Satellite-Based Aboveground Biomass Estimation in Mountain Pastures of Armenia

<jats:p>Accurate estimation of aboveground biomass (AGB) is essential for sustainable pasture management, but remains challenging in heterogeneous mountain environments. This study evaluated a remote-sensing framework for estimating AGB across five landscape zones of the Aragats Volcanic Massif, Armenia, using PlanetScope imagery, terrain variables and machine learning. Biomass was measured in 30 plots comprising 90 nested quadrats during five field campaigns from April to July 2025. Predictor selection, algorithm comparison, preprocessing and tuning were performed within a fully nested leave-one-plot-out cross-validation framework. The model achieved an out-of-fold R2 of 0.596 (RMSE = 54.0 g m−2; MAE = 33.1 g m−2) at the 20 × 20 m plot level, decreasing to 0.503 when the same predictions were evaluated against individual quadrats, which isolates the effect of field-observation support. Variance partitioning showed that weak zone-level performance arose from two distinct causes: dominant within-plot variability in the Meadow Steppe, and a strong soil background under sparse early-season cover in the Dry Steppe. Fine-resolution UAV imagery, acquired without reflectance calibration, did not improve prediction in a matched cross-sensor comparison. Applied to 18 pastures covering 1822.5 ha, the model yielded an aggregate campaign-date stock of 1651.45 t (95% uncertainty interval 1207.70–2092.80 t).</jats:p>