Author(s)

A. Sfetsos, C. Mita, N. Gounaris & D. Vlachogiannis

Abstract

This paper introduces a methodology for the determination of weather types and their representative days for a specific location. Observed as well as model derived information are used to identify groups of days that have common characteristics. The proposed approach consists of a rotated principal component analysis (r-PCA) module and the subtractive clustering algorithm. For the purposes of this study, this methodology was applied on an area of 280,000 km2 between the SW coast of Norway and the Shetland Islands. Using two years worth of data from the Norwegian Meteorological Institute (DNMI) operational forecasting model, ten representative days were identified. The examination of the transitional probability showed high probability of

Keywords