We analyze several instrumental time series (temperature data and atmospheric blocking indexes) by using techniques originally developed for the analysis of extreme values of dynamical systems. By assuming that atmospheric dynamics is chaotic, we show how such datasets can be treated as the output of dynamical systems perturbed with noise and by instrument errors. The techniques provides a criterion to discriminate whether the recurrence of a certain event belongs to the normal variability or can be considered as rare events with respect to a specific timescale fixed as parameter.  The analysis of temperature data allows for building a map of expected and unexpected temperature ranges that can be adapted according to the desired time window. The analysis of blocking indexes suggests that the transition between zonal and meridional circulations in the northern hemisphere could be linked to the presence of an unstable fixed point of the atmospheric dynamics, rather than the often invoked bistability. Such a behavior is observed by comparing the dynamics of blocking indexes  with the one of  toy systems (Lorenz equations, Henon attractor) and even with  experimental datasets produced in a Von Karman Swirling flow experiment.