Tree crowns are usually simulated as basic geometrical shapes in geometric-optical models. However

the selection of specific shape often based on visual judgment and may induce uncertainties

and the uncertainties would propagate to the practical application of the model

retrieving Leaf Area Index（ LAI）

for example. This study aimed to address the potential effect of crown shape by conducting a simulation experiment. In this study the Nilson’s gap probability model was enhanced to ensure compatibility with commonly encountered crown shapes

such as ellipsoid

cone and cone + cylinder. The enhanced Nilson’s gap probability model can be used to calculate volumes and projected area profiles of the three shapes. This enhanced model was then compared with a gap probability model derived from Beer’s Law. Clumping index was then formulated. This can be described as a function of canopy structure parameters

such as crown volume

projected area

and stem density. On the basis of the enhanced Nilson’s gap probability model and formulated clumping index

we analyzed the sensitivity of gap probability

clumping index

and LAI retrieval to crown shape. In this study

the gap probability was found sensitive to crown shape. This result could be attributed to the volumes and projected areas of different crown shapes. A high difference in gap probabilities was observed between various crown shapes in the middle of the canopy. Below the canopy

crowns with ellipsoid and cone + cylinder shapes exhibited very similar（ relative error< 9. 7%） gap probabilities. By contrast

these gap probabilities were very different from those of cone-shaped crown（ largest relative error > 36. 7%）. Our results further showed that clumping index is a function of crown shape. Similar to gap probability

the clumping indexes of crowns with ellipsoid and cone + cylinder shapes were very similar below the canopy. By contrast

these indexes differed from those of cone-shaped crown. Hence

crown shape should be considered when LAI is retrieved using ground measurements. Relative error likely reached > 25% if crowns were set in wrong shapes. Gap probability and clumping index underneath the canopy were very similar when crowns were modeled in ellipsoid or cone + cylinder shapes; however

a considerable difference was observed if crowns were modeled in a cone shape. The volumes and projected areas of different crown shapes could be a ccounted for similarities or differences in results. Based on the retrieved LAI

crown shapes could account for 25% of the relative retrieval error. Therefore

crown shapes should be carefully specified to obtain a satisfactory result. Our analysis results were d erived from simulation experiments; in our future work

ground experiments would be performed to verify our conclusions.