Data-derived spatial-resolution errors of Antarctic sea-ice thickness

Abstract

Uncertainties due to spatial resolution (E R) of gridded sea-ice thickness (z) distributions remain largely unquantified. We address this issue using remotely sensed and in situ observations of the Southern Ocean (south of −60°) to determine appropriate data sources based on length scales for continental-scale sea-ice studies. Sea-ice thickness is not normally distributed such that the mean, median, and mode are distinct from each other. Averaging only retains the mean value, reducing bias and natural variability as z is aggregated to coarse resolutions. The rate of smoothing as a function of resolution influences the sea-ice thickness distribution represented by each product. Analysis of E R for 1° through 5° resolutions shows absolute E R increasing with grid-cell size and sea-ice extent in both datasets. The absolute E R for gridded thickness distribution ranges from 0.02 to 0.40 m at 1° and from 0.03 to 0.78 m at 5°. The E R value and slope (m) is lower (E R≤0.20 m, m<0.05) for ice charts than the ship dataset (0.10 m<E R<0.80 m, m<0.17) with relative E R between datasets remaining consistent. From these results, recommendations for dataset use and future sea-ice observation frequency and distribution schemes are discussed.