Closing the Granularity Gap: A Multi-Sensor UAV Framework for Detecting Agronomic and Ecological Microstructures in Agricultural Landscapes

Modern precision agriculture increasingly demands the detection of small-scale field features—termed microstructures—that fall below the spatial resolution of conventional satellite-based remote sensing. These microstructures, whether biotic or abiotic, carry significant implications for farm operations, biodiversity conservation, and sustainable land management. This contribution presents a cumulative research framework comprising three complementary UAV-based case studies, each targeting a distinct sensor domain and detection challenge within Central European agricultural landscapes.

The first study investigates stone detection on arable land using UAV-mounted thermal cameras. Laboratory experiments and field trials demonstrate that surface stones exhibit significantly higher temperatures than surrounding soil during cooling phases, driven by differential thermal inertia—particularly under high soil moisture and declining air temperature—enabling reliable, non-invasive identification of stones that pose risks to agricultural machinery.

The second study employs UAV-based LiDAR to detect burrows of the European hamster (Cricetus cricetus), a critically endangered species whose habitat intersects with intensive farming. A geostatistical algorithm for identifying local elevation minima in high-density 3D point clouds achieved detection rates of up to 87%, offering a scalable alternative to manual surveys and supporting the reconciliation of species conservation with agricultural practice.

The third study addresses the detection of invasive plant species (including Heracleum mantegazzianum and Fallopia spp.) using multiscale, multi-sensor remote sensing from UAV and satellite platforms. By integrating phenological field observations with object-based classification algorithms, optimal detection windows and sensor configurations for mapping invasive neophytes were identified.

Collectively, the three studies illustrate how domain-specific UAV sensor deployments—thermal, LiDAR, and multispectral—can systematically address the resolution gap limiting conventional remote sensing at the sub-field scale. The framework demonstrates that transitioning from area-based to object-based monitoring is technically feasible and practically relevant for diverse agricultural and ecological management tasks.