Synergy effect detecting Harmful algal blooms (HABs) using multi-sensor data

Abstract

The satellite-based red tide detection algorithm was mainly developed for the ocean color sensor and it was developed considering the wavelength range of the specific sensor. Since sensors have different characteristics, there is a limit to red tide detection with a single-sensor. Therefore, it is necessary to estimate the optimal red tide area through fusion of multi-sensor. The objective of this paper is to identify synergy effect detecting HABs using multi-sensor and to quantify HABs distribution and area coverage. The red tide areas which are different due to SNR, spatial resolution and spectral resolution were investigated. In terrestrial sensor with high spatial resolution, narrow red tide patches are expressed clearly, but red tide detection is limited due to seawater flow, image noise, and ship trajectory. On the other hand, ocean-colour sensor was advantageous in observing the low concentration of red tide patch due to its excellent spectral resolution and signal-to-noise ratio (SNR). As a result of evaluation of cross-sensor consistency, the similarity of MODIS and VIIRS among the color sensor was high (R2=0.95), but GOCI and MODIS showed a coefficient of determination of 0.65. The similarity of ocean-colour sensor and terrestrial sensor was relatively low (R2=0.4). We applied Spectral Shape algorithm to each sensor and compared the extracted red tide areas and calculated the integrated area where they are red tide detection with a single-sensor. Therefore, it is necessary to estimate the optimal red tide area through fusion of multi-sensor. The objective of this paper is to identify synergy effect detecting HABs using multi-sensor and to quantify HABs distribution and area coverage. The red tide areas which are different due to SNR, spatial resolution and spectral resolution were investigated. In terrestrial sensor with high spatial resolution, narrow red tide patches are expressed clearly, but red tide detection is limited due to seawater flow, image noise, and ship trajectory. On the other hand, ocean-colour sensor was advantageous in observing the low concentration of red tide patch due to its excellent spectral resolution and signal-to-noise ratio (SNR). As a result of evaluation of cross-sensor consistency, the similarity of MODIS and VIIRS among the color sensor was high (R2=0.95), but GOCI and MODIS showed a coefficient of determination of 0.65. The similarity of ocean-colour sensor and terrestrial sensor was relatively low (R2=0.4). We applied Spectral Shape algorithm to each sensor and compared the extracted red tide areas and calculated the integrated area where they are