Ongoing global warming and local warm extremes: a case study of two record mild winter months in Helsinki, Finland

In winter 2006-2007, both December and March were record warm in large parts of northern Europe. In Helsinki, for example, the previous record values for these two months were exceeded by over 1°C. How unusual are such extremely warm months? Traditionally, this question would be answered on the basis of past observations, by fitting a continuous function to the observed distribution of temperatures and extrapolating it beyond the observed range. However, in a changing climate this approach is problematic. Although it is still hard to reliably detect changes in local extremes directly from observations, the ongoing global warming alone makes it physically unlikely that statistics derived from past data are valid for the actual present-day climate. This is particularly true for temperature, for which the signal-to-noise ratio between the expected greenhouse-gas-induced changes and interannual variability is much higher than for (e.g.) precipitation (Raisanen 2001, Harvey 2004). Raisanen and Ruokolainen (2008a) proposed a model-based approach for estimating the actual present-day temperature climate. In this approach, past temperature observations are “extrapolated” forward in time, by combining observed changes in the global mean temperature with model-based regression coefficients that characterize the simulated relationship between the global mean temperature and the local mean temperature and amplitude of interannual variability. An example of the results of this extrapolation is given in Fig. 1. Two time series of December mean temperature in Helsinki are shown – one obtained directly from observations for 1901-2005 and the other from observations extrapolated to the present (2006) climate conditions. The difference between the time series increases towards the beginning of the 20th century, when the global mean temperature was further below its present-day level than it has been in the last few decades. Furthermore, the extrapolation increment is slightly larger for cold (e.g., 1915) than mild (e.g., 1914) December months. This is because most climate models suggest a decrease in the interannual variability of December mean temperatures in Finland in a warmer climate. Note that the extrapolated curve in Fig. 1 is based on results averaged for 22 global climate models – in reality slightly different curves are obtained from each single model.