An automatic method is presented for extracting a microphysical index (related to the ratio of cloud absorption
coefficients) of cirrus clouds from thermal infrared AVHRR data (11 jtm and 12 im). The processing scheme makes use of
the optical properties of ice particles at the two wavelengths and of the relationship between emittance and transmittance of
semitransparent cirrus clouds. The retrieved microphysical index is characteristic of the optical properties of ice particles and
allows to explain the high Brightness Temperature Differences measured between AVHRR channel 4 and 5. Assuming that
cirrus clouds are composed of ice spheres, the effective diameter of the particles can be deduced from this microphysical
index. The retrieval scheme has obvious limitations. First, in order to perform the processing over a geographic zone, both (i)
overcast semitransparent cimis cloud pixels, (ii) overcast opaque cirrus cloud pixels, and (iii) cirrus free pixels must be
observed in this region. Such systems are frequently found on (50x50) - (200x200) Km2 regional scale. Secondly, for
multilayered cloud systems, the processing is only able to derive the microphysical index related to the semi-transparent part
of the coldest cirrus layer that contains the smallest particles. Finally, the determination of effective diameter beyond 60 .tm is
no more accurate.
The treatment is applied to 21 NOAA1 1-AVHRR images acquired during International Cirrus Experiment (ICE'89)
on the North Atlantic and Europe region from 10 to 20 October 1989. This statistical approach shows that for cloud top
temperature less than about 230 K, effective mean particle diameter is always greater than 20 jim, while warmer cirrus clouds
contain, on an average, smaller particles.