The method of algal bloom extraction in Lake Chaohu waters based on FAI-L method

近年来随着人类的活动日益加剧,水体富营养化问题已经严重威胁到湖泊生态安全。为了快速并准确地获取藻华爆发的范围,本文提出浮游藻类指数线性拟合模型 (FAI linear fitting model,FAI-L)。在以往的研究中,NDVI(normalized difference vegetation index)已经广泛应用于藻华的识别中,且采用坡度计算获取NDVI阈值的方法也得到验证,相对于NDVI,FAI对环境条件的改变敏感度较低,且由于FAI增加了短红外波段,能够有效地降低部分大气和薄云的影响,对藻华的识别有较高的精度,但是FAI识别藻华的阈值如何确定的问题没有有效的解决办法。本文通过建立NDVI与FAI的线性拟合方程,利用NDVI阈值确定FAI阈值,能够有效地解决FAI阈值确定问题。通过Landsat8和Sentinel-2的提取结果显示:(1)FAI-L相对于NDVI提取结果在精度上有较大提升。采用该方法对于Landsat8影像的藻华提取精度为97.16%,相对于NDVI的提取精度(91.72%)提高了5.44%。(2)以Sentinel-2数据为基础探究FAI-L的适用性情况,结果显示对于高分辨率的遥感影像,FAI-L的提取精度达到97.10%,对高分辨率影像具有良好的适用性。(3)FAI-L可以很好地排除云层对藻华提取的干扰,且对藻华的边缘区域和较小的零星区域有更好的提取效果。通过FAI-L能够较好地获取巢湖水域藻华的空间分布情况,可以有效降低由于云层造成的误判现象,具有较高的精度和适用性,可为藻华提取光谱敏感指数的阈值确定提供新方法。;In recent years, with increasing human activities, water eutrophication has seriously threatened the ecological security of lakes. In order to obtain range algal bloom quickly and accurately, FAI model planktonic algae index (FAI-L) is proposed in this paper. previous studies, NDVI (normalized index) been widely used identification, method obtaining threshold by slope calculation also verified. Compared NDVI, less sensitive changes environmental conditions higher accuracy identification blooms due addition short infrared bands effectively reduce impacts atmosphere thin clouds. However, there no effective solution problem how determine for identifying FAI. This paper established equation between threshold, which can solve determination. The extraction results Landsat8 Sentinel-2 showed that: (1) FAI-L was greatly improved. from images 97.16%, 5.44% than that (91.72%). (2) Based on data, applicability explored. high-resolution remote sensing images, 97.10%, had good applicability. (3) could eliminate interference clouds a better effect edge areas small sporadic blooms. spatial distribution Lake Chaohu waters, misjudgment caused clouds, high It provide new determining spectral sensitivity extraction.