A functional regression model for predicting optical depth and estimating attenuation coefficients in sea-ice covers near Resolute Passage, Canada. Issue 69 (2015)

• Record Type: Journal Article
• Authors: Mcdonald, Shaun; Koulis, Theodoro; Ehn, Jens; Campbell, Karley; Gosselin, Michel; Mundy, C.J.
• Is Part Of: Annals of glaciology. Volume 56:Issue 69(2015)
• Page Start: 147
• Page End: 154
• Publication Date: 2015
• Subjects: sea ice, -- sea-ice ecology, -- snow; Glaciology -- Periodicals; 551.3105 qA613
• DOI: 10.3189/2015AoG69A004
• Languages: English
• ISSNs: 0260-3055

Abstract

Abstract: The spectral dependence of natural light transmittance on ice algae concentration and snow depth in Arctic sea ice provides the potential to study the changing bottom-ice ecosystem using optical relationships. In this paper, we consider the use of functional data analysis techniques to describe such relationships. Specifically, we created a functional regression model describing spectral optical depth as a function of chlorophyll a concentration, snow depth and ice thickness. Measurements of the aforementioned covariates and surface and transmitted spectral irradiance were collected on landfast first-year sea ice in the High Arctic near Resolute Passage, Canada, during the spring of 2011 and used as model input. The derived model explains 75–84.5% of the variation in the observed spectral optical depth curves. No prior assumptions of snow/sea-ice optical properties are required in the application of this technique, as the model estimates the attenuation coefficients of each covariate using only the measurements mentioned above. The quality and simplicity of the model highlight the potential of functional data analysis to study the Arctic marine ecosystem.