MODIS-aqua standard chlorophyll image for April 5, 2011, over the Korean Peninsula is shown in Fig. 3(a) while the corresponding standard GOCI chlorophyll image retrieved through the GDPS is shown in Fig. 3(d). In Figs. 3(a) and 3(d), the observed chlorophyll concentration is higher in the coastal regions, particularly along the western and southern parts of the Korean Peninsula. Yet this higher chlorophyll concentration may not represent true chlorophyll because chlorophyll-like features may arise from CDOM plumes, sediments plumes and bottom reflectance when using blue–green band ratio algorithms.13–16 Thus, in turbid waters, chlorophyll product is not adequate to identify bio-optical features. Instead, we used the FLH and the RBD for the surface features detection. Since FLH gives positive readings in both algae rich and sediment rich waters, while the RBD gives positive readings only in the algae rich waters,3 we can distinguish sediment rich waters from the algae rich waters by combining these approaches. An example is shown in Fig. 3, where Figs. 3(b) and 3(e) show the FLH images from MODIS and GOCI, respectively, while Figs. 3(c) and 3(f) show the RBD images from MODIS and GOCI, respectively. The features in the FLH images [Figs. 3(b) and 3(e)] are different from their respective chlorophyll images [Figs. 3(a) and 3(d)], particularly in the region shown with the green box. These false chlorophyll features are due to the aforementioned uncertainties associated with blue–green band ratio algorithms. Agreement between the features in MODIS and GOCI FLH images is much better than the corresponding chlorophyll images (Fig. 3). Uncertainties from imperfect atmospheric correction (which affects the blue/green wavelengths more strongly than the red/NIR wavelengths9), CDOM absorption, sediment absorption and bottom reflectance all explain why chlorophyll images have different features compared with FLH images. On the other hand, FLH overcomes most of these issues and provides more precise feature information from algal and nonalgal particles in the water surface. Since the water absorption is much stronger in the red-NIR regions, algorithms based on these spectral regions can only see the first few meters of the surface waters. However, for surface feature detection and monitoring purposes, this could be beneficial because interference from bottom reflectance will be reduced, unlike with blue–green algorithms. Even though the FLH has many advantages over the blue–green band ratio algorithms, it breaks down in highly scattering waters, where high red peak values in the reflectance are primarily due to contributions from elastic scattering modulated by chlorophyll absorption rather than the fluorescence, thus falsely indicating possible chlorophyll rich areas.48–50 In contrast, the RBD technique easily differentiates between the two effects, giving positive values under true bloom conditions and negative values in highly scattering waters.3 The RBD approach is for high chlorophyll concentration () waters and it depends on the chlorophyll fluorescence quantum yield and the backscattering properties of the particles in the water.3,32 Corresponding RBD images are shown in Figs. 3(c) and 3(f) for MODIS and GOCI, respectively, where the agreement between the features in both images is very reasonable. Additionally, the sediment rich regions in the FLH images (brown box in Fig. 3) are no longer present in the RBD images. This is not surprising since FLH cannot differentiate between sediment rich waters and chlorophyll rich waters, while the RBD only detects chlorophyll rich waters (red circle in Fig. 3). The RBD and FLH results (Fig. 3) are also consistent with our previous study based on the west coast of Florida.3 Since MODIS RBD and FLH results have been validated, we used MODIS to validate GOCI results. The GOCI image shown in Fig. 3 was acquired on April 5, 2011, at 04:16 GMT, which was approximately 16 min after the MODIS acquisition [Figs. 3(a)–3(c)]. Considering that data are from two different sensors and that they are processed through two different processing systems, the agreement between GOCI and MODIS results is very good for both FLH and RBD products. Note that the MODIS and GOCI scales differ slightly because MODIS and GOCI red bands are situated at slightly different wavelengths. MODIS red bands are about 11 nm apart (centered at 667 and 678 nm) while GOCI red bands are located about 20 nm apart (centered at 660 and 680 nm). Thus, GOCI values are slightly higher than the MODIS values.