Cloud phase recognition is important for cloud studies. Ice crystals corres
pond to physical process and properties that differ from those of liquid wa
ter drops. The angular polarization signature is a good mean to discriminat
e between spherical and nonspherical particles (liquid and ice phase, respe
ctively). POLDER (Polarization and Directionality of Earth Reflectances) ha
s been launched on the Japanese ADEOS platform in August 1996. Because of i
ts multidirectional, multispectral, and multipolarization capabilities this
new radiometer gives useful information on clouds and their influence on r
adiation in the shortwave range. The POLDER bidirectional observation capab
ility provides the polarization signatures within a large range of scatteri
ng angles in three spectral bands centered on 0.443, 0.670, and 0.865 mu m
with a spatial resolution of 6.2 km x 6.2 km. These original features allow
to obtain some information both on cloud thermodynamic phase and on cloud
microphysics (size/shape). According to POLDER airborne observations, liqui
d cloud droplets exhibit very specific polarization features of a rainbow f
or scattering angles near 140 degrees. Conversely, theoretical studies of s
cattering, by various crystalline particles and also airborne measurements
show that the rainbow characteristics disappear as soon as the particles de
part from the spherical shape. In the paper the POLDER algorithm for cloud
phase classification is presented, as well as the physical principle of thi
s algorithm. Results derived from the POLDER spaceborne version are also pr
esented and compared with lidar ground-based observations and satellite clo
ud classification. This cloud phase classification method is shown to be re
liable. The major limitation appears when thin cirrus clouds overlap the li
quid cloud layer, In this case, if the cirrus optical thickness is smaller
than 2, the liquid phase may be retrieved. Otherwise, the ice phase is corr
ectly detected as long as cloud detection works.