INVESTIGATION OF NONLINEAR CHAOS IN TEMPERATURE TIME SERIES

An approach to phase space reconstruction is used to understand and predict the average monthly temperature dynamics. The predictions are made using a nonlinear local polynomial procedure by stepwise representing the dynamics recorded in the phase space in the local neighborhood. The time series of temperature data was reconstructed in phase spaces with embedding dimensions from 1 to 10 and a delay time from 1 to 14 months, and forecasts for a period of more than 1 year were made. The results showed a good correspondence between the predicted and observed temperatures.