Red Tide Detection Based on High Spatial Resolution Broad Band Satellite Data: A Case Study of GF-1

Abstract

Liu, R.-J.; Zhang, J.; Cui, B.-G.; Ma, Y.; Song, P.J., and An, J.-B., 2019. Red tide detection based on high spatial resolution broad band satellite data: A case study of GF-1. In: Jung, H.-S.; Lee, S.; Ryu, J.-H., and Cui, T. (eds.), Advances in Remote Sensing and Geoscience Information Systems of Coastal Environments. Journal of Coastal Research, Special Issue No. 90, pp. 120-128. Coconut Creek (Florida), ISSN 0749-0208.Traditional red tide remote sensing detection methods are based on ocean color satellite data with high spectral resolution (spectral band width < 20 nm), but low spatial resolution. Clearly, fine-scale remote-sensing monitoring of red tide requires high spatial resolution satellite data. Yet, satellite data with high spatial resolution often lack high spectral resolution, with spectral band width > 50 nm. Moreover, systematic research has yet to demonstrate effective detection of red tide using high spatial resolution, but low spectral resolution, satellite data. In this paper, high spatial resolution (16 m), but low spectral resolution, data from the satellite, GF-1, are analyzed to determine if effective fine-scale detection of red tide is possible. The spectral response characteristics of red tide in GF-1 data are analyzed using a 2014 red tide event that occurred in Guangdong Province, China. It was found that, despite broad spectral band widths, GF-1 WFV imaging spectrometer data still exhibits a clear tide detection algorithm for GF-1 WFV data based on this red tide spectral signal. The detection accuracy of red tide for this algorithm is better than 92 %, and the F1-Score value is better than 87 %. Spatial analysis of the Guangdong 2014 red-tide event based on this algorithm showed that the maximum red tide area extracted from GF-1 satellite data is 10 km2, and that the red tide area decreases greatly during the extinction phase. A comparative analysis of MODIS satellite data indicates that, despite the availability of multiple ocean color bands, no ocean color anomaly is detected because of coarse spatial resolution. The developed method was also successfully used to detect red tide occurred in Rizhao, Shandong Province.