Nonlinear and Frequency Analysis of the Weather in Bialystok

Abstract

This work presents the use of nonlinear analysis for the daily meteorological data from Bialystok, Poland. The analysis is based both on daily temperature and rainfall fluctuation data. The data for the period of 43 years (1964-2006) have been analyzed. The temperature and rainfall fluctuation frequency in Bialystok has been analyzed using Fourier transform and the wavelet power spectrum, as well as fractal analysis, including attractor reconstruction, calculation of the correlation dimension and the largest Lyapunov exponent. The analysis allows us to identify short, medium and long-term cycles both for the temperature and rainfall data. As a result of attractor reconstruction made on the temperature data the 3D torus was obtained. It has been shown, that the correlation dimension D2 as well as the largest Lyapunov exponent varies in time. The parameters mentioned above increase rapidly in years when more medium-term cycles of temperature fluctuation are identified. The maximum period of the believable weather forecast has been also shown. It varies from 5 to 20 days in the analyzed years.