Development of a Spectral Reflectance Dataset for Macroalgae in the Coastal Waters of Ulleungdo and Dokdo

The coastal waters of Ulleungdo and Dokdo form a transition zone where cold and warm currents intersect, resulting in highly variable marine conditions and sensitive changes in macroalgal distribution. Improving the accuracy of remote-sensing-based macroalgae classification in such environments requires a region-specific dataset that systematically compiles taxon-level spectral reflectance. In this study, we designed a passive underwater observation system (type-1) and an active multi-purpose observation system (type-2), and acquired spectral reflectance using three observation methods applied to underwater and ground environments: T1U (type-1, underwater, natural sunlight), T2U (type-2, underwater, LED illumination with shading cover), and T2G (type-2, ground, natural sunlight). The objective of this study was to build spectral reflectance data for dominant macroalgae in the Ulleungdo–Dokdo coastal region and to quantitatively evaluate observation-condition–dependent spectral characteristics and their implications for classification. Based on reflectance spectra obtained from the three observation methods, taxon-specific spectral shapes of Phaeophyceae, Chlorophyta, and Rhodophyta reproduced characteristic features consistent with previous reports, including reflectance humps near to 575, 600, and 650 nm for Phaeophyceae, a relatively flat spectral shape for Chlorophyta, and an absorption feature within 400-550 nm with reflectance humps near to 600 and 650 nm for Rhodophyta. Cross-method comparisons using two species measured in all three methods (<i>Eisenia bicyclis</i> and <i>Codium fragile</i>) showed that species-specific spectral shapes were consistently reproduced in both reflectance and first-derivative reflectance, confirming interoperability among T1U, T2U, and T2G. In T2U, localized noise was observed in the 400-500 nm range due to insufficient LED radiance, indicating the need for signal correction in the blue band region. Cosine-similarity analysis based on first-derivative reflectance indicated generally high within-taxon similarity among species (approximately 0.90-0.99), whereas mean between-taxon similarity was lower (approximately 0.75-0.89), demonstrating that taxon-level spectral shapes are quantitatively distinguishable. The spectral reflectance dataset and observation protocols developed in this study provide hyperspectral reflectance reference information for dominant macroalgae and benthic substrates in the Ulleungdo-Dokdo coastal region, supporting remote-sensing–based classification and distribution monitoring, observation-method design, and future standardization of macroalgal spectral data.