Comparison of the climatologies of high-level clouds from HIRS and ISCCP

Abstract

Comparison of individually matched analyses of high-level cloudiness from the High-Resolution Infrared Sounder (HIRS) CO2-slicing analysis and the International Satellite Cloud Climatology Project (ISCCP) analysis of satellite data for 4 months shows that the former reports about 0.12 more high-level clouds than the latter. Almost all of the difference in high-level cloud amounts occurs as differences of thin cirrus, defined by infrared emissivity ∈ < 0.5 or τvis < 1.3, consistent with a previous comparison of Stratospheric Aerosol and Gas Experiment II and ISCCP. Some of this difference may be caused by the large field of view of the HIRS instrument. Over oceans the differences in cirrus cloud amounts are caused by the higher sensitivity of the HIRS analysis to optically thin clouds, aided by a small high bias of the sea surface temperatures used in the HIRS cloud detection step. Over land the higher detection sensitivity of the HIRS analysis was partially offset by the effect of large low biases in the surface temperatures used in the HIRS cloud detection step, most especially over high mountainous terrains. From these two datasets the authors conclude that about one-third of the earth is covered by high-level clouds (tops above the 440-mb level) and more than two-thirds of these clouds are cirrus, defined as those clouds that have a net radiative heating effect (i.e., infrared ∈ < 0.84 or τvis < 3.6). About half of all cirrus clouds are optically very thin (∈ < 0.5 or τvis < 1.3). Optically thicker (τvis > 3.6) high-level clouds appear to be more frequently associated with each other than with cirrus. Notable concentrations of cirrus in the Tropics mark regions of frequent deep convective activity. However, there are also prominent features associated with the subtropical jet streams. In midlatitudes, cirrus concentrations occur in the oceanic cyclone tracks, but they are even larger over major mountain complexes. Although the quantitative uncertainties of both datasets are large in the polar regions, the agreements and disagreements between them can be explained by the presence of large amounts of cirrus over both polar regions.