GLOBAL CLIMATE CHANGE STUDY THROUGH TIME SERIES FOR PREDICTING CHRONOLOGICAL TEMPERATURE INCREASE

Global warming is a concern for humans and the effects of climate change generate rapid changes on the planet that have been perceived in recent decades. The aim of this paper is to present the behavior of the temperature on the planet in the last century to the present day, and to show quantitatively the predictions of temperature changes globally, predicting the increase or decrease of temperature on our planet for the next 50 years using time series analysis with stochastic models.

The information used includes data from the first month of 1880 to the last month of 2012, which is cross-cutting and is divided in months. Information was provided by the National Climatic Data Center with abnormal ballooning of temperatures recorded on the land surface and ocean surface. Covering a total of 133 years, in increments of 12 months, 1596 values represent historical data on which to develop the prediction models. Analysis was performed using time series methods for horizontal pattern trends and seasonal trends. The best results were obtained with the methods of single exponential smoothing with alpha = 0.48 MSE=0.009 and forecasting to 0.5445 anomaly and linear exponential smoothing method two or method parameters Holt with alpha = 0.48 and beta = 0.0, MSE = 0.009 and 0.5445 forecasting anomaly CME both have the lowest compared to other criteria for the study were handled. Where methodologies found no decisive or consistent trend, beta = 0, or a consistent seasonal factor gamma = 0.33. So according to models with fewer margins for error we can say that the average global temperature over the next 50 years will be easily above the 0.5445 oC. But do consider a pattern with a higher tendency error MSE = 0.0275 using the linear regression method would have an increase of 1.1 oC for the next 100 years and within 50 years of the study an increase of 0.8 oC, very similar to those obtained by Goddard Institute for Space Studies and NASA, providing a model y = -0.04091 + 0.000539 x, with a correlation coefficient of 0.6910. This implies that if man does not perform specific actions against global warming, climate change will continue to generate increasing devastation to nature making it increasingly difficult to predict the weather worldwide.