Reconstrução da dinâmica não linear da temperatura do ar em Cuiabá-MT

Abstract

Investigate and understand the nature of the phenomena and physical processes that regulate the climate of our planet has been a concern for people, which currently due to intense modification of the natural environment and global climate change, has been motivated to research such phenomena. The local and global impact of human activities on the natural environment that surrounds are still objects of study with many uncertainties. All around the world many researches has been done about these physical phenomenas, however little is known about their dynamics. This is due to the fact they are open systems and off-balance, involving deterministic and stochastic processes. Understanding the dynamics that govern the climate can be done through analysis of nonlinear dynamical systems because the phenomenon involved has chaotic behavior. The reconstruction of the system dynamics that originated the possible climate changes patterns, with only one measurement scale, it is possible, through specific techniques of time series analysis. The description of the level of complexity or irregularity of time series can be made through the analysis of their dimensionality, Lyapunov exponents, sample entropy, multiscale sample entropy analysis and singular spectrum analysis. The study was conducted with data from the meteorological station of Cuiabá, provided by INMET (National Institute of Meteorology) through BDMEP (Meteorological Data Bank for Education and Research) on the 1961-2013 periods. The evidence of a possible connection between air temperature dynamic states and the Oceanic Niño Index - ONI, was verified by fractal dimensionality analysis of the series, Lyapunov exponents and estimates of Sample Entropy. The results point to the existence of a climate system with climate dynamic regulation of low dimensional and presence of deterministic chaos. The temporal evolution of nonlinear parameters (dimensionality, entropy and Lyapunov exponents) present a complex dynamics that may be relate and influenced by the fluctuations in ONI index. The air temperature has a self-similarity 5-8 days in the multiscale data, which may constitute a reliable time prediction models. The singular spectrum analysis indicates a climate system that has a strong component of nonlinear trend and seasonality that influence the pattern of fluctuation of the series. The reconstruction of the series through the SSA was able to model the trend and the periodic behavior of the series, with two seasonal components that can be linked to the annual and semiannual cycle variable.