In this paper a simple but objective statistical method is developed in order to determine the characteristics of the structure of a time series in the case of internal instability.Due to the random character, at least at the monthly scale, of the climatological series of observations, the method consists in testing the randomness of the series (stability of the statistical distribution and independence of its redken shades 9gi elements) against trend and serial correlation.Moreover, the trend analysis, performed in a progressive onward and backward way, enables to detect eventual anomalies inside the series.
The description of the method is followed by a critical review of the usual methods applied to climatological records for the same purpose.Moreover, its application to some local, mesoscale and global climatological series reveals that, at least at the seasonal scale, as well for temperature as for rainfall series, perturbations occurred in the form of abrupt changes characterized by changepoints separating the different sequences of the total series into stable random subseries with different statistical properties read more (mean and eventual variance).Conclusions suggest that the present impossibility of detecting the enhanced CO2 "greenhouse effect" in the series of observations may result as well from the weaknesses involved in the theoretical prediction models as in the absence of consideration of the actual statistical properties of the climatological series of observation.