Detecting red tides in the eastern Seto inland sea with satellite ocean color imagery

A multi-spectral classification scheme is proposed to identify water with red tide(s) using satellite ocean color imagery obtained by the Sea-viewing Wide Field-of-view Sensor (SeaWiFS). The study area was the eastern Seto Inland Sea in Japan, where serious red tides frequently occur. "Background Ocean Colors" (BOCs hereafter), or colors of water around a red tide or those of the water before/after a red tide, are calculated as the monthly climatological average of normalized water-leaving radiances (nLw) with 0.01 degree spatial resolution with SeaWiFS imagery. Criteria for detecting red-tide pixels are established from analyses of characteristics of the nLws (in the 443, 490, 510, and 555 nm bands) anomalies from BOCs and the nLw spectra together with the red-tide records in Osaka Bay. The proposed scheme can efficiently indicate the presence or absence of red tides for independent match-ups with 83% accuracy. Additional validations of specific events indicate that the algorithm performed well in the study area. These results suggest that the scheme is appropriate to detect red tides in the optically complex coastal water of the eastern Seto Inland Sea.