Sort:
Issue
Estimation of forest carbon stocks in Liaoning Province based on a multi-scale geographically weighted regression model
Journal of Central South University of Forestry & Technology 2025, 45(9): 130-140
Published: 25 September 2025
Abstract PDF (5.3 MB) Collect
Downloads:4
【Objective】

Based on the multi-scale geographically weighted regression (MGWR) model, combined with remote sensing data, the above ground carbon (AGC) of forests in Liaoning Province was estimated, and the impacts of spatial heterogeneity on the estimation of AGC were assessed through the selection of different characteristic variables.

【Method】

Using 1 068 forest inventory sample plots, digital elevation model (DEM) data and Landsat 8 OLI images in Liaoning Province in 2015 as data sources, 63 feature variables (2 topographic features, 6 band features, 7 vegetation indices, 48 texture features) were extracted and selected the feature variables using stepwise regression and random forest (RF) algorithms, and construct ordinary least squares (OLS), geographically weighted regression (GWR), MGWR and random forest regression (RFR), MGWR model and Random Forest Regression (RFR) model to estimate AGC, compare the differences between the prediction accuracy of different models, and realize the accurate inversion of AGC in the study area.

【Result】

1) The prediction accuracies of each model under the stepwise regression method of selection feature variables are higher than those of the RF method of selecting feature variables, and the MGWR model has the highest accuracy (R2=0.74, RMSE=11.66 t/hm2); the GWR model is the next highest (R2=0.72, RMSE=12.30 t/hm2); and the RFR model is the next highest (R2=0.37, RMSE=16.32 t/hm2). The OLS model is the lowest (R2=0.26, RMSE=16.41 t/hm2); 2) In the stepwise regression of the feature selection method, the MGWR model improves the R2 by 0.48, 0.02, and 0.37, respectively; and decreases the RMSE by 4.75 t/hm2, 0.64 t/hm2, and 4.66 t/hm2, respectively. The R2 of the MGWR model is improved by 0.46, 0.05, and 0.33 compared with those of OLS, GWR, and RFR, respectively, while the RMSE is reduced by 6.95 t/hm2, 1.23 t/hm2, and 5.36 t/hm2, respectively, in the selection of feature variables by the RF method; 3) The prediction of aboveground carbon stock in forests of the study area ranges from 37.25 t/hm2 to 190.22 t/hm2, totaling 135.165 Tg. Spatially, the high value of forest aboveground carbon stock is mainly distributed in the west and a small area in the east, while most of the eastern area decreases, and the central plains area is the lowest.

【Conclusion】

Stepwise regression selection of feature variables to construct the MGWR model effectively solves the problem of spatial heterogeneity through flexible scale (bandwidth) settings, reveals the multi-scale spatial distribution characteristics of AGC in Liaoning Province, which is applicable to the accurate estimation of large-scale AGC, and provides a scientific basis for the enhancement of regional carbon management and ecosystem services.

Issue
Stand spatial structure characterization of broad-leaved Pinus koraiensis forest based on multi-source LiDAR data
Journal of Central South University of Forestry & Technology 2023, 43(8): 36-50
Published: 25 August 2023
Abstract PDF (9.6 MB) Collect
Downloads:9
Objective

Taking Pinus koraiensis forest as an example and the fusion of LiDAR data as the key, the non-destructive multivariate distribution of stand spatial structure parameters was explored in this study, which provided suggestions for stand spatial optimization and adjustment for Pinus koraiensis forest and promoting forest succession.

Method

Taking Unmanned Aerial Vehicle LiDAR (UAV-LiDAR), Terrestrial Laser Scanning (TLS) data and field inventory data of a 100 m×100 m (1 hm2) sample plot of Pinus koraiensis forest in Maoershan Experimental Forest Farm of Northeast Forestry University as the data source, the UAV-LiDAR and TLS data were fused, the segmentation of individual tree crown was performed, and the individual tree parameters (tree position, DBH, crown width, the angle and distance between trees, etc.) were extracted. The stand spatial structure parameters of the sample plot (including uniform angle index, neighborhood comparison, mingling degree and crowding degree) and the frequency distribution were calculated and counted. The multivariate distribution maps of stand spatial structure parameters were established by Origin 2018.

Result

The accuracy of individual tree segmentation based on the fusion of UAV-LiDAR and TLS data was F=0.70 in the stand with a tree density of 1 269 trees /hm2. The spatial structure of the broad-leaved red pine forest was characterized by a random distribution state, moderate state, highly dense state and medium and above mixed state of most trees in Pinus koraiensis forest, and the tree size was uniform. The competition among trees was fierce, and the crown of the target tree was mostly connected with the crown of three to four neighboring trees. There were unreasonable distribution combinations in Pinus koraiensis forest, and the frequency values were 0.40%, 0.75% and 3.50%, respectively.

Conclusion

Compared with the traditional method, the stand spatial structure parameters could be calculated and evaluated with the non-destructive method based on the individual tree parameters extracted from the fused LiDAR data. The multivariate distribution of stand spatial structure parameters can show the characteristics of stand internal spatial structure, and the accurate irrational structure combination can be obtained to adjust and optimize the stand structure.

Issue
Estimation of individual tree parameters of larch plantations based on UAV-LiDAR and error-in-variable regression
Journal of Central South University of Forestry & Technology 2023, 43(7): 52-64
Published: 25 July 2023
Abstract PDF (4.6 MB) Collect
Downloads:11
Objective

Taking the artificial larch plantation as an example, the influence of the error and extraction of tree height based on the individual tree detection of unmanned aerial vehicle lidar (UAV-LiDAR) on the inversion of diameter at breast height (DBH) was explored and calibrated, so as to realize the adjustment and the accurate measurement of individual tree parameters (DBH and tree height), which provided a new idea for the large-scale efficient and convenient estimation of individual tree parameters in the future.

Method

Based on the UAV-LiDAR data of 13 larch plantation plots of 4 age groups (young forest, middle-aged forest, near mature forest and mature forest) in Maoershan experimental forest farm of Northeast Forestry University, the tree height was extracted based on the individual tree detection of UAV-LiDAR, and ordinary least squares (OLS) and three error variable regression (Standard major axis (SMA), Ranged major axis (RMA) and Maximum likelihood estimate (MLE) were used to study the influence of errors on the inversion of DBH of artificial larch plantation of each age group and its calibration.

Result

The relative root mean square error (rRMSE) of tree height extracted by individual tree detection using UAV-LiDAR data ranged from 3.41% to 5.14%. In terms of DBH-tree height model prediction results, the three error variable regressions were better than OLS, and RMA had the best prediction effect. The rRMSE of the inversion of individual tree DBH decreased by 2.21% to 3.58% in the 4 age groups.

Conclusion

When the model assumptions are satisfied, the error variable regression performs better than OLS in predicting the response variable, which is also an ideal method for estimating unbiased model coefficients. In this study, the RMA method performs the best. The inversion of the DBH model has high prediction accuracy, and all errors are kept within a reasonable range for larch plantations, which can achieve the purpose of using UAV-LiDAR for the efficient and convenient estimation of large-scale forest individual tree parameters, and it can be promoted in practice.

Total 3