Answers are grounded in this paper and cite their sources.
Background This study presents a hierarchical, multi‐sensor remote sensing framework that explicitly links fine‐scale unmanned aerial system (UAS) observations to regional‐scale satellite‐based heterogeneous vegetation cover (HVC) mapping in rangelands. Methods We integrate the UAS multispectral imagery, UAS‐derived light detection and ranging (LiDAR) structural metrics, and Sentinel‐1/2 satellite observations within a staged modeling design that evaluates how spectral–structural fusion and spatial upscaling influence classification performance. Four machine‐learning models were developed to isolate the contribution of UAS‐only, UAS + LiDAR, and satellite‐fusion strategies. Results Results show that UAS + LiDAR integration provides the highest discriminatory power (overall accuracy 95% with the XGBoost model), with canopy height alone contributing up to 32% of model importance and substantially improving shrub–woody separation. UAS‐only models perform well for spectrally distinct classes, but degrade in structurally complex vegetation. Satellite‐based models enable regional scalability but show lower accuracy (73%–82%) and boundary confusion due to mixed pixels and limited training density. Conclusions The primary contribution of this work is a transferable, sensor‐agnostic framework for scaling ecological information across platforms, where high‐resolution UAS models supply structured training data for satellite monitoring rather than acting as standalone products. This approach supports operational rangeland decision‐making by combining precision diagnostics with regional surveillance capacity.
No takes yet. Share an insight, caveat, or question.
Shermin et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: