Forest vertical structure is an important parameter of terrestrial ecosystems.The improvement of the inversion accuracy of forest vertical structure with remote sensing has a great significance on improving forest biomass,estimating the accuracy of leaf area index and carrying out researches on forest succession,the carbon cycle and the primary productivity.Laser radar technology is a type of active remote sensing technology developed rapidly in recent years internationally,and has a different working mechanism compared to passive optical remote sensing and a strong detection capability for spatial structure of vegetation and terrain and acquire successful application in the quantitative measurement and inversion of forest parameters,especially for forest height and detection capability of vertical structure,with advantages superior to traditional optical remote sensing data.Airborne laser radar has the ability to acquire large areas of forest vertical structure,but not detailed on the canopy vertical structure description;ground-based laser radar can obtain fine forest vertical structure,especially the canopy vertical structure,ground-based laser radar has the potential to train data for Airborne Laser radar,but limited by the scope of data acquisition.Based on the above content,this paper does some researches on stand in protected areas by using ground-based laser radar as follow:(1)Registration of TLS data and Extraction of the digital elevation model(DEM).Laid reflective target and the reflective ball before the TLS scanning radar plots,and measured the relative position for reflective target with Total Station and correcting and stitching these target point using post-processing software(within TLS),the final step is to convert the ground-based laser into the coordinates of the point cloud successfully.Using TerraSolid software of point cloud data to classify ground points with the ground,the ground points to generate DEM nearest neighbor interpolation method.(2)Extraction of individual tree DBHs of different forest types based on TLS data.Using Hough transform detection algorithm to estimate each DBHs within plots,then according to the morphological characteristics of trees,by setting a series of threshold to rule out fitting round at the non-trunk,it can improve the position of single tree and recognition accuracy of DBHs.Done a regression analysis with measured data showed that:the estimating result of natural broadleaf forest DBHs estimation result is 1.002x+0.568,R2= 0.866.Natural conifer(the dominant conifer species)Individual DBHs estimation resultis y=1.054x-0.447,R2= 0.964.Natural coniferousforest(with dominant broadleaf trees)Individual DBHs estimation result is y=1.027x-0.082,R2 = 0.970.Coniferous fir forest Single DBHs estimate measurement result is y=0.935x+0.210,R2 = 0.919.Estimation result of bamboo DBHs estimation result is y=1.002x-0.048,R2= 0.996.So,These results suggest that the order estimating accuracy of DBHs by TLS is: artificial bamboo forest> natural conifer forest> coniferous forest> natural broadleaf forests.(3)Extraction of individual tree Height of different forest types based on TLS data.It extracts the individual tree height by individual tree trunk growth direction obtained by TLS scanning radar data and a single wooden structure of the vertical direction.Done a regression analysis with measured data showed that: the estimating result of natural broadleaf forest Heights estimation result is y=0.790x+2.534,R2 = 0.820.Natural conifer(the dominant conifer species)Individual Heights estimation result is y=1.009x-0.153,R2 =0.894.Natural coniferousforest(with dominant broadleaf trees)Individual Heights estimation result is y=0.995x-0.004,R2 =0.964.Coniferous fir forest Single Heights estimate measurement result is y=0.953x+0.548,R2 = 0.908.Estimation result of bamboo Heights estimation result is y=1.020x-0.399,R2=0.500.The order estimating accuracy of tree height is: natural conifer> coniferous forest> natural broadleaf forest> artificial bamboo forest.(4)Combined taper equation to evaluate the feasibility of TLS.Putting the data of sample parsing wood into the taper equation to obtain the predicted values of taper equation;using the determined coefficient R2 and RMSE to compare the measured value of parsing wood for test and model values of taper equation in the study,the results show using TLS point cloud data to extract taper equation at 1m-6m fits better. |