Study on the detection of red-tide outbreaks using big satellite database

Abstract

Satellite remote sensing has been successfully employed to monitor and detect the increasing incidence of harmful algal blooms (HABs) under various water conditions. In this study, to establish a comprehensive monitoring system of HAB outbreaks (particularly Cochlodinium polykrikoides blooms) in the southern coast of Korea (SCK), we tested the several proposed red-tide detection methods using SeaWiFS and MODIS ocean color data. Temporal and spatial information of red tide events from 2002 to 2013 were obtained from the National Fisheries Research and Development of Korea (NFRDI), which were matched with synchronously obtained satellite-derived ocean color data. The spectral characteristics of C. polykrikoides red tides were that increased phytoplankton absorption at 443 nm and pigment backscattering 555 nm resulted in a steeper slope between 488 and 555 nm with a hinge point at 488 (or 490) nm. On the other hand, non-red tide water, typically were presented by broader radiance spectra between the blue and green bands were associated with reduced pigment absorption and backscattering. The analysis of ocean color imageries that captured C. polykrikoides red tide blooms showed discolored waters with enhanced pigment concentrations, high chlorophyll, fluorescence, absorption at 443 nm. However, most red tide detection algorithms found a large number of false positive but only a small number of true positive areas. These proposed algorithms are not useful to distinguish true red tide water from complex non-red tide water. Our proposed method substantially reduces the false signal rate (false positive) from strong absorption at short wavelengths and provide a more reliable and robust detection of C. polykrikoides blooms in the SCK from the space.