Dense mountain forests are important carbon reservoirs and biodiversity habitats. However, complex terrain and canopy occlusion make it difficult for conventional optical remote sensing to accurately capture vertical forest structure, creating considerable uncertainty in aboveground biomass estimates under challenging environmental conditions.
To address this problem, a team led by researcher Wenjian Ni from the State Key Laboratory of Remote Sensing and Digital Earth has made new advances in probing the aboveground biomass of dense mountain forests using the spaceborne, large-footprint Global Ecosystem Dynamics Investigation lidar, or GEDI.
The team quantified the impact of terrain on national-scale forest biomass estimates and independently developed a waveform-adaptive algorithm for identifying signal cutoff points in dense forests.
Regional Forest Aboveground Biomass Estimation Using GEDI-SAWA
Terrain effects are a major challenge when using spaceborne lidar to probe mountainous forests. Although extensive research has examined these effects at the footprint scale, their contribution to uncertainty in national-scale aboveground biomass estimates has remained poorly understood.
In earlier research, the team independently developed the Slope-Adaptive Waveform Area, or SAWA, metric for large-footprint lidar waveforms. After validating its applicability to GEDI data, the researchers developed the GEDI-SAWA product.
They subsequently established a GEDI-SAWA model for estimating forest aboveground biomass and systematically assessed the effects of terrain on regional biomass estimates.
The results showed that terrain introduced approximately 1.11–3.93 petagrams of uncertainty into aboveground biomass estimates across the contiguous United States, or CONUS. In China, the estimated uncertainty ranged from approximately 2.01 to 10.14 petagrams.
The findings were published in Remote Sensing of Environment under the title “Mitigating Terrain Effects of GEDI Data on Regional Accounting of Aboveground Biomass Utilizing Slope-Adaptive Waveform Metrics.”
Figure 1. Changes in bias in GEDI-SAWA aboveground biomass estimates across different terrain slope classes. AGBD L4A represents the official GEDI product; AGBDori and AGBDadj represent SAWA estimates produced using different signal cutoff points; and AGBDREF represents the reference data. AGBD denotes aboveground biomass density.
Waveform-Adaptive Ground Detection Method
Accurately estimating aboveground biomass in dense tropical forests is essential to global change research. However, insufficient laser penetration can cause biomass to be significantly underestimated.
Based on new insights into the penetration characteristics of lidar return waveforms, the research team moved beyond conventional ground-detection methods that rely on fixed thresholds and developed a waveform-adaptive ground detection method, or WAED, for weak-signal conditions.
The team selected the dense Amazon rainforest in Brazil as a representative study area. Validation results showed that WAED significantly outperformed the official GEDI product overall.
For canopy-height retrieval, the method reduced mean bias from −3.94 meters to 0.23 meters and lowered the root mean square error from 7.58 meters to 4.29 meters.
WAED also demonstrated strong stability and generalization across different waveform sensitivities and laser beam types, effectively addressing the underestimation of forest carbon stocks by spaceborne, large-footprint lidar.
The findings were published in Remote Sensing of Environment under the title “Waveform-Adaptive Estimation of Canopy Heights Using GEDI Data in Dense Amazonian Forests.”
Figure 2. WAED ground-detection results for representative waveforms with different penetration levels. Panels (a–d) show airborne laser scanning point clouds, candidate ground elevations, waveform characteristics and ground-detection results for strongly penetrating waveforms; panels (e–h) show the corresponding results for moderately penetrating waveforms; and panels (i–l) show the results for weakly penetrating waveforms.
Paper Links:
- Mitigating terrain effects of GEDI data on regional accounting of aboveground biomass utilizing slope-adaptive waveform metrics
- Waveform-adaptive estimation of canopy heights using GEDI data in dense Amazonian forests
The above content was provided by Wenjian Ni of the State Key Laboratory of Remote Sensing and Digital Earth.
These advances also demonstrate how satellite data and advanced analytics can improve forest carbon assessment and environmental monitoring in challenging terrain. Explore STARPATH GLOBAL’s environmental monitoring solutions, or apply for our limited-availability, complimentary FDE initiative, the Pioneer Partner Program, to develop a solution tailored to your project.
