679
4.SNOW POLARIZATION FEATURES- COMPARISON BETWEEN SNOW AND CLOUD
OPTICAL PROPERTIES.
The polarization properties of snow are expected to be very weak according to the few existing
measurements (Egan and al, 1985), (Roger and al, 1993). No rainbow has been observed over
snow. The RACER campaign gave us the first opportunity to observe snow covers in clear sky
conditions with the POLDER airborne version. In this case, we did not observe intrinsic snow
signature, but the addition of polarized contribution from snow (assumed weak) and
atmospheric one (air molecules + aerosol).
The problem is important because snow/ice and clouds affects strongly, yet differentially,
surface radiation balance. Confusing snow/ice and clouds may yield large errors in
estimates of the solar energy reaching the surface. Using spectral measurements in the
visible and near-infrared to distinguish snow/ice from clouds is difficult, however, because
both types of target exhibit a high reflectance and are quite white spectrally. Thermal-
infrared data are not appropriate either, because snow/ice and clouds can have the same
apparent temperature. Although other techniques may be suitable (e.g, those that exploit
diffrences between the textural properties of snow/ice and clouds), as well as other types of
measurements (e.g., in the microwave spectral region), the photopolarimetric
measurements offer a unique opportunity to investigate wether polarization information at
visible and near-infrared wavelengths can be used to make the distinction efficiently. In
the following, we demonstrate, by analyzing a few POLDER images, that polarization is ,
indeed, a useful property of light to determine the nature of the targets.
4.1 observations
We analysed two POLDER scenes : one over a bank of stratocumulus clouds (figure 6), the
other over a spatially homogenous snow-ice field (figure 7). The aircraft altitude was 3.5
km altitude. The ground resolution is about 17x17m.
Figure 6 : Reflectance image over cloud covers
at 450 nm.
