摘要：Mexico is a country vulnerable to extreme climatic events. However, their impact is not uniform in all its regions. This study presents an analysis of extreme temperatures in 12 Mexican cities, modelled under the assumption of a non-stationary climate. Temporal trends were estimated from an available climatological base of maximum and minimum temperatures, with the non-parametric tests of Mann-Kendall and Sen's slope method, and a Generalized Extreme Value (GEV) distribution was used to model both temperatures. A likelihood ratio test and Akaike and Bayesian information criteria were used to evaluate the optimal model choice with incorporation of a covariate. Using the best model, return levels and confidence intervals for future scenarios were estimated. A trend towards urban warming was detected from both the non-parametric tests and the GEV distribution, although with heterogeneous behavior. In the series of the maximum temperatures, half of the cities analyzed were non-stationary, and of those, the city of Guadalajara, located in the center-west of the country had a negative trend. The trend for minimum temperatures was more uniform, as 90% of the cities were non-stationary with a positive trend, and only 10%, in an urban area to the east of metropolitan area of the Valley of Mexico (Milpa Alta) and a coastal city of the Gulf of Mexico (Veracruz), showed stationary series. It is therefore concluded that return periods of thermal extremes estimated in a changing climate temporarily showed a significant variation, so statistical modelling must consider this behavior, due to its importance for risk assessments and adaptation purposes.
关键词：Extreme Temperatures; Non-Stationary Climate; Generalized Extreme Value Distribution; Return Periods; Cities of Mexico
