Annual Minimum Exposed Snow and Ice Cover: More information

Note: The MODICE data presented here are preliminary data and are subject to further quality control.

The map shows annual minimum exposed snow and ice cover for each of the years of the NASA Earth Observing System era (2000-2013). This product indicates those areas that never reveal complete soil or vegetation cover across a pixel during the particular year. The interannual variability of snow cover around the globe makes individual acquisitions unreliable for indicating the proper extent of time change of the cryosphere. The MODICE product instead uses the complete daily time series of MODIS for each year to effectively shrink-wrap those land areas that remain snow and ice throughout the year (Painter et al., 2012). In turn, this gives the first available annually resolved indication of the baseline extent of the cryosphere on land.  In the map, the annual exposed ice is indicated by blue pixels.

This sample image from Annual Minimum Exposed Snow and Ice Cover shows snow and ice cover on Greenland in 2000.

Maps for each year are generated using the MODICE algorithm, which analyzes daily fractional snow and ice cover images generated using the MODIS Snow Covered Area and Grain Size (MODSCAG) model (Painter et al., 2009). Unlike many glacier products, no interpretation is needed – this is simply each year’s minimum extent of snow and ice.  That is, we are straightforwardly detecting (to the degree that the extensive time series and cloud cover occlusion permits) those footprints on the ground that keep snow and/or ice at the surface all year. The resolution of the data is higher than the other maps presented here, so changes may be better observed by zooming in, particularly to Alaska.

Now, because we are looking for exposed snow and ice, MODICE does not map debris-covered glaciers and on occasion will not map the dark Forbes bands of ogives.  There are also times when cloud cover is too persistent for us to see the surface reliably (with adequate view geometry) within a year and as such we leave those pixels indicated in yellow as “Not Seen”.  Finally, land/water masks are needed to filter out oceans and lakes where shallow water can appear as snow and ice.  However, land/water masks are imperfect due to classification and geolocation issues.  We are now using the MOD44W land/water mask, which is considered the best available for the MODIS data stream (Carroll et al., 2009).

See the reference below for further details on the MODICE algorithm and MODSCAG model.


Carroll, M., J. Townshend, C. DiMiceli, P. Noojipady, and R. Sohlberg (2009), A new global raster water mask at 250 meter resolution, International Journal of Digital Earth, 2(4). 

Painter, T. H., M. J. Brodzik, A. Racoviteanu, and R. L. Armstrong (2012), Automated mapping of Earth’s annual minimum exposed snow and ice with MODIS, Geophys. Res. Lett., 39(L20501).

Painter, T. H., K. Rittger, C. McKenzie, P. Slaughter, R. E. Davis, and J. Dozier (2009), Retrieval of subpixel snow covered area, grain size, and albedo from MODIS, Remote Sensing of Environment, 113, 868-879.

Return to map >>