Cumulative distribution function data (CDF data) is one of the basic products available to UKCP09 users. Each set of CDF data is provided for a certain variable at a specific:
CDF data is used to form the CDF and graphs, and available from the UKCP09 User Interface (external website). Maps available within UKCP09 (publications, maps, and within the User Interface) are derived using values for a specified probability level from the CDF data.
CDF data is available for:
Similar CDF data are also available for some variables as future climate (absolute) projections, in which the changes have been combined with the observed 1961–1990 climatology, rather than climate change (projections relative to the baseline). The variables for which these are available are restricted to those for which there is observed 1961–1990 climatology.
CDF data provides values of future climate changes at which the probability of being less (or greater) than these values is a corresponding set of 107 pre-defined probability levels. The 107 pre-defined probability levels are: 0.1, 0.25, 0.5, 0.75, and 1%, then at 1% intervals (approximately) to 99%, then 99.25, 99.5, 99.75 and 99.9%.
The probability levels 1–9% and 91–99% are less robust and should be used with caution. Probability levels of less than 1% and above 99% should not be used and are only supplied in order to produce a
The CDF data are generated by a slightly different procedure than that for the sampled data (see Section of 3.2.12 of the Climate change projections report). They represent marginal posterior probabilities for individual climate variables for each 25 km grid or aggregated region, 30-year time period and emission scenario.
CDF data is provided as comma-separated values (*.csv) and CF-netCDF (*.cv) format files. Note that the comma-separated files also allow the data to be imported into, and manipulated using, a standard desktop spreadsheet package (e.g. Excel).
- Preparing maps and graphs
- Exploring the nature of the projections relative to single variable thresholds
- Input into impacts models that only require one variable
- CDF data should not be used where information on more than one variable is required