Stationarity

A statistical property that means that a time series of data (such as can be produced from the Weather Generator) will exhibit little statistical variability over the time series.

In detail

Stationarity in the time series of data produced from the Weather Generator, means that whilst the series will contain realistic day to day and year to year weather variability, over the long term this series will contain little variation in the statistical description of that variabilty e.g. mean, variance, autocorrelation. This stationarity property arises because of the way in which the Weather Generator has been developed.

• An overview of how the Weather Generator works is provided in Section 1.3 of the UKCP09 Weather Generator report.

Stationarity can also be assumed in the UKCP09 Probabilistic projections for a given 30 year time period. Clearly, climate is changing over time, so while it is reasonable to assume that climate is stationary within a given 30 year time period, (and this is indeed the case in the UKCP09 Probabilistic Projections), this assumption would be violated if two series from different future time periods were combined.

• An overview of how probabilistic projections are created and presented in UKCP09 is described in Box 3 of the UKCP09 Briefing report. More details are provided in Chapter 3 of the UKCP09 Climate change projections report, including an overview of the methodology in Section 3.2.12.