Monitoring Carbon Hotspots in Patchy Landscapes from Sparse Lidar and Dense Satellite Time Series
Year: 2026
Type: preprint
Abstract: Monitoring aboveground biomass (AGB) stocks as an indicator of carbon storage is essential for climate mitigation, adaptation, and land management. Across much of Earth’s land surface, carbon stocks are concentrated in fine-scale hotspots, including isolated trees in pastures, shrub patches in tundra, and riparian vegetation in drylands. These carbon hotspots are vulnerable to change but are difficult to monitor with remote sensing at broad scales due to three interconnected challenges: (1) spatial heterogeneity, (2) temporal change, and (3) modeling uncertainty. To address these challenges, we developed a workflow that integrates moderate-resolution satellite sensors, aerial lidar collections, and forestry-derived allometric models to produce annual maps of AGB in riparian corridors in the semi-arid western United States. To resolve spatial heterogeneity, we modeled canopy height and vegetation class at 10-m resolution, achieving canopy height mean absolute error of 1.56 m and overall vegetation classification accuracy of 84%. To measure uncertainty, we quantified prediction intervals using conformal prediction for canopy height and classification, ensemble disagreement for epistemic uncertainty, and Bayesian posterior distributions for allometric biomass models. The resulting annual maps provide pixel-level AGB estimates with statistically verified uncertainty bounds. These products reveal gradual changes in riparian biomass and can guide management decisions, from prioritizing restoration efforts to identifying early warning signs of ecosystem degradation. This framework could be adapted for monitoring AGB in other patchy landscapes, incorporated into different or changing land cover categories for future work, and updated as more and better data sources become available.
DOI: 10.31223/x56j7f
Institutions: Boise State University, Conservation International
Language: English
Subfield: Forestry
Domain: Life Sciences
SDGs: Life on land, Climate action
Open Access status: green
OpenAlex ID: W7222218373