On Friday, 06 November 2015 from 8:00 a.m. to 5:00 p.m. (USA Mountain Time), our FTP services, which also includes automated data requests from services such as Polaris and the GLAS Subsetter, will be unavailable because of system maintenance. We apologize for any inconvenience this may cause you.
The SMMR land mask and coastline overlay grids provided in the FTP directory TOOLS, for use with the SMMR brightness temperature grids, are not identical to the land mask and overlay grids distributed by NSIDC with the SSM/I brightness temperature and ice concentration grids. This is because the land mask and coastal overlays were developed in two different environments. The SSM/I overlays developed at NSIDC in 1985 used the CIA World Data Base I (WDB-I) coastlines. The SMMR land mask, developed at NASA/GSFC, uses the CIA World Data Base II (WDB-II). A detailed description of the NASA/GSFC process is found in Gloersen et al. 1992, sec. 2.2.3, pp. 22-24. The differences in the land masks result in about a 3 % difference in the number of ocean pixels on the North Polar grid. To facilitate intercomparison of SSM/I and SMMR data, it is suggested that you select either the WDB-I or WDB-II land mask and lat/lon overlay for your work, and use that selection exclusively. Otherwise, areal intercomparisons may lead to inconsistent results.
The land mask is created as follows: First, a high-resolution coastline file [adapted from the Central Intelligence Agency (CIA) World Data Base II] is used to produce land boundaries on a high-resolution map with twice the resolution (four times as many pixels) of the projections used here, and with all pixels initially designated as ocean. Next, the high-resolution pixels containing a boundary are designated as coastline, and the areas within the coastline are designated as land. .... The pixels in the higher-resolution map are then grouped into 2 x 2 arrays, corresponding to the single pixels in the lower-resolution map. The determination of the proper designations for the pixels (land, ocean, or coastline) in the lower-resolution map involves performing two classifications of each 2 x 2 higher-resolution array . . . . (Gloersen et al. 1992, sec. 2.2.3, pp. 22-24)
-- one emphasizing ocean and the other land, and then tallying the results over each lower-resolution pixel.