Research

A systematic evaluation of multi-resolution ICESat-2 ATL08 terrain and canopy heights in boreal forests

Across North American boreal forests, ICESat-2 ATL08 canopy heights agreed with airborne LVIS lidar to within about 4 m RMSE at both resolutions, providing a benchmark for selecting high-quality spaceborne height data.

Sub-meter tree height mapping of California using aerial images and LiDAR-informed U-Net model

A U-Net model mapped tree canopy height across all of California at 0.6 m resolution with a mean error of 2.9 m and no saturation up to 50 m, outperforming existing global canopy height products.

Deep point cloud regression for above-ground forest biomass estimation from airborne LiDAR

Deep learning operating directly on full LiDAR point clouds estimated above-ground forest biomass significantly more accurately than state-of-the-art statistical approaches, and did so without requiring a digital terrain model.

Mapping temperate forest tree species using dense Sentinel-2 time series

Dense Sentinel-2 time series mapped 17 temperate tree species, with accuracy for the nine main species ranging from 66.8 to 98.9 percent.

Nearly ninety percent of Indonesian coffee area is classified as deforestation-free under the European Union Deforestation Regulation (EUDR)

87.9 percent of Indonesia's coffee area in 2024 meets the EUDR deforestation-free criterion, while total coffee area grew by roughly 15 percent from 2016 to 2024.

Accelerated global glacier mass loss in the early twenty-first century

The world's glaciers lost 267 gigatonnes of ice per year from 2000 to 2019, accounting for 21 percent of observed sea-level rise, with loss accelerating by 48 gigatonnes per year each decade.