Publications

Hall, DK, Key, JR, Casey, KA, Riggs, GA, Cavalieri, DJ (2004). Sea ice surface temperature product from MODIS. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 42(5), 1076-1087.

Abstract
Global sea ice products are produced from the Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS) onboard both the Terra and Aqua satellites. Daily sea ice extent and ice surface temperature (IST) products are available at 1- and 4-km resolution. Validation activities during the cold period (when meltwater is generally not present) in the Northern Hemisphere, defined here as October through May, have been undertaken to assess the accuracy of the I-km resolution MODIS IST algorithm and product. Validation was also done at the South Pole station in Antarctica. In the Arctic Ocean, near-surface air temperatures from the National Oceanic and Atmospheric Administration (NOAA) National Ocean Service (NOS) Center for Operational Oceanographic Products and Services (CO-OPS) Alaska tide stations and from drifting buoys from the North Pole Environmental Observatory (NPEO) buoy program were compared with MODIS-derived ISTs. Using the standard MODIS sea ice product, which utilizes the MODIS cloud mask, results show a bias (mean error) of -2.1 K and a root mean square (RMS) error of 3.7 K. The negative bias means that the satellite retrieval is less than the air temperature. With the bias removed, the RMS error is 3.0 K. When additional visual cloud screening is performed to eliminate MODIS pixels thought to be contaminated by fog, results improved, with a subset of the larger dataset showing a bias of -0.9 K and an RMS error of 1.6 K. Uncertainties would be reduced in the Arctic Ocean dataset if the skin temperature of the sea ice were reported instead of the near-surface air temperatures. With the bias removed, the RMS error for the Arctic Ocean dataset is 1.3 K. Results from the South Pole station in Antarctica show that under clear skies as determined using lidar measurements, the MODIS ISTs are also very close to those of the near-surface air temperatures with a bias of -1.2 K and an RMS error of 1.7 K. With the bias removed, the RMS error for the South Pole dataset is 1.2 K. Thus, the accuracy (RMS error) of the IST measurement is 1.2-1.3 K. It is not possible to obtain an accurate IST from MODIS in the presence of even very thin clouds or fog, and this is the main limitation of the MODIS ice surface temperature product. MODIS sea ice products may be ordered from the National Snow and Ice Data Center in Boulder, CO.

DOI:
10.1109/TGRS.2004.825587

ISSN:
0196-2892