Statistical Modeling of an Air Temperature Time Series of European Cities

Abstract

In the last decades, the world has been confronted with the conse- quences of global warming. The rise in the global temperature has been an increasing concern of several authorities. According to the Intergov- ernmental Panel on Climate Change, the world’s greenhouse gas emis- sions are continuing to increase and on the present path, global tempera- ture rise will far exceed the limit goal of two degrees Celsius that coun- tries have agreed in order to avoid the most dangerous impacts of climate change. In the United Nations Framework, countries adopted the Paris Agreement, on 12 December 2015 in France, at the UN Climate Change Conference where parties committed to take ambitious actions to keep global temperature rise below 2 degrees Celsius by the end of the century.However, this global phenomenon is not reflected equally at every part of the globe. Hence, the monitoring and the analysis of the temperature rise, at a global, regional or at a local level, have been a challenge for the scientific community. The warming phenomenon must be monitoring at a smaller scale. This work examines long-term time series of monthly mean air temperatures in several European cities (available at the Climate Data Online (CDO) https://www.ncdc.noaa.gov/cdo-web/). Suitable statistical time series models must be developed in order to accommodate particular characteristics of this type of correlated data. These models can be sim- ple multiple regression models with errors with a correlation structure or, for instance, models that are more versatile as state space models. In a second step, modeling results are used to identify homogeneities between different cities.