周心卉,邹志刚,唐余,等.基于无人机和卫星影像的光伏电站植被协同管理研究[J].中国环境管理,2024,16(4):118-130.
ZHOU Xinhui,ZOU Zhigang,TANG Yu,et al.Collaborative Vegetation Management of Photovoltaic Power Plants Based on UAV and Satellite Imagery[J].Chinese Journal of Environmental Management,2024,16(4):118-130.
基于无人机和卫星影像的光伏电站植被协同管理研究
Collaborative Vegetation Management of Photovoltaic Power Plants Based on UAV and Satellite Imagery
DOI:10.16868/j.cnki.1674-6252.2024.04.118
中文关键词:  农光互补电站  生态修复  环境监测  光伏+  植被指数
英文关键词:agrophotovoltaic power plant  ecological restoration  environmental management  photovoltaic+  vegetation index
基金项目:国家自然科学基金项目(72204220)。
作者单位E-mail
周心卉 浙江大学环境与资源学院, 浙江杭州 310058  
邹志刚 浙江大学环境与资源学院, 浙江杭州 310058  
唐余 浙江大学环境与资源学院, 浙江杭州 310058  
丁倩 浙江大学环境与资源学院, 浙江杭州 310058  
魏宇帆 浙江大学环境与资源学院, 浙江杭州 310058  
周金莺 浙江大学环境与资源学院, 浙江杭州 310058  
杨武 浙江大学环境与资源学院, 浙江杭州 310058 wyang@zju.edu.cn 
摘要点击次数: 211
全文下载次数: 240
中文摘要:
      农光互补电站作为一种现代生态农业发展的新模式,亟须一种高精度、全过程的遥感技术以实现其对局域尺度光伏电站植被状况的准确监测和实时管控。提供准确的遥感信息有助于优化电站环境管理、推动生态修复,并深入了解光伏电站对周边生态环境的影响。这不仅可为农光互补电站的可持续发展提供技术支持,也为智慧农业和生态电站建设提供了重要的科学依据。本研究以浙江省各类典型农光互补电站为研究对象,利用无人机获取超高空间分辨率多光谱遥感数据对电站内不同土地利用类型的归一化差值植被指数(NDVI)进行精确评估,并结合Sentinel-2 MSI、Landsat-8/9 OLI卫星的同期影像,以均值、动态范围和标准差三个指标定量分析了三类影像反演植被状况的能力,基于回归分析构建了电站全域范围内无人机和卫星影像NDVI数据的转换方程,实现了无人机NDVI数据在不同尺度的应用,并验证了其在浙江省的区域适用性。结果表明:光伏电站内地物空间异质性大、植被长势差异显著,农业种植对提升电站内的NDVI具有重要意义;无人机在全域和分区尺度上均表现出更强的植被探测能力,卫星影像则低估了电站的NDVI值;与Landsat相比,Sentinel-2和无人机之间的NDVI转换方程更有利于实现二者监测结果的互补(R2=0.86,RMSE=0.10)。研究结果拓展了无人机与卫星影像在农光互补电站中的协同应用场景,可为中高分辨率卫星影像在局域尺度碎片化的土地利用方式下的植被状况精准监测和管理提供技术参考。
英文摘要:
      As a new mode of modern ecological agriculture development, agrophotovoltaic power plants urgently need a high-precision and wholeprocess remote sensing technology to achieve accurate monitoring and real-time control of crop conditions in local-scale. Providing accurate remote sensing information can help optimize the environmental management of power plants, promote ecological restoration, and gain insights on the environmental impacts of photovoltaic power plants. It can not only provide technical support for the sustainable development of agrophotovoltaic power plants, but also provide an important scientific basis for future intelligent agricultural management decisions and ecological construction of power plants. However, significant uncertainties and biases exist in monitoring of agrophotovoltaic power plants due to the scale mismatch and spatiotemporal topological errors between the coarse resolution of satellite images and the limited area. The emergence of Unmanned Aerial Vehicle (UAV) makes up for this disadvantage, but it is difficult to promote to large-scale applications. This study took various typical agrophotovoltaic power plants in Zhejiang Province as the research object. The ultra-high spatial resolution multi-spectral remote sensing data was obtained by UAV to accurately estimate the Normalized Difference Vegetation Index( NDVI) of different land use types within PV power plants. Combining the contemporaneous images of Sentinel-2 MSI and Landsat-8/9 OLI satellites, the ability of the three types of images to invert vegetation conditions was quantitatively analyzed using three indicators: mean, dynamic range and standard deviation. A regression analysis was conducted to establish a conversion equation for UAV and satellite image NDVI data across the entire extent of PV power plants. This equation allowed for the application of UAV-derived NDVI data in different scales, and validated its regional applicability within Zhejiang Province. The results showed that there were large heterogeneity of physical space and significant differences in vegetation growth inside the PV power plants. And agricultural planting was of great significance to improve NDVI at PV power plants. UAV showed better vegetation detection capabilities at both overall and zonal scales, while satellite underestimated the NDVI value. Compared to Landsat, the NDVI conversion equation between Sentinel-2 and UAV was more conducive to complementing the monitoring results of the two. The goodness of fit(R2 ) between Sentinel-2 and UAV was 0.86, the Root Mean Square Error (RMSE) was 0.10, which was higher than Landsat(R2 = 0.76, RMSE = 0.11). The results might extend the collaborative application scenarios of UAV and satellite images at agrophotovoltaic power plants, and provide a technical reference for the accurate monitoring and management of vegetation status of medium- and high-resolution satellites to achieve fragmented land use at the local scale.
HTML  查看全文  查看/发表评论  下载PDF阅读器
关闭