Study On The Volume Estimation Model Of Cunninghamia Lanceolata Plantation Based On Airborne LiDAR Data

Forest volume is one of the basic indicators for measuring forest resource management,and it is also an important aspect that reflects the status of forest resources and the carbon sequestration capacity of forests.Light Detection And Ranging(Li DAR)technology has advantages such as practicality,high accuracy,and strong penetration.Combining airborne Li DAR point cloud data with measured data from sample plots can effectively improve the accuracy of stand volume estimation.The paper focuses on 33 sample plots of Cunninghamia lanceolata plantations located in Yangjiao Mountain Forest Farm,Jinji Forest Farm,and Lianshan Forest Farm in Qingyuan City,Guangdong Province.Based on Airborne Li DAR point cloud data collected from three forest farms in 2021 and field measured data,characteristic variables of different scales of individual trees and stands were extracted,Pearson correlation coefficient method and variance expansion factor(VIF)are used to screen characteristic variables,compare the accuracy of multiple linear regression and nonlinear regression,as well as the estimation models of stand volume at the individual tree level and stand level;Analyze the accuracy of forest stand volume estimation models for different age groups and forest farms.Due to the close correlation between stand height cross-sectional area,stand average height and stand volume,a simultaneous equation system estimation model for stand volume was developed to explore the optimal fitting method of the simultaneous equation system model;Using the optimal fitting method of the simultaneous equation system model,establish a simultaneous equation system model for forest stand height cross-sectional area,stand average height,and stand volume of different age groups and forest farms.The main conclusions are as follows:(1)Comparative analysis of the impact of multiple linear estimation models and nonlinear estimation models on the accuracy of estimating stand volume of Cunninghamia lanceolata plantation.The results shows that the multiple nonlinear estimation model has better estimation accuracy than the linear estimation model.The determination coefficient R~2of the multivariate nonlinear estimation model is 0.718,the F-value is 12.40,and the RMSE is 4.27 m~3/hm~2.The determination coefficient R~2of the multivariate linear estimation model is 0.682,the F-value is11.44,and the RMSE is 4.44 m~3/hm~2.(2)The effects of the airborne Li DAR arithmetic average height and airborne Li DAR dominant tree average height factor on the estimation accuracy of the stand volume of Cunninghamia lanceolata plantation are compared.The results shows that adding the factor of airborne Li DAR arithmetic average height could improve the estimation accuracy of the stand volume of Cunninghamia lanceolata plantation.The determination coefficient R~2of the model is 0.746,the F value is 14.80,and the RMSE is 3.98 m~3/hm~2.The decision coefficient R~2of the airborne Li DAR dominant tree average height factor model is 0.720,the F-value is 12.81,and the RMSE is 4.22 m~3/hm~2.(3)Analyze the effectiveness of forest volume estimation models developed based on single tree and stand levels.The results indicate that the forest volume estimation model based on stand level has better estimation effect.The determination coefficient R~2of the model is0.746,the F-value is 14.80,and the RMSE of the test data is 3.98 m~3/hm~2.The R~2of the correlation between the measured value of stand volume based on stand level and model estimation is 0.760,and the R~2of the correlation between the measured value of stand volume based on single tree level and the model estimation value is 0.439.(4)The paper developed a stand volume estimation model for different age groups and forest farms,and the results shows that both age group and forest farm modeling can improve the accuracy of the stand volume estimation model.The determination coefficient R~2of the stand volume estimation model based on the middle and young age forest group(young forest,middle age forest)is 0.899.The determination coefficient R~2of the stand volume estimation model based on the near mature forest group(near mature forest,mature forest)is 0.596.The determination coefficient R~2of the stand volume estimation model based on non grouped sample plots is 0.746.The determination coefficient R~2of the stand volume estimation model based on Yangjiao Mountain Forest Farm is 0.863.The determination coefficient R~2of the stand volume estimation model based on Jinji Forest Farm is 0.790.The determination coefficient R~2of the stand volume estimation model based on Lianshan Forest Farm is 0.753.The R~2of the correlation between the measured values of the stand volume estimation model based on different age groups and different forest farms and the model estimation values is 0.783 and0.801,respectively.The R~2of the correlation between the measured values and the estimated values of the stand volume estimation model based on non grouped sample plots is 0.760.Overall,the estimation models for stand volume developed by age groups and forest farms are better than those for non grouped sample plots.(5)A simultaneous equation set estimation model of stand height cross-sectional area,stand average height and stand volume is established.The results show that the simultaneous equation set estimation model based on ordinary least squares(OLS)fitting has the highest precision and the best fitting effect.The determination coefficient R~2is 0.794,the F-value is37.81,the RMSE is 2.74 m~3/hm~2.Comparing the simultaneous equation system models based on different age groups and different forest farms,the results show that the simultaneous equation system model improved the estimation accuracy of forest volume.Among them,the simultaneous equation system model of forest stand volume in the forest farm has the highest estimation accuracy,and the linear correlation R~2between the measured and estimated forest stand volume is 0.853;the estimation accuracy of the simultaneous equation system model for stand volume in age groups is second,and the linear correlation R~2between the measured and estimated values of stand volume is 0.835;the estimation accuracy of the stand volume simultaneous equation system model based on non grouped sample plots is the lowest,and the linear correlation R~2between the measured and estimated values of stand volume is 0.781.