Probabilistic regional precipitation analysis

The random nature of rainfall events obliges the hydrologist to use statistical methods in order to quantify precipitation heigths and their associated recurrence intervals. But the available data represent only limited samples of the theoretically infinite parent population. The extrapolation of a statistical distribution fitted to such a sample may hence be very erroneous. The sampling error for each pluviometric post can be reduced by relating it to neighbouring stations. Supported by a mathematical theory, the Poisson Process, the relationship is established as the annual average number of events. This number should not vary throughout vast regions because the precipitation is due to large scale meteorological events.