Premium
Arctic‐scale assessment of satellite passive microwave‐derived snow depth on sea ice using Operation IceBridge airborne data
Author(s) -
Brucker Ludovic,
Markus Thorsten
Publication year - 2013
Publication title -
journal of geophysical research: oceans
Language(s) - English
Resource type - Journals
eISSN - 2169-9291
pISSN - 2169-9275
DOI - 10.1002/jgrc.20228
Subject(s) - sea ice , snow , environmental science , sea ice thickness , satellite , remote sensing , radiometer , arctic , sea ice concentration , arctic ice pack , special sensor microwave/imager , cryosphere , climatology , meteorology , geology , microwave , oceanography , brightness temperature , geography , aerospace engineering , engineering , physics , quantum mechanics
Snow depth on sea ice (SD) is a key geophysical variable, knowledge of which is critical for calculating the energy and mass balance budgets. Moreover, accurate knowledge of the SD distribution is important to retrieve sea‐ice thicknesses from altimetry data. So far, only space‐based microwave radiometers (e.g., Advanced Microwave Scanning Radiometer for Earth Observing System; AMSR‐E) provide operational SD on seasonal sea‐ice retrievals. A thorough assessment of these retrievals is needed on a large scale and on a variety of sea‐ice types. Our study presents such an assessment on Arctic sea ice using NASA's airborne Operation IceBridge (OIB) SDs, retrieved from radar measurements. Between 2009 and 2011, ∼610 12.5 km satellite grid cells were covered by seasonal sea ice where both satellite SD retrievals and OIB data were available. Using all the available data, the difference between the AMSR‐E product and the averaged OIB snow‐radar‐derived SD is 0.00±0.07 m. Satellite‐derived SD was accurate in the Beaufort Sea and the Canadian Archipelago but underestimated (∼0.07 m) in the Nares Strait. The RMSE between the two products ranges between 0.03 and 0.15 m. The RMSE is less than 0.06 m over a shallow snow cover (<0.20 m), in areas where satellite‐retrieved ice concentrations are higher than 90%, surface smooth, and ice thicker than ∼0.5 m. Locally the AMSR‐E algorithm can significantly underestimate SD. Several regions where the retrievals were less accurate (error >0.10 m) have been identified and related to the presence of either low ice concentration or significant fraction of multiyear ice within the grid cell that has not been flagged.