关键词: 草地, 无人机, 植物多样性, 调查方式, 物种识别

Abstract

Aims: The biodiversity of grassland is essential for maintenance of the ecosystems’ services. Unmanned aerial vehicle (UAV) remote sensing has attracted much attention in the survey and research of grassland species in recent years with its advantages of flexibility, high resolution, high efficiency, and low cost. However, current research on UAV surveys of grassland plant diversity are still insufficient. This study aims to explore the impacts of expanding survey area on the results of grassland plant diversity survey, and optimal aerial photography parameters and efficiency for UAV survey of grassland plant diversity, so as to analyze the application potential of UAV survey on grassland plant diversity.  

Methods: UAV aerial photography and artificial ground survey were used to investigate plant diversity in the Hulunbuir meadow. By comparing the number of species observed and the survey time under different layout methods of quadrat and different shooting methods, the influence of expanding the survey area on the survey results was analyzed. The best aerial photography parameters and the time efficiency of using UAV to carry out grassland plant diversity survey were analyzed as well.

Results: (1) In the survey of grassland plant diversity, the number of species increased as the survey area increased within a certain range. Compared with the artificial ground survey of limited quadrats, more species could be observed using UAV in a large area survey. (2) In the survey of plant diversity in a semi-arid meadow grassland, the spatial resolution of UAV RGB image used for species identification should be better than 0.50 mm, and approximately 67.16% of species could be identified at a resolution of 0.45 mm. In terms of observation angle, it was better to shoot in two ways, which could increase 45° or 60° oblique shooting on the basis of 90° vertical shooting. (3) UAV survey could significantly shorten the time spent in species survey, but high-precision intelligent species identification model should be used to improve the survey efficiency.

Conclusion: By comparing the number of species observed by UAV and traditional artificial ground survey, this study further verifies the feasibility of UAV remote sensing for grassland plant diversity survey, and explores the key parameters of using UAV to obtain visible light images for grassland species survey. The results promote further implementation and prioritization of UAV in grassland monitoring, investigation, and protection.

Key words: grassland, UAV, plant diversity, method of survey, species recognition