Sub-hectare resolution forest biomass mapping from Copernicus DEM with low-dimensional models

Highlights

• Over forests, the Copernicus DEM typically represents average canopy elevation.
• DEM height above ground is correlated with biomass across 11 test sites (R² >= 0.49).
  50.5 t/ha (21 %) biomass estimation error at 0.25 ha scale with one-parameter model.
• 100 m resolution digital terrain model sufficient to remove topography from the DEM.
• Global application will be possible with upcoming near-global DTM from P-band SAR.

Abstract
Accurate, sub-hectare resolution mapping of above-ground biomass density (AGBD) of the global forests is needed for many applications, including carbon accounting, forest resource management, and biodiversity research. Surveys based on airborne lidar and in situ data fulfill the accuracy requirements but lack global applicability, while the globally available optical, lidar, and radar backscatter data suffer from poor sensitivity to AGBD and/or spatial resolution. We here assess the global applicability of an alternative approach that exploits the publicly available, interferometric radar-based Copernicus digital elevation model (DEM) paired with a coarse-resolution digital terrain model (DTM). For eleven boreal, temperate and tropical test sites on four continents, we show that the DEM-DTM difference represents the average canopy elevation above ground and is well-correlated with AGBD maps derived from airborne lidar and in situ data. Within each climate zone, the estimated power law model parameters varied moderately and indicated an almost linear relationship between AGBD and the DEM-DTM difference. Using site-specific power law models resulted in an ensemble root-mean-square difference (RMSD) at 50 m resolution of 43.7 t/ha (18 % of the average AGBD of 246 t/ha) and a coefficient of determination (R2) of 0.90. Climate zone-specific and a single, global scaling model performed almost equally well, with an overall RMSD of 20–21 % and an R2 value of 0.87. The global scaling model performed substantially better than the ESA Climate Change Initiative Biomass dataset on AGBD and indicated high potential for global application with the upcoming near-global DTM from ESA's BIOMASS mission (launched on 29 April 2025).

Keywords
Above-ground biomass density (AGBD) mapping; Interferometric synthetic aperture radar (InSAR); Digital elevation model (DEM); DTM; TanDEM-X; Scaling; model; Power law model.

• Sub-hectare resolution forest biomass mapping from Copernicus DEM with low-dimensional models
  Soja, M.J. , Santoro, Maurizio , Banda, F. , Tebaldini, S. , Lisiewicz, Maciej , Stereńczak, Krzysztof , Quegan, Shaun , Janssen, S.J.C. , Reiche, J. (2025) Science of Remote Sensing (2025), Volume: 12 - ISSN 2666-0172 - p. 100250-100250.