Abstract
The apparent precision of estimated areal rainfall is affected by persistent areal pattern. The areal pattern and random variations can be separated by correlation between records in two periods of a network of gauges.