摘要
Sentinel-2卫星兼具了空间分辨率高、重放周期短、谱段丰富三方面特点,为蓝藻水华爆发阶段及时准确的蓝藻水华提取提供了影像基础,但目前在大型湖泊蓝藻水华提取中的应用报道较少。为此,文章以2018—2020年巢湖的Sentinel-2遥感影像为例,开展包括浮游藻类指数(FAI)在内的多指标蓝藻水华提取方法研究,针对FAI阈值难以确定的典型问题,提出了基于回归分析的FAI阈值确定方法。结果表明,(1)与低分辨率卫星MODIS和GF-1影像的蓝藻水华提取结果相比,Sentinel-2提取到了面积小至100 m2的蓝藻水华区域,提取结果差异主要集中在蓝藻水华区域的边缘部分以及面积较小的零星区域,证明了Sentinel-2遥感影像能够更精确地估算蓝藻水华面积。(2)以NDVI的阈值0为基准,通过回归分析确定FAI的阈值为-1.152(决定系数r2达到0.982 3,显著性检验P<0.001),对2019年6—11月无云雾遮挡的Sentinel-2遥感影像,分别采用归一化植被指数(NDVI)与FAI指标提取蓝藻水华面积,结果表明蓝藻水华的分布情况基本一致,且蓝藻水华面积的相对偏差小于5%,证明了FAI阈值确定方法的有效性。(3)选取云雾遮挡但无大规模蓝藻水华爆发的Sentinel-2遥感影像,分别采用NDVI和FAI指标提取蓝藻水华面积,在无云雾遮挡的区域,两个指标的蓝藻水华分布情况一致,而有云雾遮挡的区域,FAI指标的提取面积仅为NDVI的53.89%,证明FAI指标受云雾影响更小。Sentinel-2遥感影像的高空间分辨率使巢湖蓝藻水华提取更为准确,在未来的水质监测领域将彰显更大的价值。
Sentinel-2 satellite is characterized by its high resolution,rich spectrum and short revisit period,providing abundant images and allowing to extract cyanobacteria blooms timely and accurately.However,few researches on applications of the satellites images’for cyanobacteria blooms extraction have been published recently.In this study,Sentinel-2 remote sensing images of Chaohu Lake during the period of 2018–2020 is adopted to detect the cyanobacteria blooms in Chaohu Lake using the several indexes including Phytoplankton Alga Index(FAI).A method is proposed to determine the threshold of FAI by regression analysis.Experiments on Sentinel-2 images are performed and results show that(1)Compared with the results of low-resolution satellites MODIS and GF-1,Sentienl-2 allows to extract cyanobacteria blooms with an area as small as 100 m2,and the key different between Sentinel-2 and other satellites mainly concentrated in the edges of cyanobacteria bloom area and those scattered small bloom areas.It is proved that Sentinel-2 remote sensing images could estimate cyanobacterial bloom area more accurately.(2)Taking the NDVI threshold 0 as the benchmark,the FAI threshold is calculated by linear fitting to be−1.152(the coefficient of determination r2 reaches 0.9823,and the significance test P<0.001).For the Sentinel without cloud and fog occlusion from June to November 2019 Sentinel-2 remote sensing images use NDVI and FAI indicators to extract the area of cyanobacteria blooms.The results show that the distribution of cyanobacteria bloom area is less than 5%,which proves the effectiveness of FAI threshold determination method.And(3)the Sentinel-2 images with cloudcovered areas where there are no large-scale cyanobacteria blooms are adopted to calculated cyanobacteria bloom’areas by using NDVI and FAI.In areas without cloud and fog,the distribution of cyanobacteria blooms in the two indicators is the same,while in areas with cloud and fog,the extraction area of FAI indicators is only NDVI 53.89%of the total,proving that FAI indicators are less affected by clouds and fog.The high spatial resolution of Sentinel-2 remote sensing images makes the extraction of cyanobacteria blooms in Chaohu Lake more accurate,while will show greater value in the field for water quality monitoring in the future.
作者
刘海秋
任恒奎
牛鑫鑫
夏萍
LIU Haiqiu;REN Hengkui;NIU Xinxin;XIA Ping(School of Information and Computer,Anhui Agricultural University,Hefei 230036,China;Department of Agricultural Machinery,Anhui Agricultural University,Hefei 230036,China)
出处
《生态环境学报》
CSCD
北大核心
2021年第1期146-155,共10页
Ecology and Environmental Sciences
基金
国际科技合作计划项目(1604b0602029)
国家自然科学基金项目(61805001)
安徽省自然科学基金项目(1808085QF218)
智慧农业技术与装备安徽省重点实验室自主创新研究基金项目(APKLSATE2019X007)
安徽农业大学研究生创新基金项目(2021yjs-51)。