Detection of Cochlodinium polykrikoides red tide based on two-stage filtering using MODIS data

Abstract

This study proposes a red tide detection algorithm based on two-stage filtering to provide positional information of red tide for design of desalination system. Based on MODIS Red tide Index (MRI), this algorithm conducts additional two-stage filtering processes, sea surface temperature, and 667 nm wavelength filtering, and finally detects red tide by applying the local Moran's I that is a spatial statistical method. We conducted visual interpretation and quantitative accuracy assessment using error matrix and introduced the F-measure, and Jaccard and Yule coefficients as measures of detection accuracy. The algorithm effectiveness was verified by the detection of a large Cochlodinium polykrikoides red tide event that appeared in Korean coastal waters. Therefore, it is expected that this algorithm could help plant designers and operators to choose an optimal place for a desalination system.