UAV and Satellite-Based Sensing to Map Ecological States at the Landscape Scale
UAV and Satellite-Based Sensing to Map Ecological States at the Landscape Scale作者机构：School of Natural Resources & Environment University of Arizona Tucson Arizona USA Southwest Watershed Research Center USDA Agricultural Research Service Tucson Arizona USA Bio5 Institute Tucson Arizona USA
出 版 物：《Open Journal of Ecology》 (生态学期刊（英文）)
年 卷 期：2023年第13卷第8期
页 面：560-596页
主 题：Ecological Sites Ecological States Rangeland Largest Patch Index UAV Remote Sensing
摘 要：Mapping ecological states in semi-arid rangelands is crucial for effective land management and conservation efforts because it identifies difference in the ecological conditions across a landscape. This study presents an innovative approach for mapping two ecological states, Large Shrub Grass (LSG) and Large Shrub Eroded (LSE), within the Sandy Loam Upland and Deep (SLUD) ecological sites using a combination of drone and satellite data. The methodology leverages the Largest Patch Index (LPI) as a proxy metric to estimate eroded areas and classify ecological states. The integration of unmanned aerial vehicle (UAV) data with satellite-based remote sensing provides a scalable approach that can benefit various stakeholders involved in rangeland management. The study demonstrates the potential of this methodology by generating spatial layers at the landscape scale to inform on the state of rangeland ecosystems. The workflow showcases the power of remote sensing technology to map ecological states and addresses limitations in spatial coverage by integrating UAV and satellite data. By utilizing the bare ground LPI metric, which indicates the connectedness of bare ground, the methodology enables the classification of ecological states at a regional scale. This cost-effective approach potentially offers a standardized and reproducible method applicable across different sites and regions. The accuracy of the classification process is evaluated by comparing the results to ground-based polygons, dirt roads, and water locations. While the model performs well in identifying eroded areas, misclassifications occur in regions with mixed vegetation cover or low biomass. Future research should focus on incorporating temporal information from historical remote sensing archives to improve understanding of ecological state dynamics. Additionally, validation efforts can be enhanced by incorporating more ground-truth data and testing the methodology in diverse rangeland areas. The workflow
维普期刊数据库