RESUMO
Forest aboveground biomass (AGB) and its biomass components are key indicators for assessing forest ecosystem health, productivity, and carbon stocks. Light Detection and Ranging (LiDAR) technology has great advantages in acquiring the vertical structure of forests and the spatial distribution characteristics of vegetation. In this study, the 56 features extracted from airborne LiDAR point cloud data were used to estimate forest total and component AGB. Variable importance-in-projection values calculated through a partial least squares regression algorithm were utilized for LiDAR-derived feature ranking and optimization. Both leave-one-out cross-validation (LOOCV) and cross-validation methods were applied for validation of the estimated results. The results showed that four cumulative height percentiles (AIH 30, AIH 40, AIH 20, and AIH 25), two height percentiles (H 8 and H 6), and four height-related variables (H mean, H sqrt, H mad, and H curt) are ranked more frequently in the top 10 sensitive features for total and component forest AGB retrievals. Best performance was acquired by random forest (RF) algorithm, with R 2 = 0.75, root mean square error (RMSE) = 22.93 Mg/ha, relative RMSE (rRMSE) = 25.30%, and mean absolute error (MAE) = 19.26 Mg/ha validated by the LOOCV method. For cross-validation results, R 2 is 0.67, RMSE is 24.56 Mg/ha, and rRMSE is 25.67%. The performance of support vector regression (SVR) for total AGB estimation is R 2 = 0.66, RMSE = 26.75 Mg/ha, rRMSE = 28.62%, and MAE = 22.00 Mg/ha using LOOCV validation and R 2 = 0.56, RMSE = 30.88 Mg/ha, and rRMSE = 31.41% by cross-validation. For the component AGB estimation, the accuracy from both RF and SVR algorithms was arranged as stem > bark > branch > leaf. The results confirmed the sensitivity of LiDAR-derived features to forest total and component AGBs. They also demonstrated the worse performance of these features for retrieval of leaf component AGB. RF outperformed SVR for both total and component AGB estimation, the validation difference from LOOCV and cross-validation is less than 5% for both total and component AGB estimated results.
RESUMO
Forest structure plays an important role in forest biomass inversion using synthetic aperture radar (SAR) backscatter. Synthetic aperture radar (SAR) sensors with long-wavelength have the potentiality to provide reliable and timely forest biomass inversion for their ability of deep penetration into the forest. L-band SAR backscatter shows useful for forest above-ground biomass (AGB) estimation. However, the way that forest structure mediating the biomass-backscatter affects the improvement of the related biomass estimation accuracy. In this paper, we have investigated L-band SAR backscatter sensitivity to forests with different mean canopy density, mean tree height and mean DBH (diameter at breast height) at the sub-compartment level. The forest species effects on their relationship were also considered in this study. The linear correlation coefficient R, non-linear correlation parameter, Maximal Information Coefficient (MIC), and the determination coefficient R2 from linear function, Logarithmic function and Quadratic function were used in this study to analyze forest structural properties effects on L-band SAR backscatter. The HV channel, which is more sensitive than HH to forest structure parameters, was chosen as the representative of SAR backscatter. 6037 sub-compartment were involved in the analysis. Canopy density showed a great influence on L-band backscatter than mean forest height and DBH. All of the R between canopy density and L-band backscatter were greater than 0.7 during the forest growth cycle. The sensitivity of L-band backscatter to mean forest height depends on forest canopy density. When canopy density was lower than 0.4, R values between mean forest height are smaller than 0.5. In contrast, the values of R were greater than 0.8 if canopy density was higher than 0.4. The sensitivity SAR backscatter to DBH fluctuated with canopy density, but it only showed obvious sensitivity when canopy density equals to 0.6, where both the linear and non-liner correlation values are higher than others. However, their effects on L-bang HV backscatter are affected by forest species, the effects on three forest structural parameters depend on tree species.