Uncertainty Analysis of Remote Sensing Estimation of Chinese Fir (Cunninghamia lanceolata) Aboveground Biomass in Southern China DOI Open Access
Yujie Hu, Liyong Fu, Bingsheng Qiu

et al.

Forests, Journal Year: 2025, Volume and Issue: 16(2), P. 230 - 230

Published: Jan. 25, 2025

Forest aboveground biomass (AGB) is not only the basis for forest carbon stock research, but also an important parameter assessing cycle and ecological functions of forests. However, there are various uncertainties in estimation process, limiting accuracy AGB estimation. Therefore, we extracted spectral features, vegetation indices texture factors from remote sensing images based on field data Landsat 8 OLI Southern China to quantify uncertainties. Then, established three models, including K Nearest Neighbor Regression (KNN), Gradient Boosted Tree (GBRT) Random (RF). Uncertainties at plot scale models were measured by using error equations analyze influences different scales Results as follows: (1) The R2 per-tree model Cunninghamia lanceolata was 0.970, while uncertainty residual parameters 4.62% 4.81%, respectively; transferred 3.23%. (2) methods had most significant effects models. RF more accurate than other two methods, highest (R2 = 0.867, RMSE 19.325 t/ha) lowest (5.93%), which outperformed both KNN GBRT (KNN: 0.368, 42.314 t/ha, 14.88%; GBRT: 0.636, 32.056 6.3%). Compared GBRT, enhanced 0.499 0.231, decreased 8.95% 0.37%, respectively. associated with remains primary source when compared scale. On scale, best effect. This study examines impact plot-scale model-scale methodologies lanceolata. findings aim offer valuable insights considerations enhancing estimations.

Language: Английский

Uncertainty Analysis of Remote Sensing Estimation of Chinese Fir (Cunninghamia lanceolata) Aboveground Biomass in Southern China DOI Open Access
Yujie Hu, Liyong Fu, Bingsheng Qiu

et al.

Forests, Journal Year: 2025, Volume and Issue: 16(2), P. 230 - 230

Published: Jan. 25, 2025

Forest aboveground biomass (AGB) is not only the basis for forest carbon stock research, but also an important parameter assessing cycle and ecological functions of forests. However, there are various uncertainties in estimation process, limiting accuracy AGB estimation. Therefore, we extracted spectral features, vegetation indices texture factors from remote sensing images based on field data Landsat 8 OLI Southern China to quantify uncertainties. Then, established three models, including K Nearest Neighbor Regression (KNN), Gradient Boosted Tree (GBRT) Random (RF). Uncertainties at plot scale models were measured by using error equations analyze influences different scales Results as follows: (1) The R2 per-tree model Cunninghamia lanceolata was 0.970, while uncertainty residual parameters 4.62% 4.81%, respectively; transferred 3.23%. (2) methods had most significant effects models. RF more accurate than other two methods, highest (R2 = 0.867, RMSE 19.325 t/ha) lowest (5.93%), which outperformed both KNN GBRT (KNN: 0.368, 42.314 t/ha, 14.88%; GBRT: 0.636, 32.056 6.3%). Compared GBRT, enhanced 0.499 0.231, decreased 8.95% 0.37%, respectively. associated with remains primary source when compared scale. On scale, best effect. This study examines impact plot-scale model-scale methodologies lanceolata. findings aim offer valuable insights considerations enhancing estimations.

Language: Английский

Citations

0