基于海岸带高光谱成像仪影像的太湖蓝藻水华和水草识别

水华与水草的同步监测对于研究湖泊水环境、生态特性以及水循环具有重要意义,相对于传统监测方法如实地调查,利用遥感手段具有大范围、长时间周期、高效率以及低成本等优势。基于海岸带高光谱成像仪HICO(Hyperspectral Imager for the Coastal Ocean)影像,利用叶绿素a光谱指数和藻蓝蛋白基线的水华和水草识别模型,提取太湖水华和水草分布图,经过检验,水华和水草的平均提取精度为93%和95%;通过水华和水草分布图的叠加分析了2010~2014年太湖水华和水草的分布规律,与相关文献的分析结果一致,进一步证实了识别方法的可靠性;将平均阈值与最佳阈值进行对比分析,提取水华与水草面积的精度分别为75.7%和84%,在对精度要求不高但对效率要求较高的情况下,可以利用平均阈值提取水华和水草,便于实现水华和水草的自动化提取及批处理。

Distinguishing Cyanobacteria Bloom and Aquatic Plants in Lake Taihu based on Hyperspectral Imager for the Coastal Ocean Images

The synchronous monitoring for the cyanobacteria bloom and aquatic plants is of great significance for the study of lake water environment、ecology and the water cycle.Compared with traditional monitoring methods,for instance,field investigation,using remote sensing technology with the advantages of large scope,long cycle,high efficiency and low cost.A model based on “Chlorophyll a Spectral Index” and “Baseline of Phycocyanin” was built to distinguish cyanobacteria bloom and aquatic plants in Lake Taihu by using Hyperspectral Imager for the Coastal Ocean (HICO) images.The average accuracy of cyanobacteria bloom and aquatic plants are 93% and 95% respectively.By overlapping the distribution maps of cyanobacteria bloom and aquatic plants,the distribution rules of cyanobacteria bloom and aquatic plants in Lake Taihu from 2010 to 2014 were analyzed,which are consistent with the former results in the literatures.The average thresholds were used to extract cyanobacteria bloom and aquatic plants,and the accuracy are 75.7% and 84.0% respectively.If the efficiency is more desired than accuracy,then average thresholds can be used to extract cyanobacteria bloom and aquatic plants.Which is convenient for realizing batch processing and the automation extraction of cyanobacteria bloom and aquatic plants.