Multidimensional trends: The example of temperature
Laboratoire de Mathématiques, Université Paris 11, Orsay, France
2 EDF/R&D, 6 quai Watier, 78401 Chatou Cedex, France
Our goal is to discuss the notion of multidimensional trend in the climate change context. Temperature series over Europe are used to derive non parametric trends in mean, variance and extremes. A criterion is proposed to compare the different trends, and a general methodology is applied to test if the trends in extremes are only due to the trends in the central field, as mean and variance ones. Examples are given which show that the temporal evolution in mean and variance of the temperature time series are linked, and that generally these evolutions explain a large part of the temporal evolution of the extremes.
© EDP Sciences, Springer-Verlag, 2009