An increasing number of satellite cloud detection methods include the use o
f Numerical Weather Prediction model (NWP) surface temperatures as a thresh
old for the thermal infrared cloud detection test. The NWP model surface te
mperature and the satellite apparent brightness temperature are assumed to
correlate well for clear sky. Investigations over Europe in April 1997 indi
cate that the correlation over land is highly dependent on location and tim
e of day. In this paper, it is shown that this variation of the correlation
has a distinct impact on the quality of the infrared cloud detection test.
As a result, cloud cover fractions which are retrieved using this thermal
infrared test are biased by this effect. This can have serious impact on th
e quality of cloud climatologies, especially with respect to the diurnal va
riation of cloud cover fraction. A new method is introduced to equalize the
quality of the infrared cloud detection test throughout the day. Threshold
values are allowed to be smaller than commonly used and therefore the qual
ity of the infrared test is improved. The method may be applied in both cli
mate research and near real-time processing.