Fusion Of TLS And UAV Lidar To Estimate Leaf Area Index Of Forest

Lidar technology can directly obtain three-dimensional information of ground objects and has unique advantages in inverting forest parameters.LAI is one of the key botanical parameters and evaluation indicators,which can describe the changes of carbon and water fluxes,and is of great significance to ecological research.This paper takes two plots of mixed coniferous and broadleaf forest and pure coniferous forest in Maoershan Forest Farm in Heilongjiang Province as the research object.Bell’s law and multiple linear regression methods were used to estimate the forest LAI.Then,the fused data is thinned at different sampling rates,and the effective LAI is estimated based on the thinned fused data,and exploring the effect of fusion data on the estimated LAI results.The results show:1)The voxel of TLS data for the two types of plots of coniferous pure forest and mixed coniferous and broadleaf forest estimated the accuracy of the effective LAI areR~2=0.79,RMSE=0.69、R~2=0.74,RMSE=0.63.2)The accuracy of estimating true LAI of the coniferous pure forest and mixed coniferous and broadleaf forest by using UAV lidar data voxel areR~2=0.76,RMSE=0.83、R~2=0.75,RMSE=0.97.The RMSE is large,and the result is not ideal.According to Bell’s law of UAV lidar data,the accuracy of estimating effective LAI for two types of plots of pure coniferous forest and mixed coniferous and broadleaf forest areR~2=0.78,RMSE=0.55、R~2=0.75,RMSE=0.66.The estimation accuracy of LAI of pure coniferous forest by both methods was higher than that of mixed coniferous and broadleaf forest.3)The accuracy of estimating the effective LAI of the mixed coniferous and broad-leaved forest by voxel of fusion data isR~2=0.84,RMSE=0.54.The accuracy of estimating the true LAI by the multiple regression analysis method isR~2=0.86,RMSE=0.62.The fused data was thinned at different sampling rates and the effective LAI was estimated by the voxel.When the sampling rate is 0.8,the estimation accuracy was the highest(R~2=0.93,RMSE=0.63).In summary,the fused data can make up for the imperfection of the forest stand information obtained by a single type of lidar data,which can more fully display the three-dimensional information of the forest stand and improve the accuracy of estimating the leaf area index,but the fusion data will cause data redundancy,and the estimation of the effective leaf area index estimation accuracy is the highest after the fusion data is diluted at a sampling rate of 0.8.