www.elsevier.com/locate/rse Linescan camera evaluation of SSM/I 85.5 GHz sea ice retrieval

• Main Authors: Caren Garrity A, Dan Lubin B, Stefan Kern C, Leif Toudal Pedersen D
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Published: 2002
• Subjects: Greenland; Sea ice
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.381.9072; http://spsp.ucsd.edu/Researchers/Dan_Lubin/pdfs/02rem_sens_env_v83.pdf

Retrievals of total sea ice concentration from four algorithms using the 85.5 GHz vertically and horizontally polarized channels of the Special Sensor Microwave Imager (SSM/I) over the marginal ice zone in the Barents and Greenland Seas are compared with retrievals of total sea ice concentration from helicopter-borne linescan camera observations made during a cruise of the R/V Polarstern during May–June 1997. The goals are to evaluate (1) SSM/I 85.5 GHz retrievals of total sea ice concentration for climatological purposes, and (2) the ability of 85.5 GHz data to show the sea ice edge through cloud cover, for operational purposes. The SSM/I 85.5 GHz channels offer a spatial resolution of 12.5 km, which is sufficient to resolve ice edge features and small polynyas; however, there is generally more atmospheric contamination of the sea ice signal at 85.5 GHz than at the lower frequencies (19 and 37 GHz) traditionally used for sea ice remote sensing. A self-adjusting algorithm that performs a nonlinear correction for atmospheric moisture, without explicit atmospheric input data, yields the best accuracy over total sea ice concentrations greater than 30%. However, this algorithm can misclassify clouds over open water as sea ice, and is therefore unreliable for locating the sea ice edge. The best algorithm for locating the sea ice edge is found to be the SEA LION algorithm, which explicitly uses meteorological reanalysis data to correct for atmospheric contamination. For total sea ice concentrations in the range 20–70%, empirical 85.5 GHz hybrids of lower-frequency algorithms developed at the NASA Goddard Space Flight Center can