Three different estimates of the percent of fixed geographical regions (160×160 km) either free of cloud (clear) or covered by low, middle, and high (opaque) cloud have been intercompared. The estimates were derived by analysts interpreting geosynchronous satellite images, with concurrent meteorological observations; from NIMBUS 7 temperature humidity infrared radiometer (THIR) CLOUDS ERB (CLE) data; and from Air Force three-dimensional (3-D) nephanalysis (3DN) data. Standard linear regression and correlation statistics have been used to estimate the systematic and random errors of the CLE and 3DN cloud amounts (independent variable) at different altitudes relative to the analyst (dependent variable). It has been concluded that (1) Air Force 3-D nephanalysis agrees better with the analyst than THIR/CLE, except for high cloud amount; (2) the CLE and 3DN results show the tendency to overestimate clear amount when clear amount is large and underestimate it when clear amount is small, by 10–20% for CLE and by 5–10% for 3DN, and both agree well with the analyst in the mean; (3) 3DN and CLE systematic errors relative to the analyst are largest for low cloud, typically underestimating the amount by 15%; (4) both 3DN and CLE random errors range between 6 and 30%, smallest for high clouds and greatest for low clouds; (5) CLE estimates of cloud amount over land at night should not be used for scientific investigations unless restricted to high cloud amount.

Evaluation of Nimbus 7 THIR/CLE and Air Force three-dimensional Nephanalysis estimates of cloud amount. [Temperature-Humidity Infrared Radiometer/Clouds Earth Radiation Budget Experiment

Global satellite-derived cloud climatologies are considered and the results of an automated cloud analysis model are presented. The zonal distribution of cloud is applied to the radiative fluxes as predicted by an energy balance model. Special channel cloud discrimination is discussed with reference to Landsat thematic mapper.

A. Henderson-Sellers

Papers

Cloud Climatologies from Space and Applications to Climate Modelling