Author(s): Edmond Hansen; Juliane Borge; Knut Vilhelm Hoyland
Linked Author(s): Knut Hoyland
Keywords: No Keywords
Abstract: Sea ice thickness is typically observed with a variety of methods having different footprints, ranging from centimeters to kilometers in scale. Great care must be exercised when comparing or merging datasets obtained with different techniques. Here we describe an effort to highlight systematic differences between ice thickness data sets as obtained by four different methods. Spatially referenced series of upward looking sonar (ULS) data (footprint of 2 m, spacing 1 m) obtained in Fram Strait are taken as ground truth. The ULS thickness series are used to mimic the resolution of low-resolution time referenced ULS data (sampling rate 240 s), as well as EMbird (footprint 50 m, spacing 36 m) and ICESat (footprint 64 m, spacing 172 m) data. The time referenced ULS data are mimicked by sub-sampling using observed drift distributions. EMbird and ICESat data are mimicked by constructing synthetic spatial fields of ice draft by statistical models. These synthetic patches of sea ice, derived from the statistical properties of the original ULS thickness series, are used to create synthetic series of EM bird and ICESat ice thickness observations. The result is thickness observations of the same sea ice, with four different methodologies. Comparing the statistics derived by the four methods reveals any systematic differences. Preliminary results indicate that over a sufficiently long patch of ice, the four methods yield similar mean and modal thicknesses. Both the EMbird and the ICESat observations fail to observe many of the features of the tail of the thickness distribution, and therefore miss important features of the ridged volumes of sea ice. The systematic differences identified in this way may prove useful when interpreting differences in ice thickness datasets obtained by the different methods.