Abstract
Sea ice algal aggregates represent a poorly characterized but crucial component of the Arctic marine ecosystem during the melt season, contributing to productivity and playing a key role in sympagic-benthic coupling. However, their patchy distribution and temporary nature make them difficult to study using traditional coring methods. Here, we present a study from the Arctic Ocean that uses underwater hyperspectral imaging to detect and estimate chlorophyll a of floating aggregates and other ice algal communities. Using an under-ice L-arm for calibration of a bio-optical model and a small remotely operated vehicle for spatial surveys (approximately 100 m2), we applied two models—normalized difference index and relative biomass index—to estimate chlorophyll a concentration from beneath the ice across three different habitats: level ice, under melt ponds, and ridged ice. The relative biomass index (root mean square error = 0.35 mg m−3) outperformed the normalized difference index (root square mean error = 0.91 mg m−3) in mapping algal aggregates, likely due to its ability to utilize the whole spectrum instead of a ratio of two wavelengths. Our results showed differences of aggregate distribution and size at the sea ice-ocean interface depending on the characteristics of overlying sea ice cover, with a higher number and larger sized aggregates near melt ponds versus level and ridged ice. We identified both floating aggregates and a spectrally distinct ridge community; however, only the aggregates were captured effectively by either model. This study highlights the potential of underwater hyperspectral imaging in scaling observations from point measurements (approximately 60 cm2) to larger spatial areas (up to 100 m2), enabling estimates of variable sea ice algal biomass distribution across habitats that are difficult to study, while also emphasizing the importance of thorough calibration of the model.