inversion of vegetation BRDF model in trubid media assumption. At the same time

the data used in model inversion were observed from soybean canopy which was close to ideal condition. The consideration of complex canopy was not included in this study. The potential applications of these results with satellite remote sensing data still need verifications.