Statistical modelling of small-scale precipitation data

Statistical analysis of small-scale precipitation data is often done using model techniques, which are different from techniques used when data arises from larger areas or across longer periods of time. By this the conclusions drawn from one study may be difficult to transfer to other studies analyzed under different scale conditions. The paper deals with this model-continuity problem in an experimental design where small-scale data are analyzed by a classical statistical model technique and by a more complex model, which has proved valid for precipitation analyses in larger-scale conditions. Special attention is given to spatial characteristics of the correlations between the measurements.