<b>Correlation in temperature time series in Cascavel, Paraná State</b> - DOI: 10.4025/actascitechnol.v29i2.710

Abstract

Temperatures can be correlated by power-law kind functions, and the persistency term can be characterized by a self-correlation function C(t) of temperature variations, where t is the time between observations. This function decreases as C(t) ~ t^-γ, where γ represents the correlation exponent. In this work this exponent is obtained by fluctuations analysis of temperature time series in Cascavel area, from 1972 to 2004. It is applied the Detrended Fluctuation Analysis - DFA method, which becomes useful to detect long range correlations and allows to obtain the exponents for local scale rules. These exponents can be compared with those obtained in other areas of the planet. Through this method the Hurst exponent H is obtained which is related with γ by γ = 2 (1 - H). The value γ =0,74 obtained here is in the superior limit of the values range presented in the literature for other places around the world.