Temperature variability in a changing climate

Temperature variability in a changing climate

Here we examine probability density functions (PDFs) of temperature variability over multi-decadal periods under idealised global warming scenarios. Such PDFs are expected to be more useful to managers of infrastructure and systems with a finite multi-decadal lifetime, than would PDFs for a specific future year (2030 or 2100 say). We suppose that the PDF of temperature variability at any given time t1 > t0 retains its initial normally distributed shape, i.e. PDF(T(t = t0)), with the same standard deviation, but is shifted to the right by an amount μ = μ(t). We refer to each of these PDFs, valid for individual times, as the individual PDFs. The PDF representing variability over the entire period t0 ≤ t ≤ t1 and not just at some instant within this period, i.e. the aggregated PDF (APDF), is broader than the instantaneous PDF and so the range of temperatures experienced over a finite period is greater than in an unchanging climate. The APDF is normally distributed if warming occurs at a linear rate, but is skewed towards higher temperatures if warming accelerates over the period of interest. Changes in APDFs of Australian mean minimum temperatures are generally consistent with this simple conceptual model. Idealised warming in which the variance of the individual PDFs increases is also considered. Popular conceptual models of the impact of global warming on temperature variability are relevant for instantaneous rather than aggregated PDFs. Such models neglect the straight-forward but important issues that arise when finite periods and non-linear warming rates are considered.