Landsat Snowiest Pixels Determination
Related pages: Snow mapping example; NDSI Snow map development details
A way to apply Landsat for areas that are not in shadows is to extract the whitest pixels. The following example shows Landsat 7 ETM+ imagery from 10/29/2011; Landsat 7 has stripings of missing data but still can be used effectively for snow mapping areas (details about Landsat 7 missing data are included in the Crop imagery / Landsat page). In regards to snow mapping, pixels that are saturated have reached the highest value possible (255), in the blue, green, and red bands and represent the whitest surface. In the following image, cyan represents pixels that are saturated in all visible bands. The example is of the San Juan Mountain area in Colorado. (Lighter shades represent higher values in all grayscale images below.)
San Juan Mountains, CO area
Pixels that are saturated in all three visible bands need to also represent low enough values in band 5 (SWIR); snow should have SWIR maximum reflectance threshold ranging from .20 to .25 (Dozier, 1989; pdf); see the Snow map development details link above for details. Below, the blue represent the saturated pixels, it can be seen that, overall, the pixels are located in darker band 5 areas.
The saturated pixels can be mapped over a recent aerial image; the white areas below are the saturated pixels.
Red Mountain Pass area saturated pixels
Mapping can be done on any scale; below the extent is zoomed in to the Red Mountain Pass area which is to the east of Telluride. CO. The image below is Landsat band 5 (short wave infrared); snow should have relatively low reflectance with a maximum reflectance ranging from .20 to .25 (Dozier, 1989). The stripings are missing values from Landsat 7 ETM+.
The image below shows pixels saturated in visible bands that also have band 5 values less than 0.225.
The pixels from the previous graphic are shown over hillshade.
Reference
Dozier, J. 1989. Spectral signature of alpine snow cover from the Landsat Thematic Mapper. Remote Sensing of Environment 28: pp. 9-22.