Forest Non-Lambertian Modeling Error Correction Based on DART

As a response to the Lambertian assumption modeling error in the inversion of surface reflectance, a discrete anisotropic radiative transfer model (DART) is employed to model surface scenes in a variety of terrains and vegetation situations. It is determined that the non-Lambertian feature changes in surface reflectance resulting from terrain, vegetation canopy distribution, and terrain superimposed vegetation are fitted, and that a conversion model (MBRDFS) for Lambertian surfaces is established. Then, an approach is proposed for correcting the error of the Lambertian assumptions in modeling. The results show that compared with the measured reflectance, the absolute errors of the surface reflectances corrected by the MBRDFS model are 0.0146, 0.0033, and 0.0274 in the near-infrared, red, and green bands, respectively, which are 0.34%, 1.47%, and 1.16% higher than the accuracy of the surface reflectances before correction. Compared with the uncrewed aerial vehicle (UAV) surface reflectance observations, the absolute errors in the red and green wave bands are 0.0165 and 0.0063, respectively, increasing the surface reflectance accuracy by 1.97% and 1.45%, respectively. The surface reflectance of the MBRDFS model after Lambertian modeling error correction is more consistent with the real surface reflectance. The MBRDFS model proposed in this study expresses the relationship between Lambertian characteristics and changes in surface features from a new perspective, which can provide a reference for the expansion of non-Lambertian radiative transfer modeling and canopy radiative transfer models coupled with terrain in the future.