基于卫星和无人机影像的浮叶植被覆盖度反演  

INVERSION OF FLOATING LEAF VEGETATION COVERAGE BASED ON SATELLITE IMAGES AND DRONE IMAGES

作　　者：罗奇斌 何亮[2,3] 郭诗园 谢钟书 王昭颖 葛刚[1,2] 李述[2,3] LUO Qi-Bin;HE Liang;GUO Shi-Yuan;XIE Zhong-Shu;WANG Zhao-Ying;GE Gang;LI Shu(School of Life Science,Nanchang University,Nanchang 330031,China;Key Laboratory of Poyang Lake Environment and Resource Utilization,Ministry of Education,Nanchang University,Nanchang 330031,China;School of Resources and Environment,Nanchang University,Nanchang 330031,China)

机构地区：[1]南昌大学生命科学学院,南昌330031 [2]南昌大学鄱阳湖环境与资源利用教育部重点实验室,南昌330031 [3]南昌大学资源与环境学院,南昌330031

出　　处：《水生生物学报》2025年第4期111-119,共9页Acta Hydrobiologica Sinica

基　　金：国家自然科学基金(72364023、32260290和32360292)资助。

摘　　要：为了快速、准确地监测浮叶植被的覆盖度,文章以鄱阳湖中浮叶植被广泛分布的碟形湖和隔断湖汊作为研究区域,首先基于Sentinel-2卫星影像计算像元的NDVI(Normalized Difference Vegetation Index)指数,然后利用无人机航拍影像计算对应像元范围浮叶植被的实地覆盖度,并建立卫星影像NDVI指数和实地浮叶植被覆盖度的回归模型,最后以大伍湖为实验区,分别使用文章方法和传统的像元二分法,反演实验区的浮叶植被覆盖度,并对两种方法的反演精度进行了比较。结果表明:(1)文章基于NDVI指数与浮叶植被实地覆盖度的回归模型具有非常好的拟合效果,其回归方程的决定系数R2达到了0.9;该回归模型的均方根误差RMSE为5.75%,平均相对误差MRE仅为9.0%;(2)用传统的阈值二分法反演浮叶植被的覆盖度时,当NDVI阈值为0.081时,均方根误差取得最小值20.25%,平均相对误差取得最小值53.68%,均远大于文章方法的均方根误差和平均相对误差;(3)相比于传统的像元二分法,该模型能更精确地反演浮叶植被的覆盖度,特别是在浮叶植被稀疏的区域。文章基于无人机航拍数据和卫星影像NDVI指数构建回归模型的方法,可以为水生植物群落快速监测和定量反演提供借鉴。In order to quickly and accurately monitor the coverage of floating leaf vegetation,this article focuses on the sub-lakes and isolated lakes of Poyang Lake which with widespread floating leaf vegetation.First,the normalized difference vegetation index(NDVI)of the pixels were calculated based on Sentinel-2 satellite imagery.Then,corresponding field coverage of floating leaf vegetation(FCFLV)in the pixels were assessed using drone aerial imagery,and a regression model between NDVI and FCFLV is established.Finally,using Dawu Lake as an example,floating leaf vegetation’s coverage(FLVC)of Dawu Lake was inverted using both the method presented in this paper and the traditional pixel dichotomy method,and the inversion accuracy of the two methods was compared.The results indicate:(1)The regression model based on the NDVI and FCFLV demonstrates a very good fit,with a coefficient of determination(R²)reaching 0.9;the model’s root mean square error(RMSE)is 5.75%,and the mean relative error(MRE)is only 9.0%;(2)Using the traditional threshold dichotomy method to invert FLVC yields a minimum RMSE of 20.25%and a minimum MRE of 53.68%when the NDVI threshold is 0.081,both of which are significantly higher than the RMSE and MRE obtained from the method in this paper;(3)Compared to the traditional pixel dichotomy method,this model can more accurately invert FLVC,especially in areas where the vegetation is sparse.The method presented in this paper,which constructs a regression model based on drone aerial data and satellite NDVI,can provide a reference for the rapid monitoring and quantitative inversion of aquatic plant communities.

关 键 词：鄱阳湖 浮叶植被覆盖度 归一化植被指数 卫星影像 无人机影像 

分 类 号：S932.8[农业科学—渔业资源]