Search

A method for the detection of inhomogeneities in time series was used to analyze the presence of breaks in the regime and climate shifts in monthly mean time series of single thermal tropopause pressure, height and temperature, at radiosonde stations located in India and southern South America, for the period 1973-2011. The method is based on fitting the data with a function that gets closer to a step function depending on a modifiable parameter, and to estimate the signal-to-noise ratio within a mobile window of variable width that is centered at each point of reference in the time series. There will exist an inhomogeneity at a given point of reference anytime exceeds a theoretical critical value γc. Since this value is hard to be estimated, an alternate calculation that depends on the analyzed data was carried out. Unlike similar methods for which γc = 1 is established, the analysis here determines γc for each month and for different window timescales. It is found that γc decreases with an increasing width of the timescale, fulfilling the condition γc = 0,5 i.e. a constant value, only for the longest timescales. From a joint analysis that included all cases for which γ> γc and certain time windows, inhomogeneities with climatic shift features were found over the extratropical region of South America in 1993, and over the subtropical region of India around 2001. [ABSTRACT FROM AUTHOR]