Estimation Of Poplar Stand Parameters And Structure Research Based On Airborne LiDAR

In the forest resource survey,airborne LiDAR can effectively obtain the three-dimensional structure information of forest,accurately estimate the stand parameters and detect the forest structure.In this study,three methods were used to develop a stand parameter estimation model by using two airborne LiDAR data collected from Poplar Plantation in Chenwei forest farm,Sihong County,Suqian City,Jiangsu Province in late March 2019 and late November 2019.The first method is to extract the height and crown of a single tree by using local maximum value and marker controlled watershed segmentation algorithm,and then get the height and crown width of each tree.The arithmetic mean height(AvgH)and the average crown width(AvgCW)of the sample plots were used as the prediction variables for regression analysis with the stand condition mean height(?H),the stand breast height sectional area(BA)and the stand volume(M);The second method was based on the stepwise regression between the 36 cloud characteristic variables of airborne LiDAR and the stand parameters;The third method was based on the average height which extracted by sampling of vertical angle gauge points(Vt H)and stand parameters were extracted from airborne LiDAR data for regression analysis.The purpose of this paper is to study and build the best estimation model of stand parameters based on airborne LiDAR data,and to get the changes of the estimated values of each stand parameter according to the best estimation model of each stand parameter and the airborne LiDAR data of two periods,and to make linear regression with the changes of the measured values of each stand parameter,analyze the accuracy,and then discuss the influence of four different afforestation densities on the changes of each stand parameter in the study area.Finally,the stand structure rules of different afforestation density was studied,the following conclusions were obtained:(1)Based on the canopy height model(CHM),the local maximum algorithm is used to mark the top of tree tip,extract the height of single tree,and then use the watershed algorithm based on the marker control to segment the single tree,obtain the crown boundary and determine the crown size of single tree.In this study,1845 poplars were divided,1922 of them were measured,and the accuracy of segmentation was 95.99%.Linear regression between extracted tree height and measured tree height,the coefficient R~2 is 0.9116,and the coefficient R~2 is 0.8374 by linear regression between the extracted crown width and the measured crown width.It shows that the height and crown width precision of single tree extracted in this study are high.(2)In this study,three methods were used to estimate stand parameters based on airborne LiDAR data.It was found that the accuracy of the model was the highest when VtH was used to estimate the mean height of stand condition(?H).The optimal model is?H=0.7675×VtH+5.5539,R~2 is 0.8171,andρis 92.24%.Vt H was used to estimate the BA.The accuracy of the model obtained from base area(BA)is high.The bivariate model with AvgCW factor is better than the univariate model with Vt H only.The optimal model is BA=0.7612×VtH-0.2645×AvgCW-2.291,the determination coefficient R~2 is 0.6746,and the model accuracyρis 86.28%.Three methods are used to estimate the stand volume(M),It is found that the accuracy of binary regression model with AvgCW and Vt H is still the highest.The optimal model equation is lnM=2.2544×lnVtH+0.03×lnAvgCW-1.008,R~2 is 0.6875,and the model accuracyρis 87.71%.(3)Using the airborne LiDAR data of two periods,the change of the estimated value of each stand parameter is obtained by the optimal estimation model of each stand parameter,and the regression analysis is carried out with the change of the measured value of each stand parameter of two periods.It is found that the correlation is high,among them:the decision coefficient R~2 of the model fitting of the average height(?H)of two periods change of stand condition reaches0.8172;the base area of two periods stands(BA)coefficient R~2 of variation model fitting is 0.7817;the coefficient R~2 of two-stage stand volume(M)variation model fitting is 0.7888.(4)According to the change of stand parameters in two periods,the influence of different afforestation density on the change of stand parameters was analyzed.It was found that the change of stand parameters in two periods decreased with the increase of afforestation density,such as the average height(?H),the base area(BA)and the stand volume(M).The change of stand parameters in sample plot with afforestation density of 6m×6m was the largest.The mean value of variation of mean height(?H)under different conditions is 0.87m;The mean value of variation of basel area(BA)is 0.75m~2/ha;The mean value of variation of stand volume(M)is16.05m~3/ha.(5)The relative diameter method and Weibull distribution density function were used to fit the diameter distribution of four kinds of afforestation density in the study area,and the fitting effect was very close.The best fitting effect of the two methods was that the stand density was3m×8m,R~2 reached 0.9983 and 0.9997 respectively.(6)Using the height and crown width data extracted from airborne LiDAR data,the competition indexes of different afforestation densities based on crown factors were calculated.Compared with the simple competition indexes of different afforestation densities calculated from the measured DBH data,it was found that the two results were close and the change trend was the same,both decreased with the decrease of afforestation density,and the competition indexes of four afforestation densities were less than 2,indicating that In this study area,Poplar was less disturbed by the competition of adjacent trees,while the competition index of sample plot with afforestation density of 6m×6m was the smallest,which indicated that the sample plot with afforestation density of 6m×6m had the advantage in the growth of DBH and tree height,and was least disturbed by the competition of adjacent trees.