1: \begin{abstract}

2: One of the goals of climate science is to characterize the statistics

3: of extreme and potentially dangerous events in the present and future

4: climate. Extreme events like heat waves, droughts, or floods due to

5: persisting rains are characterized by large anomalies of the time

6: average of an observable over a long time. The framework of Donsker\textendash Varadhan

7: large deviation theory could therefore be useful for their analysis.

8: In this paper we discuss how concepts and numerical algorithms developed

9: in relation with large deviation theory can be applied to study extreme,

10: rare fluctuations of time averages of surface temperatures at regional

11: scale with comprehensive numerical climate models. We study the convergence

12: of large deviation functions for the time averaged European surface

13: temperature obtained with direct numerical simulation of the climate

14: model Plasim, and discuss their climate implications. We show how

15: using a rare event algorithm can improve the efficiency of the computation

16: of the large deviation rate functions. We discuss the relevance of

17: the large deviation asymptotics for applications, and we

18: show how rare event algorithms can be used also to improve the statistics

19: of events on time scales shorter than the one needed for reaching

20: the large deviation asymptotics.

21: 

22: % \PACS{PACS code1 \and PACS code2 \and more}

23: % \subclass{MSC code1 \and MSC code2 \and more}

24: \end{abstract}

25: