Ground-based calibration for remote sensing of biomass in the tallest forests

Forest biomass is a critical component of the terrestrial carbon cycle. The highest-biomass forests are those dominated by the tallest species, Sequoia sempervirens. We use ground-based measurements and allometric equations to estimate tree biomass in primary Sequoia forests (40–42° N latitude) recently subjected to spaceborne and airborne laser scanning (GEDI and ALS, respectively), and we develop new biomass allometry using GEDI and ALS predictors. The best GEDI equation for tree (live + dead) aboveground biomass in these forests, which is based on the 88th percentile of relative height by pulse return energy (N = 200 pulses, R² = 0.37, RMSE = 48%), predicts average per-hectare values statistically indistinguishable from those predicted by a previously published GEDI equation (916 ± 74 vs. 928 ± 11 Mg ha⁻¹, mean ± 1 S_E). The best ALS equation, which is based on the height and crown size of tree approximate objects (dominant trees plus subordinates) segmented from lidar datasets (N = 503 segments, R² = 0.64, RMSE = 49%), predicts significantly higher live tree biomass than GEDI across 37465 ha of primary forest surveyed (1384 ± 77 vs. 885 ± 73 Mg ha⁻¹, mean ± 1 S_E). Underestimation by GEDI occurs because height alone is a poor predictor of biomass in the tallest forests. The new ALS equation also moderately underestimates biomass, in part because neither height nor crown size can adequately account for giant trunks. Despite these shortcomings, we demonstrate how a hierarchy of allometric equations can be used to map the distribution of biomass across forests with global maximum biomass and carbon density. Among primary forests of seven reserves, average estimated per-hectare biomass exceeds 2000 Mg ha⁻¹ in three, and ultrahigh-biomass (> 3000 Mg ha⁻¹) hectares are sparsely distributed (1%) with the largest concentration occurring on low-elevation alluvial terraces (460 ha) of Humboldt Redwoods State Park. The ALS-predicted biomass map provides realistic and context-specific benchmarks for ongoing restoration management of previously logged forests inside these reserves.