The 3D model is an important part of simulated remote sensing for earth observation. Regarding the small-scale spatial extent of DART software, both the details of the model itself and the number of models of the distribution have an important impact on the scene canopy Normalized Difference Vegetation Index (NDVI).Taking the phragmitesaustralis in the Yangtze Estuary as an example, this paper studied the effect of the P.australias model on the canopy NDVI, based on the previous studies of the model precision, mainly from the cell dimension of the DART software and the density distribution of the P.australias model in the scene, As well as the choice of the density of the P.australiass model under the cost of computer running time in the actual simulation. The DART Cell dimensions and the density of the scene model were set by using the optimal precision model from the existing research results. The simulation results of NDVI with different model densities under different cell dimensions were analyzed by error analysis. By studying the relationship between relative error, absolute error and time costs, we have mastered the density selection method of P.australias model in the simulation of small-scale spatial scale scene. Experiments showed that the number of P.australias in the simulated scene need not be the same as those in the real environment due to the difference between the 3D model and the real scenarios. The best simulation results could be obtained by keeping the density ratio of about 40 trees per square meter, simultaneously, of the visual effects.