Independent Component Analysis: A Tool for Algal Bloom Detection
Cameron Penne, Joseph Landon Garrett, T. A. Johansen, Milica Orlandić, R. Heggebø · 2023
Independent component analysis decomposition of hyper-spectral signals has characteristics ideal for detecting harmful algal blooms in coastal waters. In this proceeding, independent component analysis is used to process images taken by the HYPSO-1 satellite at two different locations: Lake Erie and the Salish Sea. For each location, a set of components are produced and compared to two common water quality indices. The results demonstrate that some of the spatial-spectral features uncovered by independent component analysis resemble common water quality indices but also highlight the need for improved interpretation of the recovered features.