Deadwood and necromass carbon pool estimation in forest projects
Deadwood is a mandatory IPCC carbon pool that forest projects routinely omit or guess. Spaceborne lidar return density and waveform shape offer a partial but auditable proxy, with honest caveats about what closed canopy hides.
Sensors
- GEDI (Global Ecosystem Dynamics Investigation): Full-waveform lidar mounted on the ISS. 25-metre circular footprints, ~60-metre along-track spacing, coverage between 51.6° N and S. Waveform shape and return energy distribution proxy canopy gap fraction and sub-canopy surface roughness, both correlates of coarse woody debris accumulation. Cannot resolve individual logs; detects aggregate structural signal only.
- ICESat-2 ATL08: Photon-counting lidar at 532 nm. 11-metre footprints along 91-day repeat tracks. The ATL08 canopy-height product separates ground and canopy photon returns; elevated ground returns in gap zones can indicate debris piles, though the signal is ambiguous without field calibration.
- Airborne discrete-return lidar (ALS): Point densities of 4 to 25 returns per square metre are achievable at survey altitudes below 2,000 metres. At these densities, individual logs above roughly 20 cm diameter can be detected and volume estimated from point-cloud cross-sections. This is the only remote-sensing method with published detection of individual coarse woody debris pieces; it serves as ground truth for spaceborne products.
- Sentinel-2 MSI: 10-metre multispectral imagery at up to 5-day revisit. Canopy gap fraction derived from red-edge and near-infrared bands correlates weakly with debris accumulation in open-canopy systems. Useful as a spatial stratification layer to direct lidar sampling, not as a standalone deadwood estimator.
Why deadwood is the pool that keeps disappearing from project accounts
The IPCC 2006 Guidelines for National Greenhouse Gas Inventories define five forest carbon pools: above-ground biomass, below-ground biomass, litter, soil organic carbon, and dead organic matter. Dead organic matter itself splits into deadwood and litter. Project-level methodologies under Verra's VM0015 and similar standards require all five pools to be reported or formally excluded with a conservative default penalty. In practice, deadwood is the pool most frequently excluded, because field measurement is genuinely difficult: logs are spatially scattered, decay classes span a wide density range (roughly 0.08 to 0.6 g/cm³ for boreal species, depending on decay stage), and systematic plot-based surveys are expensive.
The consequence is not trivial. Published studies in boreal and temperate forests have found deadwood pools equivalent to 10 to 30 percent of above-ground live biomass, particularly in old-growth and post-disturbance stands. Omitting this pool and claiming a conservative default does not always penalise the project developer enough to reflect reality, especially where stands have a history of natural disturbance or selective logging.
What a waveform gives away about the forest floor
GEDI fires laser pulses and records the full time-series of returned energy, not just a single range measurement. In a closed-canopy stand, the waveform shows a canopy peak followed by a ground return. Where the canopy is gappy, energy reaches the forest floor and the ground return broadens. Coarse woody debris sitting on the ground raises the apparent surface, scatters energy laterally, and adds low-amplitude secondary returns between the main canopy and ground peaks. Researchers at the University of Maryland and collaborators have used GEDI Level 2A and 2B products to extract metrics such as plant area index, canopy cover, and foliage height diversity, and have explored correlations between waveform complexity and surface roughness as a proxy for debris accumulation.
The honest limit: GEDI's 25-metre footprint integrates everything within that circle. A single large log and a cluster of small branches produce similar waveform perturbations. Without field calibration plots, the waveform signal cannot be decomposed into debris volume with any reliability. Published work treats this as a stratification and scaling tool, not a direct measurement. GEDI also has no coverage above 51.6° latitude, which excludes large portions of the boreal zone where deadwood pools are largest.
The closed-canopy problem no sensor has solved
Coarse woody debris accumulates most heavily on the forest floor beneath intact canopy, precisely where spaceborne optical sensors are blind and where even lidar struggles. At GEDI's footprint size, a dense canopy intercepts most of the outgoing pulse before it reaches the ground. The fraction of energy that penetrates depends on canopy cover and leaf area index. Published penetration rates in tropical closed-canopy forest can fall below 20 percent of pulse energy reaching the ground, which means the ground return is noisy and the debris signal within it is largely undetectable.
Airborne lidar at high pulse density partially solves this. At 10 or more returns per square metre, enough photons penetrate canopy gaps to reconstruct the ground surface with sufficient detail to detect large logs. Studies in Pacific Northwest old-growth and Scandinavian boreal stands have demonstrated individual log detection at diameters above roughly 15 to 20 cm using point-cloud cross-section methods. The trade-off is cost and coverage: airborne surveys suitable for this purpose run over specific project areas, not continental extents, and must be repeated to detect change.
Combining sensors to build a defensible estimate
No single sensor resolves the deadwood pool cleanly. The practical approach, reflected in published remote-sensing literature, is a stratified sampling design. Sentinel-2 gap-fraction maps stratify the project area into canopy-density classes. GEDI waveform metrics provide a continuous spatial signal correlated with surface roughness across the strata. Airborne lidar surveys, flown over a statistically designed sample of plots, provide direct debris-volume estimates that calibrate the GEDI signal. Field plots ground-truth the airborne data. The result is an area-weighted estimate with quantified uncertainty, which is what a carbon registry auditor actually needs.
ICESat-2 ATL08 adds value in open-canopy systems, particularly boreal stands with lower canopy closure, where photon returns from the ground surface are less attenuated. Its 91-day repeat cycle also allows change detection over time, which matters for monitoring decay-class transitions and debris removal after salvage logging. The two lidar missions are complementary rather than redundant: GEDI's full waveform is richer per footprint; ICESat-2's denser track network gives better spatial coverage in some regions.
What an audit package needs to include
A registry-grade deadwood estimate requires four things: a documented sampling design with stratification logic, a clear uncertainty budget at the 90-percent confidence level, a description of which pool fractions are included (standing dead trees, fallen logs, stumps, fine woody debris) and which are excluded with justification, and a change-detection protocol for monitoring periods. Remote sensing contributes most to the first two. It cannot replace field measurement of decay class and wood density, which remain the largest sources of uncertainty in converting volume to carbon mass.
Satellize structures deadwood analytics as a calibrated spatial layer: GEDI and Sentinel-2 inputs produce a project-wide surface-roughness and gap-fraction stratification map, airborne lidar data (client-supplied or procured) are processed to debris-volume estimates per stratum, and the combined output is delivered as a GIS layer with per-pixel uncertainty bounds and a methodology annex written to Verra or Gold Standard audit conventions. The approach mirrors the workflow developed for the Tonga crop-estimation programme in one respect: the analytic output is only as credible as the field data used to calibrate it, and we say so in the report.
Honest limits, stated plainly
Spaceborne lidar cannot detect fine woody debris (pieces below roughly 7 cm diameter) under any conditions. It cannot reliably detect coarse woody debris under closed canopy at GEDI's footprint size. ICESat-2 track spacing (roughly 91-kilometre cross-track at the equator) means large areas of any project receive no direct measurement. Cloud cover does not affect lidar but does affect the Sentinel-2 gap-fraction inputs, which can be gap-filled using temporal compositing at the cost of some precision.
Decay class, the variable that converts volume to carbon, is not observable from any current spaceborne system. It requires either field measurement or airborne hyperspectral data, neither of which is cheap. Projects in early decay stages (fresh deadfall) carry far less carbon per unit volume than advanced-decay material, and conflating the two introduces systematic bias. Any analytics product that does not propagate this uncertainty honestly is not fit for a third-party audit.
Typical figures
| GEDI footprint diameter | 25 m (circular) |
| GEDI along-track spacing | ~60 m; cross-track beam separation ~600 m |
| GEDI latitudinal coverage | 51.6° N to 51.6° S |
| ICESat-2 footprint diameter | ~11 m |
| ICESat-2 repeat cycle | 91 days |
| ICESat-2 cross-track spacing (equator) | ~91 km (dense near poles) |
| Airborne lidar point density (survey-grade) | 4 to 25 returns/m² depending on altitude and pulse rate |
| Minimum detectable log diameter (airborne lidar) | ~15 to 20 cm at ≥10 returns/m²; not detectable spaceborne |
| Sentinel-2 gap-fraction resolution | 10 m (visible/NIR bands), 20 m (red-edge) |
| Sentinel-2 revisit | Up to 5 days (two-satellite constellation) |
Analytics Satellize can run
| Canopy gap-fraction stratification map | Sentinel-2 red-edge and NIR band ratio compositing; temporal gap-fill for cloud | GeoTIFF layer at 10 m resolution, delivered per project boundary, updated annually or on demand |
| GEDI surface-roughness index | GEDI Level 2A waveform metric extraction (RH metrics, ground return width); spatial interpolation across project area | Gridded raster at 100 m resolution with per-cell coefficient of variation; GIS layer with methodology annex |
| Stratified debris-volume estimate with uncertainty | Calibrated scaling of airborne lidar point-cloud cross-section volumes against GEDI strata; Monte Carlo uncertainty propagation | Tabular estimate by stratum (m³/ha ± 90% CI), formatted to Verra VM0015 or equivalent reporting table |
| Change-detection layer for debris dynamics | ICESat-2 ATL08 multi-epoch ground-return comparison in open-canopy strata; flagging of significant surface-height change | Per-monitoring-period change report with flagged areas as GIS polygon layer |
| Audit-ready methodology annex | Documentation of sampling design, sensor specifications, calibration data, uncertainty budget and exclusion justifications | PDF report written to third-party audit conventions, citing public sensor documentation and published calibration studies |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.