Classification of rainfall into small- and large-scale events by statistical pattern recognition

摘要: Abstract Scale properties of daily areal rainfall in south-eastern Norway are investigated, and the events classified into small- large-scale by statistical pattern recognition. The coefficient variation standard deviation, computed from all precipitation gauges for one event, considered to be mutually independent assigned gamma densities. two parameters form a feature vector which is used as discriminatory variable classification algorithm. Empirical theoretical semivariograms were fitted sets events, semivariogram (nugget, sill range) investigated each set. Different values found, differences range, describing scale phenomena, further use cross validation with MSIE (mean squared interpolation error) criterion. Optimal range small-scale was found 16 km (19 without nugget effect), 33 (70 effect) precipitation. These optimal however, proved small consequence estimated using kriging interpolation. Areal catchments size 50 km2 1200 km2, insignificant when applying class-specific instead mean calculated regardless class. However, reduction factors (ARFs) derived analytical expressions semivariograms, considerable ARF curves types. difference became more marked low probability occurrence.