CR 25:3-13 (2003)  -  doi:10.3354/cr025003

How often can we expect a record event?

R. E. Benestad*

The Norwegian Meteorological Institute, PO Box 43, 0313 Oslo, Norway

ABSTRACT: This study applies a simple framework for analysing the incidence of record events. A test of this method on the global mean temperature yields results consistent with a global warming, where record-warm events are more frequent than for a stationary series. The record event analysis suggests that the number of record-warm monthly global mean temperatures is higher than expected, and that the number of record events in the absolute monthly maximum and minimum temperatures in the Nordic countries is slightly higher than expected from a null hypothesis of a stationary behaviour. Because the different station series are not strictly independent, it is difficult to resolve whether there is a significant trend in the warmest absolute monthly minimum temperatures in the Nordic countries. The behaviour of the maximum monthly 24 h precipitation is not distinguishable from the null hypothesis that the series consists of independent and identically distributed random variables.

KEY WORDS: Record-value statistics · Extremes · Temperature · Precipitation

Full text in pdf format