GLOBAL WARMING OR GLOBAL WARNING? THE PROBLEMS OF TESTING FOR A TREND IN ENVIROMENTAL TIME SERIES.

Abstract

Nowadays there is a contentious debate about the causes of the observed inc rease in the global temperature as well as the predictions about its future evolution. Some researches have found evidence of a determinist increasing trend using well-known temperature databases. In this work some of the techniques used to study the global warming debate are revised using sea level time series. The implicit idea is that one of the direct consequences of the global warming is the rise of the global sea level, thus if the temperatures are systematically increasing so should the sea level. The main conclusions are that researches possess certain discretion when modelling this type of time series and that, unless this discretion is cautiously used, conclusions may be questionable. Also, distinguishing for long term trends when the analysed data are reasonably described by ARIMA ( ) q p , 1 , models can be very difficult. This type of models exhibit temporary stochastic trends that can not be reasonable expected to continue in the future. Thus, predicting that the present rise in the sea level will persist may not be appropriate.