What lies beneath: National aboveground carbon stock changes and under-canopy disturbance detection from 3D point clouds

Accurate and spatially explicit quantification of forest biomass and carbon stock changes over time is necessary for carbon accounting and verification at different scales. Here, we applied deep learning to full 3D point-cloud data to produce a high-resolution time series of forest attribute maps for Denmark. The maps include predictions for biomass and canopy height with estimated uncertainties, spanning a decade from 2014 to 2024. By comparing to forest inventory estimates, we demonstrate that the maps can be used for national estimates of aboveground tree carbon stocks and stock changes. We benchmarked our maps against existing global and regional maps at varying scales and find that our maps are able to capture landscape-scale variation in both managed and unmanaged forests while retaining plot-level accuracy. Existing global and regional biomass mapping products often reduce forest structure to canopy-height metrics when estimating biomass, yet reliability of these proxies over time is largely untested. Given that most forests globally are managed, our analysis suggests biomass loss over time with steady or increasing canopy height is widespread, demonstrating that height change alone cannot reliably represent changes in forest carbon stocks.