Algorithms to estimate Antarctic sea ice algal biomass from under-ice irradiance spectra at regional scales

Abstract

The presence of algal pigments in sea ice alters under-ice irradiance spectra, and the
\nrelationship between these variables can be used as a non-invasive means for estimating ice-
\nassociated algal biomass on ecologically relevant spatial and temporal scales. While the influence
\nof snow cover and ice algal biomass on spectra transmitted through the snow-ice matrix has been
\nexamined for the Arctic, it has not been tested for Antarctic sea ice at regional scales. We used
\npaired measurements of sea ice core chla concentrations and hyperspectral-transmitted under-ice
\nirradiances from 59 sites sampled off East Antarctica and in the Weddell Sea to develop algorithms
\nfor estimating algal biomass in Antarctic pack ice. We compared 4 approaches that have been
\nused in various bio-optical studies for marine systems: normalised difference indices, ratios of
\nspectral irradiance, scaled band area and empirical orthogonal functions. The percentage of vari-
\nance explained by these models ranged from 38 to 79%, with the best-performing approach being
\nnormalised difference indices. Given the low concentrations of integrated chl a observed in our
\nstudy compared with previous studies, our statistical models performed surprisingly well in
\nexplaining variability in these concentrations. Our findings provide a basis for future work to
\ndevelop methods for non-invasive time series measurements and medium- to large-scale spatial
\nmapping of Antarctic ice algal biomass using instrumented underwater vehicles.