In this paper we propose novel statistical decompositions of climatic time series into long-term trend, seasonal and cycle components to analyze the existence of changes in rainfall and temperature patterns. The main contribution is the generalization of these decompositions to extreme values, probability of rain and the duration of dry days. We apply these methods for the metropolitan region of São Paulo. This region is characterized by a rapid urbanization process and a high population density, which is related to increased danger and vulnerability to hydrometeorological hazards. We analyzed daily series of temperature and precipitation between 1933 and 2019 from Institute of Astronomy, Geophysics and Atmospheric Sciences of University of São Paulo (IAG-USP) and the results indicate a significant increase in the trend component of daily temperature and in the rain patterns. In addition, the analysis for annual temperature records and annual records of daily rainfall show support for the hypothesis of permanent changes in observed climatic patterns. On the other hand, there is no significant evidence of changes in the pattern of duration of dry days nor changes in the trend component for the annual maximum of dry days, indicating the stability of this component.

在本文中，我们提出了将气候时间序列新的统计分解为长期趋势、季节和周期成分，以分析降雨和温度模式变化的存在。主要贡献是将这些分解推广到极值、下雨概率和干旱日的持续时间。我们将这些方法应用于圣保罗大都市区。该地区的特点是快速的城市化进程和高人口密度，这与水文气象灾害的危险性和脆弱性增加有关。我们分析了天文研究所 1933 年至 2019 年间的每日温度和降水系列，圣保罗大学地球物理和大气科学 (IAG-USP) 的研究结果表明，日气温和降雨模式的趋势分量显着增加。此外，对年温度记录和年降雨量记录的分析显示支持观测到的气候模式永久变化的假设。另一方面，没有明显证据表明干旱日持续时间模式发生变化，年度最大干旱日趋势分量也没有变化，表明该分量具有稳定性。