Automated Ice Concentration Analysis

by R. W. Grumbine


An automated ice concentration processing scheme, which has been run in a developmental mode daily since May 1, 1994, has been approved for operations. Two types of fields are produced, analysis fields, and the global fields. The analysis fields are suitable for further manual operational ice concentration analysis, and have been used for the purpose by the National Ice Center since February, 1995. These fields are also used for starting the sea ice forecast model. The global fields are also used to create global grids for making ice/no ice determinations in weather models, and have been used for that purpose by the NCEP/NCAR Climate Reanalysis Project.


The ice concentrations are derived from passive microwave observations from the DMSP F-13 SSMI instrument. The NASA Team algorithm (Cavalieri, 1992) is used to make the derivation. This algorithm is based on the contrast in the microwave between open water (cold) and ice (warm), with three frequencies used to allow for differences in the microwave between old and young ice. Heavy rain and high seas can deceive the algorithm into believing that there is ice even in places that have none. North Atlantic storms routinely appear as false patches of ice. To overcome this problem, a weather filter [Gloersen and Cavalieri, 1986] is applied in an extended mode [Grumbine, 1996]. The extended mode weather filter clears most of the false ice reports, but not all. Since fairly extreme conditions are required for the algorithm to report weather as being ice, analysts should be able to identify these errors by considering the current weather situation.

The global grid for models cannot apply this intelligence to the false ice reports, so a different method is used to generate the final product. Ice concentrations are averaged from the native polar stereographic grids on to a half-degree latitude-longitude grid. Then if the sea surface temperature on the OI SST analysis nearest the given point is greater than 2 C, the ice concentration is forced to zero. This clears false reports fairly effectively.


Sea ice concentration is taken to be the fraction of the area of a pixel (25.4 km by 25.4 km on the analysis grid) that is covered by ice. The satellite does not detect ice less than 5-10 cm thick, so satellite ice concentration is the concentration of ice thicker than this.


Sea ice concentrations are computed to a precision of 1% coverage. The accuracy is 3-5%. Flags are used for special cases in the analysis grid: weather is flagged at 1.77, bad data is 1.66, no data is 2.24. The ice concentrations from the algorithm may exceed 1.00 (100% coverage). There is some thought that there is information present in the over 100% coverage points, so the algorithm does not adjust these high values. Analyst commentary is invited! The global grid for models does not use these flags as bad, missing, or weather points are filled in with the last good observation. It seldom requires more than three days to get another good point on the ice pack. Samples, in black and white, of the two hemispheric (analysis) and the global (model) grid are given in figures 1, 2 and 3. The data are far easier to view and use in color, as they are displayed on the OMB web page at (capitalization is important). The automated analysis fields are disseminated daily in GRIB format, via ftp to various machines at NCEP.

a. NCEP Computers and File Names

Table 2 gives the NCEP computer or workstation and data set name for the experimental global ice drift point forecasts and the Alaska subset of these data.

Currently, the only way to receive these data will be from the Cray machines or by ftp (file transfer protocol) to the address given in Table 1 over the Internet.


The NASA Team algorithm has been subject to objective analysis for several years [c.f. Cavalieri, 1992]. Subjective analysis at the National Ice Center [D. Helms, personal communication 1995, M. Sullivan, 1996] also supports this algorithm as the preferred.


Abdalati, W., K. Steffen, C. Otto and K. C. Jezek 1995: Comparison of brightness temperatures from SSMI instruments on the DMSP F8 and F11 satellites for Antarctica and the Greenland ice sheet, Int J. Rem. Sensing, 16, 1223-1229.

Cavalieri, D. J. 1992: Sea ice algorithm in NASA Sea ice Validation Program for the Defense Meteorological Satellite Program Special Sensor Microwave Imager: Final Report NASA Technical Memorandum 104559, pp. 25-32.

Gloersen, P. And D. J. Cavalieri 1986: Reduction of weather effects in the calculation of sea ice concentration from microwave radiances J. Geophys. Res., 91, 3913-3919.

Grumbine, R. W. 1996: Automated Passive Microwave Sea Ice Concentration Analysis at NCEP, DOC/NOAA/NWS/NCEP/EMC/OMB Technical Note 120, 13 pp.

National Weather Service
Last Modified: November 03, 1997