Modeling extreme minimum air temperature series under climate change conditions

Abstract

Considering the presence of non-stationary components, such as trends, in the extreme minimum air temperature series available from three locations of the State of São Paulo-Brazil, the aim of this research was to describe the probabilistic structure of this variable by using a non-stationary model (based on the general extreme value distribution; GEV model) in which the parameters are estimated as a function of time covariate. The Mann-Kendall test has proven the presence of significant increasing trends in all analyzed series. Furthermore, according to the Pettitt (changing-point) test, 1991 is the initial year of these trends (in the three locations). The applied selection criteria indicated that a GEV model in which the location parameter is estimated as a function of time is recommended to describe the probability structure of the variable under evaluation. The others two parameters of this model remained time-independent. According to this non-stationary model, the detected trends in the climate conditions of these locations have shown the same rate of change (0.04°C per year).