Abstract
We present results of an investigation into the errors induced in estimates of the daily mean infrared and visible fluxes due to the effects of temporal sampling over a range of spatial scales. For polar orbiting satellites, the errors are a direct consequence of the fixed local time sampling from the sun-synchronous orbit, and the poor sampling of cloud variability. Geostationary satellites, whilst providing good temporal sampling, produce vast amounts of data leading to constraints on the data handling facilities. Sampling of this geostationary data provides a means of overcoming this problem, but leads to errors similar to those experienced with polar orbiter data. In this paper, we present results of studies of the errors due to sampling at fixed intervals in universal time and those due to sampling at fixed local times, appropriate to both types of satellite system, for a wide range of surfaces and latitudes. These errors were calculated by comparing results from an hourly METEOSAT dataset with suitably sampled subsets, for April 25th, 1978.
The results show a strong latitudinal variation of error, associated with differing circulation regimes, together with a strong dependence on the local time of sampling, minimum errors occurring when sampling at 0900 and 2100 LT. Errors are most significant for the visible region of the spectrum, and are primarily due to variations in cloud cover.