Cyanobacterial bloom recognition and spatiotemporal change analysis based on remote sensing images

摘要

In recent years, lake eutrophication has led to a large bloom of cyanobacteria, which not only brings serious ecological disaster but also restricts sustainable development of regional economy in China. The research on recognition and change detection method of cyanobacterial bloom based on remote sensing images plays a vital role in quickly monitoring the distribution information of cyanobacterial bloom and effectively preventing the abnormal growth. In this article, a normalized cyanobacteria index (NDI_CB) is constructed successfully. Then, the feature space consisting of different vegetation indexes and textures is optimized through combining variable precision rough sets with gray incidence decision making. Then a dual-weight Support Vector Machine (SVM) classification model based on the wavelet kernel is built with the theories of wavelet and machine learning. This model has been shown to be effective to extract the information of cyanobacteria. Therefore, the proposed remote sensing analysis method in this article provides a new remote sensing interpretation tool for time-space change monitoring of cyanobacterial bloom distribution in Dianshan Lake and provides sound scientific method for the prevention and treatment of the cyanobacterial bloom as well.