This paper analyzes several near-global datasets of low cloud cover, including the the International Satellite Cloud Climatology Project (ISCCP) satellite observations, C. J. Hahn et al. surface-derived observations, and the National Centers for Environmental Prediction (NCEP) and ECMWF reanalysis products (ERA). The magnitudes of annual-mean ISCCP and C. J. Hahn observations of low cloud fraction are found to differ by up to about 0.4 for a number of locations. These differences are largely attributable to the fact that ISCCP low clouds are only those low clouds that are not obstructed by higher cloud. Those of both the NCEP and ERA low clouds, which should be comparable to the Hahn low cloud dataset, have magnitudes up to about 0.3 less than the latter. The dominant EOFs of the seasonal variation of ISCCP and Hahn observations low cloud differ substantially over much of the Northern Hemisphere, where there is a sizable number of observations in Hahn. The pattern of the dominant seasonal EOF of NCEP low clouds has a number of qualitative similarities with that of Hahn between approximately 10°° and 40°°N. That of the ECMWF low clouds is less similar and has much larger amplitudes in the high latitudes of both hemispheres than any of the other datasets. The calculated regression coefficients between interannual variations of Niño-3 SST variations and low cloud departures in the equatorial central Pacific have positive magnitudes of about 0.02 (°°C) −−1 for the C. J. Hahn et al. and NCEP data, but negative values of similar magnitudes for the ISCCP and ECMWF low cloud fractions. These results suggest a need for improved observational estimates and model specifications of the three-dimensional structure of clouds.
Journal of Climate – American Meteorological Society
Published: Dec 15, 1998