MONITORING CHANGES OF CLOUDS

An analysis of the spatial and temporal scales of cloud variability an d their coupling provided by the results from existing cloud observing systems allows us to reach the following conclusions about the necess ary attributes of a cloud monitoring system. (1) Complete global cover age with uniform density is necessary to obtain an unbiased estimate o f cloud change and an estimate of the reliability with which that chan ge can be determined. (2) A spatial sampling interval of less than 50 km is required so that cloud cover distributions will generally be hom ogeneous, or statistically homogeneous, within a sample. (3) A samplin g frequency of at least six times a day ensures not only that the diur nal and semi-diurnal cycles are not aliased into long term mean values , but also that changes in them can be monitored. (4) Since estimated climate changes are only evident on a decadal time-scale, unless cloud monitoring is continuous with a record length greater than 10 years a nd has very high precision (approximate to 1%) instrument calibration with overlapping observations between each pair of instruments, it wil l not be possible either to detect or to diagnose the effects of cloud changes on the climate.