An image off the web showing the range of visible and near infrared that I am working with (300-2500 nm). Blue is around 475 nm, red around 740nm, and infrared is everything higher than red. More info at http://eosweb.larc.nasa.gov/EDDOCS/Wavelengths_for_Colors.html.
The top line is transmittance (not reflectance) at 13 cm, the bottom line at 39 cm depth in white glacier ice. In both cases it's almost all blue light, and the infra-red region is completely gone, even at 13 cm.
Reflectance for different glacier surfaces: white ice, blue bubble-free ice, and wet sand on the surface. Viewing angle is approx. east with 45 deg zenith and sun is to the north.
Incoming and reflected solar radiation spectra over white glacier ice. The ratio of these curves would give spectral albedo. Much of the infrared is completely gone in the reflected signal.
Detail of an absorption feature I investigated to see if it could be used to detect the presence of a layer of water on top of the ice. The minimum value should move to the left to 972 nm for water. A water depth of 1 cm (bottom curve) only moves the location of the minimum value from 1028 nm (top curve) to 1021 nm. So there is a slight difference, but you would need much deeper water levels to be able to detect it clearly. Note: The actual value of the minimum is not so important, just the location on the x-axis. Note: The jump at 1000 nm is a seam where the instrument switches between 2 different sensors.