The statistical climate generator

Abstract

GLOWA-Danube requires high spatiotemporal resolution time series of meteorological inputs. A spatially distributed statistical climate generator was developed, which produces a network of synthetic time series of meteorological measurements. It is based on a thorough analysis of historic weather pattern, for which the basic element was chosen to be 1 week. It uses the GLOWA-Danube climate trends and a random number generator to produce time series of weekly weather pattern for the period from 2011 to 2060 including selectable climate trends. Mahalanobis distance is used as criterion to select the most similar week from historic records to build up a future dataset. The statistical climate generator is successfully validated with four realisations of past climate, for which consistent climate parameters as well as reproduction of runoff behaviour using the hydrological model ROMET were proven.