Integrating UAV visible and multispectral imagery to assess grazing-induced vegetation responses in sandy grasslands

Qiang Guan¹Mingyang Jiang¹Wen Du^2*Xueyan Chen¹Baolong Yan¹

• ¹Inner Mongolia Minzu University, Tongliao, China
• ²Shenyang Agricultural University, Shenyang, China

Monitoring grazing intensity is crucial for maintaining ecological balance and promoting the sustainable management of sandy grasslands. Traditional ground surveys and single-source remote sensing often lack the spatial resolution, spectral richness, and robustness required to accurately characterize heterogeneous grazing impacts, especially under moderate disturbance. Unmanned aerial vehicle (UAV)-based multi-source remote sensing provides fine-scale, repeatable observations that can overcome the limitations of traditional field surveys. In this study, grazing experiments were conducted in the sandy grasslands of Inner Mongolia, China, using UAVs to capture visible and multispectral imagery across plots subjected to different grazing intensities. Spectral responses were analyzed using mean–variance statistics and Tukey's multiple comparison tests. To overcome the limited sensitivity of conventional vegetation indices and the redundancy problem commonly encountered in multi-index feature sets, we constructed a series of novel spectral indices based on separability analysis and integrated them with traditional vegetation indices. Furthermore, we developed an automatic incremental feature selection (AIFS) algorithm to adaptively optimize the feature subset and enhance model robustness. Results revealed distinct spectral responses to grazing disturbance: visible bands increased with grazing intensity due to enhanced soil background effects, whereas the red-edge and near-infrared bands effectively captured reductions in chlorophyll content and canopy structure under moderate to severe grazing. Traditional vegetation indices (e.g., NDVI, GNDVI) were sensitive to extreme grazing, while the proposed indices (idx7, idx9, idx14) exhibited superior performance in distinguishing moderate grazing levels. The optimized feature subset (idx9, idx8, idx7, NDVI, idx2, idx3, NDRE), identified through AIFS, significantly improved model accuracy while reducing redundancy. When combined with a support vector machine classifier, this approach achieved the highest recognition performance (OA = 92.13%, Kappa = 88.99%), outperforming models using all features or single-source data. These findings demonstrate that integrating UAV visible and multispectral imagery with intelligent feature selection enhances the detection of grazing-induced vegetation responses, providing a robust framework for high-precision grassland monitoring and sustainable ecological management in arid and semi-arid regions.