Abstract: The leaf area index (LAI)plays an important role for describing the growth of the cotton. It is necessary to obtain the canopy structure parameters accurately for the inversion of leaf area index. And the light detection and ranging (LiDAR)provide a way to obtain the parameters accurately. So far,domestic and international studies about LAI are mainly focused on forest. The researches on extracting the crop parameters with point cloud data are still in the initial stage,and the method of using laser scanner to retrieve the leaf area index of crop needs further verification. On the other hand,there are little researches on extracting the crop parameters with low density of point cloud data. How about using high density of point cloud data to extract the crop parameters? In this paper, the researchers obtain high-density point cloud data from farmland by keeping the unmanned aerial vehicle (UAV)flying low,and in this way,the relationship between the vegetation parameters and the LAI is explored, the point cloud density is 299 pts·m^-2. By using the Scout B-100 oil-powered single-rotor UAV as the flight platform, RIEGL VUX-1 obtained high accurate point cloud data. Firstly,using the laser scanner to get point cloud data,the density of which is 299 pts·m^-2 from Mushroom Lake Regional Farmland in northwest of Shihezi City, and the point cloud filtering is used to separate the ground and the vegetation;Secondly,the digital surface model (DSM)and digital elevation model (DEM)of the study area were affected by point cloud data. Thirdly,the canopy height model (CHM)was obtained by calculating the difference between DSM and DEM,after which,the effective canopy structure parameters were extracted from the CHM by the watershed algorithm. In order to construct LAI cotton inversion model,correlation analysis,selecting the correlation coefficient,which is greater than 0.2,including laser penetration index (LPI),echo point cloud density (D),gap fraction (f_gap)and value of normalized elevation (V_nDSM)has been made. Finally,verify the accuracy and evaluation with the measured leaf area index. The experimental results show that R² between the estimated LAI and the measured LAI is 0.824,and the RMSE is 0.072,which verifies the reliability of the model. The filtering method of point requires further investigation, and the method of collecting high-density point cloud data in the low vegetation area is still not explicit. Considering the small scale of the cotton,the research does not choose the leaf angle as the model input parameters, and the results are satisfactory. Compared with other crops LAI parameter inversion method,it is more efficient and faster than the others. By showing that using high density of point cloud data to extract the cotton leaf area index in the study area,it is concluded that this method can also be applied to point cloud data with high precision satisfactorily.

Key words: airborne LiDAR, canopy height model, cotton, LAI